From participation to diagnostic assessment: a systematic scoping review of the role of the primary healthcare sector in the National Bowel Cancer Screening Program

Additional keywords: preventive medicine, quality of health care, secondary prevention.

Screening of average-risk adults (from age 50 to 74 years) for colorectal cancer (CRC) contributes to reduced mortality (Australian Institute of Health and Welfare 2019). Acknowledging that the National Bowel Cancer Screening Program (NBCSP) is not fully implemented, participation has yet to reach the desired rate to achieve maximum benefit, particularly with some population groups being under-screened or never screened (Australian Institute of Health and Welfare 2019). A range of external constraints restricted the implementation process (Flitcroft et al. 2010), with limited involvement of primary health care (PHC) in the program design despite the eligible patient cohort (50–74 years) visiting a GP at least six times each year (Australian Institute of Health and Welfare 2018). To enhance the vital role that PHC plays in realising the benefits of screening (Cole et al. 2002; Zajac et al. 2010), more practical guidance is needed to support the fundamental role of the PHC sector in preventive and quality care (NBCSP 2016) along the CRC screening pathway.

The CRC screening pathway is characterised by multiple interfaces of care across different providers and settings, creating complexities in implementation (Zapka et al. 2010). In Australia, this is compounded by Federal and State Government boundaries implicit in a patient’s participation in the NBCSP. The role of PHC in CRC screening is similar irrespective of whether CRC screening is undertaken as routine quality care or part of an organised population-based screening program, with identification of eligible patients and endorsement and completion of screening consistent with evidence-based clinical guidelines (Emery et al. 2014). Numerous systematic reviews provide information to assist the PHC sector to improve screening participation, but many only review interventions targeting one stage of the screening pathway; for example, recruitment, whereas identifying interventions that have relevance along the entire screening pathway is expected to enhance the effectiveness of the population-based screening program.

This systematic scoping review examines provider- and practice-based interventions that support the role of the PHC sector that align with stages of the NBCSP and require the explicit involvement of GPs and their practice staff; namely, identifying and reminding patients who have not responded to CRC screening (non-adherence) and follow up of an iFOBT (immunochemical faecal occult blood test) and referral to diagnostic services, if required. A systematic scoping review was considered the most appropriate form of review to address the research question: What are the patient, professional and system-level interventions implemented in PHC settings (Interventions) that improve CRC screening completion (Outcomes) of non-adherent, eligible patients (Population), compared with baseline or a control group (Comparison). The review purpose was to identify future practice and research priorities to improve the effectiveness of CRC screening through strategies that allow better integration of the NBCSP with PHC in Australia.

Five databases (MEDLINE, PsycINFO, Embase, CINAHL and PubMed) were selected for the scoping review (Arksey and O’Malley 2005; Levac et al. 2010), as these were expected to contain relevant studies. The latest search was undertaken in November 2019 for articles from 2005, to coincide with the NBCSP implementation, to the date the search was re-run (19 November 2019). The keywords and medical subject headings specified in Appendix 1 were used. Additional studies were identified through reference tracking of systematic reviews, clinical guidelines and other key papers identified through the database searches.

For ~10% articles, two investigators (C. A. Holden, J. Caruso) reviewed the same subset of titles and abstracts, achieving 92% agreement. Both reviewers read full-texts to determine their eligibility when the inclusion and exclusion criteria were applied. Disagreements between reviewers were resolved by consensus-based discussion.

The NBCSP Quality Framework (NBCSP 2016) was used to identify stages that specifically require involvement of the PHC sector; namely, optimised recruitment participation (i.e. recruitment of non-adherent, eligible patients) and follow-up assessment (following a positive iFOBT). Interventions that were directed at these stages of opportunistic or organised CRC screening programs were included as eligible studies (Fig. 1). Studies with no family physician/GP involvement were excluded.

Fig. 1.  Primary care involvement across the National Bowel Cancer Screening Program (NBCSP 2016). Schematic diagram modified from the National Bowel Cancer Screening Program Quality Framework, version 2 (NBCSP 2016). ^AA positive (+) result means that blood was detected in the completed immunochemical Faecal Occult Blood Test (iFOBT). ^BA positive colonoscopy is identified by reporting one of the following: tubular adenoma, tubulovillous adenoma, villous adenoma, sessile serrated adenoma, traditional serrated adenoma, adenoma not otherwise classified, or carcinoma.

