Improving colorectal cancer screening in rural primary care: Preliminary effectiveness and implementation of a collaborative mailed fecal immunochemical test pilot

Abstract

Background:

Mailed fecal immunochemical test (FIT) outreach can improve colorectal cancer (CRC) screening rates. We piloted a collaborative mailed FIT program with health plans and rural clinics to evaluate preliminary effectiveness and refine implementation strategies.

Methods:

We conducted a single-arm study using a convergent, parallel mixed-methods design to evaluate the implementation of a collaborative mailed FIT program. Enrollees were identified using health plan claims and confirmed via clinic scrub. The intervention included a vendor-delivered automated phone call (auto-call) prompt, FIT mailing, and reminder auto-call; clinics were encouraged to make live reminder calls. Practice facilitation was the primary implementation strategy. At 12 months post mailing, we assessed the rates of: (1) mailed FIT return and (2) completion of any CRC screening. We took fieldnotes and conducted postintervention key informant interviews to assess implementation outcomes (eg, feasibility, acceptability, and adaptations).

Results:

One hundred and sixty-nine Medicaid or Medicare enrollees were mailed a FIT. Over the 12-month intervention, 62 participants (37%) completed screening of which 21% completed the mailed FIT (most were returned within 3 months), and 15% screened by other methods (FITs distributed in-clinic, colonoscopy). Enrollee demographics and the reminder call may encourage mailed FIT completion. Program feasibility and acceptability was high and supported by perceived positive benefit, alignment with existing workflows, adequate staffing, and practice facilitation.

Conclusion:

Collaborative health plan-clinic mailed FIT programs are feasible and acceptable for implementation in rural clinics and support CRC screening completion. Studies that pragmatically test collaborative approaches to mailed FIT and patient navigation follow-up after abnormal FIT and support broad scale-up in rural settings are needed.

INTRODUCTION

Colorectal cancer (CRC) is the second leading cause of cancer-related deaths in the United States.¹ Yet, CRC is 90% curable with timely detection and appropriate treatment of precancerous growths.² Numerous CRC screening modalities are recommended for adults aged 45–75 years old, including colonoscopy every 10 years or fecal immunochemical tests (FITs) annually for asymptomatic adults at average risk (eg, no family history of CRC).³ Despite growing support for at-home FIT testing, only about 69% of the population has been screened for CRC,⁴ and the majority with colonoscopy.^5,6

Screening participation is 15–30 percentage points lower in rural and low-income (Medicaid) populations, with additional disparities for racial/ethnic population subgroups.^7–9 These disparities are likely due to complex interactions of access to and utilization of health care, screening behavior differences, and risk factors for CRC, such as obesity, tobacco use, eating red meat, and physical inactivity.^8,10–12 In rural areas specifically, lack of prevention attitudes, privacy concerns, shortage of specialists, and distance to test facilities are identified barriers to CRC screening.¹³ Uninsured and Medicaid-insured cancer patients are found to present with more advanced disease and receive appropriate follow-up cancer care less often, and many providers do not accept Medicaid patients at all.^14–16

Mailed FIT outreach programs reduce structural barriers to screening completion and may help alleviate observed CRC screening disparities.^9,17–19 However, the ability of clinics and health plans to implement mailed FIT programs and the effectiveness of these programs may be impacted by clinic characteristics (eg, size and geographic region) or program adaptations (eg, reminder calls and incentives).^9,20,21 Most studies on mailed FIT outreach programs have been conducted in urban settings and large health systems.⁹ A handful of studies suggest that mailed FIT is effective in rural settings when delivered by research staff.^22–24 Our team previously reported on a mailed FIT outreach program using a centralized health-plan-led model; findings showed lower 6-month CRC screening completion rates in rural compared to urban residents (18% vs 22%).²⁵ Few studies explore mailed FIT implementation in alignment with existing structures in rural primary care or suggest how to tailor implementation support during key project phases.²⁶

