Abstract
Objective:
Effectiveness of fecal occult blood test (FOBT) for colorectal cancer (CRC) screening depends on annual testing, but little is known about patterns of repeat stool-based screening within different settings. Our study’s objective was to characterize screening patterns and identify factors associated with repeat screening among patients who completed an index guaiac FOBT (gFOBT) or fecal immunochemical test (FIT).
Design:
We performed a multi-center retrospective cohort study among people who completed a FOBT between January 2010 and December 2011 to characterize repeat screening patterns over the subsequent 3 years. Logistic regression analyses were used to identify factors associated with repeat screening patterns.
Setting:
Four large healthcare delivery systems in the United States
Participants:
We included individuals aged 50–71 years who completed an index FOBT and had at least 3 years of follow-up. We excluded people with a history of CRC, colonoscopy within 10 years or flexible sigmoidoscopy within 5 years before the index test, or positive index stool test.
Main outcome measure:
Consistent screening was defined as repeat FOBT within every 15 months and inconsistent screening as repeat testing at least once during follow-up but less than consistent screening.
Results:
Among 959,857 eligible patients who completed an index FIT or gFOBT, 344,103 had three years of follow-up and met inclusion criteria. Of these, 46.6% had consistent screening, 43.4% inconsistent screening, and 10.0% had no repeat screening during follow-up. Screening patterns varied substantially across healthcare systems, with consistent screening proportions ranging from 1.0% to 54.3% and no repeat screening proportions ranging from 6.9% to 42.8%. Higher consistent screening proportions were observed in health systems with screening outreach and in-reach programs, while the safety-net health system, which uses opportunistic clinic-based screening, had the lowest consistent screening. Consistent screening increased with older age but was less common among racial/ethnic minorities and patients with more comorbidities.
Conclusions:
Adherence with annual FOBT screening is highly variable across healthcare delivery systems. Settings with more organized screening programs performed better than those with opportunistic screening, but evidence-based interventions are needed to improve CRC screening adherence in all settings.
Keywords: Colon cancer, screening, FOBT
INTRODUCTION
Colorectal cancer (CRC) is the fourth leading cause of cancer-related death worldwide.1 CRC screening has the potential to significantly reduce CRC incidence and mortality.2 In randomized trials, annual fecal occult blood tests (FOBT) reduced CRC mortality by 15–33%.3–5 When high rates of adherence are achieved, the efficacy of annual FOBT in terms of life-years gained is estimated to be equal to that of colonoscopy performed every 10 years.6 Therefore, annual high-sensitivity guaiac FOBT (gFOBT) or fecal immunochemical test (FIT) screening are among the CRC screening strategies recommended by the US Preventive Services Task Force (USPSTF) for average-risk patients with no history of CRC or adenomas.7 These stool-based screening tests are particularly attractive for meeting the 80% screening goals recommended by the American Cancer Society because they are non-invasive, readily available, deliverable to patients who lack frequent health care contact and avoid logistical limitations of limited colonoscopy capacity in some settings.8, 9
The long-term effectiveness of stool-based testing as a CRC screening strategy depends on regular testing. Although the proportion of US patients who are up-to-date with CRC screening has improved to 58%, data from the National Health Interview Survey demonstrate Hispanics (42% up-to-date), low-income patients (44%), and uninsured patients (24%) are less likely to complete screening.10 Studies have also suggested nearly 25–70% of patients who complete an initial stool-based test fail to undergo repeat testing.11–14 However, most prior studies evaluated 3-sample FOBT-based screening and were restricted to single healthcare delivery systems or international screening programs; therefore, it is unclear if these results can be generalized to one-sample FIT-based screening programs in the US. FIT has many advantages over gFOBT, including being a one-sample test without the need for dietary or medication restrictions, leading to higher patient acceptance in organized CRC screening programs.15 After trials demonstrated equivalent efficacy between 3-sample gFOBT and one-sample FIT,16 many US healthcare systems switched to FIT.
Additionally, little is known about different patterns of repeat stool-based screening, e.g., intermittent compliance (2 out of 3 years) but not performed annually, and how patterns of screening vary across different settings. Finally, repeat FOBT screening may be particularly low among understudied subgroups, such as the elderly, racial and ethnic minorities, the uninsured, and people who are not well connected to primary care. Identifying factors associated with consistent stool-based screening may help inform interventions to improve the long-term effectiveness of programs that use gFOBT or FIT screening for large proportions of patients. Conversely, it may also help identify patients who, because they are not likely to repeat a stool test, may be better served by colonoscopy-based screening, if they find it as a more acceptable alternative. Therefore, the aims of our study, conducted within 4 large US healthcare systems, were to 1) characterize FOBT screening patterns over a 3-year period and 2) identify factors associated with repeat screening among patients who completed an index FOBT.
