Cultural competence education for health professionals

  • Review
  • Intervention

Authors

  • Lidia Horvat,

    Corresponding author
    1. Department of Health, Sector Performance, Quality and Rural Health Branch, Melbourne, VIC, Australia
    2. La Trobe University, Cochrane Consumers and Communication Review Group, School of Public Health and Human Biosciences, Bundoora, Vic, Australia
    • Lidia Horvat, Sector Performance, Quality and Rural Health Branch, Department of Health, 50 Lonsdale Street, Melbourne, VIC, 3000, Australia. lidia.horvat@health.vic.gov.au.

    Search for more papers by this author
  • Dell Horey,

    1. La Trobe University, Faculty of Health Sciences, Bundoora, VIC, Australia
    Search for more papers by this author
  • Panayiota Romios,

    1. Australian Red Cross, Carlton, VIC, Australia
    Search for more papers by this author
  • John Kis-Rigo

    1. La Trobe University, Cochrane Consumers and Communication Review Group, School of Public Health and Human Biosciences, Bundoora, Vic, Australia
    Search for more papers by this author

Abstract

Background

Cultural competence education for health professionals aims to ensure all people receive equitable, effective health care, particularly those from culturally and linguistically diverse (CALD) backgrounds. It has emerged as a strategy in high-income English-speaking countries in response to evidence of health disparities, structural inequalities, and poorer quality health care and outcomes among people from minority CALD backgrounds. However there is a paucity of evidence to link cultural competence education with patient, professional and organisational outcomes. To assess efficacy, for this review we developed a four-dimensional conceptual framework comprising educational content, pedagogical approach, structure of the intervention, and participant characteristics to provide consistency in describing and assessing interventions. We use the term 'CALD participants' when referring to minority CALD populations as a whole. When referring to participants in included studies we describe them in terms used by study authors.

Objectives

To assess the effects of cultural competence education interventions for health professionals on patient-related outcomes, health professional outcomes, and healthcare organisation outcomes.

Search methods

We searched: MEDLINE (OvidSP) (1946 to June 2012); Cochrane Central Register of Controlled Trials (CENTRAL, The Cochrane Library) (June 2012); EMBASE (OvidSP) (1988 to June 2012); CINAHL (EbscoHOST) (1981 to June 2012); PsycINFO (OvidSP) (1806 to June 2012); Proquest Dissertations and Theses database (1861 to October 2011); ERIC (CSA) (1966 to October 2011); LILACS (1982 to March 2012); and Current Contents (OvidSP) (1993 Week 27 to June 2012).

Searches in MEDLINE, CENTRAL, PsycINFO, EMBASE, Proquest Dissertations and Theses, ERIC and Current Contents were updated in February 2014. Searches in CINAHL were updated in March 2014.

There were no language restrictions.

Selection criteria

We included randomised controlled trials (RCTs), cluster RCTs, and controlled clinical trials of educational interventions for health professionals working in health settings that aimed to improve: health outcomes of patients/consumers of minority cultural and linguistic backgrounds; knowledge, skills and attitudes of health professionals in delivering culturally competent care; and healthcare organisation performance in culturally competent care.

Data collection and analysis

We used the conceptual framework as the basis for data extraction. Two review authors independently extracted data on interventions, methods, and outcome measures and mapped them against the framework. Additional information was sought from study authors. We present results in narrative and tabular form.

Main results

We included five RCTs involving 337 healthcare professionals and 8400 patients; at least 3463 (41%) were from CALD backgrounds. Trials compared the effects of cultural competence training for health professionals, with no training. Three studies were from the USA, one from Canada and one from The Netherlands. They involved health professionals of diverse backgrounds, although most were not from CALD minorities. Cultural background was determined using a validated scale (one study), self-report (two studies) or not reported (two studies). The design effect from clustering meant an effective minimum sample size of 3164 CALD participants. No meta-analyses were performed. The quality of evidence for each outcome was judged to be low.

Two trials comparing cultural competence training with no training found no evidence of effect for treatment outcomes, including the proportion of patients with diabetes achieving LDL cholesterol control targets (risk difference (RD) -0.02, 95% CI -0.06 to 0.02; 1 study, USA, 2699 "black" patients, moderate quality), or change in weight loss (standardised mean difference (SMD) 0.07, 95% CI -0.41 to 0.55, 1 study, USA, effective sample size (ESS) 68 patients, low quality).

Health behaviour (client concordance with attendance) improved significantly among intervention participants compared with controls (relative risk (RR) 1.53, 95% CI 1.03 to 2.27, 1 study, USA, ESS 28 women, low quality). Involvement in care by "non-Western" patients (described as "mainly Turkish, Moroccan, Cape Verdean and Surinamese patients") with largely "Western" doctors improved in terms of mutual understanding (SMD 0.21, 95% CI 0.00 to 0.42, 1 study, The Netherlands, 109 patients, low quality). Evaluations of care were mixed (three studies). Two studies found no evidence of effect in: proportion of patients reporting satisfaction with consultations (RD 0.14, 95% CI -0.03 to 0.31, 1 study, The Netherlands, 109 patients, low quality); patient scores of physician cultural competency (SMD 0.11 95% CI -0.63 to 0.85, 1 study, USA, ESS 68 "Caucasian" and "non-Causcasian" patients (described as Latino, African American, Asian and other, low quality). Client perceptions of health professionals were significantly higher in the intervention group (SMD 1.60 95% CI 1.05 to 2.15, 1 study, USA, ESS 28 "Black" women, low quality).

No study assessed adverse outcomes.

There was no evidence of effect on clinician awareness of "racial" differences in quality of care among clients at a USA health centre (RR 1.37, 95% CI 0.97 to 1.94. P = 0.07) with no adjustment for clustering. Included studies did not measure other outcomes of interest. Sensitivity analyses using different values for the Intra-cluster coefficient (ICC) did not substantially alter the magnitude or significance of summary effect sizes.

All four domains of the conceptual framework were addressed, suggesting agreement on core components of cultural competence education interventions may be possible.

Authors' conclusions

Cultural competence continues to be developed as a major strategy to address health inequities. Five studies assessed the effects of cultural competence education for health professionals on patient-related outcomes. There was positive, albeit low-quality evidence, showing improvements in the involvement of CALD patients. Findings either showed support for the educational interventions or no evidence of effect. No studies assessed adverse outcomes. The quality of evidence is insufficient to draw generalisable conclusions, largely due to heterogeneity of the interventions in content, scope, design, duration, implementation and outcomes selected.

Further research is required to establish greater methodological rigour and uniformity on core components of education interventions, including how they are described and evaluated. Our conceptual framework provides a basis for establishing consensus to improve reporting and allow assessment across studies and populations. Future studies should measure the patient outcomes used: treatment outcomes; health behaviours; involvement in care and evaluations of care. Studies should also measure the impact of these types of interventions on healthcare organisations, as these are likely to affect uptake and sustainability.

Plain language summary

Cultural competence education for health professionals

This review assessed the effects of cultural competence education for health professionals. Five studies were included involving 337 health professionals (including general practitioners, primary care teams and counsellors) and 8400 patients. Three studies were conducted in the USA, one in Canada and one in The Netherlands. At least 3000 patients were from culturally and linguistically diverse (CALD) backgrounds and some of the health professional groups were also from CALD backgrounds. Due to differences in terminology and their cultural meanings, and the evolving nature of concepts and practices in this area, in this review we use the term 'CALD participants' when referring to minority, CALD populations as a whole. When referring to participants in included studies we describe them in the terms used by the relevant study authors.

The studies differed in how the education was provided and which outcome measures were used. Involvement in care improved in one study in The Netherlands measuring mutual understanding between minority CALD background patients and their doctors. Health behaviour also changed in one small study in the USA, in which women whose counsellors had received education were three times more likely to attend planned counselling sessions. However there was no evidence of an effect on a range of treatment outcomes or evaluations of care. None of the five included studies examined the effect of cultural competence education on healthcare organisations, or assessed adverse outcomes.

The review findings showed some support for cultural competence education for health professionals. These findings are tentative, however, as the quality of the evidence was low and more data are needed. Future research on cultural competence education for health professionals should seek greater consensus on the core components of cultural competence education, how participants are described and the outcomes assessed.

Laienverständliche Zusammenfassung

Schulungen in interkultureller Kompetenz für Gesundheitsfachleute

Dieser Review bewertet die Wirkung von Schulungen in interkultureller Kompetenz für Gesundheitsfachleute. Wir schlossen dazu fünf Studien mit 337 Fachpersonen (darunter Allgemeinmediziner, Teams für die medizinische Grundversorgung und Berater) und 8400 Patienten ein. Drei Studien wurden in den USA durchgeführt, eine in Kanada und eine in den Niederlanden. Mindestens 3000 Patienten und auch einige Gruppen von Gesundheitsfachpersonen hatten einen Migrationshintergrund ("culturally and linguistically diverse", CALD). Aufgrund von Unterschieden in der Verwendung und kulturellen Bedeutung bestimmter Begriffe sowie der Weiterentwicklung von Konzepten und Praktiken in diesem Bereich verwenden wir in diesem Review den Begriff „CALD-Teilnehmer“, wenn wir uns auf ethnisch und sprachlich verschiedene Minderheiten als Ganzes beziehen. Sind Teilnehmer in eingeschlossenen Studien gemeint, beschreiben wir sie mit den Begriffen, die auch die Autoren der entsprechenden Studie verwenden.

Die Studien unterschieden sich im Hinblick darauf, wie die Schulungen durchgeführt und welche Endpunkte gemessen wurden. Eine Studie in den Niederlanden untersuchte das gegenseitige Verständnis zwischen Patienten, die einer Minderheit mit CALD-Hintergrund angehören, und ihren Ärzten und zeigte, dass sich die Mitarbeit der Patienten verbesserte. In einer kleinen Studie in den USA veränderte sich auch das Gesundheitsverhalten. Hier erschienen Frauen, deren Berater geschult worden waren, mit dreimal höherer Wahrscheinlichkeit zu den geplanten Beratungssitzungen. Es gab jedoch keine Evidenz für eine Wirkung auf eine Reihe von Behandlungs-Endpunkten oder auf die Bewertung der Versorgung. Keine der fünf eingeschlossenen Studien untersuchte die Wirkung von Schulungen in interkultureller Kompetenz auf die Organisation der Gesundheitsversorgung oder bewertete negative Folgen.

Die Ergebnisse des Reviews unterstützen in gewissem Maße das Konzept von Schulungen in interkultureller Kompetenz für Gesundheitsfachleute. Jedoch handelt es sich dabei um vorläufige Ergebnisse, da die Qualität der Evidenz gering war und mehr Daten benötigt werden. Zukünftige Forschungsarbeiten zum Thema Schulungen in interkultureller Kompetenz für Gesundheitsfachleute sollten sich darum bemühen, die zentralen Bestandteile solcher Schulungen, die Beschreibung der Teilnehmer und die bewerteten Endpunkte stärker zu vereinheitlichen.

Anmerkungen zur Übersetzung

S. Schmidt-Wussow, Koordination durch Cochrane Schweiz.

Summary of findings(Explanation)

Summary of findings for the main comparison. 
Cultural competence training for health professionals compared with no training

Patient or population: CALD patients and their health professionals (Primary care settings in high income countries)

Intervention: Cultural competence training for health professionals

Comparison: No training

Outcomes ImpactNo of Participants
(studies)
Quality of the evidence
(GRADE)

Treatment outcomes

(Different measures)1

No evidence of effect on treatment outcomes in two studies; the proportion who achieved cholesterol control target over 12 months and weight loss over six months were assessed.27672
(2 studies)
⊕⊕⊝⊝
low 3
Health behavioursClient concordance with attendance significantly improved for the intervention group across three counselling sessions. Women in intervention group were 1.5 times more likely to attend the third counselling session (RR 1.53, 95% CI 1.03 to 2.27).

282

(1 study)

⊕⊕⊕⊝
low 4

Involvement in care

(Mutual understanding)5

One study in The Netherlands reported improved mutual understanding between one in five patients (described as "mainly Turkish, Moroccan, Cape Verdean and Surinamese patients") and their largely "Western" GPs (mostly Dutch) (SMD 0.21, 95% CI 0.00 to 0.42).1092
(1 study)
⊕⊕⊝⊝
low 6

Evaluations of care

(Different measures)7

Three studies showed mixed outcomes. There was no evidence of effect on evaluations of care between intervention and control group participants in two studies but a third study showed significant improvements in client perceptions of their health professional after cultural competence training.1952
(3 studies)
⊕⊕⊝⊝
low 8

Health professionals knowledge & understanding

(Awareness of racial differences)9

No evidence of effect on clinician awareness of racial differences in the quality of diabetes care for "black" clients was found in one study among the proportion of clinicians acknowledging racial disparities in care occurred "very often" or "somewhat often" (RR 1.37, 95% CI 0.97 to 1.94), with no adjustment for clustering.87
(1 study)
⊕⊕⊝⊝
low 10
Adverse eventsNone of the included studies measured adverse outcomes.0 

GP: General Practitioner; CI: Confidence interval; RD: Risk Difference; SMD: Standardised Mean Difference CALD: Culturally and Linguistically Diverse

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1. Rate of achieving control target of LDL cholesterol < 2.59mmol/L (< 100mg/dL) in previous 12 months and change in patient weight (pounds). Data in both studies collected from patient records.

2. Study population adjusted to take account of design effect of clustering in one study. Assumed ICC of 0.2. Sensitivity analyses using different values for ICC did not substantially alter magnitude or significance of summary effect sizes.

3.Allocation concealment and attrition unclear. One study did not take account of clustering and reported outcomes for all patients not CALD population.

4. Single study with small sample especially after adjustment for the design effect of clustering. Outcome not likely to be affected by lack of blinding.

5. Validated scale to measure mutual understanding by comparing GP and patient assessments of consultation. Responses could range from -1 (total misunderstanding) to +1 (complete mutual understanding). GPs completed the questionnaire immediately after the consultation and patient interviews were conducted 3 to 8 days after a consultation.

6. Single study. Baseline imbalance and high attrition for CALD population. Blinding of assessment unclear.

7. Measures include dichotomous measure of Patient satisfaction with consultation, which was measured in patient interviews at home 3–8 days after GP consultation. There were two continuous measures: Patient reported physician cultural competency, which asks patients about 13 physician behaviours using 5-point scale with score transformed to a 0 to 100 scale, a single dimension (attractiveness) from validated scale with twelve 7-point bipolar items, Client perception of counselors ('attractiveness') .

8. Allocation concealment and attrition were high or unclear. One study did not take account of clustering and reported outcomes for all patients and not the CALD population separately. Sample sizes were small after adjusting for clustering.

9. Clinician awareness of racial differences in care measured with a 5-point Likert scale (very often to very rarely).

10. Single study with small sample size. Allocation concealment unclear. No adjustment for clustering in computation of relative risk.

Background

Increasingly, health professionals and healthcare organisations are required to deliver care to diverse population groups. There is evidence of health disparities, structural inequalities, and poorer quality health care and outcomes among people from minority cultural and linguistic backgrounds (Betancourt 2003; Smedley 2003; Flores 2005; Divi 2007). Cultural competence has gained attention from healthcare policy makers, professionals and educators as a strategy to improve quality and outcomes in health care (Betancourt 2010), particularly for diverse populations in high-income English-speaking countries.

Embedding cultural competence in healthcare systems enables systems to provide appropriate care to patients with diverse values, beliefs, and behaviours, including meeting patients’ social, cultural and linguistic needs.  The concept of cultural competence "builds on the strengths of earlier concepts such as cultural awareness, cultural security, cultural respect, and cultural safety, extending them further to facilitate changes in all dimensions of practice, including the levels of the practitioner, the organisation and the system" (Grote 2008). It has become a key feature in standards and accreditation processes for quality health care. This has occurred most notably in the United States (US), as exemplified by the recently enhanced National Standards on Culturally and Linguistically Appropriate Services (CLAS 2013) developed by the Office of Minority Health, and the work of the Joint Commission (Joint Commission 2010), which has developed standards for healthcare organisations that address effective communication, cultural competence and patient- and family-centred care.

Despite apparent broad agreement as to the raison d’être for the term 'cultural competence', the absence of a commonly agreed definition and standardised frameworks (Grant 2013; Thackrah 2013) particularly in the context of education interventions (Kleinman 2006), has made cultural competence a contested concept. Sue and colleagues (Sue 2009) identified several problems with the term, asserting that "(a) it has various meanings, (b) includes inadequate descriptors, (c) is not theoretically grounded, and (d) is restricted by a lack of measurements and research designs for evaluating its impact in treatment". For the purposes of this review we will use the seminal and most commonly cited definition of cultural and linguistic competence developed by Cross et al: "Cultural and linguistic competence is a set of congruent behaviours, attitudes, and policies that come together in a system, agency, or among professionals that enables effective work in cross-cultural situations. 'Culture' refers to integrated patterns of human behavior that include the language, thoughts, communications, actions, customs, beliefs, values, and institutions of racial, ethnic, religious, or social groups. 'Competence' implies having the capacity to function effectively as an individual and an organization within the context of the cultural beliefs, behaviours, and needs presented by consumers and their communities" (Cross 1989). Cultural competence is seen as a complex framework, and a systemic developmental process that evolves over time (Goode 2006).

Cultural competence has become a mainstream education issue applicable to all health professionals (regardless of their own cultural or linguistic backgrounds) to ensure that all people receive equitable, effective, and culturally and linguistically appropriate health care. As a strategy to achieve this, education interventions for health professionals seek to develop cultural capabilities alongside clinical capabilities (Johnstone 2006).  However, understanding and identifying what works best, where, for whom, and how, in improving health outcomes remains a challenge. Cultural competence education interventions are numerous and diverse in foci, but whilst there is evidence that they improve the knowledge, skills and attitudes of health professionals, there is a paucity of evidence about their effects on patient health outcomes (Beach 2005).

Increasingly, health professionals and healthcare organisations are required to respond appropriately to multiple and concurrent aspects of diversity (Chiarenza 2012) in the consumer and patient groups they serve, and across the contexts in which they work.  Additional layers of complexity from an increasingly diverse and internationalised health workforce due to immigration and movement of professionals challenge previous modes of cultural competence education and practice in health care. Often, the primary focus of education interventions is minority population groups, which can include health professionals themselves. Accordingly, education interventions need to consider health professionals who may belong to a minority population group; have been trained in other countries; and who may face additional challenges, such as familiarising themselves with a different healthcare context, as well as the diversity of patients in their care.  

Adding to this complexity is the lack of an internationally agreed term to describe patients from diverse population groups. Different conceptualisations and usage of the terms identity, culture, ethnicity and race abound. The imprecision of definitions due to factors such as the fluidity of self-identification and heterogeneity (between and within groups and communities), together with a lack of systematic reporting, illustrate the conceptual, ethical and methodological challenges posed by the absence of an agreed terminology (Baer 2012; Lewis-Fernández 2013; Mir 2013). For example, in Australia the term 'culturally and linguistically diverse' (CALD) is commonly used to encompass broad immigrant and refugee communities. CALD refers to “the range of different cultures and language groups represented in the population who identify as having particular cultural or linguistic affiliations by virtue of their place of birth, ancestry or ethnic origin, religion, preferred language or language spoken at home" (Department of Health 2009). The term does not include Aboriginal or Torres Strait Islander people who are the first peoples of Australia. Examples in other parts of the world include the US, where minority population groups may be identified by various terms such as African-American and Latino, and in Europe terms such as 'migrant', 'ethnic minorities' and 'vulnerable populations' are used (Schulze 2003; LBISM 2004).

Additionally, these concepts and associated practice and research are evolving rapidly in response to the socio-cultural issues in each country, and continue to be debated. For example, in Europe, a focus on ‘intersectionality’ has emerged where all variables that create inequity in health care are “conceptualised as a set of processes" and not as the"possessive characteristics of individuals”(Cattacin 2013) as a challenge to the ‘traditional’ focus on ethno-cultural diversity characteristics that may risk stereotyping individuals, groups and communities.

This complex situation required a decision about terminology for this review. We use the term 'CALD' to indicate the broad group of populations characterised by ‘minority’ status, and racial and ethnic disparities in health care. However, for specific studies in all places we agree with the challenges described by Kaplan 2003 and follow suggested guidelines of citing the language and terms used by study authors when reporting data (Kaplan 2003).

Whilst this review focuses primarily on outcomes for patients, we will also examine outcomes for health professionals and healthcare organisations.

Description of the condition

The international research literature recognises the significant influence of culture in shaping people's perceptions of health and well-being, as well as their experiences of health care (Kleinman 1978; Brach 2002; Johnstone 2006). Long-standing disparities exist in the health status of people from diverse cultural, linguistic and socio-economic backgrounds in high-income English speaking countries (Betancourt 2003; Flores 2005; Divi 2007), and there is mounting evidence that people from a minority cultural and linguistic background, particularly patients with low English proficiency, receive poor quality health care compared with mainstream patients, and are more likely to experience a "trajectory of accident opportunity" and/or adverse events in their journey through the health system (Divi 2007). This problem is not limited to English-speaking countries alone. As one example, in Europe, the 'Migrant-friendly hospitals' (MFH) project asserts that in non-English speaking countries, the health status of immigrant, refugee, and ethnic minority populations is "often worse than that of the average population" (Schulze 2003; WHO 2010 ; Ingleby 2012 ; Mladovsky 2012). Current initiatives in Europe include the development of Equity Standards to address disparities in health care (Task Force on MFCCH 2014; Cattacin 2013).

A link between culture, language and patient safety outcomes has been demonstrated (Johnstone 2006), and health disparities and lower quality care are exacerbated when healthcare professionals fail to address ethnicity, culture and language in the provision of health services (Wilson-Stronks 2008). A key aim of cultural competence for health professionals is to enhance their capacity to provide effective and appropriate health care to diverse population groups and patients. The effect on healthcare delivery is likely to be more apparent when the differences between the health professional and patient involved are greatest, but enhanced cultural competence is increasingly viewed as likely to benefit all interactions, including those where differences are not as obvious. The congruence between cultural competence and co-aligned concepts such as person- and family-centred care, consumer participation, and a ‘rights-based’ approach to healthcare equity are increasingly apparent (Beach 2006; Like 2011; Saha 2008).

The terms 'health disparities' and 'health inequities' are frequently used to describe the type of adverse health outcomes experienced by ethnic minorities within a context of broad historic and contemporary social and economic inequality. Cultural competence may be one way to reduce disparities in health care for these population groups (IOM 2002; Goode 2006; Betancourt 2010).

Description of the intervention

The components of cultural competence education for health professionals generally include cultural awareness, cultural knowledge and cultural skills (Sue 2009). These are derived from the often-cited principles of cultural competence that include: valuing diversity and similarity; understanding and effectively responding to cultural differences; engaging in cultural self-assessment at individual and organisational levels; adapting service delivery and supports; and institutionalising cultural knowledge and policies across the organisation (Cross 1989). Over the last two decades, cultural competence has become a more comprehensive, skill-based concept that involves the system in addition to the patients, and has been conceived of as an on-going quality improvement process, relevant across individual, organisational, systemic and professional levels (NHMRC 2005).

Teaching and learning methods and the content of cultural competence in education interventions range from simple reductivist ‘culture-facts’ with ‘do’s and don'ts’ (Kleinman 2006) to approaches that specify aspects of cultural self-awareness and intercultural communication skills, to more complex and nuanced understandings that show deeper analysis of socio-cultural barriers to health care at the clinical level (healthcare professional-patient encounter), organisational level (leadership and workforce), and structural level (processes of care) (Betancourt 2003). These have led to a plethora of education models, frameworks and assessment tools, mostly developed in the US, the United Kingdom or Canada.

The proliferation of cultural competence education and training options is accompanied by: a lack of uniformity in specifying requisite knowledge and skills, and teaching and learning approaches; a dearth of standards or consistent measures for implementation and assessment; and poor quality studies (Beach 2005; Bean 2006; Lie 2010; Grant 2013). These issues pose key challenges for this review. We have devised a four-dimensional conceptual framework derived from a synthesis of key cultural competence models and educational intervention frameworks described in the literature, to provide a consistent approach to describe and assess the interventions.

Our conceptual framework (see Table 1) comprises four key domains that describe core ingredients of cultural competence education interventions:

Table 1. Conceptual framework
  1. See Description of the intervention for details of the conceptual framework.

1.      Educational content 2.      Pedagogical approach 3.      Structure of the intervention 4.      Participant characteristics

a. Types of knowledge

b. Assessment and application

c. Skills

a. Teaching and learning method

b. Theoretical constructs and principles

a. Delivery and format

b. Frequency and timing

c. Assessment and evaluation of intervention

d. Organisational support

a. Delivering the intervention (teacher/ facilitators)

b. Engaging in the intervention (target audience)

  1. educational content;

  2. pedagogical approach;

  3. structure of the intervention; and

  4. participant characteristics.

Within each domain a number of components are considered, that can be interdependent at times.