To overcome barriers to adoption of cancer control interventions, experts recommend using community-based participatory research (CPBR) approaches.^27–31 Participatory implementation science blends CBPR and implementation science to support iterative, ongoing engagement between community stakeholders and researchers to create system change and ultimately achieve health equity.²⁹ Partnerships between academics and clinic/community members via practice-based research networks (PBRNs) support community engagement with the goal of enhancing the identification, uptake, adaptation, and sustainability of effective interventions in routine practice.^28,32–34 Many PBRNs employ practice facilitators to assist clinics in research and quality improvement projects to move knowledge into practice.^35–37 Practice facilitation has been shown to increase the awareness of evidence-based practices, engage leadership, improve team function, and support implementation progress.^35,36,38,39 In addition to direct partnerships with clinics, cross-sector collaborations with regional health plans can help overcome challenges to mailed FIT implementation and sustainment by enabling patient identification, program coordination, and may alleviate costs or create economies of scale for small, rural, or independent clinics.^25,40

Additional research is needed to explore how mailed FIT outreach can be successfully implemented as part of routine care in rural settings through collaborative partnerships. Therefore, we conducted this mixed-methods pilot study to refine strategies to support the implementation of a mailed FIT outreach program in rural settings using a participatory implementation science approach.²⁹ The study team facilitated the implementation of a collaborative payer-clinic mailed FIT intervention with 1 regional health plan and their rural primary care clinics. We assessed mailed FIT response rates and overall CRC screening completion, descriptively assessed patient and clinic characteristics and test completion, and evaluated program feasibility and acceptability and factors influencing implementation.

METHODS

We conducted a single-arm study using a convergent, parallel mixed-methods design between August 2017 and January 2020 titled “Using Context to Improve Implementation of Evidenced-based Interventions for Colorectal Cancer Screening in Rural Primary Care.” In a convergent, parallel mixed-methods design, quantitative and qualitative elements are conducted in the same phase of the research process, weighted equality, analyzed independently, and interpreted together.⁴¹ Study activities were approved by the Oregon Health & Science University (OHSU) Institutional Review Board (17952 and 20089).

Study setting/partners

This study was developed collaboratively by an OHSU researcher (MMD) and the Community Health Advocacy and Research Alliance (CHARA) and conducted as a partnership between the Oregon Rural Practice-based Research Network (ORPRN), the regional health insurance plan (PacificSource), and rural primary care practices in alignment with a participatory implementation science approach.²⁹ CHARA is a coalition of community members, health system stakeholders, and academics who use research to address regional health priorities,^42–44 including improved access to fecal testing in Medicaid enrollees.⁴⁵ ORPRN is a PBRN with a mission to improve health outcomes and equity for all Oregonians through community-partnered research.²⁸ PacificSource is a not-for-profit regional health plan that insures nearly 524,000 people in Oregon, Idaho, and Montana, including nearly 300,000 enrollees under Oregon’s Medicaid Coordinate Care Organization (CCO) model.^6,40,46 At the time of this study, over 60,000 Medicaid enrollees were located in the 2 participating PacificSource CCO regions: Central Oregon and the Columbia Gorge. During recruitment and implementation (Fall 2017–2018), financial incentives for achieving performance targets for CCO and for the Medicare STARS program provided external motivation to increase CRC screening rates.

Participating clinics

We worked with the health plan to identify and recruit clinics for study participation; our target recruitment goal for the pilot was 4 clinics. Eligible clinics met the following criteria: (1) located in a rural area as defined by rural-urban commuting area codes (≥ 4);⁴⁷ (2) had 10 or fewer primary care clinicians (small- to medium-sized); (3) provided care to adult Medicaid patients; (4) used an electronic health record (EHR); and (5) were not currently implementing a mailed FIT outreach program.

Patient eligibility

We initially defined patients as eligible for the outreach program if they were Medicaid or dual eligible enrollees (Medicaid-Medicare) with PacificSource, aged 50–75, spoke English or Spanish, and were not up-to-date with CRC screening according to the US Preventive Services Task Force guidelines, and with no prior history of CRC or colectomy.^1,48 We expanded to include Medicare enrollees based on the request of health plan and clinic partners. PacificSource pulled an initial list of eligible enrollees using claims data, which was subsequently reviewed by clinics to produce the final mailed FIT deployment list (see detail below).