METHODS
Study Setting and Population
This study was conducted as part of the NCI-funded Population-based Research Optimizing Screening through Personalized Regimens (PROSPR) consortium. The overall aim of PROSPR is to conduct multi-site, coordinated, trans-disciplinary research to evaluate and improve cancer-screening processes.17 The ten PROSPR research centers reflect the diversity of US delivery system organizations. The four PROSPR colorectal research centers, Parkland-UT Southwestern (Parkland), Kaiser Permanente Washington Health Research Institute (KPWA, formerly Group Health), Kaiser Permanente Northern California (KPNC), and Kaiser Permanente Southern California (KPSC), contributed data for this study. Details of the research centers’ populations and screening practices have been described elsewhere.18
We restricted our study population to individuals aged 50–71 years who completed FOBT (gFOBT or FIT) for CRC screening between January 1, 2010 and December 31, 2011. Patients from Parkland were limited to those aged 50–64 years given incomplete data capture on Medicare-eligible patients. A patient’s first FOBT during this study period was considered the index test. We excluded patients whose first FOBT in the study occurred when they were ≥72 years old because they would not have 3 full calendar years of ascertainment time before reaching age 75 when screening is no longer recommended. Similarly, we excluded patients who had a history of CRC, colonoscopy within 10 years or flexible sigmoidoscopy in the 5 years before their index stool test, or positive index stool test result because repeat screening would not be indicated in these patients. Index tests performed in-office or as an inpatient were considered not to be for screening purposes and were therefore also excluded. All activities were approved by the Institutional Review Board at each Research Center and the PROSPR Statistical Coordinating Center.
Data collection
Data, including receipt of screening, were collected from the electronic health records (EHR) and administrative databases at each site.18 Receipt of FOBT screening was extracted using EHR laboratory data at each site. Patient information included age, sex, race/ethnicity, insurance status, body mass index (BMI), Charlson comorbidity score, and primary care contact. Sex and race/ethnicity were extracted from the EHR at time of cohort enrollment. Age and insurance status were determined from the year of the index FOBT. BMI and Charlson comorbidity index were calculated from the calendar year of the index test when available; otherwise, data from the year after the index test was used. Primary care contact was defined as the number of primary care visits during the 3-year follow-up period. We used primary care contact as a measure of health care contact because primary care visits are the most common site of CRC screening inreach at each study site. Data were submitted by each research center to the PROSPR Statistical Coordinating Center for analysis.
Analysis
We characterized patients based on receipt of repeat FOBT screening, our primary outcome of interest, during the follow-up period. The follow-up period was defined as time from the index negative FOBT result to the first occurrence of one of the following: receipt of colonoscopy or flexible sigmoidoscopy, positive FOBT result, CRC diagnosis, disenrollment from the health plan, death, or end of the study period (December 31, 2013).
Our primary analysis was restricted to patients with at least 3 full additional years of follow-up, i.e. patients who had index FOBT screening in 2010 and completed follow-up through 2013. We excluded patients with colonoscopy, flexible sigmoidoscopy, CRC diagnosis, or disenrollment from the health plan during the subsequent 3-year follow-up. Similarly, patients with a positive FOBT during the first 2 years of the follow-up period were excluded because diagnostic colonoscopy, rather than repeat stool-based screening, would be indicated.
Patterns of FOBT screening were categorized as consistent, inconsistent, or no repeat testing. Consistent screening was defined as repeat screening within every 15 months during the 3-year period following the index FOBT, such that the interval between two FOBTs never exceeded 15 months. Inconsistent screening was defined as screening at least once during the follow-up period but less than consistent screening, and no repeat screening was defined as no further screening after the index test.
We performed a sensitivity analysis in which consistent and inconsistent screening patterns were defined using two definitions for annual screening: a) within every 12 months and b) every calendar year. Consistent screening was defined as repeat screening within every 12 months or every calendar year during the 3-year period, respectively. Similar to the primary analysis, inconsistent screening was defined as screening at least once during the follow-up period but less than consistent screening.
