First-year implementation of mailed FIT colorectal cancer screening programs in two Medicaid/Medicare health insurance plans: qualitative learnings from health plan quality improvement staff and leaders

Background Colorectal cancer screening rates remain low, especially among certain racial and ethnic groups and the uninsured and Medicaid insured. Clinics and health care systems have adopted population-based mailed fecal immunochemical testing (FIT) programs to increase screening, and now health insurance plans are beginning to implement mailed FIT programs. We report on challenges to and successes of mailed FIT programs during their first year of implementation in two health plans serving Medicaid and dual eligible Medicaid/Medicare enrollees. Methods This qualitative descriptive study gathered data through in-depth interviews with staff and leaders at each health plan (n = 10). The Consolidated Framework for Implementation Research, field notes from program planning meetings between the research team and the health plans, and internal research team debriefs informed interview guide development. Qualitative research staff used Atlas.ti to code the health plan interviews and develop summary themes through an iterative content analysis approach. Results We identified first-year implementation challenges in five thematic areas: 1) program design, 2) vendor experience, 3) engagement/communication, 4) reaction/satisfaction of stakeholders, and 5) processing/returning of mailed kits. Commonly experienced challenges by both health plans related to the time-consuming nature of the programs to set up, and complexities and delays in working with vendors. We found implementation successes in the same five thematic areas as well as four additional areas of: 1) leadership support, 2) compatibility with the health plan, 3) broader impacts, and 4) collaboration with researchers. Commonly experienced successes included the ability to adapt the mailed FIT program to the individual health plan culture and needs, and the synchronicity between the programs and their organizational missions and goals. Conclusions Both health plans successfully adapted mailed FIT programs to their own culture and resources and used their strong quality management resources to maximize success in overcoming the time demands of setting up the program and working with their vendors. Mailed FIT programs administered by health plans, especially those serving Medicaid- and dual eligible Medicaid/Medicare-insured populations, may be an important resource to support closing gaps in colorectal cancer screening among traditionally underserved populations.


Background
Colorectal cancer (CRC) screening decreases mortality [1], yet CRC screening rates remain well below the Healthy People 2020 target of 70.5%, especially among certain racial and ethnic groups, the uninsured, and those who are Medicaid-insured. In 2018, 65.2% of adults aged 50-75 in the United States (U.S.) were up to date on CRC screening, but falling far below this overall rate were American Indian/Alaska Native (56.5%), Hispanic (57.7%), Asian American (57.1%), and uninsured (29.7%) adults [2]. In 2015, the most recent year that these data are available, the rate of CRC screening among Medicaid-enrolled adults aged 50-64 was only 47.4% [3].
Mailing fecal immunochemical test (FIT) kits to patients who are due for CRC screening is an evidence-based strategy for increasing CRC screening that has been tested by clinics and health care systems [4][5][6][7][8][9] (e.g., CRC screening rates have increased from 26.3 to 50.8% in primary care clinics in a large nonprofit health care delivery system [8], from 17.8 to 56.5% in rural family physician offices [9]). Health insurance plans are now beginning to implement mailed FIT screening programs [10].
Critical to the expansion of such programs is the need for guidance on program implementation. Published literature describes barriers to and benefits of implementing clinicand health system-based mailed FIT programs [7,11,12], but limited data exist on health insurance plan-based programs. In this study, we explore implementation challenges and successes specific to two health insurance plans (hereafter referred to as health plans) that serve enrollees in U.S. Medicaid and Medicare programs.
BeneFIT is an implementation study of two health plans' mailed FIT programs. Each plan designed its own implementation model-one collaborative and one centralized [13]. Under the collaborative model, the health plan worked collaboratively with the health centers. The health plan coordinated and administered the mailing of FIT kits while partnering with health center staff to customize the materials and workflows. Under the centralized model, all elements of implementation were coordinated and executed at the health plan level. In this study, we used data from interviews with staff (e.g., administrative quality improvement leaders and program managers) of these two health plans and our research team to examine the successes and challenges the plans experienced in implementing their mailed FIT programs in the first year. This work provides critical information that can help health plans understand how to best launch mailed FIT programs.