1. The educational content domain has three key components: types of knowledge; assessment and application; and the skills needed for culturally competent practice by a health professional.

a. Types of knowledge: A key challenge for cultural competence education is to clarify the relationship between specific knowledge about particular communities, and generic knowledge required in any clinical encounter at any time and with any/all patients, without resorting to stereotypes or a need for an encyclopedic knowledge of many cultures. This requires knowledge content to cover core concepts or models deemed central to understanding cultural competence. These might include: general concepts of culture; cultural competence; models of health and illness; the socio-cultural context of health disparities; epidemiology; social determinants; constructs of racism and prejudice; and specific theoretical models such as explanatory models of health and illness; patient centeredness and consumer participation.

b. Assessment and application: This component refers to an intervention addressing the capacity to use the acquired knowledge in a reflective and transformative manner. The inclusion of this component in an intervention aims to help participants apply acquired theoretical knowledge to relevant environmental, population, organisational, professional and systemic contexts.

c. Skills: This component comprises two subsets: inter- and intra-personal skill development. An intervention can teach inter-personal skills to help people work with others, such as advanced communication, negotiation or collaboration. An example in this context is building skills to work effectively with interpreters. An intervention can also teach intra-personal skills to help individuals cope with situations. This can include cultural self-assessment and reflection, and deconstructing stereotypes.

2. The pedagogical approach domain in our framework is concerned with the teaching and learning method used in an intervention, and any theoretical constructs and principles that underpin it. For example, a culturally inclusive approach examines cultural biases in instructional practices that may be experiential, instructive, or didactic.

3. The structure of the intervention describes the intervention using four components:

a. Delivery and format, such as face to face, group or online participation;

b. Frequency and timing, such as single events, series of workshops, or ongoing training;

c. Assessment and evaluation of the intervention, which includes the method of quality assessment such as pre-training surveys or pre- and post-tests; and the evaluation method used, such as specific assessment tools;

d. Organisational support, including links to organisational policies and procedures; whether the cultural competence education intervention is embedded within a professional development program of the organisation; the use of mentoring and peer support processes; links to formal professional accreditation points or qualifications; and if participation is voluntary or mandatory.

4. The participant characteristics domain is concerned with describing the participants involved in cultural competence education interventions, including facilitators, teachers or experts delivering the intervention as well as the participants engaged in the intervention as the target audience (health professionals).

How the intervention might work

Professional development, education or training aimed at improving the cultural competence of health professionals or in healthcare organisations is predicated on the assumption that certain skills and understanding will enable them to become more responsive to the healthcare needs of diverse population groups, and thus improve health outcomes (Ahmann 2002; Anderson 2003; Betancourt 2003). By teaching health professionals how to identify and assess the cultural and linguistic needs of each patient and of themselves, health professionals may be able to:

  • elicit patients' explanatory models; that is, identify patients' knowledge, beliefs and perspective of their illness, including expectations of treatment and therapeutic goals (Kleinman 1978);

  • compare their own goals with those of their patients, to identify different understandings or discrepancies;

  • reflect on their cultural context (ethnicity, country of origin) and what this brings to the clinical encounter;

  • modify their behaviour and approach to maximise patient engagement and education in clinical encounters;

  • respect patients' values, wishes and preferences;

  • adapt advice and care to meet patients' social, cultural values and economic needs; and

  • connect individual patient health care and service delivery to broader organisational structures and environmental contexts.

For patients, this might facilitate cultural concordance with health professionals, and improved:

  • communication, understanding and expectations, increasing participation in healthcare decisions and treatment options;

  • sense of feeling listened to and respected, and consequent increased satisfaction with their care and change in their health behaviours (Stewart 2006);

  • health literacy capability, leading to improved treatment concordance and reduced delays in seeking care and treatment;

  • access to health services;

  • involvement of family and carers in care;

  • adherence to treatment and follow up; and

  • patient safety.

For healthcare organisations, the intervention might improve:

  • quality and safety processes and measures;

  • data collection, reporting and tracking of adverse events; and

  • cost efficiencies.

Due to the increasingly globalised nature of the health workforce, curricula needs to address the different cultural backgrounds and countries of origin of health professionals, including past training experience, and different health system experiences.

Why it is important to do this review

It was important to do this review because healthcare professionals and organisations deliver care to diverse population groups, and poorer outcomes are recognised among minority culturally and linguistically diverse groups. Discrimination in healthcare delivery at the system and patient/healthcare professional levels may also exacerbate health disparities (IOM 2002). These concerns have led governments and organisations to invest in interventions to develop cultural competence knowledge and skills in the health workforce, in order to deliver culturally appropriate and effective health care.

This review differs from existing reviews in three significant ways:

  1. It focuses explicitly on assessing the effects of cultural competence interventions on patient-related outcomes, in particular on those patients from minority CALD backgrounds.

  2. Unlike other reviews we have found so far (see below), we sought to include studies published in languages other than English, to bring an international perspective.

  3. We aimed to include education interventions for healthcare professionals working in varied health settings. 

Several reviews, including a Cochrane protocol and review, address similar issues. The protocol (van Nuland 2005) is also concerned with the effects of education interventions on patient-related outcomes, but differs from this review in several ways. It focuses exclusively on improving communication skills; the target audience is General Practice trainees during clinical consultations only; the interventions are those that occur specifically within the vocational training phase; and patients speaking a different first language from the General Practitioner are excluded. Forsetlund 2009 is a Cochrane review focusing on the effects of continuing education meetings on professional practice and healthcare outcomes. A key difference between this and our review is Forsetlund's broad definition of education; and examination of continuing educational delivery models. Our review focuses on cultural competence content as described by our conceptual framework (Table 1).

An important non-Cochrane review concluded that there is good evidence that cultural competence training improves staff knowledge, attitudes and skill, and that the provision of training to staff improves patient satisfaction (Beach 2005). While the review also concluded that evidence to show improvements in patient adherence was poor and studies that evaluated patient health status were absent, this was not a stated aim of the review. Accordingly, a key point of difference with our review is our primary aim to assess the effects of cultural competence training of health professionals. We also sought studies in languages other than English.

A Campbell systematic review and meta-analysis (Hasnain 2009) showed culturally adapted competency interventions improved rehabilitation outcomes for ethnically and culturally diverse people with disabilities residing in high-income countries. Our review has a broader scope in terms of the intervention and patient groups included.

A systematic review by Lie 2010 sought to address the effects of cultural competency training on patient-centred outcomes, but where Lie critically assessed the quality of studies, and sought to develop a research approach to assessing quality, the review did not describe or assess the educational content of the interventions and excluded non-English language studies. Our review provides a conceptual framework to do this and aimed to include studies in all languages.

An update of a Cochrane systematic review (Dwamena 2012) assessed interventions to promote a patient-centred approach in clinical consultations. The review, which included 43 randomised controlled trials of training interventions aimed at improving communication in healthcare settings, found that training health professionals in patient-centred consulting styles can be effective, leading to improved communication and, in some cases, better health outcomes. Whilst the Dwamena review focused on patient-centred care and not on cultural competence, there is some congruence between the two concepts and approaches.

More recently an overview, or systematic review of published reviews, of interventions to improve cultural competency, concluded that "the evidence base is relatively weak, and there continues to be uncertainty in the field" (Truong 2014). The overview included 19 reviews comprising a mix of study designs and intervention types. Six of the 19 reviews specifically addressed education or training for health practitioners, while one looked at interventions targeting both health professionals and patients. The authors found evidence for 'patient/client outcomes' to be weaker than for provider outcomes, healthcare access or utilisation outcomes. Several areas of uncertainty were identified in the overview including: how cultural competency is defined and understood; how language and terminology is used; the choice of outcomes and how they are measured; and the robustness of the studies included. The authors recommended that future reviews use explicit definitions or frameworks for cultural competency and assess outcomes across individual, organisational and system levels.

Objectives

To assess the effects of cultural competence education interventions for health professionals on patient-related outcomes, health professional outcomes, and healthcare organisation outcomes.

Methods

Criteria for considering studies for this review

Types of studies

We included randomised controlled trials (RCTs), including cluster RCTs, and controlled clinical trials (CCTs), also known as quasi-RCTs (trials where randomisation is attempted but subject to potential manipulation).

Types of participants

The primary participants were qualified healthcare professionals working in varied health settings (hospital, community health, and aged care). This included health professionals from minority populations, who delivered health care to specific communities.

Secondary participants were education facilitators (teachers or experts delivering the intervention), patients and health organisations.

The focus of the review was practising and credentialed health professionals working in a variety of health settings providing ongoing care. Undergraduate students and lay health workers were excluded.

Types of interventions

We included educational interventions for health professionals working in health settings (hospital, community health, and aged care) aimed (either impliedly or explicitly) at improving:

  • health outcomes of patients/consumers of minority cultural and linguistic backgrounds

  • knowledge, skills and attitudes of health professionals in delivering culturally competent care

  • healthcare organisation performance in culturally competent care.

We anticipated considerable heterogeneity in the educational interventions. We describe the interventions using our aforementioned conceptual framework based on the core domains and their components: educational content (knowledge, assessment and application, skills); pedagogical approach; intervention structure (delivery and format, frequency and timing, assessment and evaluation, organisational support), and participants.

We excluded studies in which the intervention could not be described in terms of any of the domains in our conceptual framework (see Description of the intervention). We excluded interventions that had a cultural competence component as part of a broader ‘orientation program’ for new staff and where cultural competence education was not the main focus.

We also excluded interventions that were part of undergraduate, entry-level, entry to practice, or pre-licensure education programs.

We anticipated that the comparison group would be no intervention but would have included any comparison group for all relevant trials.

Types of outcome measures

Outcome measures were related to patients, health professionals and health organisations, and were prioritised as below. Studies not reporting primary patient-related outcomes were excluded.

Primary outcomes

Primary patient-related outcomes:

  • Treatment outcomes (e.g. clinical assessments)

  • Health behaviour (e.g. attitudes, concordance)

  • Involvement in care (e.g. involvement in decision-making)

  • Evaluation of care (e.g. communication enhancement, perceptions of care)

In line with the Methodological Expectations of Cochrane Intervention Review (MECIR) standards introduced by the Cochrane Collaboration in 2012, the following healthcare organisation outcome was redefined as a primary outcome:

  • Adverse events (e.g. complaints, reporting of adverse events)

Secondary outcomes

Secondary outcomes incorporate the outcomes of cultural competence education on health professionals, including:

  • Knowledge and understanding (e.g. attitudes, behaviour, confidence)

  • Consultation processes (e.g. levels of patient-centred care, provision of an accredited interpreter)

  • Evaluation of processes and outcomes (e.g. training program assessment, provision of care, job satisfaction).

Finally, additional outcomes of interest were healthcare organisation related, and included:

  • Quality and safety measures (e.g. provision of accredited interpreter)

  • Service utilisation (e.g. use of services)

  • Health economic outcomes (e.g. cost of intervention, cost of care).

Search methods for identification of studies

Electronic searches

We searched the following electronic databases:

  • MEDLINE (OvidSP) (1946 to 18 June 2012);

  • Cochrane Central Register of Controlled Trials (CENTRAL, The Cochrane Library) (18 June 2012);

  • EMBASE (OvidSP) (1988 to 18 June 2012);

  • CINAHL (EbscoHOST) (1981 to 18 June 2012);

  • PsycINFO (OvidSP) (1806 to 18 June 2012);

  • Proquest Dissertations and Theses database (1861 to 13 October 2011);

  • ERIC (CSA) (1966 to 13 October 2011)

  • LILACS (1982 to 12 March 2012)

  • Current Contents (OvidSP) (1993 Week 27 to 18 June 2012)

The search strategy was amended from the protocol to include Current Contents in place of Index to Theses and Sociological Abstracts because of its currency and accessibility. We used the strategy presented in Appendix 1 to search MEDLINE (OvidSP), in Appendix 2 to search CENTRAL, in Appendix 3 to search EMBASE, in Appendix 4 to search CINAHL, in Appendix 5 to search PsycINFO, in Appendix 6 to search ERIC, in Appendix 7 to search LILACS and in Appendix 8 to search Current Contents. There were no language restrictions.

Searches in MEDLINE, CENTRAL and PsycINFO were updated again on 10 February 2014 and EMBASE, Proquest Dissertations and Theses, ERIC and Current Contents were updated again on 17 February 2014. Searches in CINAHL were updated on 10 March 2014.

Searching other resources

For all included studies, we also searched reference lists, and conducted author and citation searches in SciVerse Scopus database. We also searched the Campbell Library, the International Clinical Trials Registry Platform Search Portal and Current Controlled Trials metaRegister of Controlled Trials to identify other trials and contacted the relevant investigators for further information.

We wrote to each corresponding author of our included studies and sought additional information about their study and any relevant published, unpublished or ongoing studies.

Data collection and analysis

Selection of studies

Search results were combined in an Endnote database and duplicate records removed. Two review authors (LH and DH), together with three researchers (Megan Prictor, Dianne Lowe, Natalie Willis) screened all titles and abstracts to assess which studies met the inclusion criteria. Studies that did not clearly relate to the inclusion criteria (Criteria for considering studies for this review) were excluded. We retrieved in full text those articles deemed potentially relevant. Two review authors (LH and DH) independently assessed the retrieved articles for inclusion. Studies which appeared to be relevant but which were excluded are listed in the table Characteristics of excluded studies, with the reason for exclusion given. Three review authors (LH, PR, DH) verified the final list of included studies. If there had been any disagreements they would have been resolved through discussion between these three review authors.

In February 2014, another researcher (Anneliese Synnot) ran updated searches and screened the output for potentially relevant studies. Two review authors (LH and DH) retrieved and independently assessed these studies for potential inclusion. No additional studies were included in the review.

Data extraction and management

Two review authors (LH and DH) independently extracted the data using a data extraction form based on the template developed by the Cochrane Consumers and Communication Review Group. It included the following components:

Details of study

We recorded: study design; description of comparison group; aim of intervention; aim of study; methods of recruitment of participants; inclusion/exclusion criteria for participation in study; informed consent obtained (yes/no/unclear); ethical approval (yes/no/unclear); funding source and amount (if stated); statistical methods and their appropriateness (if relevant) and consumer involvement (in the design of study and/or intervention; in delivery of intervention; in evaluation of intervention; in interpretation of study findings). We also recorded whether participation was voluntary.

Participant characteristics

Health professionals are the primary participants. From each study we recorded: number; age: range, mean (standard deviation); gender; ethnicity; professional qualifications (including country of origin, recognition of qualification in dominant country), type (i.e. allied health staff), other; language(s), previous cultural competence training (yes/no/unclear).

We also recorded the following data reflecting our conceptual framework:

  • Facilitators/trainers: professional characteristics and qualifications; cultural competence experience (yes/no/unclear and how described); health professional background (yes/no/unclear);

  • Patients: number, principal health problem; stage of problem or illness; one-off treatment or ongoing treatment; social demographic details including health literacy; literacy ability/level, education level, language, ethnicity, age; dominant language proficiency; principal communication mode (English/dominant language/other language);

  • Healthcare organisations: type; size (in number of beds or places); financial status (public, profit, not-for-profit, unclear); links between cultural competence education and organisational policies (described, not described), structures and support (yes/no/unclear, description).

We used the terminology used by authors to describe study participants e.g. black, Non-western, and the broader term culturally and linguistically diverse (CALD) to refer to these populations across studies, and GPs, physicians etc to describe the healthcare professionals.

Geographical location

We included: country; type of healthcare system (universal access, specified access for low income/immigrant or refugee patients); dominant language; descriptions of minority populations.

Setting

We recorded hospital, community health, and aged care settings primarily, and noted other settings.

Intervention

We used our conceptual framework (see Description of the intervention) as the basis for data extraction. Prior to the data extraction two authors (LH and DH) independently assigned information about the intervention used in each included study to the conceptual framework. Differences in assignment were resolved through consensus and the final allocation of these parameters was confirmed with a third author (PR). The conceptual framework includes the following four domains:

1. Education content:

a. Knowledge: We recorded reporting of the following, including any descriptions used: culture; cultural competence; models of health and illness; the socio-cultural context of health disparities, epidemiology; and social determinants; constructs of racism and prejudice; and specific theoretical models such as explanatory models of health and illness; patient centeredness; consumer participation; and other topics.

b. Assessment and application: We recorded reporting of the following, including any descriptions used: application of knowledge to relevant environmental, population, organisational and professional and systemic contexts.

c. Skills – inter- and intra-personal: We recorded reporting of the following, including any descriptions used: cultural self-assessment; communication skills; collaboration skills; non-verbal communication skills; deconstructing stereotypes; trust and partnership establishment.

2. Pedagogical approach:

We recorded reporting of the following, including any descriptions used: the teaching and learning method; key theoretical construct and principles.

3. Structure:

We recorded reporting of the following, including any descriptions used:

a. Delivery and format: face to face; group, online; other.

b. Frequency: a one-off event; series of workshops; on going training.

c. Assessment and evaluation: method of assessment (pre-training surveys, pre- and post-tests, other); evaluation method used such as specific assessment tools.

d. Organisational support: links to organisational policies and procedures; whether embedded within professional development program (yes/no/unclear); use of mentoring and peer support processes (yes/no/unclear); links to formal professional accreditation points or qualifications (yes/no/unclear); participation (voluntary, mandatory, unclear).

4. Participants: as outlined at 'Participant characteristics' above.

Intervention quality

We recorded any information on the quality of the intervention as assessed by the study authors; including any information regarding the fidelity/integrity of the intervention, such as if it was delivered as intended or not, and rate of attrition if clearly attributable to quality of the intervention.

For risk of bias of the study we utilised the seven domains of the 'Risk of bias' assessment tool (see Assessment of risk of bias in included studies below).

Outcomes

We have listed all the outcomes reported in each included study and how they were measured, but our analyses is confined to those outcomes selected a priori as described in Types of outcome measures. The selection of outcomes from each study is described below in Measures of treatment effect.

Study conclusions

We recorded the conclusions drawn by the study authors.

Where any details were not included in the published study, or were unclear, we contacted authors requesting further information. We contacted authors of four studies via email for further information (Majumdar 2004; Harmsen 2005; Thom 2006; Sequist 2010). We were unable to determine current contact details for the fifth study (Wade 1991) and attempts to contact the author were unsuccessful. Three authors generously provided additional information.

Assessment of risk of bias in included studies

Two review authors (LH and DH) assessed the risk of bias for each included study independently, using the ‘Risk of bias’ tool described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011).

We assessed risk of bias associated with RCTs, CCTs (quasi-RCTs) and cluster RCTs for seven domains from the ‘Risk of bias’ tool: sequence generation; allocation concealment; blinding of participants and personnel; blinding of outcome assessors; incomplete outcome data; selective outcome reporting; and other sources of bias (Higgins 2011). We also assessed the risk of bias associated with cluster RCTs for an additional domain: selective recruitment of participants.

We judged whether a study has a low, high, or unclear risk of bias for each domain using the criteria specified in the Cochrane Handbook for Systematic Reviews of Interventions. We contacted study authors for additional information about the included studies, or for clarification of the study methods as required. We incorporated the results of the 'Risk of bias' assessment into the review through standard tables, and systematic narrative description and commentary about each of the elements, leading to an overall assessment the risk of bias of included studies and a judgement about the internal validity of the review’s results.

Measures of treatment effect

The primary outcomes of this review focused on outcomes for patients of minority culturally and linguistically diverse (CALD) backgrounds, and included: treatment outcomes (e.g. clinical assessments); health behaviour (e.g. attitudes, concordance); involvement in care (e.g. involvement in decision-making); and evaluation of care (e.g. communication enhancement, perceptions of care).

We adopted the approach to measures of treatment effect as outlined by Brennan and colleagues (Brennan 2009) and included any primary outcome identified by study authors that fell within each of our four outcome categories. In addition, as a post-hoc process, two authors (LH and DH) independently assigned the outcomes reported in each included study to the outcome categories determined for this review, with the intention of resolving any differences in categorisation, if they had occurred, by the involvement of a third author (see Table 2). We made three other post-hoc decisions relating to the review outcomes. First, where there were outcomes reported for different time periods, we used time points furthest from baseline. Second, where a selected outcome was reported in terms of the dimensions of which it was comprised, we calculated the effect estimate for each dimension and chose the dimension whose effect estimate was ranked n/2 (Wade 1991). Third, where the study population was not stratified for participants of minority CALD backgrounds, we used the whole population (Thom 2006).

Table 2. Outcomes by category for included studies
  1. # indicates a primary outcome in the included study * indicates an outcome selected for this review +This outcome was comprised of different dimensions that were reported individually. No overall measure was reported so we calculated the effect estimate for each dimension and chose the dimension whose effect estimate was ranked n/2.

Included studies Harmsen 2005 Majumdar 2004 Sequist 2010 Thom 2006 Wade 1991
Primary outcome categories
Treatment outcomesNone
  • Client health outcomes (OARS)

  • Physical & Mental Health Assessment

#Proportion achieving clinical control targets within preceding 12 months for:

  • Haemoglobin level (HbA1c level less than 7.0%)*

  • LDL cholesterol (level less than 2.59 mmol/L [100 mg/dL])

  • Blood pressure (less than 130/80 mm Hg)

Change in patient:

  • Weight*

  • Systolic blood pressure

  • Glycosylated haemoglobin

None
Health behaviourNoneNoneNoneNoneClient attrition
Involvement in care #Mutual understanding between GP and patient*Expenditure of Healthcare & Social ServicesNoneNoneNone
Evaluation of care
  • Patient satisfaction with consultation*

  • Patient’s feeling that consideration shown

  • Patient’s perception of quality of care

#Patient satisfaction*None
  • Patient satisfaction with consultation

  • Patient perception of physician consideration

  • #Patient reported physician cultural competency*

  • Client perception of counselling process

  • Client perception of counsellors*+

Secondary outcome categories
Health professionals
Knowledge and understanding #Mutual understanding between GP and patient*^ #Improved understanding cultural attitudes*

#Clinician acknowledgement of racial differences in the quality of diabetes care for black clients over 12-month period:

  • Across health service

  • Within health centre*

  • Among their patients

NoneNone
Consultation processesNoneNoneNoneNoneNone
Evaluation of processes and outcomesNoneNoneNoneNoneNone
Healthcare organisations
Adverse eventsNoneNoneNoneNoneNone
Quality and safety measuresNoneNoneNoneNoneNone
Service utilisationNoneNoneNoneNoneNone
Health economic outcomesNoneNoneNoneNoneNone

Where multiple primary outcomes were identified within any category, we ranked the reported effect estimates for each of these outcomes and selected the outcome with the median effect estimate.

If there was no primary outcome within our categories we adopted the following strategy. First we used any outcome within our categories specified in sample size calculations; then, as necessary, we ranked relevant intervention effect estimates, as reported, and selected the median effect estimate. If the number of outcomes was even (Cn), we intended to include the outcome whose effect estimate was ranked n/2.

In Table 2 we report the rationale for the outcomes selected, including whether we used the primary outcome, the outcome with the median effect estimate in the results tables or another selection process.

We also intended to verify whether the specified primary outcomes in included studies were consistent with those identified in trial protocols and/or trial registry entries but we unable to locate either protocols or trial registry entries for the included studies.

We extracted the intervention effect estimate reported for all included outcomes with their associated P value and confidence intervals, and the method of statistical analyses used to calculate them. If an inappropriate statistical method was used we have presented only re-analysed data, where this has been possible, with the P value annotated 're-analysed'.

We calculated the effect estimates and P values for:

  • Dichotomous outcomes as relative risks adjusted for baseline or risk differences;

  • Continuous outcomes as standardised mean difference, relative percentage change post intervention, mean differences post intervention, and difference in mean change.

Effect estimates were standardised so that scales and other outcome measures were concordant. Ratios greater than one, and differences between the intervention and comparator groups greater than zero indicate benefit for the intervention group. Effect estimates were multiplied by minus one where necessary.

Unit of analysis issues

If studies were included in which clusters of individuals were randomised (cluster-RCTs, CCTs) to intervention groups, but where inference was intended at the level of the individual, we re-analysed taking account of intra-cluster correlation (ICC) to adjust the sample size. We contacted authors if the estimate of ICC was not reported but received no further information by doing so. In the absence of other information about ICCs used in analyses, we used the assumptions applied in the sample size calculations of an included study (Harmsen 2005) where the ICC was estimated to be 0.2 with an assumed cluster size of 20 patients per physician. We used the formula 1 + (M -1) ICC (where M is the average cluster size) to impute the design effect (DEFF) and estimated the cluster sizes where they were not reported by dividing the number of study participants by the number of units of randomisation. This approach necessitated two assumptions. First, we assumed that cluster sizes were evenly distributed across the study sample and second, we assumed CALD populations to be evenly distributed across clusters in each study population. We conducted post hoc sensitivity analyses to assess the impact of using different ICCs (see Sensitivity analysis).

We used the DEFF to adjust for clustering in those studies where this had not been done. For dichotomous data both the number of participants and the number experiencing the event were divided by the DEFF and rounded to whole numbers. For continuous data the sample size was reduced by dividing by the DEFF but means and standard deviations remained unchanged.

Dealing with missing data

We did not plan to undertake any imputation for missing outcome data but imputed missing summary data where possible and have reported our assumptions in the results tables. We investigated the effect of our choice of ICC on the pooled effect estimate in the meta-analysis of RCTs through sensitivity analyses (see Sensitivity analysis).

Assessment of heterogeneity

We intended to pool relative risks measuring the effects of cultural competency educational interventions compared with no education on patient outcomes by assessing if the interventions in the included trials were similar enough to draw meaningful conclusions.

We considered the similarity of interventions based on the four domains in our conceptual framework. Our primary focus was the content domain and we considered to be studies to be sufficiently similar if they had at least two components within that domain for inclusion in a meta-analysis.

Statistical heterogeneity in any meta-analyses was intended to be assessed by visual inspection of the scatter of effect estimates in the forest plots and the calculated I2 statistic (Higgins 2011).