Intervention model

A timeline and summary of intervention activities appears in Appendix S1. Some intervention components were centralized and others involved collaboration between the health plan and participating clinics.²⁵ Specifically, the health plan partnered with a third-party vendor to support centralized delivery of an automated phone call (auto-call) prompt prior to the mailing, conduct the FIT mailing, and complete reminder auto-calls. Clinics were also encouraged to make follow-up reminder calls 2 weeks after the mailing given prior evidence on optimizing FIT return rates.^9,49,50 An ORPRN practice facilitator worked with the health plan and clinics to implement all program activities. Health plan collaborators included a project lead, regional PacificSource practice facilitator, and data analysts. A clinician champion and staff point of contact was identified for participating clinics. The ORPRN practice facilitator supported clinic-level review and approval of all mailed FIT materials; provided training, monitored clinic scrubbing of the patient lists; and provided regular updates and technical assistance to clinic and health plan partners.

The health plan used claims data to identify an initial list of patients meeting eligibility criteria and subsequently refined to produce the final mailed FIT deployment list. First, the health plan removed members if they had no address or phone number on file or were on the health plan no call list. Next, similar to prior studies, over a 2-week period participating clinics “scrubbed” the list to remove patients who were up-to-date based on prior CRC screening or had medical conditions that made them poor candidates for FIT.^20,25 Clinics also removed patients who were assigned to their clinics but had not yet established care, voicing concerns about challenges reaching these patients if their FIT results were abnormal. Finally, the health plan shared the final deployment list with a third-party vendor who made an auto-call prompt about 1 week prior to the mailing followed by the mailed FIT kits (information sheet, FIT, and cover letter) to enrollees on the list. The vendor distributed OC FIT-CHEK^® (a version of OC Auto by PolyMedco [Cortland, Manor, NY]). Clinics were encouraged to make follow-up reminder calls within 2 weeks and the vendor made an auto-call to nonresponding patients 4 weeks after the initial mailing. While a 1-time mailing was planned in July 2018, during implementation, the health plan conducted 3 smaller mailings between July and December 2018 with patients from each participating clinic included in each round (see Appendix S1).

Data collection and measures

Quantitative data and preliminary effectiveness

Quantitative data were obtained from the health plan (eligible enrollee listing), participating clinics (scrubbed listing and chart audit), and vendor (mailed FIT date, return, and results). We defined preliminary effectiveness as the percentage of eligible enrollees who, as of 12 months from the initial mailing, had completed (1) the mailed FIT or (2) any CRC screening (eg, colonoscopy, Cologuard, FIT via mailing, or clinic distribution).

Health plan data and vendor data were collected between January 2018 and December 2018 in alignment with the mailings. Health plan data included patient insurance status, gender, assigned primary care provider, and patient risk level. Patient health risk was defined using Truven Verisk risk scores,⁵¹ which uses a proprietary model that combines diagnosis types, complexity, and member utilization patterns to classify patients from healthy, to struggling, to in crisis (E. Towey, personal communication, July 31, 2018). We used vendor data to determine the number of eligible enrollees who returned a mailed FIT. The chart audit was conducted between September 2019 and January 2021, 12 months after the final mailing, by a trained study team member who was not involved in delivering the intervention. The chart audit was used to confirm that mailed FIT results were recorded in the clinic EHR, to identify other CRC screening completed by eligible enrollees (eg, clinic-distributed FIT, colonoscopy, and Cologuard), and to determine CRC screening results and follow-up care for all eligible enrollees. Chart audit data were collected in REDCap (Fort Lauderdale, FL), an electronic data capture tool hosted at OHSU and accessible via a secure web-based interface.^52,53

Qualitative data and implementation outcomes

Qualitative data were used to assess program participation and implementation outcomes, including program operationalization, implementation barriers and facilitators, and feasibility and acceptability.⁵⁴ We collected fieldnotes from each clinic and health plan interaction and conducted post intervention key informant interviews. The ORPRN practice facilitator documented all interactions with participating clinic and health plan partners in an online contact-tracking database. The tracking database was informed by the work of Ritchie and colleagues and included meeting type, mode, personnel involved, and facilitation activities.^55,56 Interactions were documented between August 2017 through March 2019 in alignment with the Engagement-Preparation-Implementation-Sustainability Stages, see Appendix S1.²⁶ Post intervention key informant interviews were conducted between April 2019 and August 2019 by a trained study team member who was not involved in intervention delivery. Key informants included health plan leadership, the ORPRN practice facilitator, and a clinician and staff members from each clinic; these 1:1 interviews were conducted in person or virtually depending on participant preference. The semistructured interview guide explored factors related to participation, implementation experience, intervention adaptations, program feasibility and acceptability, and future recommendations (see Appendix S2). Participants provided verbal consent prior to each interview. Interviews lasted between 30 and 45 minutes and were audio recorded and professionally transcribed. Participants received a $50 gift card for their time.