Methods
This qualitative descriptive study was embedded within the larger BeneFIT hybrid implementation-effectiveness study [14]. For the BeneFIT study overall, our research team recruited two health plans, one in Oregon and one in Washington state, willing to design and implement programs geared towards mailing FIT kits to the addresses of age-eligible health plan members overdue for CRC screening. To identify these health plans, research team members reached out to existing contacts across health plans serving Medicaid and Medicare enrollees in Oregon and Washington. This recruitment process is described in more detail in an earlier article [13]. The Washington plan is an organization that operates in multiple states and provides Medicaid and dual Medicaid/Medicare coverage for approximately 650,000 members in Washington state. The Oregon plan is an organization that provides Medicaid and Medicare (largely dual eligible Medicare-Medicaid) coverage for about 220,000 enrollees in Oregon. The health plans formed teams, consisting primarily of administrators and program managers within their respective quality improvement departments, to conduct this work. Health plans were given the freedom to tailor the mailed FIT program to their organizational resources and needs.
One developed a collaborative model, the other a centralized model. The research team provided information and sample materials (e.g., pictorial/wordless FIT test instructions) based on prior mailed FIT programs, offered technical expertise as needed, and evaluated the program [15,16]. Detailed information on the models and health plans' motivations for implementing a mailed FIT program is provided elsewhere [13].

Program implementation
The health plans developed their own program components, including workflows; relationships and protocols with print and FIT vendors; tools and data report tracking; mailed materials; staff, department, clinic, and provider communications; and staff training (Table 1). Although health plans had their own implementation strategies, many components were similar: both plans generated lists of members due for CRC screening; scrubbed the lists for ineligible members or for those whose health systems opted out; ordered FIT kits; sent the final list of eligible members, FITs, and mailing materials to a vendor that prepared and sent introductory letters and FITs; monitored and tracked FIT completion; and accepted claims from the lab that processed the returned FITs.
Basic differences in first-year implementation, outside of the overall collaborative or centralized approach, included the delivery method for reminders following the mailing (mailed postcard vs. live phone call), type of FIT used (e.g., FIT used by individual clinics vs. a standard two-sample FIT across the health plan), where FITs were returned and how they were processed (clinic-specific labs vs. centralized lab and their processes), and how results were communicated to providers and patients (following standard clinic procedures vs. mailing hard-copy results to providers plus health plan staff delivering live calls to patients with abnormal test results encouraging them to follow up with their primary care provider).

First year health plan programs
Both health plans used 2015 Healthcare Effectiveness Data and Information Set (HEDIS) criteria to identify plan enrollees not up to date on CRC screening. One health plan included six health centers with 28 clinics in the FIT program and sent 2812 introductory letters and Table 1 BeneFIT Implementation Activities Activities 2650 FIT kits to eligible enrollees. Of those who were sent introductory letters, almost half were female (49.8%); 84.0% were Medicaid-insured, and 16.0% were dually insured by Medicare and Medicaid; and 69.2% were white, 17.6% non-white, and 13.2% missing race/ ethnicity. The second plan included all but three of the 507 health centers with which it had contracts (the three were excluded because they had their own mailed FIT programs) and sent 8551 introductory letters and FIT kits. This enrollee population was 42.5% female; 79.0% Medicaid-insured, and 21.0% dually insured by Medicare and Medicaid; and 56.1% white, 17.7% non-white, and 26.2% missing race/ethnicity.