Assessment of reporting biases

If we had found sufficient RCTs (at least 10) we would have used funnel plots and formal statistical tests to investigate funnel plot asymmetry. We anticipated heterogeneity in effect estimates because of variability in the interventions. We planned to use the approach identified by Brennan 2009, and developed by Rücker and colleagues, which proposes a statistical test for when substantial between-study heterogeneity exists when data points are proportions (Rücker 2008).

Data synthesis

We have reported summary statistics for each comparison from the included studies including: baseline and follow-up summary statistics; effect estimates and their statistical significance; and information on effect modifiers and study design. We intended to summarise effect estimates for dichotomous outcomes within each comparison and study design, including the median effect estimate, inter-quartile range, and range. We calculated the risk ratios (with 95% confidence intervals) for one study which reported raw data for one outcome. Graphs have been used to display data (see Data and analyses).

We intended to use meta-analytical methods to pool relative risks measuring the effects of different types of interventions to educate health professionals compared with no education if this was feasible. Random-effects meta-analysis was to be used to pool intervention effects because of anticipated clinical and methodological diversity, and where possible we reported an approximate 95% range of underlying effect estimates, based on the between-study variance estimate, to provide some information on the spread of effect estimates (Higgins 2011).

We prepared a 'Summary of findings' table from the results of the meta-analysis using the methods described at Chapter 11 of the Cochrane Handbook for Systematic Reviews of Interventions (Schünemann 2011). This involved rating the quality of evidence for each of the key available outcomes using the GRADE approach (Guyatt 2008; Higgins 2008). The quality of evidence was judged using the categories 'High', 'Moderate', 'Low', or 'Very Low'.

Subgroup analysis and investigation of heterogeneity

We intended to undertake a subgroup analysis to compare education interventions targeted at health professionals of minority cultural and linguistic backgrounds with those directed at all health professionals if we had found any studies. In addition, we wanted to compare hospital settings with other healthcare settings if there had been sufficient data to compare studies with and without organisational support processes. If there had been sufficient data we would have compared different aspects of cultural competence education interventions, for example those outcomes for interventions comprising an education domain with only a knowledge component, and those with a more comprehensive knowledge domain (knowledge, assessment and application and skills components).

We planned to investigate subgroups visually by using box-plots and formally through subgroup analyses, and if there had been sufficient trials we would have used random-effects meta-regression to estimate the relative change in the intervention effect (and 95% confidence interval) for each.

Sensitivity analysis

We planned to undertake sensitivity analyses based on the 'Risk of bias' assessment and remove studies at the greatest risk of bias from the analysis, but there were insufficient studies with similar outcomes to do this. We undertook post hoc sensitivity analyses to investigate the effect of our choice of ICC on the pooled effect estimate in those studies that did not adjust for clustering. We used a range of possible ICCs (0.01, 0.02 and 0.05) and re-calculated the design effect (DEFF) and subsequent impact on the effective sample sizes (ESS), effect estimates and 95% confidence intervals. These are reported in the Characteristics of included studies,

Consumer participation

Consumer participation in the review was realised through the Consumer Nominee Program, which is a program implemented through the Health Issues Centre, Melbourne, Australia. Two consumers from CALD backgrounds with experience of the health service system provided feedback on the protocol. The protocol and review also received consumer feedback via the Cochrane Consumers and Communication Review Group's editorial processes.

Results

Description of studies

See Characteristics of included studies; Characteristics of excluded studies.

Results of the search

Electronic searches were conducted in June 2012 and identified 10,915 citations. After the removal of 2166 duplicates, 8749 abstracts were reviewed for eligibility. Three additional people from the Cochrane Consumers and Communication Review Group staff were involved with screening abstracts (Megan Prictor, Dianne Lowe and Natalie Willis). Forty-eight studies were identified as potentially eligible and full copies were retrieved and reviewed by two authors independently (LH, DH). We excluded 42 studies; 1 is ongoing.

We updated electronic searches in February/March 2014 and identified 2576 citations. After the removal of 164 duplicates, 2412 abstracts were reviewed for eligibility by a researcher, Anneliese Synnot. Fourteen studies were identified as potentially eligible and full copies were retrieved. All 14 studies were assessed independently as ineligible for inclusion by two authors (LH and DH).

Included studies

We included 5 studies involving 337 healthcare professionals and 8400 patients (Wade 1991; Majumdar 2004; Harmsen 2005; Thom 2006; Sequist 2010).

The included studies varied in their stated aims, study design, settings, participants and how they were described, interventions and the outcomes measured (see Characteristics of included studies).

Decreasing inequalities in care between population groups was the primary aim for Harmsen 2005, and a secondary aim for Sequist 2010. The primary aim for Sequist 2010 and the aims of other included studies explicitly addressed specific outcomes for either patients (Thom 2006; Wade 1991) or patients and their healthcare professionals (Majumdar 2004). The study aims and settings are described in detail in Table 3.

Table 3. Comparison of study aims, settings and CALD descriptions of participants used
Study nameStudy aimsSettingCALD description of health professionalsCALD description of patients
Harmsen 2005

1) To decrease inequalities in care provided between “Western” and “non-Western” patients

2) To assess the effectiveness of an educational intervention on intercultural communication aimed to decrease inequalities in care provided between “Western” and “non-Western” patients.

GP clinics in The Netherlands2 of 38 GPs had "non-Dutch (but Western) ethnic backgrounds"Assessed validated patient cultural background scale (Harmsen 2006). The "non-Western group" comprised "mainly Turkish, Moroccan, Cape Verdean and Surinamese patients". The "Western” group comprised "mostly Dutch but also some patients from other Western European, North American, Canadian and Australian origin".
Majumdar 2004

1) To determine the effectiveness of cultural sensitivity training on the knowledge and attitudes of healthcare providers

2) To assess the satisfaction of patients from different minority groups with healthcare providers who receive this training

Two community home care agencies and one hospital in Southern Ontario, Canada.

The study was conducted in an urban area described as where “approximately 25% of population is foreign-born.”

Authors reported that “Providers from both experimental and control groups who completed the T3 follow-up had similar demographics.” Majority "identified themselves as Canadians, and had parents of either Canadian or British origin".

Authors also note that “To achieve a more complete cultural perspective, steps were taken to ensure the inclusion of healthcare providers from minority groups” but do not provide any further detail.

Authors reported that “Demographically, patients in the experimental and control groups were similar” and that "Both the experimental and control groups identified their ethnicity as 'Canadian', 'British' or 'European'. The first language spoken by both groups was English". No further information was reported.
Sequist 2010

1) To evaluate the effect of cultural competency training and performance feedback for primary care clinicians on diabetes care for “black” patients

2) To assess whether these efforts reduced racial differences between “white” and “black” patients for these 3 measures of diabetes care.

8 ambulatory Health Centres, Massachusetts, USAProcess for determining ethnicity not reported, however clinicians were described as "white" (82%),"Asian" (14%), "black" (2%), and"Hispanic" (2%)."Patient race was collected by self-report during the patient registration process" and participants were described as either "black" or "white" diabetic patients.
Thom 2006To develop and evaluate a brief cross-cultural curriculum for resident and practicing physicians based on a model of culturally competent physician behaviours, and to evaluate the training plus feedback compared to feedback alone with respect to changes in patient-reported physician behaviours, patient satisfaction, patient trust in his or her physician, and disease-specific patient health outcomes

Four locations in California, USA:

  • An academic medical centre-based family practice

  • A community-based primary care practice;

  • A rural family medicine residency program

  • An inner-city family medicine residency program

No information about how ethnicity determined. Sample comprised 53 physicians described as comprising 38 "White" or "Caucasian" (both terms used for physicians) (total 72%, comprising 64% of physicians in the intervention group and 80% of the control group); 8 "Latino" (total % not reported; I=27%: C=7%); 5 "Asian American" (I=9%: C=10%); 1 "African American" (I=0%: C=3%); and 2 others (not described further).

Process of assignment of ethnicity unclear. Potential participants approached on basis of self-designation in computerised records (as “Hispanic” or “Asian”) or on the basis of their surname.

Participants in one group reported to be more likely to "self-identify as Latino or Asia". Groups in the study described as comprised of the following:

"Caucasian non-Hispanic" (21.9% feedback only and 30.0% training and feedback), "Latino/Hispanic" (28.9% and 25.3%), "African-American" (20.3% and 23.2%), "Asian" (23.0% and 13.1%) and "Other" (5.9% and 8.4%).

Wade 1991

1) To examine the effects of brief culture sensitivity training for counselors and the effects of whether counselors were black or white, on black female clients' perceptions of counselor characteristics and the counseling relationship and clients' satisfaction with counseling

2) to assess client attrition across three counseling sessions

College counseling centre in midwest city USAProcess of determining ethnicity not reported. Described as "White" (4) or "Black" (4) (counselors were stratified to intervention and control groups in randomisation)Process of determining ethnicity not reported. Described as "Black".

Four of the included studies were cluster-RCTs (Wade 1991; Harmsen 2005; Thom 2006Sequist 2010) and one was an individually randomised controlled trial (Majumdar 2004). The studies were conducted in three countries: US (Wade 1991; Thom 2006; Sequist 2010), Canada (Majumdar 2004) and The Netherlands (Harmsen 2005). We also identified an ongoing US study with an estimated completion date of 2014 (Clark 2010).

Clustering was identified in four of the five included trials, but only two trials adjusted for clustering in their analyses (Harmsen 2005 and Sequist 2010), Harmsen 2005 assumed an intra-cluster correlation (ICC) of 0.2 and 20 patients per physician to calculate sample sizes (748 patients and corresponding 38 physicians were required for each measure). No study reported actual cluster sizes.

Two cluster trials were randomised at the clinician level; Harmsen 2005 randomised 38 General Practitioners and Wade 1991 block randomised 8 counsellors to get equal numbers of "White" and "Black" counsellors in each group. Two trials were randomised at the unit level. Sequist 2010 randomised 31 primary care teams and Thom 2006 randomised 4 heterogenous primary care practices in different settings (an academic family medical centre; a community primary care; and 2 family medicine residency programs, 1 rural and 1 inner-city). Majumdar 2004 randomly selected nursing and home care providers from two home care agencies and a hospital to receive cultural sensitivity training.

Study participants

Interventions were directed at healthcare professionals in all studies. Study participants were identified differently across the included studies. Sequist 2010 identified healthcare professionals as the study participants. Harmsen 2005; Majumdar 2004, Thom 2006 and Wade 1991 identified healthcare professionals and their patients as the study participants, although two of these did not include outcome measures for health professionals (Wade 1991; Thom 2006). One study (Harmsen 2005) directed an intervention to health professionals and patients. The terms used for study participants are described in detail in Table 3.

Overall the studies involved 337 healthcare professionals (including counsellors) and 8400 patients, of which at least 3463 (41%) were from culturally and linguistically diverse (CALD) backgrounds. The number of CALD participants could not be determined in one study (Majumdar 2004) and was calculated from reported proportions in the patient samples in two trials (Harmsen 2005; Thom 2006). The design effect from clustering led to an effective sample size (ESS) of at least 2755 CALD participants. All health professionals participated voluntarily. The conditions for patient participation varied. Consent for participation by patients was waived for Sequist 2010, where patient data were extracted from electronic medical records. Patient participation was described as voluntary in two studies but no details were reported for either (Wade 1991; Majumdar 2004). Thom 2006 recruited patients by mail with a follow-up phone call. Harmsen 2005 does not describe the patient recruitment process although patient participation was dependent on attendance for doctor consultation at the time of study data collection.

Three studies involved GPs (Harmsen 2005), primary care physicians (Thom 2006) or primary care teams, comprising physicians, nurse practitioners and physician assistants (Sequist 2010). Majumdar 2004 also included nurses and home care workers and did not report the number of each. Wade 1991 involved counsellors.

Two of the three US studies reported outcomes for "black patients" separately: Sequist 2010 reported outcomes for 2699 "black diabetic patients" and Wade 1991 looked at responses to counselling involving 80 "Black women" and was the only study that was confined to a single gender. The third US study (Thom 2006) included 429 patients with at least one visit for diabetes or hypertension but did not report outcomes by cultural background so the results for all patients were included in the review. Harmsen 2005 reported outcomes for 369 "non-Western" patients of GPs (and 614 Western patients) but did not specify any particular condition for inclusion. Outcomes for the 133 patients of home care agencies included in Majumdar 2004 could not be reported here because sufficient data were unable to be extracted.

Most of the healthcare professionals in the included studies were described as "white" or "Caucasian" (72% or more), except in Wade 1991 where half of the counsellors undergoing cultural competency training and in the control group were also "Black".

Intervention

There was considerable heterogeneity in the stated purpose, content, duration and nature of the interventions assessed in each study. Harmsen 2005 focused on intercultural communication. Majumdar 2004 and Wade 1991 examined the effects of cultural sensitivity training, and Sequist 2010 and Thom 2006 delivered cultural competency training with performance feedback to health professionals (see Table 4)

Table 4. Summary of intervention across domains of conceptual framework
Study ID Harmsen 2005 Majumdar 2004 Sequist 2010 Thom 2006 Wade 1991 SUMMARY - all studies
1. EDUCATIONAL CONTENT 
a) Types of knowledge 
Culture/cultural competenceIntercultural communicationCultural sensitivityCultural competencyCultural competencyCultural sensitivityTwo studies used the term cultural sensitivity (Majumdar 2004; Wade 1991).Two used the term cultural competence (Sequist 2010; Thom 2006). One study (Harmsen 2005) explicitly focused on intercultural communication. No study included explicit definitions of terminology used.
Models of health and illnessKleinmans' explanatory modelsNot reportedUnclearReported adaption of Tirado's model of culturally competent physician behaviours specifically knowledge of patients, communication skills and cultural competence [p3].Training session included "overview of issues and concerns culturally distinct individual bring to counseling" [p10]Three of the five studies reported inclusion of a description or understanding of specific models of health and illness as part of the education intervention. Two studies were unclear or did not report. 
Socio-cultural context of health disparitiesGP intervention "aimed to improve sensitivity and knowledge about culturally determined differences in views and behaviour “.Not reported

Specified goal of increasing knowledge about health disparities (p41)

 

Not reportedOverview of the issues and concerns culturally distinct individual bring to counselling (p.10)Three studies addressed socio-cultural context of health disparities but not in a common manner. Two studies did not report inclusion of this.
Epidemiology and social determinantsUnclearNot reportedCurriculum reviewed potential racial and cultural biases in healthcare, appropriate methods of collecting clinically relevant cultural data [p41]Included epidemiologic data on diabetes and hypertension in different racial/ethnic groups.Not reported

Two studies included explicit reference to epidemiology and the social determinants of health. Two studies did not report and one was unclear.

 

Constructs of racism and prejudiceUnclearNot reportedThemes from focus groups with black patients, including illustrative quotes and care recommendations, used to develop educational materials [p41]Not reportedParticipants instructed to "attend to racial and class differences between the counselor and the client" in the training practice sessions [p10]Two studies addressed constructs of racism via varied methods. Two studies did not report on this component and one was unclear.
Specific theoretical modelsKleinmans' explanatory models [p343]None reported

One of the stated goals of curriculum was to develop understanding of trust and bias.     

 

Knowledge of patients included "knowledge of patients cultural health beliefs and identification of their level of acculturation with respect to mainstream health beliefs"Pedersen's (1985) triad model of cross-cultural counsellingFour of the five studies reported on the use or inclusion of specific theoretical models as part of the education or training intervention.  One study was unclear.
Other topicsNone reportedCultural awarenessNone reported

Module 2 discussed cultural gap between healthcare professionals and patient's knowledge and belief systems [4]

 

None reportedTwo of the studies included topics in addition to what is specified in our conceptual framework. Thom 2006 reported on the cultural gap between healthcare professionals  and patients knowledge and belief systems, and Majumdar 2004 also reported on cultural awareness.
b) Assessment and application
Application of knowledge to relevant contexts

GPs trained in self-chosen strategies to solve gaps in views and culturally defined communication

style.[344]

Not reportedCurriculum reviewed ways to incorporate information into effective clinical care plans for people with diabetes.  Monthly educational materials included practical recommendations.Module 3 - cultural brokering - included "negotiating a treatment plan with patient and family, understanding community resources available to patients, and working with the healthcare system to meet the needs of culturally diverse patients”. [p4]None reportedThree of the five studies reported on this component, and two studies did not report. Thom 2006 reported most comprehensively.
c) Skills – inter- and intra-personal 
Cultural self-assessmentGPs were allowed to reflect on their own culturally defined norms, views and
communication style
Cultural Self-Awareness Questionnaire and the Dogmatism Scale instrumentsUnclearUnclearTraining included discussion on counsellor self-awareness and the minority client. Participants instructed to make explicit their own values during counselling practice sessions [p10]Three of the five studies reported on cultural self-assessment processes (Harmsen 2005; Majumdar 2004; Wade 1991). It was unclear in two studies.
Communication, collaboration and non-verbal communication skillsGPs and patients received intervention with intention of improving the intercultural communication between them. GPs: intercultural communication.Not reportedSkills building around effective cross-cultural communication and to provide action-oriented tools to address disparities.                     Module 2 “Enhancing Communication Skills for Cultural Competency - included "listening, explaining, acknowledging, providing recommendations, and working effectively with interpreters. Participants taught to recognize and deal with their own defensiveness as counsellors and recover from mistakes made during the process of counselling [p10].

Four of the five studies (Harmsen 2005, Sequist 2010; Thom 2006; Wade 1991) reported the inclusion of communication skills in the training /education intervention to improve cross cultural communication. One study did not report this.

Thom 2006 reported a specific  module on “Enhancing Communication Skills for Cultural Competency”.

Deconstructing stereotypesUnclearUnclearUnclearModule 1 objective to "teach techniques for assessing beliefs and practices of individual patients" infers that there would be an understanding and analysis of stereotypes.Training to "articulate the client's problems within a cultural framework" [p10] could infer some element of deconstruction of stereotypes but possibly insufficient time to address this.Two of the five studies inferred a deconstructing stereotypes component (Thom 2006; Wade 1991). Three studies were unclear.
Trust and partnership establishmentUnclearUnclear

Specified goal to meet understanding attitudes of trust and bias. 

 

An objective of Module 2 was to model problematic and improved physician communication [p4], through the LEARN method.Training included skills to anticipate and deal with client resistance. [p10]Three studies of the five (Sequist 2010; Thom 2006; Wade 1991) reported  addressing issues of trust and partnership establishment between patients and healthcare professionals. it was not clear on two studies.
2. PEDAGOGICAL APPROACH
Teaching and learning methodGP intervention based on Pinto’s three step method but patient intervention is not.Unclear

Unclear

Some discussion of taking a quality improvement framework.

Peer education, adult learning principles and application of LEARN mnemonic to the patient interview (Listen, Explain, Acknowledge, Recommend, Negotiate).Unclear

Two studies identified explicit  teaching and learning methods (Harmsen 2005; Thom 2006),  and  three studies were unclear.

 

Key theoretical construct and principlesKleinman’s theory of exchanging explanatory models cited.UnclearUnclearAdapted from a model developed by one of the co authors. emphasise goal of cultural versatility the concept of a cultural competency continuum used across all modules.Based on Pedersen's model which is designed to train counsellors to: (a) articulate the client's problems within a cultural framework, (b) anticipate and deal with client resistance, (c) recognize and deal with their own defensiveness as counsellors, and (d) recover from mistakes made during the process
of counselling.
Three studies (Harmsen 2005; Thom 2006; Wade 1991) reported key theoretical constructs and principles that underpinned the training. Two studies were unclear on this component.
 Comments  An underlying principle of the educational design was clearly grounding in the actual experience of black patients. The development of educational materials and recommendations tailored to caring for black patients were derived from patient focus groups and surveys. Clinicians were also given stratified patient data throughout the study.Curriculum was field tested with 18 primary care physicians and was refined based on participant feedback.  Thom 2006 reported the curriculum was field tested with 18 primary care physicians and was refined based on participant feedback.  Sequist 2010 reported that an underlying principle of the educational design was clearly grounding the curriculum in the actual experience of black patients. The development of educational materials and recommendations tailored to caring for black patients were also derived from patient focus groups and surveys.
3. STRUCTURE
a) Delivery and format
DeliveryFace-to-face impliedUnclearFace-to-face offsite delivery, including engagement with black diabetic patients to learn of barriers to managing diabetes in the community.Face-to-face implied.Face-to-faceFour of the five studies reported face to face delivery of the education/training intervention (Harmsen 2005; Sequist 2010; Thom 2006; Wade 1991). One study did not state  clearly how the intervention was delivered.
FormatFormat of GP program not described. Patients viewed 12-minute videotaped instruction in waiting room prior to consultation. Available in languages of major ethnic groups (Moroccan-Arabic, Moroccan-Berber, and Turkish).Unclear

Mix of lectures, group discussions, and community engagement activities

 

 

3 modules corresponding to the 3 areas of cultural competency model. Mix of didactic presentations, group discussion, role-playing with learners, critique, group exercises, use of trigger tapes, and handouts. [p4]Workshop with overview, group discussion and skills training by practice sessions with volunteer black clientsFour of the five studies (Harmsen 2005; Sequist 2010; Thom 2006; Wade 1991), reported a mixture of didactic and experiential learning formats including: lectures, small group discussions; video, role plays and simulation exercises.  One study was not clear.
b) Frequency and timing
Frequency and durationGPs: 2.5 day program of training which included follow-up session after 2 weeks          36 hours of cultural sensitivity training [p163]. No other details provided.Nurse practitioners and physician assistants attended 2 consecutive days. Physicians attended the second day only.  Monthly written education materials provided [p 41].Offered as either single half-day training session or 3 separate 1 to 1.5 hour sessions.4 hour cultural sensitivity training workshopAll studies reported a range of training options and duration from 2.5 days with follow up sessions, half-day workshops (3 to 4 hours) to short training sessions lasting 1 to 1.5 hour hours. Harmsen 2005 reported a 2.5 day program of training which included follow-up session after 2 weeks. Majumdar 2004 reported that the training comprised  of 36 hours  of training but did not report  further details. Sequist 2010 reported that the nurse practitioners and physician assistants attended 2 consecutive days, but that physicians attended the second day only.  Monthly written education materials provided [p41] were also provided.  Thom 2006 reported that the education intervention was offered as either a single half-day training session or 3 separate 1 to 1.5 hour sessions. Wade 1991 reported that the training entailed a 4-hour cultural sensitivity training workshop.
c) Assessment and evaluation
Method of assessmentUnclearUnclear

Not reported

 

Unclear

 

 

UnclearThree of the five studies (Harmsen 2005; Majumdar 2004; Wade 1991) were unclear on the method of assessment of the intervention. Sequist 2010 reported that a survey tool was used in the clinical trail setting to assess healthcareprofessional awareness of disparities.  Thom 2006reported the use of  a Patient -Reported Provider Cultural Competency Scale, where learners were asked post raining to rate the usefulness of the training on a 5 point Likert scale.

 

Evaluation method

UnclearUnclear

Not reported

 

 

Post-training evaluation on a reported] were rated on 5-point Likert-scale. As above. Only one study used post training evaluation.
d) Organisational support

 

Links to organisational policies and procedures

UnclearUnclearMonthly performance feedback reports provided.UnclearUnclearOnly one study (Sequist 2010) reported that monthly performance feedback reports were provided to participating clinicians that *were stratified by cultural group and focused on achieving desired BP, cholesterol, and A1c outcomes.  These were all aligned with broader organizational goals.

 

Embedded within professional development program

UnclearUnclear

Unclear

 

UnclearUnclear

Only one study (Sequist 2010) reported that the intervention was linked to clinician professional development through CME credit which was given to attending physicians.

 

 

Use of mentoring and peer support processes

UnclearUnclearMonthly "race-stratified reports" complied and given to physicians.UnclearUnclearOnly one study (Sequist 2010) reported a mentoring or support process though the provision of comprehensive race-stratified monthly reports.  The reports included:  Clinical performance metrics (A1c, BP, cholesterol control); Medication prescribing patterns (statin medications); Patient experience data collected via patient survey; Patient insight data collected via patient focus groups on topics including role of religion, race, and diet in chronic care management

 

Links to formal professional accreditation points or qualifications

UnclearUnclear* CME credit was given to attending physicians.unclearUnclear

Only one study (Sequist 2010) reported that CME credit was given to attending physicians.

 

 

Participation

Voluntary  - by letter of invitationVoluntaryVoluntary [p 45]VoluntaryVoluntary, but counsellors recommended by counselling facultyAll studies reported that participation in the education/training was voluntary.
4. PARTICIPANT CHARACTERISTICS 
a) Delivering of interventionFacilitator(s) not reported.Facilitator(s) not reported.Facilitated by off-site program, Harvard Pilgrim Healthcare FoundationGP instructors included authors of the paper, two other physicians with expertise in cross-cultural care, and experts in training and use of interpreters." [p4]Facilitator(s) not reported.

Two of the five studies reported on who facilitated the education/training interventions.

 

b) Engagement (target audience)GPs and their patients. Adolescents aged 13 to 17 excluded and parents of children up to 12 years interviewed.Nursing and home care providers and their patientsPrimary care physicians, nurse practitioners and physician assistantsGPs and their patients with diabetes or hypertension. Patients were paid $10 for completing baseline questionnaire and $5 for each of the 2 follow-up questionnaires.80 black women and 8 female counsellors (4 black and 4 white)Both health professionals and their patients were directly subjected to an intervention in one study only (Harmsen 2005), although four studies (Harmsen 2005, Majumdar 2004, Thom 2006 and Wade 1991) reported that participants were both health or welfare professionals and their patients/clients. Sequist 2010 described only health professionals as study participants.