To be eligible, the study design was limited to randomised, quasi-randomised or controlled trials published in English. Observational studies were excluded during full-text review as the search identified sufficient controlled trials (saturation point). Systematic reviews, meta-analyses and clinical guidelines were excluded, but contributed additional studies to review from their reference lists.

Studies from countries where a population-based CRC screening program is established were included if the intervention was implemented in family practice/primary care (Schreuders et al. 2015). Interventions were included if they were: (1) aimed at increasing CRC screening participation of non-adherent patients (improved adherence and diagnostic follow up); (2) implemented in primary/general practice settings; (3) focussed on asymptomatic patients eligible for population-based screening (who had not previously participated or were from underserved population groups); and (4) were not an established component of an existing organised screening program (e.g. personal invitation, advanced notification letters, iFOBT kit mail-out etc.). The primary outcome of interest was CRC screening completion, which needed to be reported quantitatively and derived from medical records or administrative data for the study to be included. We reported pilot studies or studies reporting different analyses of the same intervention as a single study.

Articles were excluded if they: (1) failed to meet the inclusion criteria; (2) were randomised trials comparing different screening methods; or (3) assessed interventions that involved surveillance colonoscopy or follow up after cancer treatment.

Data extracted for mapping and analysis included author, year, study country, study design, alignment with NBCSP stage, sample size, primary outcome measure and the population group if specifically defined. We categorised the intervention level and intervention type using a previously defined taxonomy of patient, practitioner and system-level interventions (de Silva and Bamber 2014) to allow reporting consistency. Whether the intervention was part of a quality-improvement (QI) process was also recorded. Characteristics and findings of included articles are summarised in Table 1.

Table 1.  Characteristics and findings of included articles (n = 57)
RCT, randomised control trial; CRC, colorectal cancer; FOBT, faecal occult blood test

Studies were not appraised for quality, as the primary purpose was to extract and map the available data in line with systematic scoping review methods (Arksey and O’Malley 2005; Levac et al. 2010). However, we attempted to assess effectiveness in changing the primary outcome using criteria to classify study outcomes and applied to score intervention effectiveness previously used by Leykum et al. (2007). The criteria and accompanying rating scale addressed study heterogeneity and differences in the unit of analysis and unit of randomisation between studies (e.g. comparison with baseline values or control groups). The criteria and rating scale described by Leykum et al. (2007) was used to classify study outcomes and applied to score effectiveness of interventions described. In summary, scores of 0 (no effect), 0.5 (mixed results) and 1 (effective intervention) were applied to the reported statistical significance of study outcomes. Where possible, results adjusted for potential confounders were used to determine effectiveness.

Of 2674 articles, 57 unique studies were included in the review (Fig. 2).

Fig. 2.  PRISMA flow diagram of study selection. NBCSP, National Bowel Cancer Screening Program.

The 57 eligible studies were conducted in the USA (n = 42), Canada (n = 5), Europe (n = 5), UK (n = 3), Korea (n = 1) and Australia (n = 1). Aligning to NBCSP stages, most studies reported interventions targeting non-adherence to optimised recruitment participation (n = 46), with 11 studies targeting follow-up stages, namely positive screen follow up (n = 8) and colonoscopy referral (n = 3). Four of these studies explored GP–Program interaction, in which an organised screening program supported family practice/primary care in monitoring/care processes. Approximately half (n = 28) of eligible studies focussed on interventions that improved screening participation of specific population subgroups that tend to be under-screened or never screened. An overview of study characteristics is summarised in Table 1.

A quality framework of patient, professional and system-level interventions (de Silva and Bamber 2014) was applied to categorise interventions. The 57 studies yielded 24 different interventions around screening participation; 11 directed at the patient level, six at the professional level and seven at the system/organisational level. Eighteen studies included multiple interventions at several levels.