Qualitative data collection
Research staff collected three types of data to ascertain the health plans' experiences in implementing the mailed FIT program: field notes during planning meetings between the research team and the health plans, internal research team debrief interviews, and in-depth interviews with health plan staff (see Additional file 1 for the interview guide). Research staff trained in qualitative methods (JS, JR) attended phone meetings of the research team and both health plans to record field notes and document questions, concerns, barriers, and achievements as they developed and implemented their programs. The research team met twice for internal debrief interviews. The team was interviewed as a group by JS, using a structured guide that explored the Consolidated Framework for Implementation Research (CFIR) domains that the study team felt were most relevant to evaluation of the implementation of BeneFIT's mailed FIT intervention: intervention characteristics, inner setting, outer setting, and process evaluation [17]. Information gathered from the meeting field notes and CFIR-based research team reflections was used to develop an in-depth interview guide for use with the health plan administrators and project managers who were involved in the design and implementation of the mailed FIT program. The health plan interviews were telephone-based, conducted by either JS or JR, lasted 45-60 min, and occurred between March and June 2017, roughly 6-9 months after the start of implementation. Along with questions specific to the CFIR domains, interview questions explored topical areas such as: description of implementation activities; challenges to and successes achieved in implementing the mailed FIT program; member and provider reaction/feedback; observed strengths and weaknesses of the program model; reaction to available program results; and reflections on improvements. The Human Subjects Division of the University of Washington reviewed and approved all interview procedures and materials.

Analysis of qualitative data
The health plan staff interviews were the primary source of information on challenges and successes in the first year of program implementation. Interviews were audiorecorded and transcribed for content analysis [18][19][20]. The research team's qualitative staff (JS, JR) followed an iterative content analysis approach, developing a coding dictionary following review of a subsample of health plan interview transcripts. Aided by a qualitative software program, Atlas.ti [21], staff applied codes to each transcript and met regularly to discuss application of the codes and refinement of the coding dictionary. Coding discrepancies were discussed and reconciled, and newly identified codes were applied to any prior coded transcripts. When coding agreement was reached, Atlas.ti query and retrieval functions were used to generate topical reports of coded text. These reports were reviewed by JR and JS to identify key content themes. Transcripts of debrief interviews with research staff using the CFIR domains were summarized in multiple reviews by JS, MS, and LMB. These, along with the field notes, were used to further check interpretation of the health plan interview analysis. This iterative process resulted in a set of summary themes that were reviewed with the larger research team. Any areas identified as unclear or uncertain were re-reviewed against raw interview transcripts, resulting in a set of refined, agreed-upon themes.

Results
We completed five interviews with each health plan, with all staff instrumental to the design and execution of the mailed FIT program during the first year of implementation. Collaborative-model health plan interviewees included the chief medical director, clinical quality improvement manager, senior manager of primary care projects, population supervisor, and project manager. Centralized-model health plan interviews were conducted with the local chief medical officer, local director of quality, national vice president of quality, national director of clinical interventions, and national project manager. First-year challenges and successes are presented below. The findings, while guided by CFIR in the interviews to explore a full range of implementation concepts, are organized based on the experiences of the two health plans.

Challenges to first-year implementation efforts
Both health plans experienced implementation challenges in five areas: program design, vendor experience, engagement/ communication, reaction/satisfaction of stakeholders, and processing/returning of mailed kits ( Table 2).

Program design
Staff from both plans said they were surprised by the amount of time required and the complexity of setting up their mailed FIT programs, particularly in determining accurate eligibility lists and establishing workflows

Program Design
More time-consuming and complex to set up/start than anticipated X X "At the very beginning we did have a few provider groups that opted out because they were already providing kits to their members. It was something we definitely had to deal with early on to weed out certain populations for provider groups who were already engaged in their own initiatives." -Centralized "There was a question of what to do with patients who hadn't yet been seen in their clinic, and every clinic kind of dealt with that separately. In the end we had one health center just not process a bunch of FITs that came in because they couldn't get those patients seen and they didn't want to process FITs for themso that was a pretty big challenge." -Collaborative Some health centers/provider groups less interested as prefer to "do own thing" or had other CRC screening plans X X Lack of accurate electronic health record data for member information (e.g., address) X X Unestablished patients created unprocessed kits, extra time in outreach, or differing approaches to resolve X