Data analyses

Quantitative analysis

We analyzed quantitative data descriptively to explore the variation in response patterns based on clinic and patient-level characteristics. We explored CRC screening completion over time using Kaplan-Meier failure estimates, with a goal of understanding patterns in mailed FIT return compared to CRC screening by clinic-distributed FIT or other modalities. Because this study is an exploratory pilot and cell sizes for many subgroups were small, no formal hypothesis testing (ie, tests for statistical significance) was conducted. Data were transferred into Stata version 16.1 for data management and analysis.

Qualitative and mixed-methods analysis

We uploaded transcripts and fieldnotes into ATLAS.ti version 8 for data management and analysis. We applied a template analysis approach to understand how clinic and health plan context shaped implementation outcomes and screening completion rates.⁵⁷ First, the PI/lead author (MMD) and the project manager (MP) read all interviews and then generated a priori codes associated with the research questions (eg, feasibility, acceptability, and adaptations). They coded each transcript, identifying patterns at the setting level and inductively adding codes, as needed (eg, staff transitions and prior relationships). Second, we analyzed findings across settings, noting patterns in the inner (eg, capacity and workflows) or outer setting (eg, health plan support) associated with implementation. Finally, we produced a joint display of quantitative and qualitative data to identify factors associated with program implementation outcomes, especially adaptations defined as changes made to the mailed FIT intervention to align with local workflows (eg, scrubbing process and reminder calls).⁵⁸ Emergent themes were refined with the larger project team and shared with the study’s advisory board. We also shared findings with clinic and health plan key informants for member checking.^59,60 These methods (eg, reflexivity, multiple reviewers, and audit trail) are associated with rigor in qualitative research.^59–62

RESULTS

Seven rural primary care clinics meeting eligibility criteria and recommended by health plan partners were approached to participate; 3 agreed (43%). Two clinics declined because they preferred to focus on visit-based workflows and 2 preferred other mailed FIT arrangements (eg, completed by local lab). The 3 participating clinics were all Federally Certified Rural Health Clinics located in health professional shortage areas. One clinic was independently owned and 2 clinics were part of a hospital system. Characteristics of participating clinics are summarized in Table 1.

TABLE 1.

Characteristics of participating clinics, 2018 (N = 3)^a

	Overall, N (%)	Participating clinics
		1	2	3
Ownership status
Independent	1 (33%)	X
Hospital affiliated	2 (66%)		X	X
Rurality
Rural urban commuting area (RUCA) code 4–10	3 (100)	Yes	Yes	Yes
Frontier and remote area (FAR) 2010	2 (67)	No	Yes	Yes
Medically underserved areas/populations for migrant farmworkersb
Migrant farmworkers	2 (67)	No	Yes	Yes
Governor’s exception	1 (33)	Yes	No	No
Medical clinicians, mean (range)	12 (9–14)	9	12	14
Medicaid patients, mean (range)	1,329 (497–1,825)	1,666	1,825	497
Medicaid baseline CRC screening rates, mean (range)c	48.5% (34.6–59.9%)	51.2%	34.6%	59.9%

Abbreviation: CRC, colorectal cancer screening.

^aAll 3 clinics were Federally Certified Rural Health Clinics. All were located in health professional shortage areas for primary care, dental, and mental health.

^bAs defined by the Health Resources Service Administration (HRSA), see https://bhw.hrsa.gov/workforce-shortage-areas/shortage-designation and https://data.hrsa.gov/tools/shortage-area/mua-find.

^c2017 Rates from Coordinated Care Organization (CCO) quality incentive metrics reporting.