In addition to our primary cohort of interest, which included patients with at least 3 years of follow-up after their index test, we performed a secondary analysis among those with at least 2 but less than 3 years of follow-up. Given the shorter duration of follow-up in this cohort precluding meaningful 15-month interval analyses, we characterized consistent and inconsistent annual screening per calendar year as described above.
In univariate analyses, Fisher’s exact and Mann Whitney rank-sum tests were performed to identify factors associated with patterns of repeat screening. We used multi-level mixed effect logistic regression models to identify factors associated with consistent screening (versus inconsistent or no repeat screening), accounting for clustering at the provider and system level. Intra-class correlations were used to quantify correlations at each level.29 In a secondary analysis, we used multinomial logistic regression analysis to identify factors associated with repeat screening as a 3-level outcome (consistent, inconsistent, and no repeat). Regression models included variables of a priori clinical importance. Statistical significance was defined as p<0.05 for multivariable analyses. Analyses were conducted using R version 3.2.3 (http://www.r-project.org/) and SAS version 9.3 (SAS Institute Inc, Cary, NC).
RESULTS
Patient Characteristics
We identified 959,857 patients without CRC who completed an index FIT or gFOBT between 2010 and 2011. Patients with positive index stool tests (n= 46,559), colonoscopy within 10 years or flexible sigmoidoscopy within 5 years prior to their index stool test (n= 112,043), or colonoscopy, flexible sigmoidoscopy or positive FOBT during the follow-up period (n=31,172) were excluded, leaving 770,083 who met the inclusion criteria. There were 152,798 (19.8%) patients had less than 2 years of follow-up, ranging from 16.6% at Parkland to 23.9% at KPWA, and 273,182 patients who had between 2 and 3 years of follow-up. The subset of 344,103 patients who had 3 calendar years of follow-up after their index test formed the cohort for our primary analysis. Nearly all index tests were FIT, with only 3.0% being gFOBT. Most gFOBT were performed at KPWA, which changed to FIT after December 2011. Patient characteristics are detailed in Table 1. Patient age ranged from 50 to 71 years at time of index test, with a median age of 58 years. Over half (55.3%) of patients were female, and our cohort was racially/ethnically diverse with 55.8% White, 17.9% Hispanic, 14.1% Asian, and 7.5% Black patients. Most patients had a Charlson comorbidity score of 0 (67.5%) or 1 (18.4%).
Table 1:
Characteristics of patients with 3 years of continuous follow-up after index FIT or gFOBT among four PROSPR Research Centers, 2010 to 2013 (N=344,103)
| KP Washington N=10,279 |
KP Northern California N=197,052 |
KP Southern California N=132,420 |
Parkland- UTSW N=4,352 |
|
|---|---|---|---|---|
| Age (years) 50 – 54 55 – 59 60 – 64 65 – 69 70 – 71 |
2381 (23.2%) 2574 (25%) 2658 (25.9%) 2093 (20.4%) 573 (5.6%) |
51501 (26.14%) 51564 (26.2%) 46764 (23.7%) 35680 (18.1%) 11543 (5.9%) |
38644 (28.2%) 33025 (24.9%) 28875 (21.8%) 24146 (18.2%) 7730 (5.8%) |
2151 (49.4%) 1714 (39.4%) 487 (11.2%) --- --- |
| Sex Male Female |
4641 (45.2%) 5638 (54.8%) |
88235 (44.8%) 108808 (55.2%) |
59426 (44.9%) 72992 (55.1%) |
1509 (34.7%) 2843 (65.3%) |
| Race/ethnicity White Black Hispanic Asian/PI Other/unknown |
8089 (78.7%) 337 (3.3%) 372 (3.6%) 894 (8.7%) 587 (5.7%) |
121352 (61.6%) 12422 (6.3%) 22243 (11.3%) 33184 (16.8%) 7851 (4.0%) |
61992 (46.8%) 11369 (8.6%) 37174 (28.1%) 14086 (10.6%) 7799 (5.9%) |
668 (15.3%) 1580 (36.3%) 1864 (42.8%) 212 (4.9%) 28 (0.6%) |
| Body mass index (kg/m2) BMI < 25 BMI 25 - < 30 BMI 30 - < 35 BMI ≥35 Missing |
3108 (30.2%) 3765 (36.6%) 1955 (19%) 134 (13.4%) 77 (0.7%) |
57633(29.2%) 71377(36.2%) 3850419.5%) 23523(11.9%) 6008 (3%) |