Vendor Experience
Complexity and time working with vendors to get FIT kits ordered and distributed X X "I would say the biggest challenge was the ordering of the FIT there was a shortage for a while…and we had the whole issue of expired kits…Also, physically getting kits to the vendor so they could mail them out took tremendous coordination, effort, emails, reminding and tracking." -Collaborative "The largest barrier we encountered was around patients receiving the FIT kits from the vendor…it took longer than we anticipated for the pre-letter mailing to go out [from vendor]. And per our timeline we wanted the kits to be mailed about a week after that and unfortunately the kits were not mailed out [by vendor] until a month after the pre-letter went out." -Centralized Delay in obtaining kits from vendor to mail them out (e.g., large quantity or out of date) X X

Delay in vendor mailing introductory letter and kits X
Lack of sufficient oversight with vendor so difficult to know exactly how many reminder calls were being completed or if following script X Current lab vendor requires a two-sample test which may be barrier to FIT completion for patients X

Engagement and Communication
Lack of communication with other key departments (e.g., membership services) about mailed program so less able to address patient questions X X "The customer service people at [health plan] didn't really know about the program and should have because they would get [member] phone callsso we should have written something up for them beforehand." -Collaborative "Once we started getting our first list in for members that needed to receive calls to make sure to follow-up with their provider on the positive screening, it was sort of a scramble to figure out who was going to make those calls. The department at the time of signing up for the BeneFIT project had staff to do it, but once we received the actual list they didn't have time and staff -so that staffing barrier needed to be addressed." -Centralized Lack of staffing for follow-up calls regarding positive results/initial staff designated for this work unavailable once call list was ready X

Reaction/Satisfaction of Stakeholders (provider/health center/patient)
General patient resistance to completing a FIT and doing CRC screening The delays with the timeline and vendor also impacted members who were thinking they were going to receive the kit a week after the letter. And when members didn't receive it, well, that's when the calls started coming in…[members]saying 'I haven't received it in the mail.'" -Centralized "There is still that resistance to using FIT out there in some areas…we had one health center group that had very little penetration of patients touched by the BeneFIT program, and [in those] we have a high percentage of clinics where provider groups are still pushing colonoscopy." -Collaborative Patient calls asking about whereabouts of FIT due to delay of kit mailing X Providers desiring response rates for their patients/teams but information not available at time of inquiry X Provider/health center resistance to FIT screening in some locations X

Processing/Returning of Mailed Kits
Returned FIT kits not always processed X "A big challenge was making sure all the names and dates were on the collected samples so when sent to the lab we wouldn't get any rejections…but there was a break in Lack of workflow/process to ensure returned FIT kits X and vendor expectations. Both plans were also challenged by lack of accurate data on members, such as current addresses. Additionally, both plans had provider groups or health centers opt out due to their own CRC screening programs. A unique issue that arose for the collaborative-model health plan was unestablished members-those patients assigned to their clinic but not yet seen by a provider. Leaders in some health centers expressed concerns about receiving and processing a mailed FIT kit for patients who had not first established care. Addressing this issue of unestablished clinic patients required some health centers to spend additional time and create new workflows (i.e., sending promotional letters in place of a mailed FIT kit or making phone call attempts to reach patients and encourage them to establish care with the clinic).

Vendor experience
Staff from both health plans said the communication and time spent working with vendors to order and distribute the FIT kits were challenges, which resulted in a delay in mailing the kits. Those working under the centralized model experienced additional challenges, including: the required use of the laboratory's two-sample FIT test, which may have created patient barriers to completion; the mail vendor sending the introductory letter and kit later than planned; and lack of oversight in ensuring vendors were making the expected number of reminder phone calls or properly following the reminder call script.

