Components and effectiveness of patient navigation programmes to increase participation to breast, cervical and colorectal cancer screening: A systematic review

Abstract Background Inequalities in cancer incidence and mortality can be partly explained by unequal access to high‐quality health services, including cancer screening. Several interventions have been described to increase access to cancer screening, among them patient navigation (PN), a barrier‐focused intervention. This systematic review aimed to identify the reported components of PN and to assess the effectiveness of PN to promote breast, cervical and colorectal cancer screening. Methods We searched Embase, PubMed and Web of Science Core Collection databases. The components of PN programmes were identified, including the types of barriers addressed by navigators. The percentage change in screening participation was calculated. Results The 44 studies included were mainly on colorectal cancer and were conducted in the USA. All described their goals and community characteristics, and the majority reported the setting (97.7%), monitoring and evaluation (97.7%), navigator background and qualifications (81.4%) and training (79.1%). Supervision was only referred to in 16 studies (36.4%). Programmes addressed mainly barriers at the educational (63.6%) and health system level (61.4%), while only 25.0% reported providing social and emotional support. PN increased cancer screening participation when compared with usual care (0.4% to 250.6% higher) and educational interventions (3.3% to 3558.0% higher). Conclusion Patient navigation programmes are effective at increasing participation to breast, cervical and colorectal cancer screening. A standardized reporting of the components of PN programmes would allow their replication and a better measure of their impact. Understanding the local context and needs is essential to design a successful PN programme.


| INTRODUCTION
In the last decades cancer incidence and mortality have been increasing worldwide, across high-and low and middle-income countries (LMIC). 1 One approach that has been shown to reduce breast, cervical and colorectal cancer-related mortality at a population level, is cancer screening. [2][3][4] For cancer screening to be as effective as possible, it is important that screening programmes reach high coverage of the target population. This is achieved with screening programmes easily accessible and available to everyone -regardless of their socioeconomic position. Several interventions have been described to promote equitable cancer screening and reduce structural barriers related to access (e.g., the use of mobile units), but one approach that is gaining significant interest is patient navigation. In 1990 Freeman developed patient navigation to assist low-income women in USA to overcome barriers to follow-up after an abnormal breast cancer screening result. 5 As patient navigation began to take shape, it was implemented in the screening of other cancer sites, at different levels of the cancer screening continuum and for other socially disadvantaged groups. Therefore, a patient navigation intervention is by default designed to improve access among hard-to-reach populations. Moreover, the patient navigation approach is focused on supporting people to overcome barriers and has the following characteristics: (i) occurs within a specific cancer care event (e.g., one-time screening), (ii) follows the individual until a specific endpoint is reached (e.g., a definitive diagnosis), (iii) targets the health services needed to achieve the endpoint (e.g., screening and/or diagnostic care), (iv) addresses barriers at a health system, individual, educational, and/or social and emotional level, and (v) aims to reduce delays in cancer care access and uptake. 6 As the evidence for patient navigation developed, DeGroff et al. (2014) outlined 10 key components that should be considered when designing a patient navigation programme: (1) identification of a theoretical framework and establishment of programme goals, (2) specification of the community characteristics, (3) determination of the point(s) of intervention in the cancer care continuum, (4) establishment of the setting where navigation is provided, (5) description of the services offered and the patient navigator responsibilities, (6) determination of the background and qualifications of navigators, (7) selection of the communication method between individuals and navigators, (8) design of the patient navigator training, (9) establishment of the supervision of navigators, and (10) evaluation of patient navigation. 7 Previous systematic reviews have described the positive impact that patient navigation interventions have on improving screening participation for breast, cervical and colorectal cancer, although it is acknowledged most of these studies have been conducted in the USA. The literature has reported increased participation in patient navigation programmes when compared to control (as usual care) as well as other types of interventions. 6,8,9 Although the previous reviews have been helpful to extend our understanding in this field and summarize a complex evidence base, the definition used by the studies to conceptualize a patient navigation intervention is wide-ranging and varied. This is challenging, given interventions might be based on the concept of patient navigation, but may not necessarily use this term to describe them. To the best of our knowledge, a comprehensive review on patient navigation programmes using a framework guided by the key components outlined by DeGroff et al. has not been undertaken. This work sought to address this gap by identifying the reported components of patient navigation to consider when conceptualizing patient navigation, and by assessing the effectiveness of patient navigation programmes to promote breast, cervical and colorectal cancer screening.

