Health-related empowerment in cancer: Validity of scales from the Health Education Impact Questionnaire
Elizabeth Maunsell PhD,
Research Center, Centre hospitalier universitaire (CHU) de Québec, Québec, Québec, Canada
Deschenes-Fabia Breast Disease Center, Centre hospitalier universitaire (CHU) de Québec, Québec, Québec, Canada
Departement of Social and Preventive Medicine, Faculty of Medicine, Université Laval, Québec, Québec, Canada
Corresponding author: Elizabeth Maunsell, PhD, Research Centre, Centre Hospitalier Universitaire (CHU) de Québec, 1050 Chemin Sainte-Foy, Québec, Québec, Canada, G1S 4L8; Fax 418 682-7383. email@example.com
See related article on pages 3219–27 and related editorial on pages 3108–10, this issue.
Empowerment refers to an individual's feelings of being able to manage the challenges of the cancer experience and of having a sense of control over one's life. However, empowerment questionnaires that have been validated for the cancer setting are lacking. The objective of this study was to validate scales from the Health Education Impact Questionnaire (heiQ), which assesses the effects of health education programs among individuals with chronic conditions. The heiQ scales Social integration and support, Health service navigation, Constructive attitudes and approaches, Skill and technique acquisition, and Emotional distress were identified as key dimensions of empowerment for the cancer context.
Adults who were diagnosed with cancer <3 years earlier were recruited from a population-based cancer registry and from the Canadian Cancer Society's information and peer-support programs. The 731 participants completed a mailed questionnaire, which included the heiQ scales, related constructs, and demographics. Reliability was assessed using Cronbach α values, and validity was determined using confirmatory factor analysis and scale correlations with related constructs (self-efficacy, intrusive thoughts about cancer, and mental and physical health).
The hypothesized 5-factor model fit the data adequately (chi-square statistic, 528.17; degrees of freedom, 265; root mean square error of approximation, .04; non-normed fit index, .99; comparative fit index, 1.00; standardized root mean residual, .05). Factor loadings were high (23 of 25 were ≥.70), and the factor correlations indicated separate but related constructs. Cronbach α values ranged from .75 to .90. A priori hypotheses about the correlations between heiQ scales and related constructs all were supported.
Empowerment is recognized as a positive, contextually defined, dynamic, multidimensional construct.[1-3] It can be defined both as a social process by which individuals gain mastery over their lives and as an outcome resulting from this process.[3-5] For the cancer context, we defined health-related empowerment as feelings of being able to manage the challenges of the cancer experience and of having a sense of control over one's life.[1, 3, 6] An empowered individual may better understand and participate in their care, mobilize resources, and take actions that can reduce anxiety, enhance strategies for dealing with cancer, and improve quality of life. Thus, health-related empowerment includes intrapersonal, interactional, and behavioral dimensions.
Empowering an individual who has cancer during different phases of the cancer trajectory is often an objective that informs the provision of support services. Empowering patients will be increasingly important given the growing numbers of patients and survivors, greater reliance on outpatient treatments, and financial strains in health care systems, including the shortage of health professionals (psychosocial in particular). These trends mean that considerable responsibility for management of the cancer journey now falls on patients, their caregivers, and community-based services.[6, 7] However, empowerment is rarely formally assessed, because few validated questionnaires measuring health-related empowerment exist, particularly for the cancer setting.[2, 8-12] Measures like general psychological or cancer-specific distress, illness intrusiveness, and quality of life poorly capture the potentially empowering effects individuals may gain from supportive interventions. Our own experience with trials testing supportive interventions in oncology supports this hypothesis. We observed that some supportive interventions were not effective for changing quality-of-life scores despite patients reporting qualitatively that the intervention was useful in helping them cope overall.[13-15] These apparent contradictions may be partly because of the lack of instruments that measure proximal benefits such as personal control over one's life, reduced isolation, confidence, and hope. For these reasons, we sought to validate a measure of empowerment for use in cancer research and program evaluations.