Engagement/communication
Both health plans experienced challenges in communicating with key departments (e.g., member services) about the mailed FIT program and in ensuring that patients were given the correct phone number and/or contact person when they had questions about the introductory letter or mailed kit. For the centralizedmodel health plan, vendor delays in mailing kits created additional staffing issues in conducting follow-up calls to patients with abnormal results (positive for microscopic blood), as the staff identified for this work were less available at the time follow-up calls were due.

Reaction/satisfaction of stakeholders
Staff in both health plans felt that overall, patient resistance to CRC screening and the "yuckiness" of stoolbased screening posed a challenge. Centralized-model health plan staff also fielded calls from members wondering when they would receive their FIT kit as it had not arrived in a timely fashion following the introductory letter. Additionally, centralized-model providers wanted to know the response of and outcomes for their patients sent the mailed FIT, and this information was unavailable in real time. In the collaborative-model health plan, some providers continued to prefer colonoscopy over FIT as a primary screening method. Given this preference, plan leaders felt there might be less acceptability and potential "penetration" of the mailed FIT program in these clinics.

Processing/returning of mailed kits
At the time of the interview, centralized-model health plan staff expressed slight concern about not having complete data on first-year FIT completion rates due to both the delayed mailing timeline and lag in claims data availability, making it difficult to assess the progress and impact of the program. For the collaborative model, several FIT kit processing issues may have affected return rates in the first year. For example, some completed and returned FIT kits were not processed due to mislabeling, the health plan member not being assigned to a provider, or the lab vendor determining the kit was not properly completed (e.g., missing a collection date).
There was a lack of standard workflow to ensure completed mailed FIT kits were properly labeled (e.g., calling patients who provided no collection date) before going to the lab for processing, and no system was in place to ." -Centralized were properly labeled before going to lab for processing No system in place to inform patient their completed FIT kit was not processed due to an issue (e.g., mislabeling) X Some health centers were lower performers due to staff turnover or longstanding operational issues X Assessing fully kit return rate outcomes hindered by delays and lag in claims data (e.g., up to 6 months) X

Vendor Experience
Documentation level from Vendor has been helpful X "We had great success with follow-up calls to our members and getting them appointments for positive screenings and get them into the officethat worked well." -Centralized "What we learned with the ordering and processing of kits [with vendor] was a process…like just getting the workflow of the kits sent here directly to [health plan], instead of kind of being the middleman and just shuffling things around, and that seems to be working pretty good now." -Collaborative Success completing follow up calls and making appointments for members with abnormal result X Working through initial challenges to order and mail kits valuable X

Engagement and Communication
Familiarity with health centers and strong relationships X "The provider groups were informed well in advance on the project and were happy with the information…and another success is that we are getting more provider engagement around the intervention and understanding around what we are doing here in the quality department." -Centralized "[Health centers] are invested in it, they are committed to it… they spread it to their teams. They had meetings and explained the [program]they knew what was happening… we didn't have any clinics take part that hadn't already used FIT before in their clinics." -Collaborative Staff at health centers knew what was happening regarding the program / and took work personally to work hard and achieve goal X Provider groups and their teams were engaged and informed of program X

Reaction/Satisfaction of Stakeholders (provider /health center /patient)
Positive reactions from health centers and providers (e.g., no complaints) X X "Members were very appreciative…We often don't get a sense of member [satisfaction], but through this we were able to do a lot of member outreach after a possible positive result and actually speak to members…who were telling us about their experiences of screening." -Centralized "Providers are very grateful that this is just another way that we can tap into getting them in for screening…and the resources required at clinic level were so lowit's when the FIT comes back in where most of clinic resources are and they already staff that generallyso running the program at the clinic level was low enough that clinics weren't saying they couldn't staff it." -Collaborative Positive reaction from patients (e.g., no complaints)/ expressions of appreciation inform patients that their completed and returned FIT was not processed.

Successes with first-year implementation efforts
The health plans described success in the same five areas in which they had challenges, along with four additional areas: leadership support, compatibility with the health plan, broader impacts, and collaboration with researchers (Table 3).