As detailed in Figure 1, 376 enrollees were initially identified by the health plan as eligible for CRC screening using claims data. Following a secondary review by the health plan and the clinic scrub, 169 (45%) were included in the final mailed FIT deployment list. Excluded enrollees were up to date on CRC screening (n = 90), had other medical conditions that made them a poor candidate for FIT (n = 17), had not yet established care with the clinic (n = 48), were removed by the health plan (n = 30), or were removed for other clinic reasons (n = 22).

FIGURE 1.

Mailed FIT intervention steps and patient exclusions, consort diagram. Abbreviations: EHR, electronic health record; FIT, fecal immunochemical test. ^a Two patients had an abnormal result from a clinic-distributed FIT and 5 patients had polyps or adenomas detected during colonoscopy. ^b One of these tests was for Cologuard and the remaining 7 were for FIT. For FIT, 2 of these tests had expired by the time the FIT was returned. No results were available for the 6 remaining tests during the chart audit.

Abbreviations: EHR = Electronic Health Record, FIT = Fecal Immunochemical Test)

Preliminary effectiveness: mailed FIT and any CRC screening completion

As shown in Figure 1, 21% (n = 36) of enrollees on the deployment list completed a mailed FIT and 15% (n = 26) completed either a clinic-distributed FIT or another form of CRC screening within 12 months from the mailing. Of the 62 enrollees completing CRC screening, half returned the mailed FIT (58%) and half were screened by a clinic-distributed FIT (23%), colonoscopy (16%), or Cologuard (3%). No enrollee completing the mailed FIT had an abnormal result; however, 2 had abnormal results following a clinic-distributed FIT and 5 had polyps or adenomas detected during colonoscopy. Most enrollees who returned a mailed FIT did so within 90 days (3 months) from the mailing, while the other screening (including clinic-distributed FIT) occurred more evenly across the 12-month evaluation period, see Figure 2.

FIGURE 2.

Rates of CRC screening over time (Kaplan-Meier failure estimate). Note: Each tick mark in the rug plot on the bottom of the graph shows when a patient returned a mailed FIT, whereas the rug plot on the top of the graph shows when patients were screened by another method. The Kaplan-Meier failure curve shows the overall proportion screened over time.

Table 2 summarizes the characteristics of patients on the deployment list and the 2 preliminary effectiveness outcomes, completion of: (1) mailed FIT or (2) any CRC screening (mailed FIT or otherwise). Patients were primarily Medicaid-enrolled (61%), female (59%), and non-Hispanic white (80%, data not shown). Forty percent of Medicare enrollees and 14% of Medicaid enrollees returned the mailed FIT. Similarly, 40% of patients 65+ completed the mailed FIT compared to 13% aged 50–64. Likewise, 30% of male patients completed the mailed FIT, compared to only 16% of female patients. None of the 13 eligible Hispanic enrollees returned the mailed FIT, although 2 screened by another method. A little over one-third (38%) of enrollees with a prior history of any CRC screening returned the mailed FIT. Thirty-six percent of patients classified as “at risk” (per the Truven Verisk risk score described above) returned the mailed FIT.

TABLE 2.

Patient characteristic overall and by mailed FIT or any CRC screening completed, n (col %)

	On deployment list, mailed FIT	Completed CRC screening
		Returned mailed FIT	Screened by any method
Total N	169	36 (21)	62 (37)
Insurance typea
Medicaid	103 (61)	14 (14)	29 (28)
Medicare	47 (28)	19 (40)	25 (53)
Dual	17 (10)	<10 (N/A)	<10 (N/A)
Age, mean (range)	61 (50–75)	66 (51–74)	63 (50–74)
Age 50–64	117 (69)	15 (13)	35 (30)
Age 65+	52 (31)	21 (40)	27 (52)
Genderb
Male	67 (40)	20 (30)	23 (34)
Female	100 (59)	16 (16)	38 (38)
Ethnicity
Non-Hispanic/Latinx	142 (84)	34 (24)	55 (39)
Hispanic/Latinx	13 (8)	—	<10 (N/A)
Unknown	14 (8)	<10 (N/A)	<10 (N/A)
History of any CRC screeningc	58 (34)	22 (38)	29 (50)
Colonoscopy	19 (11)	<10 (N/A)	<10 (N/A)
FIT	44 (26)	17 (39)	24 (55)
No history of CRC screening	111 (66)	14 (13)	33 (30)
Truven risk scoresd
Healthy	—	—	—
Stable	40 (24)	<10 (N/A)	13 (33)
At risk	50 (30)	18 (36)	24 (48)
Struggling	55 (33)	<10 (N/A)	16 (29)
In crisis	17 (10)	<10 (N/A)	<10 (N/A)
Missing/unknown	<10 (N/A)	<10 (N/A)	<10 (N/A)
Assigned primary care provider	139 (82)	33 (24)	55 (40)

N/A = not available due to data censoring when the sample size is <10.