32225(24.3%) 48847(36.9%) 29746(22.5%) 19720(14.9%) 1879 (1.4%) |
600(13.8%) 1266(291%) 1186(27.3%) 1125(25.9%) 175 (4%) |
| Charlson Score 0 1 2 ≥3 |
7716(75.1%) 1474(14.3%) 654 (6.4%) 435 (.2%) |
138958(70.5%) 33897 (17.2%) 14263 (7.2%) 9934 (5%) |
83519(63.1%) 26367(19.9%) 12370 (9.3%) 10164 (7.7%) |
211348.6%) 1572(36.1%) 344 (7.9%) 323 (7.4%) |
| PCP visits during follow-up 0 visits 1 visit 2 visits 3 visits 4 visits ≥ 5 visits |
154 (1.5%) 389 (3.8%) 645 (6.3%) 865 (8.4%) 961 (9.4%) 7265(70.7%) |
2913(1.5%) 4205 (2.1%) 6395 (3.3%) 8885 (4.5%) 11483 (5.8%) 163171(82.8%) |
1487(1.1%) 2479 (1.9%) 4000 (3.0%) 5423 (4.1%) 7024 (5.3%) 112007(84.6%) |
222(5.1%) 279 (6.4%) 291 (6.7) 286 (6.6%) 286 (6.6%) 2988(68.7%) |
FIT - fecal immunochemical test; FOBT – fecal occult blood test; KP – Kaiser Permanente; PCP – primary care provider; UTSW – University of Texas Southwestern Medical Center
Repeat Screening Patterns among Patients with 3 Years of Follow-up
Among patients with 3 years of follow-up after their index test, just under half (46.6%) underwent consistent screening with repeat testing at least every 15 months. Inconsistent screening was performed by 43.4% of patients, and 10.0% did not repeat screening during the follow-up period. However, screening patterns varied widely across healthcare systems, with the consistent screening proportions ranging from 1.0% to 54.3% and no repeat screening ranging from 6.9% to 42.8% (Figure 1a). In a sensitivity analysis excluding patients who underwent gFOBT, 47.6% and 42.5% of patients underwent consistent and inconsistent screening, respectively. Consistent and inconsistent screening was performed in 8.1% and 81.9% of patients, respectively, using a cut-off of within 12 months, and 55.7% and 34.1%, respectively, when defined as annual screening every calendar year.
Figure 1: Distribution of Patterns of Repeat FOBT screening, by PROSPR healthcare system.
1a. Distribution of Patterns of Repeat FOBT screening among patients with ≥3 years of follow-up after index negative FOBT (N=344,103)
1b. Distribution of Patterns of Repeat FOBT screening among patients with ≥2 but <3 years of follow-up after index negative FOBT (N=273,182)
Screening patterns significantly differed across healthcare systems, even after adjustment for patient-level variables (p<0.001). Relative to KPNC, the odds of consistent screening (versus inconsistent or no repeat screening) were significantly lower at KPSC, KPWA, and Parkland (Table 2). The intra-class corrrelation (ICC) was 0.04 and 0.02 at the provider and system levels respectively, suggesting that screening patterns were more similar at the provider level than system level. We also noted variation in screening patterns by patient characteristics, most notably age, race/ethnicity, and comorbidity. Compared to patients aged 50–54 years, higher odds of consistent screening were observed among patients aged 55–71. Compared with non-Hispanic Whites, Hispanics and Blacks had lower odds of consistent screening, with odds ratios of 0.82 (95%CI 0.80–0.84) and 0.81 (95%CI 0.78–0.83), respectively. Consistent screening was also inversely associated with comorbid illness burden, with lower odds of consistent screening among those with higher Charlson comorbidity scores (OR 0.90–0.91 for Charlson comorbidity scores 1–2 and OR 0.78 for Charlson comorbidity score 3). Compared to patients with commercial/private insurance, those with Medicaid were less likely to have consistent screening. Additionally, patients with higher BMI categories had lower odds of consistent screening compared to those with BMI < 25 kg/m2. Associations were similar when evaluated in a sensitivity analysis only including patients who underwent FIT and in a multinomial model evaluating screening patterns as a 3-level outcome (consistent, inconsistent, and no repeat) (Supplemental Table 1).