 

STUDY SUMMARY

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Education Intervention was a 2.5 day program for GPs: and their patients.  It included a follow-up session after 2 weeks.

Focus of the intervention was to improve intercultural communication between GPs and their patients. 

Education intervention focused on improving cultural sensitivity.  A 36 hour program targeting nursing and home care providers and their patients.  Used a number of validated tools.

Insufficient information across the domains of our conceptual framework.  Awaiting further info.  

Two (consecutive)s day training program  for primary care physicians, nurse practitioners and physician assistants (though physicians attended only one day). Delivered by training experienced firm. Intervention can be mapped against most domains of our conceptual  framework.

Only intervention with  reporting on the organisational support component of domain 3.

Education intervention for GPs and their patients.

One of shortest education interventions.

Only study that explicitly describes the content of the education intervention.

Reporting on most  domains (with most detail) of our conceptual framework.

Small target group of counsellors (8). Unclear across  most domains of our conceptual framework. 

All 4 domains and their 11 sub-domains of the conceptual framework developed for this review were addressed in the descriptions of the educational interventions in some way in the included studies, although no study covered all aspects of the framework, or in the same way (see Table 1). Three categories of reporting against the framework were evident. Thom 2006 and Sequist 2010 were 'high reporting' with at least two-thirds of the domains and sub-domains described. Harmsen 2005 and Wade 1991 were 'medium reporting' with about half of all domains and sub-domains described. Majumdar 2004 was 'low reporting' and described less than a third of all domains and sub-domains described (see Table 4 and Appendix 9).

This diversity of approaches to cultural competence education is reflective of the field of cultural competence itself as outlined in the 'Description of the condition' section.

Comparison

In all studies the control group was given no intervention.

Outcomes

Thirty-two outcomes from the 5 included trials were categorised (see Table 2) and 12 outcomes were identified for potential inclusion in the review following the procedure outline in the review protocol (see Table 5) including 10 in our 4 primary outcome categories. These comprised treatment outcomes from three studies (Majumdar 2004; Thom 2006; Sequist 2010); a health behaviour outcome (Wade 1991); involvement in care outcomes from two studies (Majumdar 2004; Harmsen 2005); and evaluation of care outcomes from four trials (Wade 1991; Majumdar 2004; Harmsen 2005; Thom 2006).

Table 5. Rationale for outcome selection
Outcome categoryStudy outcomeStudy IDRationale for selectionData typeAnalysis
Primary outcomes *Sample size adjusted for clustering
Treatment outcomes

Client health outcomes (OARS)

Physical & Mental Health Assessment

Majumdar 2004Choice of two outcomes - more information from study authors needed to determine which had lower effect size.UnknownExcluded (no useable data)
Rate of achieving control target of LDL cholesterol < 2.59mmol/L (< 100mg/dL) in previous 12 months Sequist 2010A primary outcome (one of three) with the median effect size for population of interest ("black" patients).Rate (dichotomous)RD calculated as generic inverse variance using reported sample sizes, difference in reported proportions and derived numbers of those reporting "very often" or "somewhat often". Standard error calculated from reported 95% CI.
Change in patient weight (pounds) Thom 2006Study primary outcome for all study participants. Other primary outcome measures relate to study subgroups and all data are not reported.ContinuousEffect estimates (SMD) were standardised for concordance (to ensure effect estimates indicated benefit to intervention group when greater than zero (change in weight loss multiplied by minus one).*
Health behaviourClient concordance Wade 1991Only outcome in this category reported by study. Reported in terms of client attrition.DichotomousRR calculated to ensure standardisation (i.e. that benefits indicated by values > 1)
Involvement in careMutual understanding between GP and patient Harmsen 2005Only primary outcome in this outcome category reported by study.AgreementSMD calculated with generic inverse variance
Expenditure of Healthcare & Social Services Majumdar 2004More information from study author needed.UnknownExcluded (no useable data)
Evaluation of carePatient satisfaction with consultation Harmsen 2005One of three non-primary outcomes in this category reported by study with the median effect size.Rate (dichotomous)RD calculated with generic inverse variance
Patient satisfaction Majumdar 2004Outcome used for sample size calculation.UnknownExcluded (no useable data)
Patient reported physician cultural competency Thom 2006One of three outcomes in this category reported by study, but this was the only primary outcome.ContinuousSMD calculated with generic inverse variance
Client perception of counsellors - attractiveness dimension Wade 1991One of two outcomes reported in this category. Scale measures 3 behavioural dimensions of counsellors: expertness, trustworthiness and attractiveness. Limited data reported for client perception of counselling so we could not apply our criteria for selection as intended. Post hoc decision to use one dimension of scale in review. The study reported results for each dimension of client perception of counsellors. The effect estimate for each dimension was calculated. "Attractiveness" was the dimension with median effect (n/2).ContinuousInverse variance (SMD)*
Healthcare professionals
Knowledge and understandingImproved understanding cultural attitudes Majumdar 2004Primary outcomeUnknownExcluded (no useable data)
Clinician acknowledgement of racial disparities in diabetes care Sequist 2010

Primary outcome.

Reported at three levels (all health service, with own health centre, and within own practice). Outcome with median effect size chosen (within health centre).

Rate (dichotomous)RD calculated with generic inverse variance
Mutual understanding between GP and patient Harmsen 2005Also categorised as primary outcome in the patient-related outcome category, involvement in care.AgreementExcluded as used for patient-related outcome
Consultation processesNo outcomes reported
Evaluation of processes and outcomesNone reported
Healthcare organisations
Adverse eventsNo outcomes reported
Quality and safety measuresNo outcomes reported
Service utilisationNo outcomes reported
Health economic outcomesNo outcomes reported

Only two potential secondary outcomes, both for the same outcome category - knowledge and understanding by health professionals - were identified in the included trials (Majumdar 2004; Sequist 2010).

Primary outcome categories
Treatment outcomes

Only two trials provided sufficient treatment outcome data for inclusion in the review. Sequist 2010 reported dichotomous outcomes (whether clinical control targets were achieved within the preceding 12 months for "black patients with diabetes"), and Thom 2006 reported changes in continuous outcomes for a mixed patient group after six months. Data in both trials were extracted from patient medical records.

Sequist 2010 included three patient health outcomes as primary outcomes: haemoglobin level (HbA1c level less than 7.0%), LDL cholesterol (level less than 2.59 mmol/L [100 mg/dL]), and blood pressure (less than 130/80 mm Hg). The outcome with the median effect size for the population of interest was the proportion who achieved the clinical control target for LDL cholesterol (level less than 2.59 mmol/L [100 mg/dL]).

Thom 2006 reported three patient health outcomes as secondary outcomes. These included changes in: weight, systolic blood pressure, and glycosylated haemoglobin. The latter two outcomes were reported only for those patients with hypertension or diabetes but the number of patients with these conditions was not reported. This study did not report data for CALD patients separately, so we made a post-hoc decision to include changes in weight for this outcome category as it applied across the whole study population.

The third trial, Majumdar 2004, reported two patient health outcomes measured at three time periods: baseline; 3 months after cultural competency training for health professionals; and at 18 months. Client health outcomes were measured using self-reported assessments including the Off-Axis-Ratio (OAR) and physical and mental health outcomes measured with the Physical & Mental Health Assessment Questionnaire. There were insufficient data available to determine which outcome to select for this outcome category, and as no data from either outcome could be extracted neither could be included in this review.                                   

Health behaviour

Health behaviours were reported in only one trial. Wade 1991 assessed client attrition and reported the numbers of participants attending each of the three counselling sessions for the intervention and control groups. This allowed the risk ratios for participants attending the third counselling session to be calculated and this health behaviour to be compared for the intervention and control groups (ensuring concordance of benefit to intervention would be indicated by values greater than one).

Involvement in care

Two trials reported outcomes related to patient involvement in care (Majumdar 2004; Harmsen 2005), but only one trial at moderate risk of bias provided sufficient data for inclusion. Harmsen 2005 reported mutual understanding between the GP and their patient by comparing answers from the GP and patient about different aspects of the consultation, using a validated scale where responses could range from -1 (total misunderstanding) to +1 (complete mutual understanding). GPs completed the questionnaire immediately after the consultation and patient interviews were conducted three to eight days after a consultation. This outcome used data from patients and GPs and was also categorised as a secondary outcome for health professionals. In this review, mutual understanding was only used in this outcome category (see Table 5).

Majumdar 2004 reported an outcome called Expenditure of Healthcare & Social Services over 18 months which appeared to be concerned with the amount of money paid by a patient for their care, but we were unable to ascertain further information about this outcome so it could not be included in the review. 

Evaluations of care

Four of the five included trials reported outcomes in the 'evaluation of care' category, including two trials that reported multiple outcomes, but it was only possible to extract data from three trials (Wade 1991; Harmsen 2005; Thom 2006).The interventions in these studies were not considered sufficiently similar to justify meta-analysis.

Two outcomes were categorised as evaluations of care in Wade 1991: client perception of counsellors and client perception of counselling. Only the former was reported in sufficient detail for data to be extracted. The Counselor Rating Form - Short Form (CRF) assesses perceptions of counsellor credibility and attractiveness with twelve 7-point bipolar items, ranging from not very (1) to very (7). Three behavioral dimensions of counsellors were measured: expertness, trustworthiness, and 'attractiveness'. The average ratings for these dimensions were reported for three time-points: baseline and following the first two counselling sessions. As summary scores were not reported, we made a post hoc decision to use only the dimension of scale with median effect in this review. Effect estimates were calculated based on change from baseline scores and the second counselling session. We selected the dimension with median effect (n/2), attractiveness (which referred to client perception of the counsellor).

Harmsen 2005 reported three outcomes categorised as evaluations of care. All were secondary outcomes (patient satisfaction with consultation; patient perception of whether physician was considerate, and patient perception of quality of care) and were all measured in patient interviews at home three to eight days after consultations. The outcome with the median effect size was patient satisfaction with consultation, which was assessed with three answer categories (yes/doubtful/no), dichotomised to (‘yes’ versus ‘doubtful’ and ‘no’).

The previously validated Patient Reported Physician Cultural Competency (PRPCC) Scale was the primary outcome measure used in Thom 2006. The PRPCC asks patients about 13 physician behaviours identified as important for cultural competency using a 5-point scale (from 'never' to 'always') with the total score transformed to a 0 to 100 scale.

Secondary outcomes
Knowledge and understanding

Three trials measured outcomes that assessed knowledge or understanding of cultural competence among health professionals (Majumdar 2004; Harmsen 2005; Sequist 2010) but only Sequist 2010 provided data for this outcome.

Harmsen 2005 reported mutual understanding between the GP and their patient. This outcome was also categorised as a primary outcome (involvement in care) for patients so was not included as a secondary outcome (see Table 5).

Sequist 2010 measured clinician awareness of racial differences in care: across all of the health service; within their own health centre; and among patients they personally treated, using a 5-point Likert scale ('very often' to 'very rarely') and analysed using ordinal logistic regression. However data are reported as the proportion of clinicians who acknowledged awareness of racial disparity in care as 'very often' or 'somewhat often' after study completion. The outcome with the median effect size was clinician perception of racial differences in care across their health centre.

Majumdar 2004 assessed understanding of cultural attitudes, described as "open-mindedness and cultural awareness of nurses" with the Self-Assessment of Cultural Awareness tool. It appears that this instrument may have been validated but not published.

Other secondary outcomes

No other secondary outcomes related to health professionals in terms of consultation processes or evaluation processes and outcomes, or related to healthcare organisations in terms of adverse events, quality and safety measures, service utilisation or health economic outcomes, were measured in the included trials.

Excluded studies

We excluded 42 studies from the initial search, primarily because they were either not RCTs, the intervention was not related to cultural competence education and/or there were no patient outcomes (see Characteristics of excluded studies). One further study is ongoing and can not be included in the review at this stage (Clark 2010). Fourteen studies identified in updated searches run in February 2014 were assessed as ineligible for similar reasons.

Risk of bias in included studies

We judged the overall risk of bias of included studies as being moderate to high (see Figure 1). All included studies had mixed assessments across the 'Risk of bias' criteria, several of which were unclear, (in particular selection bias), or were problematic (such as incomplete outcome reporting) (see Characteristics of included studies and Figure 1 for further details).

Figure 1.

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Allocation

Four of the included studies provided insufficient description of both sequence generation and allocation concealment (Majumdar 2004; Harmsen 2005; Thom 2006; Sequist 2010) so were judged to be unclear for these two domains. Wade 1991 was also unclear about allocation concealment but used a table of random numbers for assignment, which was judged to be at low risk for sequence generation.

Blinding

Three studies in which the intervention was directed at healthcare professionals specifically reported that patients were unaware of the group to which their healthcare professional had been allocated. These studies were rated at low risk of performance and detection bias due to adequate blinding (Wade 1991; Thom 2006; Sequist 2010;). In a fourth study (Harmsen 2005) interventions were also directed to the patients of GPs in the intervention group. This study was judged to have an unclear risk of bias for performance, although detection bias was judged to be low. It appears there was no blinding in Majumdar 2004, which was assessed to be at high risk of both performance and detection bias.

Incomplete outcome data

Sequist 2010, which extracted patient data from medical records, had complete data for the primary outcome (treatment outcomes), and subsequently had low risk of attrition bias. Patient data were also extracted from medical records in Thom 2006, but actual numbers were not reported with the results, although at least 25 per cent of the study sample did not return data, so the risk of bias was high. The remaining studies also experienced high rates of attrition related to primary outcomes, particularly for CALD groups. In Harmsen 2005 patients with a "non-Western" country of origin dropped out more often than "Western" patients (55% compared with 45% of the total number of failed home interviews, respectively). Majumdar 2004 progressively lost one third of participating health professionals, mostly from the intervention group, and two-thirds of patients, from both the intervention and control groups, throughout the study. Attrition was greatest among the clients in the control group for Wade 1991; 25 of 40 participants in the control groups returned for the second session and only 11 of 40 returned for the third session.

Selective reporting

No studies reported published protocols against which trials could be assessed, so it was not possible to determine whether selective reporting had been a factor in any included studies. Majumdar 2004 and Wade 1991 were rated at high risk of bias as stated study outcomes were not clearly reported or only partially reported. The remaining studies were rated as unclear risk of bias on this domain.

Other potential sources of bias

Of the four cluster RCTs (Wade 1991; Harmsen 2005; Thom 2006; Sequist 2010), only Sequist 2010 was judged to be at low risk of other bias. An imbalance in the attrition of "non-Western" patients for the intervention and control groups in Harmsen 2005 gave an unclear risk of other bias. There were high risks of bias in the other two studies. Wade 1991 gave no indication of whether the analyses took account of clustering. Thom 2006 had a number of areas of concern including unit-of-analysis errors. While adjustments were made for some possible effects, results were reported for individuals with no adjustment for clustering. There were also significant baseline differences between the intervention and control groups in terms of demographic characteristics, diagnoses and outcome assessments. In addition there was considerable heterogeneity across the participating sites and it was unclear whether physicians were recruited before or after the allocation of cluster sites. It is unclear if there are any other potential sources of bias in Majumdar 2004.

Effects of interventions

See: Summary of findings for the main comparison

See Summary of findings for the main comparison for the effects of cultural competence training for health professionals compared with no training.

Primary outcome categories

Treatment outcomes

There was no evidence of effect in treatment outcomes following cultural competence education of health professionals in two studies. Sequist 2010 reported no evidence of effect in the proportion of patients with diabetes who achieved clinical control targets for LDL cholesterol (level less than 2.59 mmol/L [100 mg/dL]) within the preceding 12 months between the intervention and control groups (the analysis was on "black" patients in the intervention and the control group) (Risk Difference (RD) -0.02, 95% CI -0.06 to 0.02; 1 study, USA, 2699 patients), see Analysis 1.1; Figure 2 . These results were reported as adjusted for clustering. Unadjusted data were not available. Thom 2006, in the USA, found no evidence of effect in the mean weight loss between patients in the intervention and control groups after six months (Standard Mean Difference (SMD) 0.07, 95% CI -0.41 to 0.55), see Analysis 1.2; Figure 3. The quality of evidence for this outcome was considered to be low because allocation concealment was unclear in both studies. In addition, attrition was unclear in Thom 2006, there was no adjustment for clustering and outcomes for CALD patients were not reported separately (see Characteristics of included studies). Sensitivity analyses using different values for the intra-cluster correlation and the subsequent design effect from clustering did not substantially alter the magnitude or significance of the summary effect size.

Figure 2.

Forest plot of comparison: 1 Cultural competence training vs no training, outcome: 1.1 Treatment outcomes.

Figure 3.

Forest plot of comparison: 1 Cultural competence training vs no training, outcome: 1.2 Treatment outcomes.

Health behaviour

One US study showed that client concordance with attendance significantly improved in the intervention group across three counselling sessions following cultural competence education of counsellors (Wade 1991). Women in the intervention group were more than three times more likely to attend the third counselling session compared with those in the control group (relative risk (RR) 1.53 (95%CI 1.03 to 2.27), see Analysis 1.3; Figure 4. The evidence for this outcome was considered to be low as it was from a single study and the study sample was small, especially after adjustment for clustering. Sensitivity analyses using different values for the intra-cluster correlation and the subsequent design effect from clustering did not substantially alter the magnitude or significance of the summary effect size. This outcome was considered to be unaffected by the high attrition that contributed to the overall low quality rating given to this study (see Characteristics of included studies).

Figure 4.

Forest plot of comparison: 1 Cultural competence training vs no training, outcome: 1.3 Health behaviour.

Involvement in care

Involvement in care by "non-Western" patients in one study in The Netherlands increased in terms of improved mutual understanding between "non-Western" patients (described as "Turkish, Moroccan, Cape Verdean and Surinamese patients"). and their largely "Western" GPs (mostly Dutch but including others) following cultural competence education of the health professionals. This outcome was reported as a mean difference of 11 per cent of the range minus 2.0 to plus 2.0 (SMD 0.21, 95% CI 0.00 to 0.42) (Harmsen 2005) (see Analysis 1.4; Figure 5). The quality of evidence for this outcome was considered to be low due to concerns about baseline imbalance and high attrition among the CALD study participants (see Characteristics of included studies).

Figure 5.

Forest plot of comparison: 1 Cultural competence training vs no training, outcome: 1.4 Involvement in care.

Evaluations of care

Evaluations of care were mixed in three studies that were not pooled because of significant differences in the content of the interventions (see Table 4).

Harmsen 2005 reported patients' satisfaction with consultation as a dichotomous outcome and found no difference between the proportion of satisfied "non-Western" patients in the intervention and control groups (RD 0.14, 95%CI -0.03 to 0.31) (see Analysis 1.5; Figure 6). The quality of evidence was considered to be low due to concerns about baseline imbalance and high attrition among the CALD study participants (see Characteristics of included studies).

Figure 6.

Forest plot of comparison: 1 Cultural competence training vs no training, outcome: 1.5 Evaluations of care (dichotomous).

Two studies reported evaluations of care as a continuous outcome (Thom 2006; Wade 1991) (see Analysis 1.6; Figure 7) . Neither study allowed for clustering so it was necessary to adjust their sample sizes based on the intra-cluster correlation used in the sample size calculations in Harmsen 2005 (ICC = 0.2). Sensitivity analyses using different values for the ICCs did not substantially alter the magnitude or significance of the summary effect sizes (see Characteristics of included studies). Thom 2006 reported patient scores of physician cultural competency and found no evidence of effect between the intervention and control groups (SMD 0.11, 95% CI -0.63 to 085). However, the study reported the outcome for all patients, both "Caucasian" (27% of study participants) and others (73%) (described as "Latino", "African American", "Asian" or other). Wade 1991 found client perceptions of their health professionals among women in the intervention group to be significantly higher than for those in the control group (SMD 1.60, 95% CI 1.05 to 2.15) (see Analysis 1.6; Figure 7). This finding was independent of whether the counsellor was "Black" or "White". The quality of evidence was considered to be low as allocation concealment and attrition were either high or unknown and sample sizes were small, particularly after adjustment for clustering (see Characteristics of included studies).

Figure 7.

Forest plot of comparison: 1 Cultural competence training vs no training, outcome: 1.6 Evaluations of care (continuous).

Adverse outcomes

None of the included studies reported adverse outcomes.

Secondary outcomes

Knowledge and understanding

Clinician awareness of racial disparities in diabetes care study completion was higher in intervention health centres compared with those in the control group (70% vs 51%) (Sequist 2010). The authors analysed differences between groups using ordinal logistic regression and found it to be statistically significant (P = 0.003), whereas the relative risk of acknowledging disparity in care as 'very often' or 'somewhat often' between the intervention and control groups, without adjustment for the clustering, showed no evidence of effect (RR 1.37, 95% CI 0.97 to 1.94, P = 0.07) (Sequist 2010) (see Analysis 1.7; Figure 8). Adjustment for clustering, if it had been possible, would reduce this effect estimate.

Figure 8.

Forest plot of comparison: 1 Cultural competence training vs no training, outcome: 1.7 Knowledge & understanding of health professionals.

Other secondary outcomes

No other secondary outcomes related to health professionals in terms of consultation processes or evaluation processes and outcomes, or related to healthcare organisations in terms of adverse events, quality and safety measures, service utilisation or health economic outcomes were measured in the included trials.

Discussion

Summary of main results

Despite methodological limitations of the included studies, this review provides the best available evidence for widely used interventions that aim to improve cultural competence among health professionals working with culturally and linguistically diverse (CALD) populations. The purpose of this review was to assess the effects of cultural competence education interventions for health professionals primarily on patient-related outcomes. The review did not set out to judge the intrinsic value or ethical principles on which cultural competence is based, which is well-established, nor did it incorporate evidence from qualitative studies as this was beyond the scope of this review. As we outlined in the Background, due to differences in terminology and their cultural meanings and the evolving concepts and practices across countries, when addressing all populations as a whole, we use the term CALD participants or a similar term but in all places when we report the data for specific populations we use the language of study authors for included studies.

Five randomised controlled trials assessing the impact of cultural competence education for health professionals on patient-related outcomes were identified. The studies involved 337 healthcare professionals (GPs, primary care teams and counsellors) and 8400 patients, of which at least 3463 (40.5%) were from CALD backgrounds, although the design effect from clustering meant an effective minimum sample size of 3164 CALD participants. The interventions and outcomes assessed in the included studies were diverse and meta-analyses could not be performed despite our efforts to maximise the data available through the use of broad outcome categories.

We found no evidence of an effect on treatment outcomes or evaluations of care as a result of the intervention. There was low quality evidence of greater patient engagement and patient behaviour. Mutual understanding between "non-Western" patients and their mostly "Western" GPs improved in one study in The Netherlands (SMD 0.21, 95% CI 0.00 to 0.42) (Harmsen 2005) and client concordance improved in one small US study where "Black" women were one and half times more likely to attend a third counselling session compared with similar women in the control group (RR 1.53, 95% CI 1.03 to 2.27) (Wade 1991). None of the five included studies assessed adverse outcomes. One American study provided information on health professional outcomes and reported no evidence of an effect on clinician awareness of racial differences in the quality of diabetes care at the study completion based on ordinal logistical analysis (Sequist 2010), but in this review we found no difference in the relative risk of those reporting awareness of racial disparities in care 'very often' or 'somewhat often'. None of the five included studies examined the effect of cultural competence education on healthcare organisations.

When the conceptual framework developed for this review was applied to the specific interventions, all of its 4 main domains and the 11 associated sub-domains were addressed in some way. However none were common to all interventions, and no study reported all aspects of the framework (see Table 4 and Table 6). This exercise suggests that it may be possible to obtain greater agreement on core components for cultural competence education interventions.

Table 6. Quantitative summary across domains
Study ID Harmsen 2005 Majumdar 2004 Sequist 2010 Thom 2006 Wade 1991 TOTAL
1. EDUCATION CONTENT:      
a. Types of knowledge
·   Culture/cultural competence5
·   Models of health and illnessNRU3
·   Socio-cultural context of health disparitiesNRNR3
·   Epidemiology and social determinantsUNRNR2
·   Constructs of racism and prejudiceUNRNR2
·   Specific theoretical modelsNR4
·   Other topics NR NR NR +2
Sub-Total 4 1+ 5 4+ 5

19/30

60%

b. Assessment and application
·   Application of knowledge to relevant contexts.NR√.NR3
Sub-Total 1 0 1 1 0

3/5

60%

c. Skills – inter and intra personal 
·   Cultural self-assessmentUU3
·   Communication, collaboration and non-verbal communication skillsNR4
·   Deconstructing stereotypesUUUNR1
·  Trust and partnership establishmentUU3
Sub-Total 2 1 2 3 3

11/20

55%

DOMAIN TOTAL 7 2 8 8 8

33/55

60%

2. PEDAGOGICAL APPROACH:      
a.   Teaching and learning methodUU3
b.    Key theoretical construct and principlesUU3
Comments   +2
DOMAIN TOTAL 2 0 1+ 2+ 1

6/10

60%

 3. STRUCTURE:      
a. Delivery and format      
·    DeliveryU4
·    FormatU4
b. Frequency and timing      
·    Frequency and duration5
c. Assessment and evaluation      
·    Method of assessmentUUU2
·    Evaluation methodUU 2
Total 3 1 5 5 3

17/25

68%

d. Organisational Support      
·    Links to organisational policies and proceduresUUUU1
·    Embedded within professional development programUUUU1
·    Use of mentoring and peer support processesUUUU1
·    Links to formal professional accreditation points or qualificationsUUUU1
·    Participation5
Total 1 1 5 1 1

9/25

36%

DOMAIN TOTAL 4 2 10 6 4

36/50

72%

4. PARTICIPANTS’ CHARACTERISTICS:      
a. Delivery of interventionNRNRNR2
b. Engagement (target audience)5
DOMAIN TOTAL 1 1 2 2 1

7/10

70%

TOTAL number of framework components reported (proportion of framework described by study)

14

(56%)

5+

(20%)

21

(84%)

18+

(72%)

14

(56%)

14/25

(58%)

The diversity of approaches to cultural competence education reflected in the included studies, in terms of the terminology, goals of the interventions and descriptions of participants, supports our earlier description of the complexity of issues inherent within the field of cultural competence itself (see Background).