^aTwo patients were not-insured as reported by the health plan but received the mailing; neither completed any form of CRC screening.

^bTwo patients were missing gender information.

^cOne patient had completed CRC screening by flexible sigmoidoscopy and 3 had screened using Cologuard.

^dTruven risk scores are assigned by the health plan using a proprietary model that combines diagnosis types, complexity, and member utilization patterns.

Implementation outcomes: themes associated with program effectiveness, acceptability, and feasibility

Ten individuals participated in key informant interviews, including clinic staff (3 staff and 2 clinicians), health plan staff and leadership (4), and the ORPRN facilitator. Only 1 clinician approached for the interview declined due to a role change during the study period (10/11, 91% response rate). The ORPRN practice facilitator also documented 28 contacts with the clinics and 16 with the health plan during program implementation. We summarize themes related to program participation and implementation outcomes.

Perceived positive benefit, prior relationships supported program adoption

The health plan was motivated to participate because they believed patients would be more responsive if the mailing came from the clinic. Clinics participated because of the desire to improve CRC screening, low perceived disruption to daily operations, limited experience with mailed FIT, and the potential for the health plan partnership to offset costs. One informant shared that some clinicians expressed concern about “sending the FIT test when really the colonoscopy is the gold standard…but most agreed that some screening is better than no screening.” (Clinic Staff, Interview 9) At this same clinic, the clinician champion commented:

“It’s kind of a win across the board…This was a way to reach the patients in a form that was cost neutral to us. [The health plan] paid for the actual kits and mailing of the kits. Our role in that way was rather passive, just waiting for the results to come in and then dealing with whatever the results were.”

(Clinician, Interview 8)

Informants noted a history of working across organizational boundaries; these prior relationships and trust supported program uptake.

Program components were easy and aligned with existing workflows (feasibility and adaptations)

Informants predominantly perceived the mailed FIT program as easy to implement. Clinician champions found the program had limited impact on their workload and provided a clear opportunity to increase their CRC screening rates. One clinician said, “….it was [the] sort of stuff we wanted to do already and it did not require a lot of effort on our part.” (Clinician, Interview 7) Participating clinic staff generally agreed, with 1 noting:

…Everybody was very open to [the program] and the providers seem to follow up as they needed to….I was able to get reports as I needed. … So overall it went well and there weren’t too many glitches in the program.

(Clinic staff, Interview 9)

Program elements were operationalized within existing clinic workflows, which led to 2 key process variations across the participating sites: scrubbing the eligibility list and making reminder calls. The independent clinic incorporated these tasks into the workflows for individual medical assistants, which was perceived as having minimal impact:

We didn’t change anything in our clinic. We already make those phone calls to patients to make sure [they] had a FIT test. Our MAs made one call to each patient just reminding them, and then that was it on our part.

(Clinic staff, Interview 1)

In contrast, in clinics affiliated with the hospital system, scrubbing and reminder calls were the responsibility of the central quality improvement team. This led to a heavy workload just before and following the mailings, which was difficult for the small team to complete. It is notable that CRC screening completion rates ranged from 27% to 48% by clinic with mailed FIT accounting for between 43% and 73% of the completed tests, see Table 3. Two clinics completed 93% and 94% of the planned reminder calls (mailed FIT return rates 24% and 26%), while 1 clinic only completed 41% (mailed FIT return rate 12%).

TABLE 3.

