key: cord-0934833-6qquut02
authors: Selby, Kevin; Jensen, Christopher D.; Levin, Theodore R.; Lee, Jeffrey K.; Schottinger, Joanne E.; Zhao, Wei K.; Corley, Douglas A.; Doubeni, Chyke A.
title: Program components and results from an organized colorectal cancer screening program using annual fecal immunochemical testing
date: 2020-09-30
journal: Clin Gastroenterol Hepatol
DOI: 10.1016/j.cgh.2020.09.042
sha: da0ba66edbf8c269c7b055de8b97fe35cbf08579
doc_id: 934833
cord_uid: 6qquut02

Background and Aims Programmatic colorectal cancer (CRC) screening increases uptake, but the design and resources utilized for such models are not well known. We characterized program components and participation at each step in a large program that used mailed fecal immunochemical testing (FIT) with opportunistic colonoscopy. Methods Mixed-methods with site visits and retrospective cohort analysis of 51-75-year-old adults during 2017 in the Kaiser Permanente Northern California integrated health system. Results Among 1,023,415 screening-eligible individuals, 405,963 (40%) were up to date with screening at baseline, and 507,401 of the 617,452 not up-to-date were mailed a FIT kit. Of the entire cohort (n=1,023,415), 206,481 (20%) completed FIT within 28 days of mailing, another 61,644 (6%) after a robocall at week 4, and 40,438 others (4%) after a mailed reminder letter at week 6. There were over 800,000 medical record screening alerts generated and about 295,000 FIT kits distributed during patient office visits. About 100,000 FIT kits were ordered during direct-to-patient calls by medical assistants and 111,377 people (11%) completed FIT outside of the automated outreach period. Another 13,560 (1.3%) completed a colonoscopy, sigmoidoscopy, or fecal occult blood test unrelated to FIT. Cumulatively, 839,463 (82%) of those eligible were up to date with screening at the end of the year and 12,091 of 14,450 patients (83.7%) with positive FIT had diagnostic colonoscopy. Conclusions The >82% screening participation achieved in this program resulted from a combination of prior endoscopy (40%), large initial response to mailed FIT kits (20%), followed by smaller responses to automated reminders (10%) and personal contact (12%).

Most deaths from colorectal cancer (CRC) are preventable with screening, but many eligible people are 2 not up-to-date on screening. 1,2 In 2006, Kaiser Permanente Northern California (KPNC) began an 3 organized program of annual mailed FIT combined with opportunistic colonoscopy. That approach 4 increased screening dramatically: the proportion of its members up-to-date with screening doubled from 5 about 40% to over 80%, 3,4 accompanied by a 52% decrease in CRC mortality. 3 Health systems wishing to 6 replicate this approach lack detailed information about the program components and required 7 resources. Prior reports noted that extensive service delivery infrastructure (e.g., program management 8 and quality assurance activities) and navigation staff were needed to increase screening uptake, 5 9 multiple methods of outreach and in-reach increased screening participation, and multicomponent 10 approaches were more effective than individual components. 6 Screening outreach has become more 11 important with precipitous drops in uptake due to the coronavirus disease 2019 (COVID-19) pandemic. 12

However, few studies have examined the specific program components, resources required, and 13 screening outcomes of simultaneous use of multiple strategies in a well-defined population to serve as a 14 model for informing such approaches. 7 15

We sought to characterize the program components, resources needed, and incremental participation at 16 each step in the screening process over a one-year period (2017) in an established KPNC program that 17 primarily uses mailed FIT for persons due for screening with colonoscopy on request. 18

Study Design: The study used a mixed methods sequential explanatory design to assess increases in CRC 20 screening uptake with a population-based programmatic approach. We thus evaluated screening 21 program quantitative data in tandem with qualitative data including direct ethnographic observations of 22 J o u r n a l P r e -p r o o f screening processes. The KPNC Institutional Review Board approved this study and waived the 1 requirement for individual informed consent. 2 Setting: We used data from KPNC, a large integrated health care delivery organization with 15 health 3 service areas that serve approximately 4.5 million members in urban, suburban and semi-rural regions in 4

California. Each service area has its own leadership, primary care offices, and gastroenterology 5 departments. KPNC's members are similar socio-demographically to the rest of Northern California, 6 except at extremes of income, but are less likely to have >5 doctor visits per year or report being in poor 7 health. 8 8