Most identified studies (n = 46) focussed on interventions that optimised screening participation (non-adherence), with most exploring patient-level interventions; that is, those interventions targeting the patient that were generated from the practice. Most patient-level interventions focussed on self-management support systems, with education activities (Walsh et al. 2005; Sequist et al. 2009; Aragones et al. 2010; Dietrich et al. 2013; Green et al. 2013; Jerant et al. 2014), lay-person support structures (mostly lay-person patient navigators) (Fiscella et al. 2011; Lasser et al. 2011; Jandorf et al. 2013; Leone et al. 2013; Shankleman et al. 2014; Reuland et al. 2017) and health coaching/counselling (Myers et al. 2007; Fiscella et al. 2011; Menon et al. 2011; Davis et al. 2013; Basch et al. 2015; Temucin and Nahcivan 2018) reported as mostly effective in improving adherence in organised CRC screening programs. Reminders for screening (Walsh et al. 2005; Myers et al. 2007; Fiscella et al. 2011; Dietrich et al. 2013; Green et al. 2013; Leone et al. 2013; Baker et al. 2014; Cohen-Cline et al. 2014; Hendren et al. 2014; Phillips et al. 2015; Benton et al. 2017; Hirst et al. 2017; Kiran et al. 2018; Sun et al. 2018; Dodd et al. 2019; using different formats, e.g. text messaging (Hirst et al. 2017), GP-endorsed mail-outs (Benton et al. 2017; Kiran et al. 2018) and automated telephone calls (Phillips et al. 2015), or a combination) were also mostly effective in improving adherence in organised CRC screening programs for both eligible patients not up-to-date with screening and under-screened population subgroups. Involving patients in decisions (e.g. via decision aids) reported mixed results (Pignone et al. 2011; Price-Haywood et al. 2014; Reuland et al. 2017; Jimbo et al. 2019). Interventions that aimed to improve access to care, such as outreach programs for vulnerable/marginal groups and young people (Potter et al. 2011; Baker et al. 2014; Shankleman et al. 2014) tended to be mostly effective in improving adherence to CRC screening.

Nineteen eligible studies included professional-level interventions to improve adherence to CRC screening programs, although most (n = 15) simultaneously included patient- or system-level interventions as a multi-component study. In a comparable way to studies that test the effectiveness of patient reminders for screening, point-of-care prompts to the physician (at the time of the patient consultation) also tended to demonstrate improvements in adherence to CRC screening in primary care (Sequist et al. 2009; Aragones et al. 2010; Fiscella et al. 2011; Hendren et al. 2014; Guiriguet et al. 2016; Dodd et al. 2019). Training initiatives targeting professionals reported mixed results (such as training in specific tools or conditions; Fiscella et al. 2011; Maddocks et al. 2011; Basch et al. 2015; Sun et al. 2018) and training in communication skills, cultural competency, patient involvement, support to self-manage etc. (Price-Haywood et al. 2014; Aubin-Auger et al. 2016; Huei-Yu Wang et al. 2018), although interprofessional training delivered by peers and through academic detailing, tended to report more effective outcomes.

Only nine eligible studies included system-level interventions to improve adherence to CRC screening programs. There were too few eligible studies to determine intervention effectiveness; however, improvement initiatives (Ling et al. 2009; Ornstein et al. 2010; Mader et al. 2016) tended to report improved adherence. Similarly, while there were too few system-level studies to determine the specific effectiveness of computer systems for sharing information within and across organisations, most studies that focussed on system-level interventions included computer systems that supported the intervention delivery, suggesting that health informatic approaches improved quality of care.

Despite the role of primary care services in ensuring that patients who receive a positive screen result are referred appropriately for further assessment, only one-quarter of eligible studies (n = 11) explored primary care interventions that improve diagnostic patient follow up and approximately half of these included a system-level intervention (n = 6).

Positive screen follow up. Only eight eligible studies focussed on improving patient follow up with a positive initial screen. Most were single-level interventions; that is, across only patient (n = 4), professional (n = 1) or system (n = 2) levels.

Most studies explored the effectiveness of patient navigators to improve positive screen follow up across both patient level (lay-person support services; Paskett et al. 2012; Raich et al. 2012; Freund et al. 2014) and professional level (as new staff roles within the family practice/primary care setting; Green et al. 2014). Other interventions that explored positive screen follow up included system-level interventions that focussed on electronic referral systems (Humphrey et al. 2011), QI and facilitation projects (Wei et al. 2005) and external reminder systems (Stock et al. 2017). Although there were too few eligible studies to determine effectiveness of some interventions, patient navigators and QI initiatives tended to report improved outcomes.