Colorectal cancer screening completion overall and by clinic, N (col %)

	Overall	Participating clinics
		Clinic 1	Clinic 2	Clinic 3
Total N	169	46	72	51
Any method	62 (37)	22 (48)	26 (36)	14 (27)
By modalitya,b
FIT, direct mail	36 (21)	11 (24)	19 (26)	6 (12)
FIT, nondirectmail	14 (8)	9 (20)	2 (3)	3 (6)
Colonoscopy	10 (6)	2 (4)	4 (6)	4 (8)
Received reminder call	132 (78)	43 (93)	68 (94)	21 (41)
Received call and returned mailed FIT	31 (86)	10 (91)	19 (100)	2 (33)

^aTwo patients (3%) completed CRC screening by Cologuard.

^bNo patients who returned the mailed FIT had an abnormal result. Two patients returning an FIT that was nondirect mail had an abnormal result (14%) and 5 patients completing colonoscopy had an abnormal result (50%).

Multiple unplanned mailings and EHR documentation contributed to clinic burden

One of the dominant challenges faced during implementation was a shift from a planned single round of mailing to 3 rounds. This shift aligned the study FIT mailing with existing workflows for the health plan’s Medicare population outreach program, but was unexpected by both the research team and clinics. The additional mailing waves required additional work by each clinic due to repeating the chart scrub. This work could be exacerbated based on the quality of data in a clinic’s EHR health maintenance summary, as illustrated in the following quote:

We scrubbed probably 600 charts last year helping prepare for this mailing…I think internally a factor that hindered us was that our health maintenance reporting didn’t exactly give us a list of patients that needed a colonoscopy or needed screening, so that…pushed it to the chart level scrubbing….It was a big undertaking, but we learned a lot in the process.

(Clinic staff, Interview 9)

Staffing, unanticipated turnover, and practice facilitation shaped implementation

Program implementation was facilitated by having clinic-level staff support with dedicated time to complete intervention elements. Generally, this consisted of a clinician champion who endorsed program activities and 1–2 clinical staff leads who operationalized program elements. Notably, every organization involved in program implementation during the study period experienced turnover. Staff transitions required retraining, re-education, and revisiting the program timing and workflows. Participants expressed appreciation for support from the ORPRN practice facilitator to guide implementation during these transitions.

Adaptations recommended to enhance program sustainability and support patient engagement/follow-up

Overall, clinics expressed desire to sustain most elements of the mailed FIT program and health plan partners wanted to keep the collaborative program going. Clinic staff reflected on acceptability from the patient’s standpoint, noting concerns that the FIT may come out of the blue and a personal touch for initial outreach could improve mailed FIT outcomes. Staff also wanted an option to coordinate colonoscopy should patients prefer that route and protocols to support follow-up after abnormal FIT.

DISCUSSION

Our findings emphasize factors influencing implementation and effectiveness of a collaborative mailed FIT program in rural primary care. Although 376 health plan enrollees were identified as eligible using claims data, only 45% (169) were included in the deployment listing after health plan review and clinic-level scrub. For enrollees who received the mailing, 21% completed a mailed FIT (most within 3 months of the mailing) and an additional 15% completed another form of opportunistic CRC screening within the year. While no formal statistical tests were conducted due to small subgroup sizes, descriptive statistics suggest that Medicare enrolled patients, males, older patients, patients categorized as “at risk” by the health plan, those with a prior history of CRC screening (especially prior FIT), and those who received a reminder call were inclined to complete the mailed FIT. Informants from the participating health plan and rural clinics perceived the program as both feasible and acceptable, noting it was easy to implement with limited impact on clinician workflows and the tailored support of a practice facilitator. Clinic staff also noted the importance of adequate staff effort and time to complete project activities.

Our findings add to a growing body of work regarding the effectiveness of mailed FIT programs and program optimization.^{9,20,21,25,50} Gupta and colleagues recently identified 9 best practices to support optimal implementation of mailed FIT outreach programs (eg, primers, FIT instructions, and data infrastructure).²¹ While these elements focus on program components for mailed FIT programs, an important contribution of the present study relates to the detailed description of the partnered approach used to select, implement, and evaluate the pilot via a partnership between CHARA, an established PBRN, the regional health plan, and participating clinics.^43–45 Thus, while optimizing mailed FIT programs, investigators and health system stakeholders are encouraged to consider collaborative models that may help overcome common barriers to mailed FIT implementation via academic community partnerships supported through PBRNs or learning health systems.²⁸