Screening Program: Overviews of the CRC screening program have been published previously. 3, 9, 10 Prior 9

to 2006, KPNC relied on visit-based physician requests for CRC screening (i.e., opportunistic screening), 10 predominantly using flexible sigmoidoscopy and guaiac fecal occult blood tests. Following pilot testing in 11 2006, KPNC established a direct-to-patient annual mailed FIT outreach program for those not up to date 12 with screening, without the need for a face-to-face office visit. Screening up to date was defined as 13 receipt of colonoscopy within 10 years, sigmoidoscopy within 5 years, or FIT within the same calendar 14

year. Completed tests are analyzed by an automated OC-Sensor Diana (Polymedco Inc, Cortland Manor, 15 NY) with a cutoff level of >20 µg hemoglobin per gram of stool for a positive result. Patients with a 16 positive test are directed to have follow-up colonoscopy. Screening colonoscopy in place of FIT is 17 available by request. 18

Framework: The overall FIT-based screening program involves 6 core functions: 1) central management 19 of FIT-based screening, 2) automated FIT outreach, 3) local FIT outreach, 4) local FIT in-reach, 5) central 20 processing of completed FIT kits, and 6) local follow-up of FIT results. 21 J o u r n a l P r e -p r o o f Data collection: We began by creating detailed process maps of the entire FIT-based CRC screening 1 program, from the identification of those due for screening, to the completion of diagnostic 2 colonoscopies for those with positive tests. This required a review of program components and site visits 3 to primary care offices, gastroenterology departments, and the regional laboratory. Data collection 4 methods included field notes, ethnographic observations, and interviews with program leaders. 5

Centralized FIT outreach activities were determined at the regional level (across all of KPNC); however, 6 for greater granularity, we measured local outreach and in-reach activities in a single KPNC service area; 7 while some details of in-reach differ between service areas, global resource utilization is similar across 8 service areas. Staff positions were described using job titles (e.g., clinical lead, project manager, etc.) and 9 training level (e.g., physician, medical assistant, etc.). 10

Screening cohort: We identified a cohort of KPNC health plan members who were CRC screening-eligible 11 in 2017, as defined earlier. The program targets people who are due for screening and are 51-75 years 12 old on December 31 of each calendar year. 13

Statistical analyses: For the quantitative analysis, we summarized the cohort characteristics and the 14 percentages of people who were eligible for screening and completed each step of the screening 15 process. We also examined the percentages who were mailed a FIT kit, completed a FIT (after initial 16 outreach, robocall reminder, mailed reminders, and telephone outreach), and completed colonoscopy 17 after a positive FIT. 18

Core Functions 20

Central management of FIT-based screening: Oversight of the CRC screening program was ensured by a 21 population health management team including a part-time clinical leader (physician), full-time lead 22 J o u r n a l P r e -p r o o f project manager, part-time data analyst health educator, and a part-time operations manager. This 1 group managed all automated outreach and provided assistance and materials to the service areas. They 2 often test changes to outreach procedures in parallel with existing procedures (A/B testing), and only 3 proceed to widespread implementation if they observe increases in FIT completion. They were 4 supported by the Information Technology Division which maintains a Patient Reminder, Outreach 5

Management & Population Tracker (PROMPT) system and the Population Health Management Division, 6 that coordinates all population-level preventive activities. PROMPT was developed by KPNC as a custom-7 built add-on to the EPIC-based electronic health record system. A Consumer Report Services group 8 manages approximately 800 to 1000 complaints and questions annually (primarily people ineligible for 9

FIT or requesting a new FIT kit). A Communications Department regularly updates information materials 10 (materials available upon request). 11

Automated FIT outreach: Pre-notices, automated FIT kit mailings, and the coordination of robocall and 12 letter reminders were performed by an outside vendor ( Figure 1 ). Each year, the PROMPT system 13 identifies people eligible for FIT outreach. Letters are then mailed near the anniversary of the prior year's 14 FIT completion date, or birthday or half birthday for those who had not completed FIT previously. The 15 vendor sends pre-letters (touch 1) to eligible individuals one week before the arrival of mailed FIT kits 16 (touch 2). The FIT kit includes information materials personalized with the primary care provider, 17 pictorial instructions, and a prepaid envelope addressed to a central laboratory. Four weeks after mailing 18 the FIT kits, a robocall (touch 3) with interactive voice response is made, and after six weeks, a reminder 19 letter (touch 4) is sent to non-respondents. These automated steps are tailored for those identified as 20 Black or Hispanic; the Communications Department created paper information materials with messages 21 that resonated more strongly with these groups during a series of focus groups. Active users of an on-22 line patient portal receive electronic messages (i.e., e-Alert) 3 days prior to mailings and are only sent a 23 reminder letter by mail if the electronic message is not opened. 24 J o u r n a l P r e -p r o o f Local FIT outreach: The names of non-respondents are automatically sent to their primary care office 8 1 weeks after FIT mailings for further local follow-up. Medical assistants make telephone calls or send 2 partially personalized electronic messages or mailings when possible (touch 5) to non-respondents to 3 complete and return the test. This activity occurs primarily between 8-13 weeks after the automated 4 processes to minimize redundancy. 5