Colonoscopy referral. Only three eligible studies focussed on interventions to improve colonoscopy referral. All reported system-level change interventions including improvement initiatives (such as continuous QI projects, including audit and feedback (Singh et al. 2009), and improvement collaboratives (Powell et al. 2011)) and service provision (such as improvements to referral letters (Lebwohl et al. 2011)), acknowledging that there were insufficient studies to determine effectiveness in the context of colonoscopy referral. Patient-level interventions (layperson-led support services, patient navigators; Lebwohl et al. 2011) targeting colonoscopy referral were implemented with other system-level activity as a multi-level intervention.

Despite many organised population-based CRC screening programs worldwide, all requiring primary care involvement (to different extents), there were relatively few eligible studies that specifically tested interventions that target GP–Program interaction activity to improve CRC screening completion. Two studies from Canada (Jonah et al. 2017; Stock et al. 2017) and two from France (Le Breton et al. 2016; Rat et al. 2017) had mixed results. All used system-level activities (i.e. reminder systems (external organisations reminding practices about specific monitoring/care processes) at different points of the screening pathway, with the Canadian studies (Jonah et al. 2017; Stock et al. 2017) also incorporating physician audit and feedback.

The scoping review identified that studies including a continuous QI element reported greater effectiveness (Table 2).

Table 2.  Intervention effectiveness relating to the quality-improvement (QI) process (Leykum et al. 2007)

Nineteen studies reported that interventions aimed at improving CRC screening participation were implemented as a QI process, but details of the QI model were not reported in one study (Cha et al. 2011). With one exception (Leone et al. 2013), all studies (n = 18) reported effectiveness or trends towards improvements in outcome measures when the intervention was implemented as a defined QI initiative or within an existing QI process. Most studies addressed optimised recruitment participation (n = 14); however, interventions addressing follow-up stages were also identified.

Where QI process detail was described (n = 18), in eight it was part of a QI framework, but it was difficult to determine which elements contributed to the overall study outcomes. The same applied to an additional four studies where the intervention was implemented within an existing named QI program. Only seven studies explicitly indicated that the CRC screening participation intervention was a defined improvement initiative and detailed the elements of the model applied.

This systematic scoping review highlights the PHC practice and research opportunities to improve CRC screening participation, particularly for non-adherent, eligible patients in the context of a population-based bowel cancer screening program. This review made a distinction between optimised recruitment participation (non-adherence) and follow up, where an organised screening program may moderate the PHC role; better integration of CRC screening into existing primary care services (prevention and quality care interventions) along the entire screening pathway may maximise the benefits of population-based screening programs.

In the large number of systematic reviews (including meta-analyses) and empirical studies that focus on PHC interventions that enhance CRC screening participation, most studies focus on identifying and reminding patients who have not responded to CRC screening (non-adherence). Few studies consider the PHC role in the full CRC screening pathway, despite the important role the primary care service has in follow up and referral for diagnostic services, if required. Emery et al. (2014) provides the most comprehensive analysis of the primary care role to support cancer screening and management, including follow-up diagnostic assessment, albeit across several cancer types. The alignment of interventions with NBCSP stages that require specific PHC involvement is a unique perspective of our review. This approach identifies practice opportunities and research gaps in ensuring patients complete the screening pathway, particularly if in the NBCSP.

Acknowledging that observational studies were excluded, a significant gap identified is the dearth of high-quality Australian studies investigating interventions that specifically address opportunities for PHC to address patient screening non-adherence and follow up. Most Australian research has focussed on interventions relevant to an organised screening program, without reference to the essential role of primary care services, with interventions that the NBCSP has already implemented (e.g. advanced notification) and/or compared screening test efficacy (which were excluded from our search criteria). Without robust Australian studies, the generalisability of the review findings to the Australian setting might be limited. However, this finding also identifies opportunities and a strong need for more Australian research in this area, specifically to study interventions that can be implemented in primary care services to complement the NBCSP rather than developing parallel systems to improve bowel cancer screening participation.

Interventions are categorised according to different quality care levels for easier incorporation into existing QI processes, which have been shown to be more effective in achieving change in routine clinical practice (Grol and Grimshaw 2003). However, most studies report interventions as discrete activities and on only one element of the screening pathway (e.g. recruitment), which may not readily integrate with existing QI practice in primary care services. Furthermore, most reviews investigate interventions aimed at earlier participation stages with fewer exploring diagnostic follow up of positive screening tests (Selby et al. 2017). Without explicit PHC engagement in screening programs, alternative and individualised practice-based processes are adopted that attempt to work alongside, but potentially diminish the effectiveness of organised screening programs. This review moves beyond studies that explore the practitioner influence on screening participation and instead focuses on how PHC can facilitate (non-adherent) eligible patients to participate in CRC screening.