Overall completeness and applicability of evidence

The review aimed to assess the effects of cultural competence interventions for health professionals on patient-related outcomes, health professional outcomes and healthcare organisational outcomes, but the small number of studies and their considerable heterogeneity posed significant challenges in addressing these objectives, and limits the generalisability of the review's findings. All included studies in the review were conducted in three high-income countries, which may also limit the applicability of the evidence globally. The majority of participants were recruited from primary health practices and the results are considered applicable to these settings but it is unclear whether they would also be applicable to acute settings. One study (Thom 2006) did not report findings for the CALD population separately, which may have led to an underestimate of the effect on this group. Three studies compared CALD and non-CALD populations and reported CALD populations separately (Majumdar 2004; Harmsen 2005; Sequist 2010), whereas one study included only CALD patients (Wade 1991).

Two studies (Thom 2006; Sequist 2010), addressed at least two-thirds of the components of our conceptual framework. These were the only studies in the review to provide data on treatment outcomes. However these studies differed in terms of the duration of the interventions (two days versus three hours), links to organisational support (see Table 4) and patient populations (‘black’ patients with diabetes, and all patients). Two studies reported about half of the conceptual framework components (Wade 1991; Harmsen 2005) and both investigated two primary outcomes of this review. Only one of these outcomes was common to both studies, evaluation of care, the most commonly reported outcome, which was included in four of the five included studies. Again the interventions in these two studies varied in terms of duration (2.5 days versus 4 hours). Only Harmsen 2005 incorporated a patient-directed component. The fifth study in the review investigated three of the four primary outcomes of this review but these were not reported in sufficient detail to enable us to extract data and this study provided the least information about the education intervention (Majumdar 2004).

There were few common outcomes across the included studies. The primary outcomes for this review were patient-related. Outcomes related to evaluations of care were most commonly reported (four of five studies), followed by treatment outcomes (three studies), involvement in care (two studies) and health behaviours (one study). However, even when outcomes were sufficiently similar, some were insufficiently reported for us to obtain data, or the education interventions were too dissimilar for meta-analyses to be performed. Follow-up times also varied across studies. It remains unclear how long it might be expected that patient treatment outcomes be influenced by the introduction of an education intervention directed at health professionals, or whether additional process or organisational considerations may precede such outcomes, such as greater involvement in care or the presence of organisational supports.

Three studies looked at outcomes related to health professionals (Majumdar 2004; Harmsen 2005; Sequist 2010), but only one provided data for this review (Sequist 2010). None of the included studies looked at organisational outcomes.The completeness and applicability of the evidence from this review was largely influenced by the differences in the scope and designs of the included studies and how the education interventions were conceptualised and implemented. The problem was exacerbated by limited descriptions of the interventions. This is not surprising given the lack of consensus in the cultural competence field. Sue 2009 pointed to the use of different meanings attributed to cultural competence, inadequate measures and "the lack of measurements and research designs for evaluating [cultural competence's] impact in treatment". All these issues were evident in the five RCTs included in this review, in which the interventions varied considerably in terms of educational content, pedagogical approach, structure and participant characteristics (see Table 4). We had anticipated considerable heterogeneity which prompted the development of our conceptual framework. Our framework proved useful in highlighting gaps in the descriptions of the interventions examined.

A lack of consistency across the included studies in how CALD populations were selected and described was also problematic. It is clear, however, that study authors identified study participants based on a conceptualisation of characteristics of cultural and linguistic differences seen to disadvantage them and contribute to health inequities.

The considerable heterogeneity and inconsistency in cultural competence education interventions mean that it remains uncertain as to what is the most effective way to provide culturally competent education for health professionals that lead to improved patient outcomes. It is somewhat surprising, given that cultural competence education is not a new concept, and has evolved over the last thirty years. It is clear, however, that the conceptual challenges identified in our review contribute to the low number of studies that met our inclusion criteria.

Quality of the evidence

The risk of bias in the five RCTs included in this review ranged from moderate to high risk, and the evidence for all reported outcomes was judged to be of low quality (see Summary of findings for the main comparison). Four of the five studies (Majumdar 2004; Harmsen 2005; Thom 2006; Sequist 2010) involved both CALD and non-CALD patients and where possible we used only data from the CALD participants, however this was not possible in two studies (Majumdar 2004; Thom 2006) . We also looked for outcomes for healthcare professionals involved in the studies. Three studies assessed an outcome in the category, knowledge and understanding, for this group, but data from only one study could be used in the review (Sequist 2010). The other two were mutual understanding (Harmsen 2005) and improved understanding of cultural attitudes (Majumdar 2004) . Mutual understanding was categorised as both a health professional outcome (knowledge and understanding) and as a patient outcome (involvement in care). The latter took precedence based on the criteria we established a priori. No data could be extracted from Majumdar 2004.

The quality of the evidence for different outcomes was also limited due to a range of other factors including: data that came from single studies; allocation and concealment that was unclear; high attrition; the inclusion of both CALD and non-CALD participants; and small sample sizes, particularly after adjustment for clustering. All studies had design limitations, and in all but one (Sequist 2010), at least one attribute was considered to be at high risk of bias.

Potential biases in the review process

We used a broad search strategy in an effort to identify all possible RCTs. Among the specialist electronic databases used were those for non-English language and educational interventions and trial registers, so we are confident that all relevant studies are likely to have been identified. We established two strategies a priori in this review that helped to prevent or minimise bias. First, we developed a conceptual framework at the protocol stage. This allowed two authors to independently identify the specific components of each intervention based on the framework. Use of a conceptual framework developed a priori enabled us to describe interventions consistently across the different studies. Our second strategy was to nominate outcome categories, also a priori. We applied a post hoc process to categorise outcomes in the included studies based on common methods for Cochrane reviews. We used an outcome category approach in anticipation that there would be considerable variation in the choice of outcomes used across studies. Our post hoc amendments, whereby two authors independently categorised outcomes in the studies, worked to strengthen this process. We also undertook sensitivity analyses based on a range of possible inter-cluster correlation coefficients for two studies that did not adjust for clustering (Thom 2006, Wade 1991). In the end we took a conservative approach to account for the design effect from clustering in these studies.

Not all relevant data could be obtained from the published studies. Additional information was sought from study authors where possible and some data were computed from the information available, which may have introduced some bias. In addition, the absence of data specific to a CALD population in one study may have diluted the effect of the intervention (Thom 2006).

Agreements and disagreements with other studies or reviews

Two other reviews include studies common to this review although there are important differences in approach and findings (Beach 2005; Lie 2010). These earlier reviews include a broader range of studies but neither include one study included this review, a cluster RCT (Harmsen 2005). This is a significant difference as Harmsen 2005 is the only study in our review in which an intervention was directed at both the health professionals and their patients. Harmsen also introduces an outcome, mutual understanding, which seeks to assess the acceptability of intercultural communication between the doctor and patient in the consultation. There are other important methodological differences. While this review has only one study (Wade 1991) in common with Beach 2005, which did not assess study quality, the review undertaken by Lie 2010 includes four studies included in this review (Wade 1991; Majumdar 2004; Thom 2006; Sequist 2010). Lie 2010 assessed study quality with two different assessment tools, a modified STROBE (Strengthening the Reporting of Observational Studies in Epidemiology: Explanation and Elaboration) criteria and the MERSQI (Medical Education Research Study Quality Instrument). While details of these assessments are not reported, the quality assessments for two of the four studies (Majumdar 2004 and Thom 2006) differed to those made in this review. Both studies were rated as low overall quality in this review, whereas Lie 2010 found their quality to be moderate. In addition both Majumdar 2004 and Wade 1991 are described by Lie as quasi-RCTs. Some of these disparities appear to be related to unit of analysis; neither Thom 2006 nor Wade 1991 are reported in Lie 2010 as cluster RCTs but in both studies randomisation occurred at the healthcare professional level (primary care practice and counsellors) and analyses were performed at the patient level. Results from both studies were re-analysed in this review using the intra-cluster correlation assumed in the sample size calculations performed by Harmsen 2005. We found attrition to be high in the third study common to both reviews (Majumdar 2004) and although the study reported the use of several validated instruments we were unable to extract any useable data for any outcome in this review.

Our review does, however, support recommendations made by Lie 2010, and the conclusions reached by Truong 2014, particularly regarding the need to address the quality of studies in this field and the need for research and resources for more rigorous examination of the content, application and outcomes of cultural competence education interventions. Of note is some congruence with our review findings from the Harmsen 2005 study, which was the only study that incorporated a patient-directed component, and a recent update of a Cochrane review that focused on interventions to promote a patient-centred approach in clinical consultations. Dwamena 2012, whose authors synthesised the evidence of effects on training health professionals to be more patient-centred, found “modest support for the observation that multifaceted interventions have an effect that is not found with training for providers only”.

Authors' conclusions

Implications for practice

Cultural competence continues to be developed as a major strategy to address health inequities. We identified five studies assessing the effects of cultural competence education for health professionals on patient-related outcomes. There was positive, albeit low quality evidence, showing improvements in the involvement of CALD patients in their care. The reported findings either showed support for the educational interventions or no evidence of effect. No studies assessed adverse outcomes. The quality of evidence however, is insufficient to draw generalisable conclusions, largely due to the heterogeneity of the interventions in terms of their scope, design, duration, implementation and outcomes selected. The implementation of culturally competent health care needs better quality evidence supported by evaluations that seek to identify the key components associated with effectiveness and acceptability to health professionals and the populations to whom care is directed.

It is important to point out that our review did not set out to judge the intrinsic value or ethical principles on which cultural competence is founded. The need to address health inequities and provide culturally and linguistically appropriate health care is well established, and present in the ever complex and diverse societies across the world. Moreover, cultural competence education for health professionals remains a key initiative for governments, policy makers, healthcare organisations, and health professionals, and is supported through policy frameworks, standards, and accreditation requirements. From the evidence assessed in this review, however, we conclude that uncertainty exists about the best and most effective way to educate health professionals in cultural competence that leads to improved health outcomes for CALD populations.

Despite this uncertainty, we believe the review provides insights for those seeking to add value to existing cultural competence education in policy and practice through the following observations:

  • Better alignment between policy frameworks, practice and research activities is needed. For example, policies that aim to reduce health inequities, by guiding healthcare practice, performance and accountability, could be more explicitly linked to, and informed by, research and evidence. Similarly, the direction, scope and methodology of research could be better tailored to address identified policy and practice imperatives.

  • Cultural competence education programs need to be better specified and described including their conceptual rationale, actual content, delivery, organisational support and approach to evaluation.

  • Routine evaluations of the impact of cultural competence education programs need to consider whether any data collected assess the impact on patients, health professionals and health organisations and how these data could better contribute to quality improvement.

  • Evaluations could consider the range of outcomes used in this review as they encompass the patient experience, the potential impact on health professionals and the potential impact on organisations.

All of these areas are important considerations for future decision-making in cultural competence education, policy and practice.

Implications for research

Our review has revealed several important and timely considerations for future research.

  • There is need for greater methodological rigour in studies of cultural competence education for health professionals. In particular, well-designed and appropriately resourced randomised controlled trials are needed that address the scope of potential impact of cultural competence education for health professionals on CALD patients, the health professionals themselves and the organisation involved.

  • Further research is required to gain agreement on the core components of cultural competence education, including how such components are described, delivered, organisationally supported and evaluated. Establishing consensus on core components of cultural competence education is timely given its prominence as a core strategy to address health inequities. The conceptual framework used in this review provides a basis for this process. The four core domains identified are: educational content; pedagogical approach; structure of the education; and participant characteristics. These domains and the 11 sub-domains of our framework enable more consistent reporting and assessment across studies and populations. This would allow different approaches to be assessed consistently, such as duration andfrequency.

  • There is a need for the development of internationally agreed upon terminology to describe people from diverse population groups, and to measure health status, inequities and outcomes. Greater consistency in conceptualisations and usage of key terms, better systematic reporting and guidance for researchers conducting secondary analyses of published studies in which minority CALD populations are described would also be beneficial.

  • Future studies should consider the range of outcomes used in this review. We used a range of patient outcome categories (treatment outcomes; health behaviours; involvement in care and evaluations of care), which are reflective of patient's experience of care and provide a more holistic assessment of the potential impact of cultural competence education. It is likely that each of these outcomes may contribute to patient decision-making. The relationship between the outcomes however, is unclear and also warrants further investigation. Longitudinal studies may need to be considered due to uncertainty about how long it may take for cultural competence education of health professionals to influence patient treatment outcomes.

  • Different structures of cultural competence education for health professionals need to be assessed, for example duration and frequency. In particular, it needs to be established whether CALD patients should also be targeted by these type of interventions, and whether CALD consumers could be included in participatory research development and evaluation.

  • The impact of cultural competency in health research should also be explored in terms of the skills, attributes and interpretations used by researchers, and also in the way that CALD populations are selected, described and included. Current resources to assist with good reporting of health research studies (such as the Equator Network Undated) do not currently include advice of this type.

Acknowledgements

We gratefully acknowledge Cath Harmer of the Department of Health (Victoria), Consumer Partnerships and Quality Standards Unit for her support of and commitment to this review. We also thank the staff and editors of the Cochrane Consumers and Communication Review Group, particularly Rebecca Ryan, Josip Car, Megan Prictor, Dianne Lowe, Anneliese Synnot, Natalie Willis and Sophie Hill, and Sharon Kramer from the Australasian Cochrane Centre for their advice and assistance in the preparation of this review. MP, NW, AS and DL helped with screening abstracts. We also thank Antonio Chiarenza for his advice on current European initiatives. Finally, we would like to thank the peer reviewers for their insights and constructive feedback.

Data and analyses

Download statistical data

Comparison 1. Cultural competence training vs no training
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Treatment outcomes (dichotomous)12699Risk Difference (Random, 95% CI)-0.02 [-0.06, 0.02]
2 Treatment outcomes (continuous)168Std. Mean Difference (IV, Random, 95% CI)0.07 [-0.41, 0.55]
3 Health behaviour (dichotomous)128Risk Ratio (M-H, Random, 95% CI)1.53 [1.03, 2.27]
4 Involvement in care1 Std. Mean Difference (Random, 95% CI)0.21 [0.00, 0.42]
5 Evaluations of care (dichotomous)1 Risk Difference (Fixed, 95% CI)0.14 [-0.03, 0.31]
6 Evaluations of care (continuous)2 Std. Mean Difference (IV, Random, 95% CI)Subtotals only
7 Knowledge & understanding (dichotomous)187Risk Ratio (M-H, Random, 95% CI)1.37 [0.97, 1.94]
Analysis 1.1.

Comparison 1 Cultural competence training vs no training, Outcome 1 Treatment outcomes (dichotomous).

Analysis 1.2.

Comparison 1 Cultural competence training vs no training, Outcome 2 Treatment outcomes (continuous).

Analysis 1.3.

Comparison 1 Cultural competence training vs no training, Outcome 3 Health behaviour (dichotomous).

Analysis 1.4.

Comparison 1 Cultural competence training vs no training, Outcome 4 Involvement in care.

Analysis 1.5.

Comparison 1 Cultural competence training vs no training, Outcome 5 Evaluations of care (dichotomous).

Analysis 1.6.

Comparison 1 Cultural competence training vs no training, Outcome 6 Evaluations of care (continuous).

Analysis 1.7.

Comparison 1 Cultural competence training vs no training, Outcome 7 Knowledge & understanding (dichotomous).

Appendices

Appendix 1. MEDLINE (OvidSP) search strategy

1. culture/

2. cultural competency/

3. anthropology cultural/

4. cultural characteristics/

5. cultural diversity/

6. cross-cultural comparison/

7. multilingualism/

8. language/

9. ((cultur* or linguistic* or language*) adj3 (competenc* or understanding or knowledg* or expertise or skill* or sensitiv* or aware* or appropriate* or acceptab* or safe* or humility or service* or communicat* or barrier* or divers* or comparison* or identity or specific or background* or value* or belief*)).tw.

10. transcultural nursing/

11. (intercultural* or inter-cultural or transcultural* or trans-cultural or cross-cultural or crosscultural or multicultural* or multi-cultural* or multiethnic or bicultural or bi-cultural or multilingual* or multi-lingual* or bilingual or bi-lingual).tw.

12. "Emigration and Immigration"/

13. "Emigrants and Immigrants"/

14. "Transients and Migrants"/

15. refugees/

16. exp population groups/

17. minority groups/

18. minority health/

19. (immigrant* or migrant* or refugee* or ethnic* or racial or indigenous or aborigin* or second language* or ((language or english) and proficien*) or interpreter*).tw.

20. (non-english or hispanic* or latino* or ((african or black or asian or native or mexican) adj american*) or inuit* or maori or islander*).tw.

21. or/1-20

22. exp health personnel/

23. attitude of health personnel/

24. ((health* or medical or nurs* or hospital) adj2 (personnel or provider* or professional* or practitioner* or worker* or staff or specialist*)).tw.

25. (doctor* or physician* or general practitioner* or gp or gps or nurse* or clinician* or dentist* or an?esthetist* or midwi* or hospitalist* or surgeon* or obstetrician* or gynecologist* or geriatrician* or gerontologist* or pediatrician* or therapist* or physiotherapist* or psychologist* or psychiatrist* or counsel?or* or social worker* or welfare worker* or medical interpreter*).tw.

26. "internship and residency"/

27. preceptorship/

28. (intern? or internship or resident? or residency or preceptor? or preceptorship).tw.

29. or/22-28

30. education/

31. exp teaching/

32. ((computer or web or internet) adj (based or assisted or mediated)).tw.

33. curriculum/

34. curriculum.tw.

35. competency-based education/

36. ((education* or teaching or learning or elearning or instruction* or training or skills or didactic or pedagogic* or online or on-line or web* or internet or cd-rom* or dvd or multimedia or multi-media or computer*) adj2 (intervention* or session* or course* or program* or activit* or presentation* or round* or material* or package* or module* or demonstration* or method* or process*)).tw.

37. exp teaching materials/

38. video recording/

39. (audiovisual or audio or video*).tw.

40. exp education continuing/

41. exp education graduate/

42. education professional/

43. exp inservice training/

44. (inservice or in service or workshop* or (discussion adj1 group*) or lectur* or seminar* or (short adj2 course*) or role play* or immersion or mentor* or lifelong learning or life long learning).tw.

45. (graduate or postgraduate or doctoral).tw.

46. ((staff or professional or workforce or work force) adj (development or training)).tw.

47. exp health personnel/ed

48. teaching rounds/

49. ((medical or continuing or residency or distance) adj2 education).tw.

50. international educational exchange/

51. transcultural nursing/ed

52. cultural competency/ed

53. ((cultural* or transcultural* or multicultural* or intercultural* or bicultural*) adj2 (education or train* or teach* or learn* or instruct* or coach* or skills or content*)).tw.

54. or/30-53

55. 21 and 29 and 54

56. randomized controlled trial.pt.

57. controlled clinical trial.pt.

58. randomized.ab.

59. placebo.ab.

60. drug therapy.fs.

61. randomly.ab.

62. trial.ab.

63. groups.ab.

64. or/56-63

65. 55 and 64

 

Appendix 2. CENTRAL search strategy

#1           MeSH descriptor: [Culture] this term only           

#2           ((cultur* near/1 anthropolog*) or *cultural-factors):kw                

#3           ((cultur* or linguistic* or language*) near/3 (competenc* or understanding or knowledg* or expertise or skill* or ability or proficien* or sensitiv* or aware* or respons* or appropriate* or acceptab* or safe* or humility or service or communicat* or barrier or bias* or factor or characteristic or divers* or differen* or comparison or identity or specific or tailor* or background or value or belief)):ti,ab,kw           

#4           MeSH descriptor: [Language] this term only       

#5           (language near/2 translat*):ti,ab,kw      

#6           (intercultural* or inter-cultural* or transcultural* or trans-cultural* or cross-cultural* or crosscultural* or multicultural* or multi-cultural* or multiethnic or bicultural* or bi-cultural* or multilingual* or multi-lingual* or bilingual* or bi-lingual*):ti,ab,kw       

#7           MeSH descriptor: [Population Groups] explode all trees               

#8           minority:ti,kw   

#9           (*migration or *migrant* or refugee* or ethnic* or racial or indigenous or aborigin* or second-language or ((language or english) and proficien*) or interpreter*):ti,ab,kw                

#10         (non-english or hispanic* or latino* or african or asian or indian or chinese or ((black or native or mexican) next american*) or eskimo or inuit* or maori or islander):ti,ab,kw     

#11         ((language or cultural or religious) next (group or minority)):ti,ab,kw      

#12         #1 or #2 or #3 or #4 or #5 or #6 or #7 or #8 or #9 or #10 or #11    

#13         MeSH descriptor: [Health Personnel] explode all trees  

#14         ((health* or *medical or nurs* or hospital or operating-room or psychiatric or pharmac*) near/2 (personnel or provider or professional or practitioner or worker or aide or assistant or staff or officer or specialist or consultant)):ti,ab,kw                

#15         (doctor or physician or general-practitioner or gp or gps or nurse or clinician or dentist or pharmacist or an*esthetist* or midwi* or hospitalist or surgeon or obstetrician or gyn*ecologist* or geriatrician or gerontologist or pediatrician or radiologist or *therapist or audiologist or dietitian or psychologist or psychiatrist or counsel*or or social-worker or welfare-worker or medical-interpreter):ti,ab,kw        

#16         (intern or internship or resident or residency or preceptor or preceptorship):ti,ab,kw    

#17         #13 or #14 or #15 or #16               

#18         #12 and #17       

#19         (educat* or teach* or train* or coach* or learn* or instruct* or competenc* or understanding or knowledg* or expertise or skill* or ability):kw,ti            

#20         MeSH descriptor: [Teaching] explode all trees   

#21         curriculum:ti,ab,kw        

#22         ((computer or web or internet) next (based or assisted or mediated)):ti,ab,kw

#23         ((education* or teaching or learning or elearning or instruction* or training or coaching or skills or didactic or pedagogic* or online or on-line or web* or internet or computer* or video* or audio* or cd-rom* or dvd or multimedia or multi-media) near/2 (intervention or session or course or program or activity or presentation or round or material or package or module or demonstration or method or process)):ti,ab,kw  

#24         MeSH descriptor: [Teaching Materials] explode all trees               

#25         (online or on-line or web* or internet or computer* or e-learning or elearning or audio* or video* or cd-rom* or dvd or multimedia or multi-media or tape or recording):kw,ti     

#26         (inservice or in-service or workshop or (discussion near/1 group) or lectur* or seminar or (short near/2 course) or role-play* or immersion or mentor* or lifelong-learning or life-long-learning):ti,ab,kw  

#27         (graduate or postgraduate or doctoral or masters):ti,ab,kw        

#28         ((staff or professional or workforce or work-force or personnel) next (development or training)):ti,ab,kw           

#29         ((health*-professional or health-care-professional or *medical or continuing or residency or distance) near/2 (educat* or train*)):ti,ab,kw     

#30         international-educational-exchange:ti,ab,kw     

#31         ((*cultural* or *lingual* or *ethnic*) near/2 (educat* or train* or teach* or learn* or instruct* or coach* or skill* or competen* or content* or intervention)):ti,ab,kw          

#32         #19 or #20 or #21 or #22 or #23 or #24 or #25 or #26 or #27 or #28 or #29 or #30 or #31    

#33         #18 and #32 in Trials       

Appendix 3. EMBASE search strategy

1. cultural competence/

2. cultural anthropology/

3. exp "ethnic or racial aspects"/

4. cultural sensitivity/

5. cultural safety/

6. cultural value/

7. cultural bias/

8. language/

9. language ability/

10. english as a second language/

11. interpreter service/

12. ((cultur* or linguistic* or language*) adj3 (competenc* or understanding or knowledg* or expertise or skill* or abilit* or proficien* or sensitiv* or aware* or respons* or appropriate* or acceptab* or safe* or humility or service* or communicat* or barrier* or bias* or factor* or characteristic* or divers* or differen* or comparison* or identity or specific or tailor* or background* or value* or belief*)).ti,ab,kw.

13. transcultural care/

14. (intercultural* or inter-cultural* or transcultural* or trans-cultural* or cross-cultural* or crosscultural* or multicultural* or multi-cultural* or multiethnic or bicultural* or bi-cultural* or multilingual* or multi-lingual* or bilingual* or bi-lingual*).ti,ab,kw.

15. exp migration/

16. immigrant/

17. refugee/

18. exp "ethnic racial and religious groups"/

19. ((language or cultural) adj (group* or minorit*)).ti,ab,kw.