| METHODS
A systematic search of the literature was conducted in Embase, PubMed and Web of Science Core Collection databases in March 2020 and then updated (January 2021). The search was limited to papers published since 2000 (as previous literature was considered not relevant to our purpose) without language restriction. The search strategies combined Medical Subject Headings and free text terms regarding screening, breast, cervical and colorectal cancer, interventions, participation and social inequalities. As an example, the search strategy used in Web of Science Core Collection is presented in Appendix 1.
The population considered was all people eligible for breast, cervical or colorectal cancer screening as defined by the eligibility criteria for that screening. The screening methods considered were those recognized and validated in IARC (International Agency for Research on Cancer) handbooks. [10][11][12] Interventions were patient navigation programmes that aimed to increase access to cancer screening. Although we did not require studies to identify K E Y W O R D S breast cancer, cervical cancer, colorectal cancer, components, effectiveness, patient navigation, screening their intervention as patient navigation, we included only those where the intervention was individualized and ready to address some type of barrier, specifying it or not. The outcome was participation in cancer screening comparing patient navigation against usual care or other interventions. Screening participation could be extracted from health service records or as a self-report. Included study designs were controlled trials, cohort studies, repeat crosssectional studies, case-control studies, before-after studies and after only studies. Studies that were not original, reported several interventions or interventions targeting only populations at high risk of developing cancer (e.g., genetic/familial disorders) were excluded from the review.
Inclusion and exclusion criteria were piloted in 100 references before their application to the whole set of identified references, discussing any discrepancies until a consensus was reached among researchers. Two researchers (IM and LZ) independently assessed titles and/ or abstracts of the identified references using Covidence software, with a third (AC) in case of discrepancy. Two researchers (IM and LZ) read 40% of the full-text references and Cohen's Kappa statistic was used to measure the interrater reliability (IRR). As Kappa was higher than 0.7, the first reviewer assessed the remaining references.
After the selection of the included studies, the following information was extracted for each study and included in an Excel spreadsheet: period of analyzed data, country, cancer site, components of patient navigation 7 -'theoretical framework', 'programme goals', 'community characteristics', 'point of intervention', 'services provided', 'communication method', 'navigator background and qualification', 'training', 'supervision', and 'monitoring and evaluation (other than screening participation)' -participants -number, age group and percentage of females when applicable -measure of socioeconomic position of the population included and measure of socioeconomic position in the analysis if applicable, study design, comparison, screening method and main findings -outcome, including baseline characteristics.
The components of patient navigation examined were 14 instead of the 10 described by DeGroff et al., 7 as 'theoretical framework and programme goals' were broken down into two components, and 'services provided' by patient navigators were divided into four components based on the categories of barriers addressed. These categories were: (a) health system barriers (including scheduling appointments, paperwork and patient-provider communication), (b) individual barriers to cancer screening (lack of transportation, financial and insurance barriers, lack of childcare or language translation, low health literacy or low literacy), (c) educational barriers related to cancer and screening and (d) social and emotional barriers. 6 The data extraction was carried out by one reviewer (IM or SBM) and revised and completed by a second reviewer (SBM or IM). Disagreements were resolved by open discussion. If a consensus was not reached, a third reviewer was consulted (AC), and the majority decision was followed. In studies providing screening participation rate, the percentage increase in participation following the intervention was calculated.
The methodological quality of the included studies was assessed independently by two reviewers (IM and KM) through the application of the study quality assessment tools of the National Heart, Lung and Blood Institute for quantitative studies. 13 Studies were classified into three quality categories (good, fair and poor) based on criteria regarding study design, including randomization, blinding, drop-out and outcome measures, among others. Discrepancies were reconciled through discussion.
A meta-analysis of the studies was planned, but their heterogeneity hindered this, and findings were synthesized by means of a narrative synthesis. 14 The systematic review was registered with PROSPERO (CRD42020193657).