Program design
Staff from both plans appreciated the flexibility to adapt core components of the mailed FIT program. They also viewed the program as an additional screening avenue outside of the clinical visit that could potentially decrease patient screening barriers (e.g., travel, time) while improving screening rates. Collaborative-model health plan staff felt their partnership approach was a benefit in that it encouraged health center participation by tailoring some components (e.g., introductory letter, type of FIT) to each health center's preference while alleviating both cost and work burden through centralization of key pieces such as mailing of kits. Collaborative-model health plan staff believed that encouraging health centers to choose their own workflow practices-such as reviewing eligibility lists for prior CRC screening or conducting reminders to encourage FIT returns-was also a successful element of their program design.

Vendor experience
While the centralized-model health plan had some oversight challenges with its vendor, staff also felt that the tracking documentation they received regarding mailings and reminder calls was helpful during the early phase of implementation. They also felt that the planned workflow of having health plan staff make outreach calls to members who received abnormal FIT results was a success. Collaborative-model health plan staff felt the effort of managing delays from the vendor in ordering and mailing the FIT was in and of itself a first-year success.

Engagement/communication
Under the collaborative-model health plan, familiarity with and strong partnering relationships with the health centers were identified as a key driver of success. Additionally, clinic leads at the health centers were informed of and invested in the program, and actively communicated about the program with their staff, who were perceived as motivated and focused on achieving the goals of the program. Centralized-model health plan staff felt that they had successfully communicated with provider groups about the mailed FIT program.

Reaction/satisfaction of stakeholders
Both health plans received positive feedback from provider groups and health centers. Positive reactions from members were also noted by both health plans, e.g., members offered expressions of gratitude to staff, and some called to share that they had completed their screening. No provider group, health center, or member complaints were received by either health plan. Members of the centralized-model health plan expressed appreciation for the follow-up call they received explaining next steps after their abnormal FIT result. Within the collaborative-model health plan, health centers reported minimal burden on staff or time in implementing the mailed FIT program.

Processing/returning of mailed kits
A variety of successes related to processing and returning of the mailed FIT kits were identified by both health plans. First, implementation of the program helped to establish workflows and identify areas of improvement for the second year. Additionally, no issues regarding access to colonoscopy for an abnormal FIT result were reported in the interviews. Overall, leadership from both health plans felt the mailed FIT program was assisting their organizations in meeting CRC screening metrics and FIT completion rates were promising enough to continue for a second year.

Leadership support
Both health plans listed as successes strong champions within their organizations who endorsed and guided the mailed FIT program at the leadership level, as well as capable staff leading the day-to-day implementation.
Collaborative-model health plan staff also felt it was beneficial that some of their health center medical directors actively promoted FIT as a primary screening method and supported the concept of a mailed FIT program.

Compatibility
The mailed FIT program was viewed by both health plans as matching well with their organizational mission and goals, including their desire to improve CRC screening rates for their membership. Additionally, both health plans had established quality improvement departments focused on programs to improve population-based measures that could house and execute the mailed FIT program. Collaborative-model health plan staff also felt the mailed FIT program aligned well with the plan's history of working in partnership with its health centers and clinics to implement new and/or innovative programs.

Broader impacts
Both health plans identified three broader impacts they believed were created by participating in the BeneFIT mailed FIT program. First, leaders felt the program helped to enhance patient involvement not only with their own health, but also with their provider, health clinic, and health plan. The program was also viewed as increasing provider and clinical staff awareness about the kinds of activities undertaken by the health plans' quality departments, including about how a mailed FIT program works. Collaborative-model health plan leaders felt health center staff gained knowledge about CRC screening in general and learned specific skills such as how to identify poorly labeled returned kits and correct them. Finally, staff from both health plans believed participation in the mailed FIT program offered a roadmap for how to address other care gaps using a populationbased mailing approach.