Table 2:
Factors associated with receipt of consistent screening versus not consistent, by length of follow-up study period among patients with ≥3 years of continuous follow-up after index negative FIT or gFOBT (N=344,103)
| Variable | Adjusted* OR (95% CI) |
N (%) with consistent screening |
N (%)with inconsistent screening |
N (%) with no repeat screening |
|---|---|---|---|---|
| Site KPNC KPSC KPWA Parkland-UTSW |
Reference 0.54 (0.49 – 0.59 0.07 (0.06 – 0.08) 0.01 (0.01 – 0.02) |
107085 (54.3%) 52059 (39.3%) 1065 (10.4%) 43 (1.0%) |
76376 (38.8%) 62673 (47.3%) 7985 (77.7%) 2448 (56.3%) |
13591 (6.9%) 17688 (13.4%) 1229 (12.0%) 1861 (42.8%) |
| Age (years) 50 – 54 55 – 59 60 – 64 65 – 69 70 – 71 |
Reference 1.30 (1.27 – 1.33) 1.65 (1.61 – 1.69) 1.91 (1.85 – 1.97) 2.08 (1.98 −2.18) |
36831 (38.9%) 39911 (44.9%) 39694 (50.4%) 32934 (53.2%) 10882 (54.8%) |
46696 (49.3%) 39786 (44.8%) 32128 (40.8%) 23616 (38.1%) 7256 (36.6%) |
11150 (11.8%) 9180 (10.3%) 6962 (8.8%) 5369 (8.7%) 1708 (8.6%) |
| Sex Male Female |
Reference 1.03 (1.01 – 1.05) |
71381 (46.4%) 88869 (46.7%) |
66212 (43.1%) 83263 (43.8%) |
16218 (10.5%) 18149 (9.5%) |
| Race/ethnicity White Hispanic Black Asian Other/unknown |
Reference 0.82 (0.80 – 0.84) 0.81 (0.78 – 0.83) 0.97 (0.95 – 1.00) 0.90 (0.84 – 0.98) |
94147 (49.0%) 24725 (40.0%) 10313 (40.1%) 24144 (49.9%) 1394 (40.4%) |
80520 (41.9%) 29611 (48.0%) 11790 (45.9%) 20122 (41.6%) 1632 (47.3%) |
17434 (9.1%) 7317 (11.9%) 3605 (14.0%) 4110 (8.5%) 423 (12.3%) |
| Body mass index (kg/m2) BMI < 25 BMI 25 - < 30 BMI 30 - < 35 BMI ≥35 |
Reference 0.94 (0.92 – 0.96) 0.90 (0.87 – 0.92) 0.85 (0.82 – 0.87) |
445858 (49.0%) 58550 (46.7%) 32040 (44.9%) 19488 (42.6%) |
38927 (41.6%) 54154 (43.2%) 31931 (44.7%) 21369 (46.7%) |
8781 (9.4%) 12551 (10.0%) 7420 (10.4%) 4885 (10.7%) |
| Charlson Index 0 1 2 ≥3 |
Reference 0.91 (0.89 – 0.93) 0.90 (0.88 – 0.93) 0.78 (0.76 – 0.81) |
110136 (47.4%) 28400 (44.9%) 12677 (45.9%) 9039 (43.3%) |
100117 (33.3%) 28015 (35.2%) 12070 (35.4%) 9280 (36.7%) |
22053 (9.7%) 6895 (11.2%) 2884 (10.8%) 2537 (12.6%) |
| Insurance Commercial/private Medicare Medicaid Other Charity/uninsured** |
Reference 1.00 (0.98 – 1.04) 0.80 (0.73 – 0.88) 2.30 (1.07 – 4.50) 0.76 (0.40 – 1.53) |
111243 (45.0%) 48082 (53.0%) 886 (34.6%) 11 (6.2%) 30 (0.9%) |
111464 (45.1%) 34738 (38.3%) 1178 (45.9%) 118 (65.9%) 1982 (59.4%) |
24502 (9.9%) 7989 (8.8%) 500 (19.5%) 50 (27.9%) 1327 (39.7%) |
| PCP visits during follow-up 0 visits 1 visit 2 visits 3 visits 4 visits ≥5 visits |
Reference 0.93 (0.81 – 1.06) 0.86 (0.76 – 0.98) 0.98 (0.86 – 1.11) 0.95 (0.84 – 1.08) 0.93 (0.83 – 1.05) |
2157(45.2%) 3207 (43.6%) 4850 (42.8%) 7026 (45.5%) 9026 (45.7%) 133986 (46.9%) |
1823 (38.2%) 3050 (41.5%) 5031 (44.4%) 6755 (43.7%) 8903 (45.1%) 123920 (43.4%) |
796 (16.7%) 1095 (14.9%) 1450 (12.8%) 1678 (10.9%) 1825 (9.2%) 27525 (9.6%) |
CI – confidence interval; KPNC – Kaiser Permanente Northern California; KPSC – Kaiser Permanente Southern California; KPWA – Kaiser Permanente Washington; PCP – primary care provider; OR – odds ratio; UTSW – UT Southwestern
Models adjusted for covariates in table
Parkland-UTSW offers a sliding fee scale program, which provides access to medical care, including CRC screening, for uninsured residents of Dallas County
Repeat Screening Patterns among Patients with 2–3 Years of Follow-up