| RESULTS
The initial systematic search identified 5540 references to screen after taking out duplicates, and the updated search found an additional 308 references. After the application of inclusion and exclusion criteria, 5508 references were deemed not relevant for the topic of interest, resulting in 340 references selected to be read full text. Finally, 51 references on patient navigation were included (Figure 1), corresponding to 44 studies.
The quality of the included studies was mainly rated as poor (Appendix 2). The most frequent shortcomings were a lack of, or not reporting the justification of the sample size, a concealed allocation and a drop out higher than 20%.

| Components of patient navigation
Most studies (n = 37, 84.1%) had at least 9 out of 14 components of patient navigation. All studies described the goals of the navigation programme, community characteristics and point of intervention (Table 1, Appendices 3 and 4). Programme goals were generally to increase screening participation among disadvantaged populations.

| Effectiveness of patient navigation
Screening participation was reported based on medical record review in the majority of included studies, but some studies estimated participation based on self-reporting only, 22,26,35,43,44,50,62 while three used a combination of both methods. 8,17,34 Screening participation was measured over a period ranging from 14 days 49 to 30 months. 46 Patient navigation increased screening participation for the three cancer sites regardless of the measure of socioeconomic measure considered (  [15][16][17][18][19]22,26,31,[43][44][45][46]49,50,53,[56][57][58][59]63 these being mainly educational interventions. 15,18,19,22,26,[44][45][46]49,50,59,63 The proportion of participants assigned to patient navigation receiving the intervention ranged from 10.5% 33 to 96.1%; 54 few studies provided this information. Overall, the comparison of patient navigation with usual care (21 studies) or other interventions (10 studies) favoured navigation. Compared to usual care, screening participation was 19.9% 21 to 250.6% 29 higher for colorectal cancer, 0.4% 54 to 160.0% 58 higher for cervical cancer, and 33.6% 64 to 45.5% 62 for breast cancer. Regarding the comparison to educational interventions, participation with patient navigation was 3.3% 26 to 36.1% 22 for colorectal cancer, and 6.6% 50 to 3558.0% 44 for breast cancer. However, there were a few exceptions: screening participation was lower in patient navigation groups in comparisons against a health-literacy informed educational intervention in USA, 15 and against control and other interventions in UK. 56,57 Another study aimed to assess the impact of introducing patient navigation on social inequalities within the national organized screening programme in France. In this study navigation was found to be more effective in affluent than in deprived strata, entailing that if it was applied to the whole population, it has the potential to aggravate social inequalities in screening participation. 24,25