Local FIT in-reach: Throughout the year, those ages 51-75 who attend office visits and are not up to date 6 with screening may receive reminders and be offered a FIT kit (in-reach) at the visit. The PROMPT system 7 alerts medical assistants during the 'rooming' process, showing which step of the outreach process has 8 been completed, allowing staff to, if possible, leverage upcoming delivery of FIT kits rather than hand out 9 additional kits in the office. If the patient cannot recall receiving a mailed FIT kit and no record exists of a 10 completed test, a kit is given to the patient at the time of the visit. 11

Laboratory processing of completed FIT kits: Whether received by mail or in-person, all tests were 12 completed at home and returned in prepaid envelopes to a designated central laboratory where staff 13 review the contents for completeness ( Figure 2 ). Tests with no date or illegible information were 14 submitted to the laboratory's Client Services Department. If the information obtained allowed further 15 processing, the required test information is manually entered into a laboratory database, an order 16 placed, and a label generated for subsequent automated processing. If the test cannot be processed, a Primary care providers or their staff contact patients individually to explain the need for diagnostic 3 colonoscopy and make an electronic referral to the gastroenterology department. Each primary care 4 department has a medical assistant assigned to track results. In most cases, gastroenterology 5 departments have a designated nurse practitioner or similar who ensures follow-up of patients with a 6 positive FIT, in addition to a designated medical assistant scheduler who contacts patients each day to 7 explain the colonoscopy procedure and schedule appointments. In some areas, this staff member calls 8 members with a positive FIT result directly, without waiting for primary care referral. 11 Patients 9 scheduled for colonoscopy are given a prescription and standardized instructions for bowel preparation 10 to be picked up from the local pharmacy. All FIT-positive colonoscopies are considered preventive 11 examinations, and therefore have no or limited co-payments. 12

For negative examinations (i.e., no biopsy), the endoscopist adds to the patient's medical record an 13 indication for "average-risk screening in 10 years." For positive colonoscopy examinations (i.e., polyp or 14 mass), the endoscopist enters "pending pathology results." Once pathology results are available, the 15 relevant guideline-recommended re-screening or surveillance interval is entered into PROMPT. If the 16 lesion was cancerous, the primary care provider is notified, an e-referral is made to a colorectal surgeon. 17

Referral is made to medical oncology if stage IV disease is found. Compared to those who completed screening, those who did not were younger and less likely to have 5 completed a FIT in 2016 (Table 1) . 6

Outcomes of automated outreach: The screening pathway and incremental participation for those 7 eligible for mailed FIT at the beginning of 2017 is shown in Figure 1 . A total of 507,401 were mailed a FIT 8 kit and 206,481 (41%) completed the test within 4 weeks. Among those eligible who were not mailed a 9 kit, many completed screening prior to their mail date, some had previously refused participation, and 10 others were excluded due to serious illness (as documented by a physician or because residing in a 11 skilled nursing facility or hospice). Of 300,920 members who received a robocall reminder at week 4, 12 61,644 (20%) completed FIT, and a further 40,438 (13%) completed the test within 2 weeks after a 13 reminder letter. Almost all of those who completed a FIT during the automated outreach had completed 14 a FIT the year prior (93% , Table 1 ). At this point, through a combination of prior colonoscopy and mailed 15 outreach efforts, 70% of the population was screening up to date. 16