20. minority health/

21. (immigrant* or migrant* or refugee* or ethnic* or racial or indigenous or aborigin* or second language* or ((language or english) and proficien*) or interpreter*).ti,ab,kw.

22. (non-english or hispanic* or latino* or ((african or black or asian or native or mexican) adj american*) or eskimo or inuit* or maori or islander*).ti,ab,kw.

23. or/1-22

24. exp healthcare personnel/

25. exp health personnel attitude/

26. ((health* or medical or paramedical or nurs* or hospital or operating-room or psychiatric or pharmac*) adj2 (personnel or provider* or professional* or practitioner* or worker* or aide* or assistant* or staff or officer* or specialist* or consultant*)).ti,ab,kw.

27. (doctor* or physician* or general practitioner* or gp or gps or nurse* or clinician* or dentist* or pharmacist* or an?esthetist* or midwi* or hospitalist* or surgeon* or obstetrician* or gyn?ecologist* or geriatrician* or gerontologist* or pediatrician* or radiologist* or therapist* or physiotherapist* or audiologist* or dietitian* or psychologist* or psychiatrist* or psychotherapist* or counselor* or counsellor* or social worker* or welfare worker* or medical interpreter*).ti,ab,kw.

28. (intern? or internship or resident? or residency or preceptor? or preceptorship).ti,ab,kw.

29. or/24-28

30. 23 and 29

31. education/

32. teaching/

33. ((computer or web or internet) adj (based or assisted or mediated)).ti,ab,kw.

34. curriculum/

35. curriculum.ti,ab,kw.

36. education program/

37. ((education* or teaching or learning or elearning or instruction* or training or skills or didactic or pedagogic* or online or on-line or web* or internet or computer* or audio* or video* or cd-rom* or dvd or multimedia or multi-media) adj2 (intervention* or session* or course* or program* or activit* or presentation* or round* or material* or package* or module* or demonstration* or method* or process*)).ti,ab,kw.

38. exp audiovisual equipment/

39. videorecording/

40. (audiovisual or audio or video*).ti,ab,kw.

41. continuing education/

42. postgraduate education/

43. masters education/

44. doctoral education/

45. training/

46. staff training/

47. in service training/

48. lifelong learning/

49. (inservice or in service or workshop* or (discussion adj1 group*) or lectur* or seminar* or (short adj2 course*) or role play* or immersion or mentor* or lifelong learning or life long learning).ti,ab,kw.

50. (graduate or postgraduate or doctoral or masters).ti,ab,kw.

51. ((staff or professional or workforce or work force or personnel) adj (development or training)).ti,ab,kw.

52. medical education/

53. residency education/

54. exp paramedical education/

55. social work education/

56. ((health* professional* or healthcare professional*) adj2 (educat* or train*)).ti,ab,kw.

57. teaching round/

58. ((medical or continuing or residency or distance) adj2 education).ti,ab,kw.

59. ((cultural* or intercultural* or inter-cultural* or transcultural* or trans-cultural* or cross-cultural* or crosscultural* or multicultural* or multi-cultural* or multiethnic or bicultural* or bi-cultural* or multilingual* or multi-lingual* or bilingual* or bi-lingual*) adj2 (educat* or train* or teach* or learn* or instruct* or coach* or skill* or competen* or content*)).ti,ab,kw.

60. or/31-59

61. 30 and 60

62. randomized controlled trial/

63. controlled clinical trial/

64. single blind procedure/ or double blind procedure/

65. crossover procedure/

66. random*.tw.

67. placebo*.tw.

68. ((singl* or doubl*) adj (blind* or mask*)).tw.

69. (crossover or cross over or factorial* or latin square).tw.

70. (assign* or allocat* or volunteer*).tw.

71. or/62-70

72. 61 and 71

Appendix 4. CINAHL search strategy

S46s45Limiters - Exclude MEDLINE records
Search modes - Boolean/Phrase
S45s29 and s44Search modes - Boolean/Phrase
S44S30 or S31 or S32 or S33 or S34 or S35 or S36 or S37 or S38 or S39 or S40 or S41 or S42 or S43Search modes - Boolean/Phrase
S43AB time series or TI time seriesSearch modes - Boolean/Phrase
S42AB (pre test or pretest or post test or posttest or preintervention or postintervention) or TI (pre test or pretest or post test or posttest or preintervention or postintervention)Search modes - Boolean/Phrase
S41TI (singl* or doubl* or tripl* or trebl*) and TI (blind* or mask*)Search modes - Boolean/Phrase
S40AB (singl* or doubl* or tripl* or trebl*) and AB (blind* or mask*)Search modes - Boolean/Phrase
S39AB (random* or trial or groups or placebo* or assign* or allocat* or volunteer* or factorial* or experiment* or control* or compar* or intervention* or chang* or evaluat* or impact* or effect?) or TI (random* or trial or groups or placebo* or assign* or allocat* or volunteer* or factorial* or experiment* or control* or compar* or intervention* or chang* or evaluat* or impact* or effect?)Search modes - Boolean/Phrase
S38PT Clinical TrialSearch modes - Boolean/Phrase
S37MH Quasi-Experimental Studies+Search modes - Boolean/Phrase
S36MH Quantitative StudiesSearch modes - Boolean/Phrase
S35MH PlacebosSearch modes - Boolean/Phrase
S34MH Crossover DesignSearch modes - Boolean/Phrase
S33MH Comparative StudiesSearch modes - Boolean/Phrase
S32MH Random AssignmentSearch modes - Boolean/Phrase
S31MH Experimental Studies+Search modes - Boolean/Phrase
S30randomi?ed controlled trial*Search modes - Boolean/Phrase
S29s13 and s28Search modes - Boolean/Phrase
S28s14 or s15 or s16 or s17 or s18 or s19 or s20 or s21 or s22 or s23 or s24 or s25 or s26 or s27Search modes - Boolean/Phrase
S27audiovisual or video* or multimedia or multi-media or computer based or web based or internet based or computer assisted or internet assisted or web assisted or onlineSearch modes - Boolean/Phrase
S26(cultural* competen* N2 educat*) or (cultural* competen* N2 train*) or (cultural* competen* N2 skill*) or (cultural* competen* N2 program*) or (cultural* competen* N2 course*) or (cultural* competen* N2 content*) or (cultural* competen* N2 instruct*) or (cultural* competen* N2 teach*) or (cultural* competen* N2 learn*)Search modes - Boolean/Phrase
S25MH cultural competence/edSearch modes - Boolean/Phrase
S24MH transcultural nursing/edSearch modes - Boolean/Phrase
S23MH transcultural care/edSearch modes - Boolean/Phrase
S22MH health personnel+/edSearch modes - Boolean/Phrase
S21(communication* skill* N2 educat*) or (communication* skill* N2 train*) or (social skill* N2 educat*) or (social skill* N2 train*)Search modes - Boolean/Phrase
S20education N1 competency-basedSearch modes - Boolean/Phrase
S19(health* professional* N2 train*) or (healthcare professional* N2 train*) or (health* professional* N2 educat*) or (healthcare professional* N2 educat*)Search modes - Boolean/Phrase
S18staff development or (staff N2 train*) or professional development or (workforce N2 train*) or (work force N2 train*) or (workforce N2 develop*) or (work force N2 develop*) or inservice or workshop* or (discussion N1 group*) or lectur* or seminar* or (short N2 course*) or role play* or immersion or mentor* or lifelong learning or life long learningSearch modes - Boolean/Phrase
S17graduate or postgraduate or doctoral or mastersSearch modes - Boolean/Phrase
S16MH students, graduate+Search modes - Boolean/Phrase
S15MH education, continuing+Search modes - Boolean/Phrase
S14MH education, graduate+Search modes - Boolean/Phrase
S13s8 and s12Search modes - Boolean/Phrase
S12s9 or s10 or s11Search modes - Boolean/Phrase
S11intern or interns or internship or resident or residents or residency or preceptor*Search modes - Boolean/Phrase
S10MH health manpower+Search modes - Boolean/Phrase
S9MH health personnel+Search modes - Boolean/Phrase
S8s1 or s2 or s3 or s4 or s5 or s6 or s7Search modes - Boolean/Phrase
S7international nursingSearch modes - Boolean/Phrase
S6non-english or hispanic* or latino* or african american* or black american* or asian american* or native american* or mexican american* or eskimo or inuit* or maori or islander*Search modes - Boolean/Phrase
S5immigra* or migrant* or refugee* or ethnic* or racial or indigenous or aborigin* or second language* or ((language or english) and proficien*) or interpreter*Search modes - Boolean/Phrase
S4minority group* or minority health* or language group* or religious group* or language minorit* or religious minorit* or religious communit*Search modes - Boolean/Phrase
S3(linguistic* N3 service*) or language service* or linguistic* proficien* or (language* N3 proficien*) or (linguistic* N3 expertise) or (language* N3 expertise) or (linguistic* N3 skill*) or (language* N3 skill*) or linguistic barrier* or language barrier*Search modes - Boolean/Phrase
S2cultural* or intercultural* or transcultural* or crosscultural* or multicultural* or multiethnic or bicultural* or multilingual* or multi-lingual* or bilingual* or bi-lingual*Search modes - Boolean/Phrase
S1MH culture+Search modes - Boolean/Phrase

Appendix 5. PsycINFO search strategy

1. exp sociocultural factors/

2. "culture (anthropological)"/

3. "racial and ethnic differences"/

4. "racial and ethnic relations"/

5. traditions/

6. language/

7. foreign language translation/

8. ((cultur* or linguistic* or language*) adj3 (competenc* or understanding or knowledg* or expertise or skill* or abilit* or proficien* or sensitiv* or aware* or respons* or appropriate* or acceptab* or safe* or humility or service* or communicat* or barrier* or bias* or factor* or characteristic* or divers* or differen* or comparison* or identity or specific or tailor* or background* or value* or belief*)).ti,ab,hw,id.

9. (intercultural* or inter-cultural* or transcultural* or trans-cultural* or cross-cultural* or crosscultural* or multicultural* or multi-cultural* or multiethnic or bicultural* or bi-cultural* or multilingual* or multi-lingual* or bilingual* or bi-lingual*).ti,ab,hw,id.

10. exp human migration/

11. immigration/

12. exp "racial and ethnic groups"/

13. minority groups/

14. exp religious groups/ or exp religious affiliation/

15. ((language or cultural or religious) adj (group* or minorit*)).ti,ab,hw,id.

16. (immigrant* or migrant* or refugee* or ethnic* or racial or indigenous or aborigin* or second language* or ((language or english) and proficien*) or interpreter*).ti,ab,hw,id.

17. (non-english or hispanic* or latino* or ((african or black or asian or native or mexican) adj american*) or eskimo or inuit* or maori or islander*).ti,ab,hw,id.

18. or/1-17

19. exp health personnel/

20. exp health personnel attitudes/

21. ((health* or medical or paramedical or nurs* or hospital or operating-room or psychiatric or pharmac*) adj2 (personnel or provider* or professional* or practitioner* or worker* or aide* or assistant* or staff or officer* or specialist* or consultant*)).ti,ab,hw,id.

22. (doctor* or physician* or general practitioner* or gp or gps or nurse* or clinician* or dentist* or pharmacist* or an?esthetist* or midwi* or hospitalist* or surgeon* or obstetrician* or gyn?ecologist* or geriatrician* or gerontologist* or pediatrician* or radiologist* or therapist* or physiotherapist* or audiologist* or dietitian* or psychologist* or psychiatrist* or psychotherapist* or counselor* or counsellor* or social worker* or welfare worker* or medical interpreter*).ti,ab,hw,id.

23. (intern? or internship or resident? or residency or preceptor? or preceptorship).ti,ab,hw,id.

24. or/19-23

25. 18 and 24

26. education/

27. teaching/

28. training/

29. ((computer or web or internet) adj (based or assisted or mediated)).ti,ab,hw,id.

30. curriculum.ti,ab,hw,id.

31. ((education* or teaching or learning or elearning or instruction* or training or skills or didactic or pedagogic* or online or on-line or web* or internet or computer* or audio* or video* or cd-rom* or dvd or multimedia or multi-media) adj2 (intervention* or session* or course* or program* or activit* or presentation* or round* or material* or package* or module* or demonstration* or method* or process*)).ti,ab,hw,id.

32. exp multimedia/

33. instructional media.ti,ab,hw,id.

34. exp audiovisual instruction/

35. (audiovisual or audio or video*).ti,ab,hw,id.

36. exp continuing education/

37. exp graduate education/

38. exp postgraduate training/

39. exp clinical methods training/

40. (inservice or in service or workshop* or (discussion adj1 group*) or lectur* or seminar* or (short adj2 course*) or role play* or immersion or mentor* or lifelong learning or life long learning).ti,ab,hw,id.

41. (graduate or postgraduate or doctoral or masters).ti,ab,hw,id.

42. ((staff or professional or workforce or work force or personnel) adj (development or training)).ti,ab,hw,id.

43. nursing education/

44. counselor education/

45. social work education/

46. exp internship programs/

47. ((medical or continuing or residency or distance) adj2 education).ti,ab,hw,id.

48. ((health* professional* or healthcare professional*) adj2 (educat* or train*)).ti,ab,hw,id.

49. communication skills training/

50. human relations training/

51. ((cultural* or intercultural* or inter-cultural* or transcultural* or trans-cultural* or cross-cultural* or crosscultural* or multicultural* or multi-cultural* or multiethnic or bicultural* or bi-cultural* or multilingual* or multi-lingual* or bilingual* or bi-lingual*) adj2 (educat* or train* or teach* or learn* or instruct* or coach* or skill* or competen* or content*)).ti,ab,hw,id.

52. or/26-51

53. 25 and 52

54. random*.ti,ab,hw,id.

55. trial*.ti,ab,hw,id.

56. control*.ti,ab,hw,id.

57. placebo*.ti,ab,hw,id.

58. ((singl* or doubl* or trebl* or tripl*) and (blind* or mask*)).ti,ab,hw,id.

59. (cross over or crossover or factorial* or latin square).ti,ab,hw,id.

60. (assign* or allocat* or volunteer*).ti,ab,hw,id.

61. treatment effectiveness evaluation/

62. mental health program evaluation/

63. exp experimental design/

64. "2000".md.

65. or/54-64

66. 53 and 65

Appendix 6. ERIC search strategy

((kw=((cultur* or linguistic* or language*) within 3 (competenc* or understanding or knowledg* or expertise or skill* or proficien* or sensitiv* or aware* or respons* or appropriate* or acceptab* or safe* or humility or service* or communicat* or barrier* or bias* or factor* or characteristic* or divers* or differen* or comparison* or identity or specific or tailor* or background* or value* or belief*))) or (kw=(intercultural* or inter-cultural* or transcultural* or trans-cultural* or cross-cultural* or crosscultural* or multicultural* or multi-cultural* or multiethnic or bicultural* or bi-cultural* or multilingual* or multi-lingual* or bilingual* or bi-lingual*)) or (kw=(emigration or immigration or immigrant* or migrant* or refugee* or ethnic* or racial or indigenous or aborigin* or second language* or ((language or english) and proficien*) or interpreter* or non-english or hispanic* or latino* or ((african or black or asian or native or mexican) within 1 american*) or eskimo or inuit* or maori or islander*)) or (DE=("ethnic groups" or "alaska natives" or "american indians" or "nonreservation american indians" or "urban american indians" or "maya people" or "reservation american indians" or "navajo nation" or "anglo americans" or "canada natives" or "chinese americans" or "eskimos" or "samoan americans" or "filipino americans" or "italian americans" or "japanese americans" or "french canadians" or "spanish americans" or "korean americans" or "pacific islanders" or "pacific americans" or "hawaiians" or "polish americans" or "mexican americans" or "hmong people")) or (DE=("language minorities" or "limited english speaking" or "non english speaking"))) and ((DE=("health personnel" or "allied health personnel" or "emergency medical technicians" or "home health aides" or "mental health workers" or "school psychologists" or "nurses" or "school nurses" or "physicians" or "foreign medical graduates" or "psychologists")) or (kw=((health* or medical or paramedical or nurs* or hospital or operating room or psychiatric or pharmac*) within 2 (personnel or provider* or professional* or practitioner* or worker* or aide* or assistant* or staff or officer* or specialist* or consultant*))) or (kw=(doctor* or physician* or general practitioner* or gp or gps or nurse* or clinician* or dentist* or pharmacist* or an*esthetist* or midwi* or hospitalist* or surgeon* or obstetrician* or gyn*ecologist* or geriatrician* or gerontologist* or pediatrician* or radiologist* or therapist* or physiotherapist* or audiologist* or dietitian* or psychologist* or psychiatrist* or psychotherapist* or counselor* or counsellor* or social worker* or welfare worker* or medical interpreter*)) or (kw=(intern or interns or internship or resident or residents or residency or preceptor*))) and (kw=(random* or trial* or placebo* or assign* or allocat* or volunteer* or ((singl* or doubl* or tripl* or trebl*) and (blind* or mask*)) or crossover or cross over or factorial* or latin square))

Appendix 7. LILACS search strategy

(mh:I01.880.143$ or transcultural or cultural* or culturais or multicultur* or mh:language or multilingu* or bilingu* or interpreter* or migra* or emigrant* or immigrant* or imigrant* or inmigrant* or mh:M01.686$ or refug* or minori* or ethnic* or etnic* or etnia* or ethno* or etno* or indigen* or nativ* or indian* or indio* or amerindio*) and (mh:M01.526.485$ or personnel or pessoal or personal or ("recursos humanos" and (salud or saude)) or "corpo clinico" or staff or provider* or professional* or profesional* or profissional* or practitioner* or worker* or trabajador* or paramedic* or doctor* or physician* or medico* or "clinicos gerais" or nurse* or enfermero* or enfermera* or enfermeir* or clinician* or dentist* or odontologo* or pharmacist* or farmaceutico* or midwi* or surgeon* or cirujano* or cirurgio* or therapist* or physiotherapist* or fisioterapeuta* or psychologist* or psicologo* or psychiatrist* or psychotherapist* or counselor* or counsellor* or "agentes comunitarios" or intern or interns* or internado or internato or residen* or preceptor* or assistant* or asistent* or assistant* or auxiliar* or aide or aides) and (mh:I02.358$ or educa* or ensin* or ensen* or formac* or curricul* or course* or curso* or program* or instruc* or train* or capacitac* or competenc* or abilit* or aptitud* or aptidao* or inservice or workshop* or lectur* or clases or aulas or seminar* or tutor* or postgrad* or "pos-graduacao" or mentor*) and (random* or aleat* or azar or trial* or ensaio* or ensayo* or ((study or estud*) and control*) or multicenter or "double-blind" or "doble ciego" or "duplo cego" or "single-blind" or crossover or "cross over" or cruzado or assign* or asign* or allocat* or aloca* or pt:("randomized controlled trial" or "controlled clinical trial" or "multicenter study"))

Appendix 8. Current Contents search strategy

1. cultural anthropology.ti,ab,kw.

2. ((cultur* or linguistic* or language*) adj3 (competenc* or understanding or knowledg* or expertise or skill* or abilit* or proficien* or sensitiv* or aware* or respons* or appropriate* or acceptab* or safe* or humility or service* or communicat* or barrier* or bias* or factor* or characteristic* or divers* or differen* or comparison* or identity or specific or tailor* or background* or value* or belief*)).ti,ab,kw.

3. (intercultural* or inter-cultural* or transcultural* or trans-cultural* or cross-cultural* or crosscultural* or multicultural* or multi-cultural* or multiethnic or bicultural* or bi-cultural* or multilingual* or multi-lingual* or bilingual* or bi-lingual*).ti,ab,kw.

4. (minority adj (group* or health*)).ti,ab,kw.

5. ((language or cultural or religious) adj (group* or minorit*)).ti,ab,kw.

6. (migration or immigration or immigrant* or migrant* or refugee* or ethnic* or racial or indigenous or aborigin* or second language* or ((language or english) and proficien*) or interpreter*).ti,ab,kw.

7. (non-english or hispanic* or latino* or african* or asian* or indian or indians or chinese or ((black or native or mexican) adj american*) or eskimo or inuit* or maori or islander*).ti,ab,kw.

8. or/1-7

9. ((health* or medical or paramedical or nurs* or hospital or operating-room or psychiatric or pharmac*) adj2 (personnel or provider* or professional* or practitioner* or worker* or aide* or assistant* or staff or officer* or specialist* or consultant*)).ti,ab,kw.

10. (doctor* or physician* or general practitioner* or gp or gps or nurse* or clinician* or dentist* or pharmacist* or an?esthetist* or midwi* or hospitalist* or surgeon* or obstetrician* or gyn?ecologist* or geriatrician* or gerontologist* or pediatrician* or radiologist* or therapist* or physiotherapist* or audiologist* or dietitian* or psychologist* or psychiatrist* or psychotherapist* or counselor* or counsellor* or social worker* or welfare worker* or medical interpreter*).ti,ab,kw.

11. (intern? or internship or resident? or residency or preceptor? or preceptorship).ti,ab,kw.

12. or/9-11

13. 8 and 12

14. (educat* or teach* or train* or coach* or learn* or instruct* or competenc* or understanding or knowledg* or expertise or skill* or abilit*).kw,ti.

15. ((computer or web or internet) adj (based or assisted or mediated)).ti,ab,kw.

16. curriculum.ti,ab,kw.

17. ((education* or teaching or learning or elearning or instruction* or training or skills or didactic or pedagogic* or online or on-line or web* or internet or computer* or audio* or video* or cd-rom* or dvd or multimedia or multi-media) adj2 (intervention* or session* or course* or program* or activit* or presentation* or round* or material* or package* or module* or demonstration* or method* or process*)).ti,ab,kw.

18. (audiovisual or audio or video*).ti,ab,kw.

19. (inservice or in service or workshop* or (discussion adj1 group*) or lectur* or seminar* or (short adj2 course*) or role play* or immersion or mentor* or lifelong learning or life long learning).ti,ab,kw.

20. (graduate or postgraduate or doctoral or masters).ti,ab,kw.

21. ((staff or professional or workforce or work force or personnel) adj (development or training)).ti,ab,kw.

22. ((health* professional* or healthcare professional*) adj2 (educat* or train*)).ti,ab,kw.

23. ((medical or continuing or residency or distance) adj2 education).ti,ab,kw.

24. ((cultural* or intercultural* or inter-cultural* or transcultural* or trans-cultural* or cross-cultural* or crosscultural* or multicultural* or multi-cultural* or multiethnic or bicultural* or bi-cultural* or multilingual* or multi-lingual* or bilingual* or bi-lingual*) adj2 (educat* or train* or teach* or learn* or instruct* or coach* or skill* or competen* or content*)).ti,ab,kw.

25. or/14-24

26. 13 and 25

27. (random* or trial* or placebo* or assign* or allocat* or volunteer* or ((singl* or doubl* or tripl* or trebl*) and (blind* or mask*)) or crossover or cross over or factorial* or latin square).mp.

28. 26 and 27

Appendix 9. Narrative summary of interventions across domains of conceptual framework

The interventions assessed in each study showed considerable heterogeneity in terms of their stated purpose, their content, duration and format or nature. The focus of the intervention used in Harmsen 2005 was intercultural communication, whereas Majumdar 2004 and Wade 1991 examined the effects of cultural sensitivity training, and Sequist 2010 and Thom 2006 looked at cultural competency training with performance feedback to health professionals.

All aspects of the conceptual framework developed for this review were found in some way in the descriptions of the five educational interventions (4 domains and their 11 sub-domains) included in the review, although no study covered all aspects of the framework, or in the same way (see Table 1). Three categories of reporting against the framework were evident. Thom 2006 and Sequist 2010 were 'high reporting' with at least two-thirds of the domains and sub-domains described. Harmsen 2005 and Wade 1991 were 'medium reporting' with about half of all domains and sub-domains described. Majumdar 2004 was 'low reporting' and described less than a third of all domains and sub-domains described (see Table 4).

The educational content domain is comprised of three key components: types of knowledge; assessment and application; and the skills needed for culturally competent practice by a health professional.

In terms of types of knowledge, all studies used terminology and concepts such as cultural competence (Sequist 2010; Thom 2006), cultural sensitivity (Majumdar 2004; Wade 1991) or intercultural communication (Harmsen 2005) but no consistent concept was used across the studies, nor did any study provide an explicit definition of the concept or terminology utilised. Four of the five studies reported use or inclusion of specific theoretical models as part of the education or training intervention, with three studies reporting use of specific theoretical models of health and illness (Harmsen 2005; Sequist 2010; Thom 2006) but not in a common manner. Three studies reported attention to the socio-cultural context of health disparities as part of their education intervention (Harmsen 2005; Sequist 2010; Wade 1991). Two studies did not report inclusion of this. Constructs of racism and prejudice were incorporated into the training curriculum in Sequist 2010 and Wade 1991, and epidemiology and the social determinants of health were explicitly referenced by Sequist 2010 and Thom 2006. Two studies included topics additional to the components of this domain: Thom 2006 reported on the cultural gap between healthcare professionals and patients knowledge and belief systems, and Majumdar 2004 reported on cultural awareness.

Three of the five studies reported on the assessment and application component. In Harmsen 2005 GPs used self chosen training styles to resolve gaps in views and culturally defined communication styles. Sequist 2010 reported that the curriculum reviewed ways to incorporate relevant information into effective clinical care processes and provided monthly written education materials and physician-level performance reports on differences in achievement of clinical goals for black and white patients. One of the three modules used in Thom 2006 focused on: understanding the importance of working with trained interpreters and how to use them effectively; negotiating treatment plans patients and families; the role of cultural broker that connecting patients to their community and resources.