Some interventions demonstrate benefits across both the screening (non-adherence) and diagnostic follow-up pathway. These include improvement initiatives (such as QI initiatives, including facilitation/audit and feedback (system level)) and self-management support initiatives (such as patient navigators (patient and professional level)). Consistent with other reviews (Klabunde et al. 2007; Zapka et al. 2010; Emery et al. 2014), reminders for screening and point-of-care prompts are important interventions for optimising recruitment participation; however, their effectiveness for subsequent screening stages is not known. The effectiveness of alternative reminder systems, such as external organisations (e.g. the National Cancer Screening Register, NCSR, or equivalent) may offer substitute reminders across the screening pathway, but their effectiveness in the context of the NBCSP needs testing. This review confirms that interventions targeting multiple levels of quality care represent more effective strategies to improve CRC screening participation (Senore et al. 2015). Opportunity exists to align CRC screening participation efforts with routine primary care QI processes. The revision of the Practice Incentive Payment (PIP) (which encourages general practices, through additional government payments, to continue providing quality care (Australian Government Department of Human Services 2019)) to include CRC screening (a national cancer priority) within a quality care model might further support a primary care role in the NBCSP. Identifying practice priorities that streamline the patient experience across the screening pathway and avoid duplication of organised screening programs, is expected to improve the NBCSP effectiveness and overall patient care.

A limitation of this review is the focus on an organised population-based screening program, rather than CRC screening more broadly for the eligible population. However, the findings are relevant to whether screening is undertaken in private practice or through an organised screening program, given the role of primary care services in non-adherence and preventive care follow up. Furthermore, limiting the search to publications post 2005 and excluding observational studies, might have resulted in potentially relevant studies being excluded. Publication bias, where studies with null results are less likely to be submitted or accepted for publication, may overestimate intervention effectiveness. However, as almost half (47%) of the included studies reported null or mixed outcomes, the effect of publication bias is likely to be very low. Studies that were not specific to CRC tended to report combined effectiveness of a single intervention across all screening programs, making it difficult to determine the effectiveness of included interventions. The effectiveness categorisation that we used was our attempt to overcome these limitations to determine the intervention effectiveness when specifically applied to CRC screening.

Most studies identified in this review evaluated single screening elements, despite evidence that interventions incorporating multi-component or QI practices tend to be more effective strategies, particularly if they do not require clinical staff involvement (Klabunde et al. 2007; Zapka et al. 2010; Senore et al. 2015). Future research needs to focus on QI practices targeting CRC screening that effectively bridge the gap between organised population-based screening programs and ‘usual care’ delivered in primary care services. In this context, the review highlights the untapped opportunities and benefits that the NCSR may offer to seamlessly engage and support the PHC sector to undertake CRC screening through digital solutions and overcome external constraints that have restricted the NBCSP implementation process to date (Flitcroft et al. 2010).

In summary, our review points to a potential opportunity to enhance the PHC role to maximise the benefits of population-based bowel cancer screening programs through existing primary care preventive and QI initiatives. As noted by Dodd et al. (2019), the possibility exists for PHC in Australia to adopt an important ‘adjunct’ role to support the NBCSP along the entire screening pathway, particularly for those asymptomatic, eligible patients who are more difficult to reach. The NBCSP cost-effectiveness warrants the investment in evidence-based strategies to improve screening adherence, particularly those that target improved CRC screening and follow up in primary care services (Worthington et al. 2020). As others have noted (Zapka et al. 2010), the NBCSP needs to invest in provider- and system-level strategies that ‘bridge the care transitions across primary and hospital-based services’, from screening to diagnosis and possible treatment.

Authors O. Frank, J. Caruso, D. Turnbull, R. L. Reed have nothing to disclose. C. L. Miller reports grants from the Medical Research Future Fund, grants from Beat Cancer Project and grants from the National Health and Medical Research Council during the conduct of the study and grants from the Australian National Data Service, outside the submitted work; C. A. Holden and I. Olver report grants from Cancer Council SA Beat Cancer Translational Research Scheme, during the conduct of the study.