Local outreach and in-reach: During local outreach to non-responders (n=198,838), primarily between 8-17 13 weeks after automated mailings, 42,753 (4.2%) completed a FIT. A further 27,631 (2.7%) of those 18 mailed a FIT completed a test by the end of 2017, presumably through local in-reach. Those who 19 completed a FIT or colonoscopy outside of the automated outreach were younger and less likely to have 20 completed a FIT the year prior than those who responded to automated outreach (Table 1 ). In 2017, 21 over 800,000 PROMPT patient alerts for CRC screening occurred and nearly 295,000 additional FIT kits 22

were given directly to patients during clinic visits. Approximately 100,000 additional FIT kits were 23 J o u r n a l P r e -p r o o f ordered through direct-to-patient calls from medical assistants. The central laboratory attempted to 1 contact approximately 18,000 members by telephone or secure email to obtain missing information, 2 allowing them to process about half of these samples. were FIT positive and 6,203 (42%) received a colonoscopy within 30 days of the result date and 11,738 5 (81%) within 6 months. Of those who received a colonoscopy within 1 year, 7,301 (60%) had ≥1 adenoma 6 resected, 1,028 (8.5%) had an advanced adenoma (i.e., advanced neoplasia on histology), and 335 (2.7%) 7

were diagnosed with CRC. 8

In a population of over one million people with a 40% screening rate at baseline, centralized, automated 10 outreach resulted in a 30 percentage point increase in screening within 8 weeks. Subsequent clinic-based 11 outreach via personalized telephone calls and messages, and in-reach through visit-based reminders 12 resulted in an additional 12 percentage point increase in coverage, yielding an overall 82% screening 13 participation rate. 14 CRC screening rates in the US are well short of the 80% goal set by the National Colorectal Cancer Round 15 Table. 2,12 Effective programs are needed to increase uptake, particularly if expansion of lower eligibility 16 age is more widely adopted. Our study findings are consistent with randomized trials showing that FIT 17 mailings provide a 28 percentage point increase in screening compared to opportunistic screening 18 alone. 13 In another meta-analysis, various types of patient navigation increased screening uptake by 17 19 percentage points, and patient reminders by 3 percentage points, 6 though inconsistent implementation 20 can diminish effectiveness. 14 In a 4-arm randomized trial that added automated electronic health record-21 linked FIT mailings to usual care, then mailings, and finally nurse navigation, 15 each addition provided 22 J o u r n a l P r e -p r o o f added benefit: usual care to automated mailings increased participation from 26% to 51%, with 1 telephone assistance increasing participation further to 58%, and navigation to 65%, though that study 2 required informed consent, limiting the representativeness of its participants. Our study shows that high 3 rates of screening completion can be achieved on a much larger scale among a diverse, community-4 based population. The proactive delivery of screening using opt-out principles likely contributes to the 5 high screening rates achieved. 16 KPNC offered colonoscopy in addition to the mailed program, which 6 over time contributes to the high screening rates. A potential pitfall of employing multiple strategies is 7 overscreening (patients already up to date who nonetheless complete a FIT). Anecdotally, overscreening 8 is rare because the electronic health record add-on, PROMPT, is continuously updated across all sites. 9

However, the use of multiple outreach strategies can create tension between providers and patients 10 who do not want CRC screening and resent repeated reminders. 11

People can encounter multiple barriers to CRC screening. 12 Systematically mailing FIT without charge 12 along with reminders addresses multiple structural barriers. However, potential participants often still 13 express feelings of fear, negative past experiences with the health system, and fatalism. 17 In-person 14 contact with a trusted provider may explain why automated outreach alone may not be sufficient for 15 reaching screening rates >80%. Repeat non-participants in the KPNC program are less intrinsically 16 motivated and more often disgusted by stool collection for FIT. 18 FIT outreach may be particularly 17 relevant with limitations in access to colonoscopy and population fears of in-person visits due to KPNC invested significant resources to track tests with incomplete information and to ensure timely 20 completion of colonoscopy following positive FIT results. 20 These quality criteria, often overlooked by 21 guidelines, 21 are critical to effective screening. Additional follow-up was necessary for 4% of tests 22 received at the laboratory, including 2% that could not be processed. Error rates as high as 20% have 23 been observed. 22 The fewer errors in the program studied may be due to automated processes including 1 pre-printed labels in contrast to handwritten identifiers and dates. KPNC also distributes illustrated 2 (wordless) FIT instructions created with input from patient focus groups, which has decreased laboratory 3 recall for specimens with incomplete data. KPNC investments in tracking systems, patient support and 4 navigation resources, and endoscopy capacity increased colonoscopy completion for positive FIT within 5 30 days from 9% to 34% between the 2006-2008 period and 2013-2016. 11 6 Strengths of this study include its mixed methods approach to provide a complete picture of 7 programmatic CRC screening in a large, diverse population. A primary study limitation is that we did not 8 have a comparison group or time-point to evaluate the precise effects of implementing the various 9 screening program components described. However, several screening components have been shown to 10 be effective in smaller randomized trials of a single intervention. Our results are from a single, large 11 integrated health system, which may limit applicability for smaller programs or individual practices; 12 however, many of the most effective elements (i.e., mailed outreach, and telephone reminders) are 13 commonly used. We do not have information about reasons for non-participation; some members may 14 have made informed decisions not to be screened. Further, we described an established program and 15 did not show the steps needed to build and launch the program. Finally, precise cost information was not 16 available due to difficulty identifying true costs vs. charges and high variability in personnel costs across 17 settings. 18