Here, we report on a cross-sectional study evaluating the reliability and validity of 5 scales from the Health Education Impact Questionnaire (heiQ), a generic measure we considered relevant to the cancer setting. It was initially developed to assess interventions for individuals with chronic diseases and has robust psychometric properties.
MATERIALS AND METHODS
Identification of Measures Potentially Assessing Key Dimensions of Empowerment
The Medline, CINHAL, and PsychINFO databases were reviewed up until January 2007 to identify studies of instruments measuring empowerment and empowerment-related constructs, including self-efficacy, coping, and perceived control. We considered generic and disease-specific instruments and, for the latter, and did not place limits on the types of disease for which the instrument had been developed. The instruments considered had to have face and content validity from the research team's perspective and based on feedback from key experts. They also had to contain items referring specifically to the individual's health in some way, because the aim of support interventions in this population is to improve the individual's ability to manage their physical and emotional health, and an explicit link to the health state is important for respondent focus on health-related empowerment. We sought measures relevant to patients in active treatment (curative or palliative) and to survivors who had completed active treatment. Finally, we considered it essential that individual scales should be short to make using them feasible in a variety of evaluation contexts.
Four team members with extensive experience in psycho-oncology research, measurement instruments, and program evaluation for cancer patients independently reviewed all instruments and their psychometric properties.[5, 9, 17-20] Face and content validity was assessed by asking staff from the Canadian Cancer Society's (CCS) Cancer Information Service (CIS) and peer-support program, CancerConnection, to review items for relevance, comprehensiveness, and applicability when assessing patient and survivor support interventions. Cognitive interviews with 21 cancer patients recruited from oncology clinics and CCS support programs who had previously self-administered the questionnaire were conducted by telephone. Synthesis of cognitive interviews, along with expert assessment of the item content and psychometric properties, resulted in retention of 5 of the 8 heiQ scales for quantitative validation.
The heiQ version 3.0 is a well validated, widely used measurement system developed in Australia to comprehensively assess the effects of health education programs on self-management among individuals with chronic conditions and has been successfully adapted for administration in other languages. The heiQ developers used a validity-driven approach to achieve full representation of the construct of self-management capability. Consequently, 8 independent scales arose from a grounded consultation with patients and stakeholders. The grounded theory and empirical analysis dictate that no overall score based on all items should be calculated. Although generic, the heiQ had not been specifically validated for use in the cancer setting.
The 5 scales we retained were selected because they were conceptualized as key aspects of empowerment in the cancer setting. These scales are Social integration and support, Health service navigation, Constructive attitudes and approaches, Skill and technique acquisition, and Emotional distress (with reverse scoring of negatively worded items). Slight adaptations to wording were made for the cancer context after discussion with the principal heiQ developer (R.H.O.). Given our intent to measure the most crucial aspects of empowerment in cancer, we did not retain scales that were not considered specific enough or clearly appropriate to all moments in the cancer trajectory (Health-directed behavior), scales in which items were not linked clearly enough to the health state (Positive and active engagement in life), or scales with some items representing situations one would not necessarily want an individual with cancer to be self-managing (Self monitoring and insight).
Canadian English-speaking cancer patients and survivors aged ≥18 years were recruited from a provincial population-based cancer registry and from CCS CIS and CancerConnection users. The ethics committees of the University of Waterloo, the University of Manitoba Bannatyne Campus, and Hôpital du Saint-Sacrement of the Centre hospitalier universitaire (CHU) de Québec approved this study.
Two random samples (one of men and one of women) targeting patients and survivors who had a histologically confirmed cancer diagnosed 3 to 27 months earlier than February 28, 2009 (the date of sample selection) were selected by the cancer registry. Individuals who were diagnosed with in situ cancer, nonmelanoma skin cancer, and neurologic tumors were ineligible as were those who had previously told the registry they did not want to participate in research studies. The samples were cross-referenced with vital statistics to decrease the probability of contacting families of individuals who had died.
Between April and July 2009, patients and survivors who called the CIS were recruited at the end of their call, and those who called CancerConnection were recruited at the end of their 10-day follow-up call, after having at least 1 contact with their peer-support person. All clients who were calling for the first time and diagnosed 3 to 27 months earlier were eligible except those who were calling from outside Canada, those judged by staff as too distressed, those not fluent in English, or those who called for reasons other than support (eg, how to make a donation).