Collaboration with researchers
Both health plans found it helpful to draw upon the expertise of the research staff as well as to learn about the other health plan's model and how it was addressing implementation challenges. Centralizedmodel health plan leaders identified the ability to use a set of pictorial/wordless instructions with their FIT kit mailing as a success [22].
Based on their experiences in the first year, the health plans made suggestions for others to consider during the early phases of planning a similar effort. Table 4 highlights their advice, categorized into four key areas: engagement, planning, member eligibility, and FIT mailing/tracking/processing. Regarding engagement, both health plans emphasized the importance of early and continuing support and buy-in from all stakeholders, such as health plan leadership and staff, clinics and provider groups, and vendors. With planning, both health plans highlighted the importance of allowing at least 6 months of planning time prior to implementation to establish protocols and vendor relationships. Additionally, both noted that adequate resources of all types (e.g., staffing, funding for the cost of kits, mailings, and reminders), along with adequate communication about the program to key departments (e.g., membership services), were critical to robust program planning and implementation. The health plans also underscored how working with vendors requires close, constant oversight and communication to ensure program components are both implemented within scope and well documented. Finally, each health plan identified specific advice in the areas of member eligibility and FIT mailing/tracking/processing according to their respective models. The collaborativemodel health plan advocated the importance of cleaning the mailing list prior to mailing (e.g., removing inaccurate addresses), while centralized-model health plan staff suggested allowing provider groups or clinics the opportunity to opt in or out of the mailed program. Using the same FIT kit that individual clinics or provider groups used, and establishing protocols for accurately labeling completed FIT kits for processing, were emphasized by the collaborative health plan. The centralized health plan recommended that there be "standing" or automatic orders for the FIT kits for eligible patients (versus individually ordered by providers) so health plans can efficiently execute the mailing of kits.

Discussion
This study found that a mailed FIT program can be adapted to the culture and needs of individual health plans. Both health plans received positive feedback about the program from providers and patients, and plan leaders felt the results were positive enough to continue the programs for a second year. In addition, they felt that the process of implementing the mailed FIT program resulted in expertise that could be transferred to other population health programs.
The two models shared several challenges that spanned two domains in the CFIR model: characteristics of the intervention and outer setting. Implementation of the program was more time-consuming than anticipated, and there were complexities and delays in working with vendors-and in one plan, conducting other program tasks. These findings are not surprising, given that this was the first time both health plans had attempted a large-scale implementation of a mailed FIT program.

Individual plan successes and challenges
Successes and challenges unique to the individual plans were most closely aligned with the CFIR domains of outer setting and inner setting. Not surprisingly, collaborative-model health plan leaders reported many successes related to their partnership with health centers. For example, the health centers were actively engaged in identifying individuals eligible for the program, and thus had a strong awareness of the program. Both staff and leaders in the clinical setting were committed to the program and could actively promote its success. Health centers were able to tailor materials (e.g., introductory letter) and workflows to their local setting without the staff burden of mailing the kits. These successes are a result of a collective commitment to the program, to sharing responsibilities and resources, and to learning from one another throughout the implementation process. These qualities are all components of collaborating and building strong partnerships, [23][24][25][26] and pave the way for collective efforts. However, there were challenges specific to the sharing of responsibilities and resources between the health plan and individual health centers, including dependence on clinic processes, preferences, and staff availability. The lack of standardized processes for all health centers complicated program management, tracking, and monitoring.
Staff with the centralized-model health plan reported success in tracking the program and its results through their vendor, which allowed the plan to follow up with individuals with abnormal FIT tests. The health plan asked the vendor to provide weekly updates on FIT kit returns, and the vendor reported FIT kit results to both the plan and the member's primary care provider. In this way, the health plan's care coordinators were able to make calls encouraging individuals with abnormal FITs to schedule a follow-up visit with their primary care provider. These efficiencies are consistent with centralized systems that prioritize both standardization and flexibility in program features, since decisions are made by a core leadership team [27]. This centralized process was also associated with reported challenges, such as questions from providers and members about the project and lack of inclusion of the local clinics in promoting and administering the program. In addition, even the centralized program's staff reported that staff support for adequately monitoring vendor activities was a challenge, as the quality improvement staff responsible for the program had many competing quality initiatives that vied for their time.