In conclusion, this study showed mailed FIT with automated outreach and targeted personalized 19 outreach and in-reach increased screening participation to 82%. Substantial resources were used for 20 laboratory quality control and the follow up of positive FIT. High-quality CRC screening can be achieved 21 on a large scale, but attention is needed for individuals requiring repeated personal contacts. 22 J o u r n a l P r e -p r o o f Management, Population Tracking (PROMPT) system. At 56 days after the FIT kit mailing, the names of 3 non-responders are transferred to responsible primary care practices for local outreach. Local outreach 4 occurs primarily within 5 weeks of transfer. All FIT completed in 2017 that were not within 91 days of a 5

FIT mailing were assumed to be due to local FIT in-reach. 6 Program components and results from an organized colorectal cancer screening program using annual fecal immunochemical testing What you need to know:

Background:

-Colorectal cancer screening rates in the United States may be plateauing at <65% -Mailed fecal immunochemical testing (FIT) can increase screening rates Findings:

-About 61% of those mailed a FIT responded to automated outreach with a pre-letter, FIT kit, automated call and reminder postcard, yielding an overall screening rate of 70% -Personalized telephone outreach and reminders during clinic visits gave an additional 12% percentage point increase in overall screening Implications for patient care:

-Automated FIT outreach provides an efficient way to reach most people eligible for CRC screening -Attention is needed for individuals requiring repeated personal contacts

Colorectal cancer statistics

A letter from the National Colorectal Cancer Roundtable

Effects of Organized Colorectal Cancer Screening on Cancer 11 Incidence and Mortality in a Large, Community-based Population

Race/Ethnicity and Adoption of a Population Health 14

Management Approach to Colorectal Cancer Screening in a Community-Based Healthcare System

Costs of Planning and Implementing the CDC's 17

Colorectal Cancer Screening Demonstration Program

Evaluation of Interventions Intended to Increase 19

Colorectal Cancer Screening Rates in the United States: A Systematic Review and Meta-analysis

Causes of Socioeconomic Disparities in Colorectal Cancer and 22 Intervention Framework and Strategies

Similarity of the Adult Kaiser Permanente Membership in Northern California to the 24 Insured and General Population in Northern California: Statistics from the 2011 California Health 25 Interview Survey

Fecal Immunochemical Test Program Performance Over 4 29 Rounds of Annual Screening: A Retrospective Cohort Study

Organized colorectal cancer 31 screening in integrated health care systems

Strategies to Improve Follow-up After Positive Fecal 33

Immunochemical Tests in a Community-Based Setting: A Mixed-Methods Study

Colorectal Cancer Facts & Figures

Mailed Outreach Is Superior to Usual Care Alone for Colorectal 38 Cancer Screening in the USA: A Systematic Review and Meta-analysis

Effectiveness of a Mailed Colorectal Cancer 40

Screening Outreach Program in Community Health Clinics: The STOP CRC Cluster Randomized Clinical 41 Trial

An Automated Intervention With Stepped Increases in 43 Support to Increase Uptake of Colorectal Cancer Screening: A Randomized Trial

A Randomized Controlled Trial of Opt-in Versus Opt-Out 1

Colorectal Cancer Screening Outreach

Patient-reported barriers to colorectal cancer 3 screening: a mixed-methods analysis

Factors associated with use and non-use of the

Test (FIT) kit for Colorectal Cancer Screening in Response to a 2012 outreach screening program: a 6 survey study

Colorectal Cancer Screening and Prevention in the COVID-19 Era

Timely follow-up of positive cancer screening results: 10 A systematic review and recommendations from the PROSPR Consortium

Reminder Calls Improve Patient Handling of Fecal Immunochemical Test Samples

7%) Asian or Pacific Islander