Potential registry participants received a prenotification letter, whereas CCS clients were verbally informed about the study and were asked for consent to share their name and contact information with the research team. A first questionnaire package that included a cover letter was sent to those who consented. A second questionnaire package was sent if the completed questionnaire had been not returned within 2 weeks of the first mailing. Finally, a reminder letter was sent 4 weeks after the second questionnaire package. The questionnaire package contained questions on sociodemographic, disease, and treatment characteristics; the 5 heiQ scales; and scales from instruments measuring related constructs (self-efficacy, intrusive thoughts about cancer, mental and physical health).[18, 20, 23, 24]
The heiQ scales were scored by summing item responses (in which 1 indicates strongly disagree; 2, disagree; 3, agree; 4, strongly agree) and dividing by the number of scale items. With reverse scoring of Emotional distress items, higher scores indicate higher levels of empowerment; scores can vary from 1.00 to 4.00. We examined the extent of missing responses for each scale and present descriptive statistics (mean, standard deviation, median, and interquartile range). Because one eventual use of these scales would be for evaluating interventions with before-and-after designs, it was important to assess possible floor and ceiling effects in the cancer population. To assess these, we examined score distributions to identify proportions of individuals whose reported empowerment level was so low or high that a meaningful deterioration or improvement—defined as a difference perceivable by an individual—could not be detected by the scale. The threshold to determine whether there is room for potentially meaningful deterioration or improvement has been identified as from 5% to 10% of the scale's breadth (which generally corresponds to half a standard deviation).[25, 26] We set the threshold to 10%, or scores ≤1.3 of 4.0 or ≥3.7 of 4.0, respectively.
For descriptive statistics and the confirmatory factor analysis (CFA), we used multiple imputation based on the expected maximization (EM) algorithm to estimate and replace missing observations when less than 20% of the heiQ items were missing at random. Reliability was assessed using Cronbach α values. The proposed 5-factor model of these heiQ scales was examined using CFA for ordered categorical measures with robust maximum likelihood estimation using calculated polychoric correlations and an asymptotic covariance matrix in LISREL 8.80 (Scientific Software International, Inc., Lincolnwood, Ill). The factor loading for the first item of each latent variable was set to 1.0 to establish the metric of the latent variable. We used multiple, robust goodness-of-fit indices to evaluate the tenability of the model: root mean square error of approximation (RMSEA), non-normed fit index (NNFI), comparative fit index (CFI), and the standardized root mean residual (SRMR).[29, 30] Values ≥.90 and ≥.95 for the CFI and the NNFI, respectively, indicate acceptable and good fit of the model. RMSEA and SRMR values ≤.08 indicate acceptable model fit, whereas values ≤.05 indicate good fit.[29, 31] The strengths of the standardized factor loadings between each item and its corresponding latent variable also were examined.
Correlations were used to assess associations between the latent scores computed within the CFA framework, and also between the latent variables and other related constructs, with directions and potential strengths of the relations hypothesized a priori.
Among individuals who were contacted through the registry, participation was 64.4% (524 of 814 individuals), with identical rates for women and men (64.1% and 64.6%, respectively). CCS participation was 72% overall (207 of 289 individuals), and participation was lower from the CIS (74 of 131 individuals; 56.4%) than from CancerConnection (133 of 158 individuals; 84.2%). Examination of the means (± standard deviations), medians, and ranges of scores were very similar in both groups (data not shown); consequently, the groups were combined for a total of 731 participants.
More participants were women (58%), aged ≥55 years (71%), had a college or university-level education (54%), were living with a partner/spouse (71.7%), and were working at the time of the study (55.4%). The median time since diagnosis was 19 months. Breast, prostate, and colorectal cancers were the most frequent disease sites (Table 1).
Table 1. Characteristics of 704 Respondents Included in the Confirmatory Factor Analysis
No. of Respondents (%)
No. of Respondents (%)
Abbreviations: IQR, interquartile range; SD, standard deviation.