| DISCUSSION
To the best of our knowledge, this is the first study to systematically describe the components of patient navigation programmes in breast, cervical and colorectal cancer screening using a specific framework of this intervention. In this systematic review we have identified studies on patient navigation as a single intervention and have described its impact on screening participation when compared against usual care and educational interventions.
Patient navigation increased participation to screening in breast, cervical and colorectal cancer in comparison with usual care and educational interventions alone, in line with the findings from previous publications, 6,8,65 suggesting patient navigation can improve the effectiveness and outcomes of screening programmes, and advance in health equity. 65 While many studies included in this review have overcome several previously described limitations (e.g., lack of control group or of randomization to treatment or comparison groups), 6 there remains an issue of not having a single definition of patient navigation, which in any case is rarely provided.
Moreover, another finding from this work is that studies describing patient navigation interventions could be better reported. Although the 'who' (nurse, social worker, lay person, etc.), 'what' (what barriers are addressed), 'how' (communication channel used) and 'where' (setting) aspects of intervention delivery were often described, the intensity of the intervention (number of interactions between navigator and individual, schedule and length) was rarely reported. 66 This lack of transparency also applied to the control intervention, frequently usual care. Such incompleteness of data hindered the possibility of conducting a meta-analysis on the impact of patient navigation on breast, cervical and colorectal cancer screening participation, and its association with key components of patient navigation. Therefore, bearing in mind the great variation in the definition of patient navigation, we recommend a standardized reporting of its components that would allow comparison between studies, external validity, replication in different settings and ultimately a better measure of its impact on cancer screening participation. The better reporting of navigation programmes together with a consistent data collection would facilitate sustainability. 67 A positive finding was that over 84% of studies reported 9 or more components out of 14, being supervision and addressing of social and emotional barriers the least reported. As in a previous review, 68 the duration of training on patient navigation was quite diverse. We did not find a specific length reported to be adequate, as opposed to 3 days previously described for lay persons in cervical cancer screening. 69 Compared to a review conducted in the USA, 68 studies included in ours reported shorter duration of training (half a day vs 12 h). The maximum length of training could not be compared, as few studies reported if training was delivered over time or massed. 22,44,62 In any case, the length of training is generally linked to the background of trainees, and their expected roles and responsibilities. Another paper from the USA reported a consensus on the domains and competencies of the patient navigation training. 70 The topics described in the papers in our review included these competencies, but no study covered all.
The local context will determine the importance of each component to achieve a successful patient navigation programme. To plan patient navigation services, we need to know what the population eligible for the selected cancer screening requires through a needs assessment, as Ruggeri et al. did, 41 which will enable identifying which services it should include. 65 It is also possible to put in place a patient navigation programme and from its evaluation identify which are the most frequent barriers. Additionally, the least approached barriers were social and emotional, which are related to a lower screening participation 71 and could entail a delay in seeking medical help. 72 Moreover, although considered a requirement, 73 supervision of navigators was seldom described, being more frequently reported in programmes addressing multiple cancer sites screening, probably because of their complexity.
All studies included in this systematic review except one were conducted in high-income countries. The implementation of patient navigation programmes has been scarcely reported outside USA, 74 including LMICs. A recent scoping review on this intervention in LMICs in cancer care focused mainly on tertiary level, with only one study on screening. The main services reported were facilitation of the linkage to follow-up services, coordination of appointments and education to ensure understanding of symptoms and signs. Interestingly, few studies labelled their intervention as patient navigation. Studies evaluating navigation in cancer care reported mainly implementation science outcomes, such as the acceptability, fidelity and feasibility of the intervention, 75 rarely described in our included studies. Due to a high variability in health care systems across the world, there are limitations to applying the results from high-income countries to LMICs. More research is needed in these settings to understand patient navigation in cancer screening with a global perspective.
Previous systematic reviews focused on USA and Canada only. 6,8 The inclusion of studies conducted in countries other than these two is one of the strengths of our systematic review. Other strengths are the search in three databases, including both medical and social databases, no language restriction and the use of a framework to assess the reporting of elements of patient navigation programmes.
The main limitation of this research is that the systematic search was not initially developed to assess patient navigation, as it included only "patient navigation" and "patient-centred care" as search terms. However, when assessing the full texts, we were broad in the consideration of the term, as screening practitioners, prevention managers and care gap analysts have been included as navigators. Additionally, the inclusion of studies measuring self-reported participation to screening may have overestimated the impact of navigation.
In conclusion, patient navigation is effective in increasing participation in breast, cervical and colorectal cancer screening, which can improve the effectiveness and outcomes of screening programmes. A standardized reporting of patient navigation and its components would allow its replication and a better measure of its impact. The local context and needs will determine the importance of each component and will enable the design of a successful patient navigation programme.

ACKNOWLEDGMENTS
This publication is supported by a grant awarded by the Norwegian Research Council (project number 288638) to the Center for Global Health Inequalities Research (CHAIN) at the Norwegian University for Science and Technology (NTNU).

CONFLICT OF INTEREST
The authors declare no conflicts of interest.

TABLE OF INCLUDED STUDIES BY CANCER SITE
The studies are provided in alphabetical order. However, when more than one paper describes a study, they have been placed consecutively.