In terms of the skills component, three of the five studies reported on cultural self-assessment processes (Harmsen 2005; Majumdar 2004; Wade 1991) and four studies reported the inclusion of communication skills training to improve cross cultural communication (Harmsen 2005; Sequist 2010; Thom 2006; Wade 1991). One of the three education modules in Thom 2006 was devoted to ‘enhancing communication skills for cultural competency'. Skills to deconstruct stereotypes were inferred in two studies (Thom 2006; Wade 1991) but were unclear in the others. Issues of trust and partnership establishment between patients and healthcare professionals were raised by Sequist 2010, Thom 2006 and Wade 1991.

Under the pedagogical approach domain three studies reported the theoretical models that underpinned the training (Harmsen 2005; Thom 2006; Wade 1991) but the actual content or principles were not reported in detail, nor did studies describe how these models influenced instructional practices. Two studies identified explicit teaching and learning methods such as ‘adult learning’ (Harmsen 2005; Thom 2006). Of note was the understanding given to the term pedagogy. In response to requests for additional information authors responded to the term pedagogy as synonymous with teaching and learning method. In our conceptual framework we use the term pedagogy as defined by (Alexander 2008) as inclusive of the practice of teaching as well as the educational values, philosophies, theories, research and evidence; curriculum content; learning technologies and tools; and social, cultural, linguistic, organisational and community contexts of education. All of these things inform, shape and explain the practice of teaching and affect the nature of learning experiences and the learning outcomes achieved (Abbey 2013).

The structure of the intervention domain is concerned with practical aspects of the intervention. At least four of the five included studies reported the education intervention was delivered face-to-face with a mix of didactic and experiential learning formats that included lectures, small group discussions; videos, role plays and simulation exercises (Harmsen 2005; Sequist 2010; Thom 2006; Wade 1991). There was considerable heterogeneity in the duration and frequency of the interventions. The intervention in Harmsen 2005 was directed at both GPs and their patients. GPs attended a 2.5 day training program with a follow-up session after 2 weeks while patients viewed a 12-minute video in the waiting room prior to their consultation. Majumdar 2004 reported that the intervention comprised 36 hours of training but did not report any further detail. In Sequist 2010 the off-site training program consisted of 2 consecutive days for nurse practitioners and physician assistants with physicians attending the second day only. Participants were also given monthly written education materials and monthly physician-level performance reports on differences in achievement of clinical goals for black and white patients. Thom 2006 offered a 3-module education intervention to physicians as either a single half-day training session or as 3 separate 1 to 1.5 hour sessions. Wade 1991 reported that the training for female counselors entailed a 4-hour cultural sensitivity workshop. The method of assessment used for participants to evaluate the intervention was unclear in three studies. Sequist 2010 used a survey tool previously used in a clinical trial setting to assess healthcare professionals' awareness of disparities, while Thom 2006 developed scale for participants to rate the usefulness of the training based on a 5-point Likert scale.

The organisational support domain, which is concerned with links between the intervention and organisational policies and procedures and whether it is embedded within a professional development program was the most under-reported aspect of the conceptual framework. Only one study (Sequist 2010) reported explicit organisational links including: provision of monthly performance feedback reports as part of a mentoring or support process; alignment of intervention with broader organizational goals; and allocation of continuing medical education points. All studies reported that the training was voluntary.

The participant domain, addressed the target of the intervention and the intervention facilitators. In one study (Harmsen 2005) both GPs and their patients received an intervention (training for GPs and video for patients). In four studies the intervention was directed at the healthcare professional with data collected from their patients or clients. Thom 2006 involved GPs and their patients (supplemented with medical record data), Majumdar 2004 targeted nursing and home care providers and their patients and Wade 1991 involved female counselors and their female clients. Sequist 2010 only included health professionals (primary care physician, nurse practitioners and physician assistants) in the intervention and collected patient data from medical records.

What's new

DateEventDescription
6 May 2014AmendedAdditional text in Implication for research section

Contributions of authors

All review authors contributed to this review. The idea for the review was initially conceived by PR and DH. LH and PR developed the scope and content for the review through their work and experience in the field of cultural competence, education and training, consumer participation and health policy development and evaluation. DH has contributed methodological expertise, knowledge and experience in conducting systematic reviews.

The design and coordination of the initial review proposal, conceptual framework and protocol development has been a collaborative process. LH led the coordination and writing of the protocol (Horvat 2011), and all authors provided significant input, comment and feedback. Specialist expertise in the methodological and technical components has been provided by DH. The guidance, development and implementation of the search strategy has been provided by JKR. The guarantor of the review is LH.

Declarations of interest

This review has been supported by the Department of Health Victoria. The lead author is the author of the Cultural Responsiveness Framework: Guidelines for Victorian Health Services and an employee of the Victorian government’s Department of Health, Consumer Partnerships and Quality Standards Unit (formerly Policy and Strategy Unit). This Unit is part of the Sector Performance, Quality, and Rural Health Branch, which funds the author’s time and work on the review, and also funds the editorial base of the Cochrane Consumers and Communication Review Group (CC&CRG).

This review was conducted at the editorial base of the CC&CRG and an editor of the group, Dr. Horey, is also a co-author of this review. Dr. Horey was not involved in the editorial process for this review.

Due to differences in terminology and their cultural meanings, and evolving concepts and practices across countries when addressing all populations as a whole in this review we have used the 'culturally and linguistically diverse' (CALD). This term is commonly used in Australia to encompass broad immigrant and refugee communities (but not Aboriginal or Indigenous communities) to indicate the broad group of populations characterised by ‘minority’ status, and racial and ethnic disparities in health care. In reporting data we have followed the suggested guidelines developed by Kaplan 2003, and cited the language of study authors ascribed to specific populations in their studies at the time they were conducted. These different usages illustrate the complexity and challenges posed by a lack of agreed terminology in this field.

Sources of support

Internal sources

  • Department of Health, Victoria, Australia.

    Salary support to LH, DH and JKR

External sources

  • No sources of support supplied

Differences between protocol and review

In line with the Methodological Expectations for Cochrane Intervention Reviews (MECIR) standards that were introduced by the Cochrane Collaboration in 2012, adverse events (e.g. complaints, reporting of adverse events), which were described as secondary outcomes in the protocol (Horvat 2011), were redefined as a primary outcome of the review. Reference to an additional Cochrane review (Dwamena 2012) was added to the Background, and updated information was added to the Description of the condition. The review makes explicit the implied exclusion criterion of studies not reporting patient-related outcomes (see Types of outcome measures and Characteristics of excluded studies).

Two authors (LH and DH) independently assigned the outcomes reported in each included study to the outcome categories determined for this review, with the intention of resolving any differences in categorisation, if they had occurred, by the involvement of a third author (see Table 2.)

Three other post-hoc decisions relating to the review outcomes were made. First, where there were outcomes reported for different time periods, we used time points furthest from baseline. Second, where a selected outcome was reported in terms of the dimensions of which it was comprised, we calculated the effect estimate for each dimension and chose the dimension whose effect estimate was ranked n/2. Third, where the study population was not stratified for participants of minority culturally and linguistically diverse (CALD) backgrounds, we used the whole population.

A number of post-hoc decisions were made related to the analyses. First we estimated the cluster sizes where they were not reported by dividing the number of study participants by the number of units of randomisation. This required two assumptions: first that cluster sizes were evenly distributed across study samples; second that CALD populations were evenly distributed across clusters in each study population. We also conducted post-hoc sensitivity analyses to assess the impact of using different ICCs and used the computed design effect (DEFF) to adjust for clustering in those studies where this had not been done. We used a range of possible ICCs (0.01, 0.02 and 0.05).

Notes

 

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Harmsen 2005

Methods

Study design: Cluster RCT; RCT randomisation at GP level with outcome measurements at patient level

Location: The Netherlands

Setting: GP Clinics

Study aims: 1) To decrease inequalities in care provided between "Western" and "non-Western" patients; 2) To assess the effectiveness of an educational intervention on intercultural communication aimed to decrease inequalities in care provided between "Western" and "non-Western" patients

Inclusion/exclusion criteria: GPs with a practice population of at least 25% of patients with a "non-Western" country of origin were invited to participate. Inclusion criteria for patients were a GP consultation on random days in February, May and November 2000. Adolescents aged 13 to 17 years excluded.

Informed consent obtained: Unclear

Ethical approval: Yes. University Ethical Commission of the Erasmus Medical Centre, University Medical Centre in Rotterdam (MEC163.267/1997/122)

Funding source and amount: Theia Foundation of Zilverenkruis Achmea (JvH/agbrf640), ZonMW: Netherlands Organisation for Health Research and Development. Fonds Aachterstandswijken, Districts Huisartsen Vereniging Rotterdam (Besluit FAW:98/09/H-O). Stichting Bevordering van Volkskrach (MK/avg/000–001). Amount not stated.

Statistical methods and their appropriateness: Multilevel multiple regression techniques adjusted for baseline values

Consumer involvement : None reported

Participants

Health professional participants

Type and number: 38 GPs (19 in each group)

Age: Not reported

Gender: 30 males and 8 females

Ethnicity: Process of determining GP ethnicity was not described. 2/38 GPs were reported to have "a non-Dutch (but Western) ethnic background" and "had lived and worked for more than 20 years in The Netherlands".

Professional qualifications: GPs - general physicians

Languages: Not reported

Previous cultural competence GP training: not reported

Method of recruitment: GPs were approached by letter followed by one repeat request by telephone

Patient participants

Type and number: 369 CALD patients described as "non-Western" patients. Total study sample = 986 patients (includes 614 Western patients)

Age: Adolescents aged 13 to 17 excluded. Parents of children up to 12 years interviewed. Non-Western patients included 7 aged 0 to 12 years (1.9%); 107 aged 18 to 29 years (29.2%); 177 aged 30 to 49 years (48.4%); 66 aged 50 to 65 years (18%); 9 aged over 65 years (2.5%)

Gender: Male = 142 (38.8%), Female = 224 (61.2%) (Western M = 223 [36.5%] and F = 388 [63.5%])

Ethnicity: Cultural background was assessed using a validated patient cultural background scale (Harmsen 2005) The "non-Western group" (comprising "mainly Turkish, Moroccan, Cape Verdean and Surinamese patients") and a '.Western' group (comprising "mostly Dutch but also some patients from other Western European, North American, Canadian and Australian origin".

Professional Qualifications: Patients - no statistical tests reported comparing education levels of non-Western and Western patients, however comparison shows significant difference in education levels (χ2= 23.5, P value < 0.001). Non-Western patients were more likely not to have completed primary school (8.0% cf 3.2% of "Western" patients (OR 0.38, 95% CI 0.20 to 0.69)).

Languages: 20.8% of non-Western patients reported to speak no Dutch or to have poor self-perceived proficiency, 32.5% reported average proficiency and 46.7% reported good proficiency

Method of recruitment: Not described but needed to attend for a consultation on random days

Estimated average cluster sizes: 16 for all participants and 10 for CALD participants only

Interventions

Education intervention: 2.5 day program for GPs with follow-up session after 2 weeks, and 12 min video-taped instruction for their patients in waiting room.

Aim of intervention: To improve intercultural communication between GPs and their patients
Comparison: No intervention

Outcomes

Primary outcome categories

Treatment outcomes: None reported

Health behaviour: None reported

Involvement in care: Mutual understanding between GP and patient [primary outcome]*

Evaluations of care: Patient satisfaction with consultation*; Patient’s feeling that consideration shown; Patient’s perception of quality of care

Secondary outcome categories

Health professionals

Knowledge and understanding: None reported

Consultation processes: None reported

Evaluation of processes and outcomes: None reported

Healthcare organisations

Adverse events: None reported

Quality and safety measures: None reported

Service utilisation: None reported

Health economic outcomes: None reported

Other outcomes: None reported.

*= selected outcome. The rationale for the selection of outcomes for each category is reported in Table 4.

Conceptual framework

Focus of the intervention was to improve intercultural communication between GPs and their patients. 

1.  Education content:· Knowledge addressed: culture, cultural competence; models of health and illness; socio-cultural context of health disparities; specific theoretical model.  Unclear if epidemiology; social determinants or constructs of racism and prejudice addressed.

Assessment and application of knowledge to relevant environmental, population, organisational and professional and systemic contexts addressed.

Skills included cultural self-assessment and communication skills; collaboration skills; non-verbal communication skills. Unclear if skills in deconstructing stereotypes or trust and partnership establishment included.

2.  Pedagogical approach: Teaching and learning method: Description for the GP intervention was based on Pinto’s three step method but the patient intervention is not. Key theoretical construct and principles: reported as Kleinman’s theory of exchanging explanatory models.

3.  Structure:

Delivery: Face-to-face

Format: 2.5 day program with follow-up session after 2 weeks

Frequency and duration: Once only

Method of assessment: Unclear

Method of evaluation: Unclear

Organisational support: Unclear

Participation: Voluntary

4.  Participant characteristics:

Target audience: GPs and their patients

Who delivered the intervention: Not reported

Notes

Contact with author: Yes - additional information regarding education intervention. Only study that included an intervention for patients.

Participant numbers in each group were computed from Figure 1 (Flowchart of levels of response of patient population [p345]) for baseline and the third measurement (T3). At baseline, there were data for 176 participants in the intervention group, which included 35% "non-Western" participants (62) and 175 participants in the control group, including 33% "non-Western" participants (58). At T3 data were reported for 151 participants in the intervention group (41% "non-Western" [62]) and 151 participants in the control group (31% "non-Western" [47]). At T1 there were 120 "non-Western" patients (Intervention = 62 and Control = 58) and at T3 there were 109 "non-Western" patients (I = 62 and C = 47).

The authors reported: "The effect on mutual understanding and perceived quality of care was analysed using multilevel multiple regression techniques adjusted for baseline values.…Differences between the two patient groups were tested by means of regression analysis with adjustment for baseline fraction, weighing cases (physicians) with the total number of patients seen at baseline plus at the measurement concerned." Thus it was judged that the study authors had taken clustering into account in their analyses so subsequent adjustments were not undertaken.

Risk differences at 6 months were taken from Table 4 (p 348), which reported difference between intervention and control groups as percentage of range adjusted for baseline with 95% CIs. Mutual understanding (involvement in care outcome) was reported as a continuous outcome with a 2% range and 11% difference to give MD 0.21, 95% CI 0.002 to 0.422. This difference was reported as adjusted for baseline differences. Neither confidence intervals nor standard deviations were reported for baseline data.

Patient satisfaction with consultation (evaluation of care outcome) was reported as a dichotomous outcome with RD 0.14 95%CI -0.031 to 0.305. SDs were calculated using RevMan.

Authors' conclusions: "Our intervention on intercultural communication for both GPs and patients was effective in the non-Western patient group, which supports our aim of Western patient group, which supports our aim of decreasing differences in outcomes of care between Western and non-Western patients. Both mutual understanding and perceived quality of care improved."

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskNo information given about sequence generation process
Allocation concealment (selection bias)Unclear riskInsufficient information about allocation concealment
Blinding of participants and personnel (performance bias)
All outcomes
Unclear riskNo blinding of GPs but authors state "Patients were ignorant about the group assignment of their GP". Patients seeing intervention GPs also exposed to a 12-min video-taped instruction prior to doctor visit, which was considered possibly to influence their expectations and assessments.
Blinding of outcome assessment (detection bias)
All outcomes
Low riskInterviewers, experts and research assistants who conducted preliminary data processing were blinded to intervention assignment (p 344).
Incomplete outcome data (attrition bias)
All outcomes
High riskMore than half the non-Western participants dropped out (55% of total failed home interviews).
Selective reporting (reporting bias)Unclear riskInsufficient information to determine
Other biasUnclear riskThere was an imbalance in attrition between the intervention and control groups between T1 and T3. No non-Western patients were lost from the intervention group but the control group reduced from 58 to 47 (16% loss).

Majumdar 2004

Methods

Study design: RCT randomisation at healthcare provider level

Location: Southern Ontario, Canada

Setting: Two community home care agencies and one hospital

Study aims: 1) To determine the effectiveness of cultural sensitivity training on the knowledge and attitudes of healthcare providers; 2) To assess the satisfaction of patients from different minority groups with healthcare providers who receive this training

Inclusion/exclusion criteria: No criteria reported for providers. Patients were eligible if they received care from the selected healthcare providers (those who did and did not participate in the cultural sensitivity training). Patients with history of cognitive impairment excluded

Informed consent obtained: Yes

Ethical approval: Unclear

Funding source and amount: Not reported

Statistical methods and their appropriateness: Not reported

Consumer involvement (in the design of study and/or intervention; in delivery of intervention; in evaluation of intervention; in interpretation of study findings): Not reported

Participants

Health professional participants

Type and number: 114 healthcare professionals (nurses and homecare workers)

Mean age: 40.09 (SD 11.03) experimental group; 42.57 (SD 10.15) control group

Gender: Not reported

Ethnicity: The majority "identified themselves as Canadians, and had parents of either Canadian or British origin". It is noted that healthcare professionals from minority groups were included in the study but this is not described in detail.

Professional qualifications: Mean education 14.72 (SD 2.85) years experimental group and 14.54 (SD 2.77) years control group

Language(s): Care providers: English (less than 15% for both groups spoke French as their native language)

Previous cultural competence care provider training: not reported

Method of recruitment: Unclear. "All study participants were financially compensated for time spent in training and measurement."

Patient participants

Type and number: 133 patients of two home care agencies and one hospital (the purpose of the study was to assess outcomes "of patients from different minority groups" but these are not described.) Unable to determine number of CALD patients from available data.

Mean age: not reported

Gender: Male = 35%, female = 65% in both groups

Ethnicity: "Both the experimental and control groups identified their ethnicity as 'Canadian', 'British' or 'European'". The study was reported to be conducted in urban area where approximately 25% of population is foreign-born.

Professional qualifications: high school or intermediate level (grades 5 to 8) reported as highest level of education by 77.2% of experimental group and 78.4% of control group.

Language(s): "the first language spoken by both groups was English"

Method of recruitment: Patients volunteered but no further details are provided

Interventions

Education intervention: 36-hour program targeting nursing and home care providers and their patients.  Used a number of validated tools (6 tools)

Comparison: No intervention

Aim of intervention: To change provider attitudes, patient satisfaction and accessibility to services (implied only)

Outcomes

Primary outcome categories

Treatment outcomes: Client health outcomes (OARS); Physical & Mental Health Assessment

Health behaviour: None reported

Involvement in care: Expenditure of Health Care & Social Services

Evaluation of care: Patient satisfaction [primary outcome] Client Satisfaction Questionnaire

Secondary outcome categories

Health professionals

Knowledge and understanding: Improved understanding cultural attitudes [primary outcome]

Consultation processes: None reported

Evaluation of processes and outcomes: None reported

Healthcare organisations

Adverse events: None reported

Quality and safety measures: None reported

Service utilisation: None reported

Health economic outcomes: None reported

Other outcomes : These were difficult to determine clearly, but reported instruments suggested that the following were included: provider cultural attitudes (open-mindedness vs closed-mindedness). This was measured with Ego Defensiveness, Open-Closed Mindedness and Nurses Attitude Toward Culturally Different Patients Questionnaire. Other instruments used to measure outcomes included: Rokeach Dogmatism Scale and Self-Assessment of Cultural Awareness.

Conceptual framework

Education intervention focused on improving cultural sensitivity.

Insufficient information across the domains of our conceptual framework.  Unable to obtain further information. 

1.  Education content: Knowledge addressed: cultural awareness and cultural sensitivity included but other topics not reported.

Assessment and application of knowledge not reported.

Skills in cultural self-assessment and communication included but unclear if deconstructing stereotypes or trust and partnership establishment included. Used a number of validated tools and questionnaires.

2.  Pedagogical approach: not reported

3.  Structure:

Frequency and duration: 36 hour program

Delivery: Unclear

Format: Unclear

Method of assessment: Unclear

Method of evaluation: Unclear

Organisational support: Unclear

Participation: Voluntary

4.  Participant characteristics:

Target audience: nursing and home care providers and their patients

Who delivered the intervention: not reported

Notes

Contact with author: Yes - but no additional information provided

Useable data relating to the review outcomes could not be extracted from this study

Authors' conclusions: The study showed “the benefits of cultural sensitivity training and indicated the feasibility of promoting awareness among healthcare providers by using culture-sensitive interventions.

Structured cultural sensitivity training programs could potentially help reduce cultural disparities in the healthcare system. The success of this training program is marked by the fact that the changes reported by the providers were not temporary.

This study also indicated that attitudinal change is a long­term process. The changes in providers were associated with improved functional outcomes among their patients."

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskNo information given about sequence generation process
Allocation concealment (selection bias)Unclear riskNo information given about allocation concealment
Blinding of participants and personnel (performance bias)
All outcomes
High riskNo blinding of participants, and outcomes judged likely to be influenced by lack of blinding
Blinding of outcome assessment (detection bias)
All outcomes
High riskNo blinding of outcome assessment, and outcome measurement likely to be influenced by lack of blinding
Incomplete outcome data (attrition bias)
All outcomes
High riskHigh rates of attrition across both patient and healthcare professional groups. More patients from control group dropped out of the study and more healthcare professionals dropped out of the intervention group compared to the control group
Selective reporting (reporting bias)High riskNot all pre-specified outcomes were reported but selected items from pre-specified instruments with statistically significant differences were reported
Other biasUnclear riskInsufficient information to determine

Sequist 2010

Methods

Study design: Cluster RCT - Randomisation at primary care team level

Location: Massachusetts, USA

Setting: 8 Ambulatory Health Centres

Study aim: 1) To evaluate the effect of cultural competency training and performance feedback for primary care clinicians on diabetes care for "black" patients; 2) To assess whether these efforts reduced racial differences between "white" and "black" patients for these 3 measures of diabetes care.

Inclusion/exclusion criteria: Primary care teams within a multispecialty group practice in eastern Massachusetts that had instituted key components of the Chronic Care Model including a common electronic medical record system

Informed consent obtained: Written informed consent obtained from clinicians, and a waiver of informed consent approved for patients

Ethical approval: Yes. The Harvard Pilgrim Health Care and Brigham and Women’s Hospital Human Studies Committees

Funding source and amount: Grant from Robert Wood Johnson Foundation (amount not stated)

Statistical methods and their appropriateness: Intention-to-treat analyses. Used generalized estimating equations to adjust SEs for clustering by primary care team and then fit multivariate ordinal logistic regression models to predict differences in clinician responses
Consumer involvement: In the development of the intervention 1084 diabetic patients responded to a survey to assess their perceived needs for information in a number of domains: right foods choice; weight loss; regular exercise; stress management; and understanding medications

Participants

Health professional participants

Type and number: 124 primary care clinicians in 31 primary care teams (comprising 91 primary care physicians and 33 nurse practitioners [NPs] or Physician Assistants [PAs]) Gender: (reported for clinicians only) Female: Physicians = 60%. NPs or PAs = 94%

Ethnicity: The process for determining health professional ethnicity was not reported, however. clinicians were described as "white" (82%),"Asian" (14%), "black" (2%), and "Hispanic" (2%).

Professional qualifications: Physicians reported to have average of 19.0 years clinical experience and NPs or PAs as 18.7 years experience

Language(s): Not reported

Previous cultural competence training: Unclear

Method of recruitment: Not reported

Patient participants

Type and number: 2699 "black" patients (36%, population used in this review). Total sample = 7557 diabetic patients (4858 "white" patients [64%])

Age: (reported for all patients, "black" and "white" combined) Mean 62.5 years (SD 13.5) intervention group; 62.3 years (13.2) control group and 4858 "white" patients 4858 (64%)

Gender: Not reported

Ethnicity: "Patient race was collected by self-report during the patient registration process" and were described as either "black" or "white" diabetic patients.

Professional qualifications: Education levels described mainly at high school and above

Language(s): Not reported

Estimated average cluster sizes = 944 for all participants and 337 for "black" participants only

Method of recruitment: Used data from patient records. No recruitment

Interventions

Education intervention: Two-day training program (consecutive) for primary care physicians, nurse practitioners and physician assistants (physicians attended only one day). Delivered by training experienced firm

Comparison: No intervention

Aim of intervention: To help primary care teams meet specified goals of: understanding attitudes of trust and bias; increase their knowledge of health disparities; gain skills to improve delivery of cross-cultural care

Outcomes

Primary outcome categories

Treatment outcomes: Rate of achieving clinical control targets within preceding 12 months for: Haemoglobin level (HbA1c level less than 7.0%); LDL cholesterol level less than 2.59 mmol/L [100 mg/dL])*; Blood pressure less than 130/80 mm Hg) [primary outcomes]

Health behaviour: None reported

Involvement in care: None reported

Evaluation of care: None reported

Secondary outcome categories

Health professionals

Knowledge and understanding: Clinician acknowledgement of racial differences in the quality of diabetes care for "black" clients over 12-month period across health services, within health centre*; among their patients [primary outcome]

Consultation processes: None reported

Evaluation of processes and outcomes: None reported

Healthcare organisations

Adverse events: None reported

Quality and safety measures: None reported

Service utilisation: None reported

Health economic outcomes: None reported

Other outcomes : None reported

*= selected outcome. The rationale for the selection of outcomes for each category is reported in Table 4.

Conceptual framework

Intervention can be mapped against most domains of our conceptual framework.