Time since first diagnosis, mo
Mean ± SD
21.6 ± 31.8
Type of cancer
Never legally married
More than 1 type
Highest completed education
College level or university
Employed at the time of study
The frequency of missing responses per scale ranged between 4.1% and 8.6%, except in Skill and technique acquisition, which had 14.6% missing responses. This was primarily because of a single item, “I have a good understanding of equipment that could make my life easier,” with 9.3% of missing responses; <5% were missing for other items in this scale. The use of the EM algorithm resulted in a group of 704 for CFA analyses.
CFA results indicated a good model fit; the hypothesized, oblique 5-factor model fit the data adequately based on the RMSEA, CFI, NNFI, and SRMR values (Santorra-Bentler scaled chi-square statistic, 528.17; degrees of freedom, 265; RMSEA, .04; NNFI, .99; CFI, 1.00; SRMR, .05) (Table 2). Standardized factor loadings between each item and their respective latent variable were all significant, ranged between .65 and .91, and exceeded the generally recommended cutoff value of .7 in 23 of 25 items (92%) overall (Table 2). The correlations between the latent factors of the heiQ were positive and of moderate to high magnitude (r = .38-.78). The lowest Cronbach α value was .75 (Skill and technique acquisition); and the others ranged from .84 to .90. Figure 1 presents the range of scores for the 5 heiQ scales. Median and mean scores were generally close to 3 of a possible 4 for all scales. Less than 3% of scores were below the potential floor (≤1.3 of 4.0). Higher proportions were in the range of scores defined as potential ceiling effects (≥3.5 of 4.0), varying from 9.1% (Skill and technique acquisition) to 22.2% (Constructive attitudes and approaches) (Table 3).
Construct validity was supported, and all a priori hypotheses about the direction and strength of correlations between related constructs were confirmed (Table 4).
Table 2. Standardized Factor Loadings and Latent Variable Correlations of the 5 Health Education Impact Questionnaire Scales (n = 704)
Abbreviations: CASE-Cancer, Communication and Attitudinal Self-Efficacy for cancer (see Wolf et al., 2005); CDSE, Chronic Disease Self-Efficacy (see Lorig et al., 1996); CI, confidence interval; heiQ, Health Education Impact Questionnaire; IES-R, Impact of Events Scale-Revised (see Weiss & Marmar 1997); SF-8, 8-item Short-Form Health Survey (see Ware et al., 2001).
Numbers are slightly less than 704 in the confirmatory factor analysis if data were missing for the compared constructs.
Correlations were defined as weak (<.3), moderate (.3-.69), or strong (≥.7); all P < .0001.
Social integration and support
CDSE: Obtain help from community, family, and friends
CDSE: Do chores
Health service navigation
CASE-Cancer: Understand and participate in care
CASE-Cancer: Seek and obtain information
CDSE: Communicate with physician
Constructive attitudes and approaches
CASE-Cancer: Maintain a positive attitude
SF-8: Mental component summary
Skill and technique acquisition
CASE-Cancer: Understand and participate in care
CASE-Cancer: Seek and obtain information
Emotional distress (reverse-scored)
CASE-Cancer: Maintain a positive attitude
SF-8: Mental component summary
Overall, our results support the validity of the 5 heiQ scales we retained as generic measures of key dimensions of health-related empowerment in the cancer population. The confirmatory factor analysis provides strong evidence that the original factor structure is robust in this population. Furthermore, all but 2 items had high factor loadings, indicating that items were strong indicators of their respective factors. Scale internal consistency was very good, with only 4 to 6 items per scale. The positive and moderate-to-high correlations between the 5 latent factors support the view that these scales measure different but related dimensions of empowerment—a multidimensional construct. Our a priori predictions concerning associations between these various dimensions of empowerment and constructs closely related to them all were confirmed. To our knowledge, no other empowerment scales have been validated as rigorously for use specifically in cancer research.