Our pilot suggests that optimizing mailed FIT response rates may require tailored program adaptations in relation to 2 levels: the clinic and the individual patient. For example, at the clinic level, we found that operationalization of the scrub and reminder calls were aligned with existing clinic workflows. Clinic structure (independent vs system-based, higher functioning vs lower functioning EHR) may, therefore, suggest different workflows to operationalize program elements and to support program implementation. Specifically, centralized models may require additional temporary staffing, while distributed models may require protocols for training and quality assurance. Our findings also suggest opportunities for tailoring mailed FIT elements at the patient level to reach those at risk for nonresponse and to conserve resources for those who are likely to respond with minimal reminders or prompts. For example, patients screened in the past are more likely to respond to a mailed FIT program and may, therefore, not need supplemental outreach.^63,64 In contrast, certain population groups may be less likely to respond (eg, younger patients, women, Medicaid-insured, at higher levels of risk, and Latino patients) and strategies could be tailored to support targeted outreach to these populations. These may include routine mailed FIT as well as strategies for bundling FIT with other preventive strategies (at home testing for diabetes, HPV, and blood pressure) (Coury J and Davis MM, In preparation).⁶⁵ Additionally, given that 27% of the patients scrubbed by the clinic (48/177) were removed because they had never been seen at the clinic (but were attributed by the health plan), proactive patient engagement strategies delivered prior to mailed FIT may also be needed. Identifying strategies to reach unestablished patients or those more likely not to respond to the mailing (newly age eligible) groups using stakeholder engagement research methods may be especially helpful in the wake of COVID-19.^7–9,65,66

Our mixed-methods approach demonstrates the importance of engaging nonphysician staff in health system transformation and evaluation efforts, documents the high frequency of turn over and the importance of retraining, and identifies opportunities to enhance mailed FIT programs implementation by advocating for appropriate resourcing of clinic staff and external practice facilitators to support implementation.^67–69 Population outreach programs like mailed FIT and visit-based strategies are both important to maximize CRC screening completion rates.^9,17 Recent work by our team suggests that activities of clinic-based support staff influence patient response rates on centralized mailed FIT programs.⁷⁰ One strength of the collaborative approach is that administrative activities and program costs are covered by the health plan and thus reduce clinic burden (eg, addressing, stamping, and stuffing envelopes). While a growing body of work emphasizes how reminder calls or alerts for mailed FIT programs can increase completion rates and lead to cost savings overall—these tasks may be challenging for individual clinics to complete.⁷¹ Pignone and colleagues estimate the direct costs of mailed FIT at $55-$73 per patient screened,⁷² but Medicare reimburses FIT at an average of $21.⁷³ Therefore, FIT reimbursement rates do not adequately cover the actual costs of mailing and promoting FIT completion by clinic-level staff. This is in contrast to the $600 wrap-around reimbursement for mailing, reminders, navigation, and processing that Cologuard has negotiated.

The current study has a few notable weaknesses. First, we conducted a single-arm pilot and thus had no control/comparison group. Thus, we could only assess FIT and any CRC completion rates in the sample population. Future research should examine the impact of collaborative mailed FIT programs to see if these workflows lead to higher rates of screening compared to a control group. Second, we only reached a subset of rural patients by partnering with a Medicaid/Medicare health plan. Many rural clinics prefer to implement standard workflows across their patient populations yet struggle to produce population-level data or to track CRC screening completion rates. Finally, we implemented the pilot program in 3 rural clinics and findings may have limited generalizability. However, these clinics did represent 2 of the dominant ownership models in rural primary care: independent and hospital-affiliated clinics.

Our study provides important data to inform future implementation studies and to encourage rapid scale-up of mailed FIT programs in rural settings through PBRN, health plan, and clinic partnerships. The collaborative health plan and clinic model was feasible and acceptable and over time may serve to attenuate persistent disparities in CRC screening. Practice facilitation also helped smooth implementation, but clinic-level staffing was still critical to operationalize program components. Future studies should explore tailored implementation support based on clinic structure and need, include rural-adapted patient navigation programs for follow-up after abnormal FIT, and support targeted outreach for patients least likely to complete a mailed FIT with further attention to social risk and race/ethnicity within rural patients. Programs to scale-up mailed FIT and other population outreach strategies are critical to overcome persistent disparities in CRC screening in rural settings.