Mitigating clinic-level challenges
Coronado et al. have previously described challenges to implementing a mailed FIT program in clinic settings, with the most commonly reported challenges being related to the CFIR domain of inner setting: difficulties in access and accuracy of electronic health record data on CRC screening, and the time burden on staff [11]. Transferring the mailing of FIT kits to the health plan clearly helps alleviate these burdens. Health plan staff used available claims data to identify potentially eligible members, and these processes are familiar to health plans and their quality improvement staff. Similarly, both health plans had prior relationships with vendors, and in the centralized-model health plan, a relationship with a lab that could supply thousands of FIT kits at one time. However, staff from both plans reported challenges in working with their vendors and noted that having designated personnel at the plan to set up and oversee the work was critical.
Clinics in our previous study [11] identified as a challenge the lack of timely or accessible data to assess the success of a mailed FIT program. However, for this study both health plans worked with vendors to develop methods for tracking early implementation status pertaining to FIT mailings and returns. For example, the centralized-model health plan's vendor routinely provided a tracking and monitoring log with information on FIT returns to assess early program implementation, while collaborative-model project staff reviewed biweekly individual health center-level claims data on FIT completion as a means of assessing program status. Additionally, both health plans eventually had access to claims data that could be used to examine FIT kit returns in relationship to mailings.
Our results are limited by the inclusion of only two health plan programs. The plans developed different models for their mailed FIT programs, and we do not know the degree to which the differences in the models influenced implementation. Additionally, we interviewed a small number of individuals at the health plans. We did not directly solicit other stakeholder opinions (e.g., providers, patients). However, given that the research focus was on program implementation, all relevant stakeholders at each health plan participated in the interviews. Additionally, we solicited their understanding of patient and provider reactions. Another limitation is the possibility of researcher bias when collecting field notes (e.g., misinterpretation of a "success" or "challenge"). Our use of CFIR domains to shape the study's interview guide ensured that we had a strong guiding framework to capture implementation challenges and successes. However, we did not systematically explore CFIR constructs in each interview, and thus have not used CFIR to organize our study results. We enlisted several measures to ensure credibility and trustworthiness of our interview findings [28,29], including: consistent use of an interview guide, formal coding and content analysis by a trained qualitative methodologist, triangulation of findings through integration and comparison to other sources of data such as field notes and research staff debrief interviews, and "member-checking" our findings by sharing and reviewing them with health plan staff.

Conclusions
This research demonstrates the feasibility of implementing mailed FIT programs tailored to health plans based on their resources and preferences, as well as the substantial successes and addressable challenges that health plans experience as they work to lower rates of CRC and cancer death among their members. The two health plans participating in this study successfully adapted the components of prior clinic-and health system-based mailed FIT programs [4,6,8] to their own culture and resources, and found that their programs were a good fit with their organizational mission and goals. The availability of quality management resources allowed the health plans to overcome the most commonly described challenges reported for clinic-and health system-based mailed FIT programs-the time needed to set up and administer the program. Even with these resources in place, however, challenges common to both health plans included spending more time than anticipated to set up the programs and to work with their vendors. As health plans increase their investment in addressing population health management, they can use the results and lessons learned from this study to optimize the design of their own mailed FIT programs to best match their own organizational structure, resources, culture, mission, and goals. Mailed FIT programs administered by health plans, especially those serving Medicaid-and Medicareinsured populations, are an important resource in closing gaps in CRC screening among traditionally underserved populations.
Additional file 1. BeneFIT Implementation Interview Guide. In-depth interview guide for use with the health plan administrators and project managers involved in the design and implementation of the mailed FIT program.