On the basis of the literature, examination of related instruments, and clinical and program experience, we are confident that these 5 dimensions represent very important aspects of health-related empowerment for cancer patients and survivors who are dealing with cancer diagnosis, treatments, and survivorship. At the time this work began, we identified no adequately validated questionnaires of empowerment for use in the cancer setting, and this was still the case in 3 recent systematic reviews.[2, 8, 10] Although we recognize that these 5 heiQ scales may not comprehensively measure all dimensions of empowerment given the multidimensional nature of the construct and its possible evaluation as both a process and outcome, these scales nonetheless cover the intrapersonal, interactional, and behavioral dimensions that are part of health-related empowerment. Furthermore, the empowerment dimensions considered key here resemble 3 of the 4 dimensions that are part of very recently published 40-item measure of empowerment for women with breast cancer. However, in 2 scales of this latter instrument (Personal strength, Social support), there are no items referring to the individual's health in some way.
We cannot determine with the current data whether the tendency to higher scores is related to the nature of the scales themselves or to the characteristics of the participants, but there are reasons to believe it may be the participants studied. Most were more than 1 year postdiagnosis. It is possible that getting beyond treatment increased feelings of empowerment. CCS participants may have been more empowered at the outset or may have felt empowered after seeking help through these programs. Still, some participants registered low levels of empowerment on each scale. Moreover, we note that heiQ scores were more variable and less skewed toward high scores than those from other widely used scales measuring related concepts (which we also administered).[18, 20] Further study to determine whether the scales are sensitive enough to capture low levels of empowerment will be important. The heiQ scale levels in newly diagnosed patients, a group generally recognized as disempowered, and change in empowerment level over time will be assessed in an ongoing randomized trial testing an e-health support intervention among patients with newly diagnosed colorectal cancer.
Overall, the 5 heiQ scales examined in this study appear to be promising for measuring health-related empowerment in both intervention research and program evaluation in the cancer setting. The items are short and easy to understand, and we demonstrated that psychometric properties were excellent in the cancer population. These scales can be self-administered, and the low frequency of missing responses suggests they are relevant, acceptable, understandable, and provide a high percentage of usable responses. Although 1 item had a nonresponse proportion approaching 15% (“I have a good understanding of equipment that could make my life easier”), it is an important question nonetheless because of its relevance to those patients who have to use different types of equipment. More generally, the limited number of items and scales (each scale also can be used individually) are advantageous for reducing respondent burden and increasing the feasibility of their use in both research and program evaluation. Our study was conducted in 2 settings with participants who were recruited from a population-based registry and from the CCS cancer information and support programs. The very great majority of respondents had scores at levels where potentially meaningful deterioration or improvement could be registered were they to complete the scales again after a change in empowerment level.
In summary, we believe that these heiQ scales represent promising measures for detecting desired empowering effects of supportive interventions and programs in the cancer setting. In the context of increased need for patients and families to self-manage during treatments, recovery, and survivorship, being able to provide evidence about whether or not support services empower patients should be a priority and would constitute valuable public health information for the individuals responsible for relevant health care planning, policies, and priorities. This study demonstrates that the heiQ scales are valid, reliable measures of key dimensions of generic health-related empowerment and could fill an important gap in the measures currently available to evaluate the effects of supportive cancer interventions.
This work was supported by the Canadian Institutes for Health Research (CIHR) (KAL-82607) and the Canadian Cancer Society Research Institute (formerly the National Cancer Institute of Canada) (010498 and 010499) through the Propel Center for Population Health Impact. Dr. Lauzier was supported by the CIHR-funded Strategic Training Initiative in Health Research (STIHR) Psychosocial Oncology Research Training Program (PORT) (postdoctoral award); by the Fonds de Recherche du Québec-Santé in partnership with L'Institut National d'Excellence en Santé et en Services Sociaux (INESSS) (Chercheur-boursier award).
CONFLICT OF INTEREST DISCLOSURES
Dr. Lauzier receives salary support from the Université Laval Chair on Adherence to Treatments (funded through unrestricted grants from AstraZeneca Canada, Merck Canada, Pfizer Canada, Sanofi-Aventis Canada, and the Prends soin de toi Program). Dr. Osborne reports that Deakin University owns the heiQ Questionnaire intellectual property and charges a license fee to cover the legal and administrative costs of managing the tool.