1.  Education content: Knowledge included culture, cultural competence, socio-cultural context of health disparities, epidemiology and social determinants, constructs of racism and prejudice and specific theoretical models such as explanatory models of health and illness; patient centeredness; consumer participation.

Unclear if models of health and illness included.

Assessment and application of knowledge to relevant environmental, population, organisational and professional and systemic contexts addressed.

Skills: Skills building around effective cross-cultural communication and to provide action-oriented tools to address disparities. Included communication skills and trust and partnership establishment. Unclear if skills in cultural self-assessment or deconstructing stereotypes included.

2.  Pedagogical approach: Teaching and learning method described but key theoretical construct and principles unclear. An underlying principle of the educational design was clearly grounding in the actual experience of black patients. The development of educational materials and recommendations tailored to caring for black patients were derived from patient focus groups and surveys. Clinicians were also given stratified patient data throughout the study.

3.  Structure:

Delivery: Face-to-face and off site , included a mix of lectures, group discussions, and community engagement activities

Format: 2 day program for Nurse practitioners and physician assistants (1 day only for physicians)

Frequency and duration: Once only

Method of assessment : Unclear

Evaluation method: Unclear

Organisational support: Links to organisational policies and procedures described. Embedded within professional development program with formal mentoring and peer support processes and professional accreditation points available. Monthly "race-stratified reports" complied and given to physicians, and monthly performance feedback reports provided

Participation: voluntary

4.  Participant characteristics:

Target audience: Primary care physicians, nurse practitioners and physician assistants

Who delivered the intervention: Harvard Pilgrim Health Care Foundation

Notes

Contact with author: Yes - additional information relating to education intervention

Study authors report adjusting for clustering, but the ICC used was not reported

Authors' conclusions: "We demonstrate that an intensive 12-month program of cultural competency training and race-stratified performance feedback increased primary care clinician awareness of the presence of racial disparities. However, we could not attribute any improvement in important measures of disease control for black diabetic patients to the program."

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskNo information given about sequence generation process
Allocation concealment (selection bias)Unclear riskNo information given about allocation concealment
Blinding of participants and personnel (performance bias)
All outcomes
Low riskUnclear if blinding of participants occurred but patient treatment outcomes judged unlikely to be influenced by lack of blinding
Blinding of outcome assessment (detection bias)
All outcomes
Low riskPatient treatment outcomes obtained from medical records. Clinician outcomes self-reported
Incomplete outcome data (attrition bias)
All outcomes
Low riskNo missing data for patient treatment outcomes. (Possible 20% attrition in clinician numbers)
Selective reporting (reporting bias)Unclear riskInsufficient information to determine
Other biasLow riskSample adjusted for clustering by primary care team but inter-correlation coefficient not reported

Thom 2006

Methods

Study design: Cluster RCT

Location: California, USA

Settings: 4 locations - Academic Medical Centre-based Family Practice; Community-based Practice; Rural Family Medicine Residency Program; Inner-city Family Medicine Residency Program

Study aims: To develop and evaluate a brief cross-cultural curriculum for resident and practicing physicians based on a model of culturally competent physician behaviors, and to evaluate the training plus feedback compared to feedback alone with respect to changes in patient-reported physician behaviors, patient satisfaction, patient trust in his or her physician, and disease-specific patient health outcomes

Inclusion/exclusion criteria: Physicians worked at one of four sites. Patients of physicians recruited to study were eligible if they had been seen in the past 12 months and had at least one visit for diabetes (ICD 9 code 250) or hypertension (ICD 0 code 401 to 405) that could be identified from computerized billing and encounter records

Informed consent obtained: Unclear but implied

Ethical approval: Yes. Approved by Institutional Review Boards at Stanford University and at the University of California

Funding source: California Endowment, grant #19991083 (amount not stated)

Statistical methods: Baseline characteristics compared using Chi2 test for dichotomous or categorical variables and Student's t-test for continuous variables. Mean changes in patient outcome measures from baseline to end of the study were calculated for each group and compared between groups using multiple analysis of variance (MANOVA) to adjust for differences in patient and physician characteristics and for differences in patient and physician characteristics and levels of each outcome variable at baseline. No adjustment for clustering reported.
Consumer involvement: None reported

Participants

Health professional participants

Type and number: 53 primary care physicians (23 in training & feedback group and 30 in feedback only group)

Age: Mean = 39.2 years

Gender: 24 females in total (45%); including 41% (9) of intervention group (Training & Feedback) and 52% (15) in the control group (Feedback only group) (figures computed from percentages)

Ethnicity: There was no information about how ethnicity was determined. The physician sample was described as comprising 38 "White" or "Caucasian" (the study uses these terms synonymously for physicians) (72%, I:C = 64%: 80%); 8 "Latino" (27%: 7%); 5 "Asian American" (9%: 10%); 1 "African American" (0%: 3%) ; and 2 others (not described further)

Professional qualifications: 33 family physicians in practice (62%) and 20 family medicine residents (38%)

Language(s): 43 another language (in addition to English), 34 spoke Spanish

Previous cultural competence training: Not reported

Method of recruitment: On site directly by one of the study investigators

Patient participants

Type and number: 429 patients with at least one visit for diabetes (ICD 9 code 250) or hypertension (ICD 0 code 401 to 405) were included in the trial. There were 247 in the intervention group (Training and Feedback) including 173 "non-Causcasian" patients and 182 in the control group (feedback only) including 142 "non-Caucasian" patients. The total number of "non-Causcasian" patients = 315 (73%)

Age: Mean = 54.9 years (SD = 11.6) (Training & Feedback group), 62.1 years (SD = 11.4) (Feedback only group)

Gender: Female 48.8% (Training & Feedback group), 63.2% (Feedback only group)

Ethnicity: Assignment of ethnicity is unclearly reported although the study reports that participants in one group were more likely to "self-identify as Latino or Asia". The groups in the study sample were comprised of the following:"Caucasian non-Hispanic" (21.9% feedback only and 30.0% training and feedback), "Latino/Hispanic" (28.9% and 25.3%), "African-American" (20.3% and 23.2%), "Asian" (23.0% and 13.1%) and "Other" (5.9% and 8.4%)

Professional qualifications: Not reported

Language(s): Recruitment letters sent to patients were in English, Spanish, and Chinese

Other: Participating patients were paid $10 for returning the baseline questionnaire, and $5 for each follow-up questionnaire

Estimated average cluster sizes: 107 for all participants and 79 for CALD participants only

Method of recruitment: Patients recruited by mail, with a follow-up phone call if needed after the second mailing. Recruitment letter and screening questionnaire in English, Spanish and Chinese. Participating patients were paid $10 for returning baseline questionnaire and $5 for each follow-up questionnaire

Interventions

Education intervention: Short education intervention (3 hours) for GPs.

Comparison: No intervention

Aim of intervention: 1) to expand knowledge of ethnic patients (knowledge and belief systems; 2) to enhance communication skills for cultural competency; 3) use of interpreters and cultural brokering

Outcomes

Outcomes for CALD participants were not reported separately

Primary outcome categories

Treatment outcomes: Change in patient weight*; systolic blood pressure; glycosylated haemoglobin [primary outcomes]

Health behaviour: None reported

Involvement in care: None reported

Evaluation of care: Patient satisfaction with consultation; patient perception of physician consideration; Patient Reported Physician Cultural Competency (PRPCC) Scale [primary outcome]*

Secondary outcome categories

Health professionals

Knowledge and understanding: None reported

Consultation processes: None reported

Evaluation of processes and outcomes: None reported

Healthcare organisations

Adverse events: None reported

Quality and safety measures: None reported

Service utilisation: None reported

Health economic outcomes: None reported

Other outcomes: Patient trust in the physician

*= selected outcome. The rationale for the selection of outcomes for each category is reported in Table 4.

Conceptual framework

Explicitly describes content of education intervention and reports on most domains of conceptual framework.

1.  Education content: Knowledge included culture, cultural competence, models of health and illness, epidemiology; and social determinants and specific theoretical models such as explanatory models of health and illness; patient centeredness; consumer participation. Socio-cultural context of health disparities and constructs of racism and prejudice not reported.

Assessment and application of knowledge to relevant environmental, population, organisational and professional and systemic contexts addressed.

Skills included communication skills; collaboration skills; non-verbal communication skills, deconstructing stereotypes, and trust and partnership establishment.

Unclear if cultural self-assessment included.

2.  Pedagogical approach: Teaching and learning method described but and key theoretical construct and principles unclear. Adapted from a model developed by one of the co authors. Peer education, adult learning principles and application of LEARN mnemonic to the patient interview (Listen, Explain, Acknowledge, Recommend, Negotiate)

3.  Structure:

Delivery: Face-to-face

Format: Offered as either single half-day training session or 3 separate 1 to 1.5 hour sessions

Frequency and duration: Once only

Method of assessment: Unclear

Evaluation method: Post-training evaluation rated on 5-point Likert-scale

Organisational support: Unclear

Participation: Voluntary  

Participant characteristics:

Target audience: GPs

Who delivered the intervention: GP instructors included authors of the paper, two other physicians with expertise in cross-cultural care, and experts in training and use of interpreters

Notes

Contact with author: Yes - additional information re education intervention

Comparison of changes in outcomes (means and standard deviations) are reported in Table 5, and participant numbers in each group provided in Table 4. (Table 5 is assumed to be mislabelled, reporting the numbers of physicians in each group rather than the number of patient participants).

Outcomes were not reported for the "non-Caucasian" participants separately so all outcomes are for all participants. The average cluster size was assumed to include 107 participants (28 "Caucasian" and 79 "non-Caucasian")

Numbers in the intervention and control groups for the continuous outcomes - change in patient weight (treatment outcome) and Patient Reported Physician Cultural Competency (PRPCC) evaluation of care outcome) were adjusted based on the design effect (DEFF=1+(1-M)ICC, where M is the average cluster size).

We used the estimated ICC of 0.2 (from Harmsen 2005 sample size calculations) as this was the only ICC we identified for this type of intervention. However this ICC is based on an average cluster size of 20. Sensitivity analyses were conducted with ICCs of 0.01, 0.02, and 0.05 using data from a study of ICCs (Campbell 2000).

The sensitivity analyses did not substantially alter the magnitude or significance of the summary effect size for either outcome.

Treatment outcome (Change in patient weight):

1) ICC = 0.2; DEFF = 22.2; ESSIntervention = 11 ESSControl = 8

SMD = 0.06 (95% CI -0.85 to 0.97)

2) ICC = 0.02; DEFF = 3.12; ESSIntervention = 79 ESSControl = 58

SMD = 0.07 (95% CI -0.27 to 0.41)

3) ICC = 0.01; DEFF = 2.06; ESSIntervention = 119 ESSControl = 88

SMD = 0.06 (95% CI -0.25 to 0.38)

4) ICC = 0.05; DEFF = 6.3; ESSIntervention = 39 ESSControl = 29

SMD = 0.07 (95% CI -0.41 to 0.55)

Evaluation of care (Patient Reported Physician Cultural Competency):

1) ICC = 0.2; DEFF = 22.2; ESSIntervention = 11 ESSControl = 8

SMD = 0.11 (95% CI -0.80 to 1.02)

2) ICC = 0.02; DEFF = 3.12; ESSIntervention = 79 ESSControl = 58

SMD = 0.11 (95% CI -0.23 to 0.45)

3) ICC = 0.01; DEFF = 2.06; ESSIntervention = 119 ESSControl = 88

SMD = 0.11 (95% CI -0.16 to 0.39)

4) ICC = 0.05; DEFF = 6.3; ESSIntervention = 39 ESSControl = 29

SMD = 0.10 (95% CI -0.27 to 0.48)

Authors' conclusions: "We did not find any measurable impact of a brief (4.5 hours) training curriculum aimed at improving physician cross-cultural knowledge and skills on any of the outcomes we chose."

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskNo information given about sequence generation process
Allocation concealment (selection bias)Unclear riskNo information given about allocation concealment.
Blinding of participants and personnel (performance bias)
All outcomes
Low riskNo information given about whether patients were blind to physician allocation but treatment outcomes judged unlikely to be affected by lack of blinding
Blinding of outcome assessment (detection bias)
All outcomes
Low riskNo information given about whether patients or assessors were blind to physician allocation but primary outcomes (treatment outcome and patient reported physician cultural competency) judged unlikely to be affected by lack of blinding
Incomplete outcome data (attrition bias)
All outcomes
High riskOf the 671 patients enrolled in the study 320 completed all surveys (48%). The proportions of respondents from the intervention and control groups are not reported
Selective reporting (reporting bias)Unclear riskInsufficient information available
Other biasHigh riskUnit-of-analysis error. Cluster trial but results reported for individuals with no adjustment for clustering. Adjustments for other possible effects. Significant baseline differences between study populations in intervention and control groups reported for demographic characteristics, diagnoses and outcome assessment. Considerable heterogeneity across participating sites. Unclear if physicians recruited at cluster sites before or after allocation

Wade 1991

Methods

Study design: Cluster RCT

Location: USA

Setting: College counseling centre in midwest city

Study aims: 1) To examine the effects of brief culture sensitivity training for counselors and the effects of whether counselors were "Black" or "White", on "Black" female clients' perceptions of counselor characteristics and the counseling relationship and clients' satisfaction with counseling; 2) to assess client attrition across three counseling sessions

Inclusion/exclusion criteria: Counselors recommended by counseling faculty. No criteria for counseling clients reported

Informed consent obtained: Unclear. Clients signed consent forms. No participants (counselors nor clients) were told about the purpose of the study

Ethical approval: Unclear

Funding source and amount: Not stated

Statistical methods: Multivariate analyses of variance (MANOVAs), with Hotelling's test of significance, performed on client ratings of counselors and counseling process. Due to considerable attrition, post-hoc analyses were conducted including MANOVAs using client ratings from last counseling session attended. Significant MANOVA results were followed by univariate F tests on each type of client rating from each counseling sessions. No adjustment for clustering reported

Consumer involvement: black female volunteers who either were former welfare recipients or defined themselves as lower-middle-class socio-economic status acted as clients in the training practice sessions. They were paid $5 per hour

Participants

Health professional participants

Type and number: 8 counselors

Age: 30 to 54 years, mean = 35.5 years. No significant differences in age, between groups P < 0.42

Gender: All female

Ethnicity: Process of determining ethnicity not reported. Described as "White" (4) or "Black" (4) (counselors stratified to intervention and control groups in randomisation)

Professional qualifications: All had a Masters degree in counseling, 5 enrolled in doctoral programs and 3 were community-based counselors. Experience ranged from 2 to 12 years with a mean of 5 years. No significant differences in counseling experience between groups P < 0.48

Language(s): Not reported

Previous cultural competence training: Unclear

Method of recruitment: Not described. Counselors were paid $5 fee per session or $10 per client, whichever amount was greater

Patient participants

Type and number: 80 "Black" clients

Age: 19 to 44 years, mean = 37.5 years with significant differences between groups P < 0.026. Clients assigned to treatment group were younger (M = 29.60 years, SD = 7.34) than those assigned to control group (M = 33.18 years, SD = 7.15). Authors judged subjects in the two groups "appeared to be in the same cohort."

Gender: All female

Ethnicity: Process of determining ethnicity not reported. Described as "Black" (80)

Professional qualifications: 70% high school diploma or equivalent, 7.5% completed 1 year of college, 11% completed associate's degree, 4% completed 3 years of college, and 7.5% completed bachelor's degree. No significant differences between number of educational years' experience between groups.

Language(s): Not reported

Other: Clients referred to college counseling centre by social service agencies or self-referred in response to radio and newspaper announcements. Clients received no financial remuneration for participation but customary fees were waived

Cluster sizes:10 "Black" women

Method of recruitment: Referred to college counseling center by social service agencies or self-referred in response to radio and newspaper notices. Participation described as voluntary and that they were allowed to terminate counseling at any time. Fees were waived

Interventions

Education intervention: Culture sensitivity training for counselors (4 hours)

Comparison: No intervention

Aim of intervention: To teach counselors to: a) attend to client's suspiciousness of social system and how this affects their perceptions of counselors; b) attend to racial and class differences between clients and counselor; c) attend to how client's attitudes and feelings of the experience of being black affect counseling process; d) attend to clients' values and to make explicit their own values during counseling

Outcomes

Primary outcome categories

Treatment outcomes: None reported

Health behaviour: Client concordance with attendance (over three counseling sessions*)

Involvement in care: None reported

Evaluation of care: Client perception of counseling process; Client perception of counselors repeated at end of three counseling interviews (This outcome was comprised of different dimensions that were reported individually. No overall measure was reported so we calculated the effect estimate for each dimension and chose the dimension [Attractiveness] whose effect estimate was ranked n/2 )

Secondary outcome categories

Health professionals

Knowledge and understanding: None reported

Consultation processes: None reported

Evaluation of processes and outcomes: None reported

Healthcare organisations

Adverse events: None reported

Quality and safety measures: None reported

Service utilisation: None reported

Health economic outcomes: None reported

Other outcomes: None reported

*= selected outcome. The rationale for the selection of outcomes for each category is reported in Table 4.

Conceptual framework

Small target group of counselors (8). Unclear across  most domains of our conceptual framework

1.  Education content: Knowledge included culture, cultural competence, models of health and illness, socio-cultural context of health disparities, constructs of racism and prejudice and specific theoretical models such as explanatory models of health and illness; patient centeredness; consumer participation.

Not reported if epidemiology; and social determinants were included

Application of knowledge to relevant environmental, population, organisational and professional and systemic contexts not reported.

Skills included cultural self-assessment, communication skills and establishment of trust and partnerships

2.  Pedagogical approach: Key theoretical construct and principles described but teaching and learning method unclear. Based on Pedersen's model

3.  Structure:

Delivery: Face-to-face

Format: Workshop

Frequency and duration: Once only, 4 hours of training

Method of assessment: Unclear

Evaluation method: Unclear

Organisational support: Unclear

Participation: Voluntary.

4.  Participant characteristics:

Target audience: Counselors

Who delivered the intervention: Not reported

Notes

Contact with author: No

The risk ratio for the dichotomous outcome, client concordance (health behaviour outcome) was calculated from client attrition data reported in the publication (numbers attending each session for intervention and control groups). Values for the three behavioural dimensions (expertness, trustworthiness and attractiveness) of the continuous outcome client perception of counseling (evaluation of care outcome) were reported at three time points (after each counseling session), but a summary measure for the outcome was not reported. A post-hoc decision was taken to use the change in one dimension of the scale between the first and final time points. The effect estimate for each dimension was calculated and the attractiveness dimension, which had the median effect (n/2), selected.

Numbers in the intervention and control groups were adjusted for the continuous outcome - client perception of counseling (evaluation of care outcome), and events and participants for the dichotomous outcome - client concordance with attendance (health behaviour outcome) based on the computed design effect (DEFF=1+(1-M)ICC, where M is the average cluster size) . We used the estimated ICC of 0.2 (from Harmsen 2005 sample size calculations) as this was the only ICC we identified for this type of intervention. However this ICC is based on an average cluster size of 20. Sensitivity analyses were conducted with ICCs of 0.01, 0.02, and 0.05 using data from a study of ICCs (Campbell 2000). The intervention and control groups both comprised 40 black women.

The sensitivity analyses did not substantially alter the magnitude or significance of the summary effect size for either outcome.

Evaluation of care (Client perception of counseling):

1) ICC = 0.2; DEFF = 2.8; ESSIntervention = 14 ESSControl = 14

SMD = 1.60 (95%CI 0.99 to 2.20)

2) ICC = 0.02; DEFF = 1.18; ESSIntervention = 34 ESSControl = 34

SMD = 1.60 (95%CI 1.05 to 2.15)

3) ICC = 0.01; DEFF = 1.09; ESSIntervention = 37 ESSControl = 37

SMD = 1.60 (95%CI 1.08 to 2.13)

4) ICC = 0.05; DEFF = 1.45; ESSIntervention = 28 ESSControl = 28

SMD = 1.60 (95%CI 0.99 to 2.20)

Health behaviour (Client concordance with attendance):

1) ICC = 0.2; DEFF = 2.8; ESS = 14

RR = 1.53 (95%CI 1.03 to 2.27)

2) ICC = 0.02; DEFF = 1.18; ESS = 34

RR = 1.48 (95%CI 1.11 to 1.96)

3) ICC = 0.01; DEFF = 1.09; ESS = 37

RR = 1.52 (95%CI 1.17 to 1.98)

4) ICC = 0.05; DEFF = 1.45; ESS = 28

RR = 1.53 (95%CI 1.12 to 2.10)

Authors' conclusions: "A major finding of this study is that Black female clients’ perceptions of counselors and the counseling process were affected more by culture sensitivity training of the counselors than by counselor race. Counselors who had received culture sensitivity training were assigned higher ratings on expertness, trustworthiness, attractiveness, unconditional regard, and empathy than were counselors who had not received the additional training. Furthermore, clients assigned to counselors the culture sensitivity training group returned for more follow up sessions and reported greater satisfaction with the counseling process than did clients assigned to counselors in the control group."

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskUsed table of random numbers to assign counselors to intervention and control groups
Allocation concealment (selection bias)Unclear riskInsufficient information about allocation concealment
Blinding of participants and personnel (performance bias)
All outcomes
Low riskCounselors and clients unaware of nature of the study
Blinding of outcome assessment (detection bias)
All outcomes
Low riskNo blinding of outcome assessment but outcome measurement judged to be unlikely to be influenced
Incomplete outcome data (attrition bias)
All outcomes
High riskImbalance in reasons for missing data across groups, and although data was imputed using an appropriate method, there were very high rates of attrition in the control group over the study (> 70%)
Selective reporting (reporting bias)High riskOne outcome - clients' perceptions of the counseling process - was not reported for each group, probably due to the high attrition in the study
Other biasHigh riskThere is no indication that analyses have taken account of clustering

Characteristics of excluded studies [ordered by study ID]

StudyReason for exclusion
Afuwape 2010Intervention directed at patients, not health professionals
Berlin 2010Patient outcomes not part of study design
Bhattacharyya 2010Intervention is not related to cultural competence
Bhui 1998Not an RCT
Bloch 2012Patient outcomes not part of study design
Bremner 2011Intervention directed at patients, not health professionals
Butterfloss 2002Not an RCT or cultural competence intervention
Choi 2012Intervention directed at patients, not health professionals
Christensen 1984Not an RCT
Cooper 2009Intervention does not involve training for health professionals
Cooper 2010Intervention does not involve training for health professionals
Cortese-Peske 2013Patient outcomes not part of study design nor assessed
Crenshaw 2011Not an RCT
Fossli Jensen 2000Intervention does not involve cultural competence training for health professionals
Gendron 2013Not an RCT
Greenberg 2013Patient outcomes not part of study design
Greer 2007Not an RCT
Heitzler 2011Patient outcomes not part of study design nor assessed
Jackson 2000Intervention is not directed at health professionals
Javier 2013Not an RCT
Khanna 2009Not an RCT
Kopke 2012Not a cultural competence education intervention
Kutob 2009Outcomes for patients is not part of study design or assessed
Kutob 2013Patient outcomes not part of study design nor assessed
Lasser 2010Intervention is not directed at health professionals
Manfredi 1998Intervention did not include cultural competence education of health professionals
Marra 2010Not an RCT
Martey 1995Not an RCT
Maxie 2006Not an RCT
McCabe 2006Not an RCT
Miranda 2003Intervention did not include cultural competence training for health professionals
Mostow 2010Not an RCT
Naeem 2011Intervention did not include cultural competence training for health professionals
Nagel 2009aIntervention did not include cultural competence training for health professionals
Nagel 2009bNot an RCT
Newland 2008Not an RCT
O'Brien 1977Outcomes for patients is not part of study design or assessed
Omer 2008Not an RCT
Palmer 2011Not an RCT
Park 2013Patient outcomes not part of study design nor assessed
Prescott-Clements 2013Patient outcomes not part of study design nor assessed
Resnick 2009Not a cultural competence education intervention and outcomes for patients is not part of study design or assessed
Roberts-Thomson 2010Not a cultural competence education intervention
Rogers 2000Not an RCT
Schim 2006Outcomes for patients is not part of study design or assessed
Schouten 2005Outcomes for patients is not part of study design or assessed
Sheikmoonesi 2011Not a cultural competence education intervention
Sheridan 1982Not an RCT
Sixta 2007Not a cultural competence education intervention
Sixta 2008Not an RCT
Smith 2001Outcomes for patients is not part of study design or assessed
Thomas 2000Not an RCT
Tozer 2010Not an RCT
Witter 2012Patient outcomes not part of study design nor assessed
Xu 2010Not an RCT

Characteristics of ongoing studies [ordered by study ID]

Clark 2010

Trial name or titleNCT01251523: Improving Asthma Outcomes Through Cultural Competence Training for Physicians
MethodsRandomized clinical trial
ParticipantsParents of pediatric subjects (children with asthma)
InterventionsThe randomized clinical trial compares two educational interventions: Physicians Asthma Care Education (PACE) and PACE plus with 90 physicians in Atlanta and the Bronx and their 1192 patients.
OutcomesPrimary outcomes measures include: emergency department visits of parents and children at 3 time points.
Starting dateNovember 2010
Contact informationLara J Thomas, MPH; ljthomas@umich.edu
Notes

This study will evaluate the effects cultural competence training for physicians on health outcomes of children and performance ratings given to physicians by the parents of the child. The estimated completion date for the study was January 2014.

Researcher Anneliese Synnot contacted Lara Thomas in February 2014 and was advised that no data are available yet.

Ancillary