Of the 273,182 patients who had between 2 and 3 years of follow-up, 18.5% (n=50,527) completed FOBT screening in 2010 and were followed through 2012, and 81.5% (n=222,655) were screened in 2011 and followed through 2013. Patient characteristics, (Supplemental Table 2) did not meaningfully differ from those of the primary cohort. Consistent screening every calendar year was performed in 46.6% of patients, inconsistent screening in 35.3%, and no repeat screening in 18.1%. Screening patterns again varied across healthcare systems, with the proportion of consistent screening per calendar year ranging from 6.1% to 57.8% and no repeat screening ranging from 13.6% to 55.8% (Figure 1b). Repeat screening patterns varied significantly across healthcare systems, even after adjustment for patient-level variables. Factors associated with repeat screening patterns were similar to those reported in the primary analysis among patients with 3 years of follow-up, although associations were weaker (Supplemental Table 3). For example compared to KPNC, the odds of patients undergoing consistent screening were lower, with odds ratios of 0.40 (95%CI 0.37–0.43) for KPSC, 0.19 (95%CI 0.17–0.21) for KPWA, and 0.03 (95%CI 0.02–0.05) for Parkland. Associations were also similar when evaluated in a multinomial model evaluating screening patterns as a 3-level outcome (consistent, inconsistent, and no repeat; data not shown).
DISCUSSION
Our cohort study is one of the first multi-health system studies to characterize repeat stool-based CRC screening patterns as implemented in real-world clinical practice settings using contemporary FOBTs in the United States. We found that, after completion of an initial stool-based screening test, only slightly less than half of all patients who opted for stool-based screening underwent consistent screening at least every 15 months over the subsequent 3-year period. Over one-third (43.4%) of patients had inconsistent screening, and 10.0% had no repeat testing. However, we found very large variation across the 4 healthcare systems, with proportions of consistent screening ranging from 1.0% to 54.3% and no repeat screening ranging from 6.9% to 42.8%. While several sociodemographic and clinical factors were modestly associated with screening patterns, we found study site was associated with screening patterns even after adjusting for patient differences.
The 2017 guidelines from the USPSTF recommend highly sensitive gFOBT, FIT, or colonoscopy as first-line modalities for CRC screening; however, the effectiveness of a stool test recommendation depends on consistent annual screening and high rates of follow-up colonoscopy among those with positive stool test results.19, 20 Stool-based screening is dependent on annual testing because early cancers and advanced adenomas may bleed only intermittently. Prior studies have demonstrated that adenoma detection remains stable during subsequent rounds of stool testing, so low rates of repeat screening may significantly increase the risk of interval cancers and CRC mortality.21, 22 Given the necessity for patients to undergo consistent annual FOBT to achieve equally effective screening as those undergoing alternative screening modalities, the USPSTF and opinion leaders/editorialists have emphasized the importance of characterizing patient adherence to screening programs as a high priority area for research.7
One notable finding from our study is the variability in the proportion of patients with consistent screening depending on how this was defined. When using a cut-off of 12 months, consistent screening was observed in fewer than 10% of patients; however, consistent screening was substantially higher at approximately 50% with a more liberal definition such as every 15 months or calendar year. More liberal definitions with a “grace period” allow for time during which FOBTs can be ordered and completed after the patient becomes due for screening. The variability in consistent screening by definition is also important when comparing results across studies, which may use different definitions for consistent screening. Furthermore, a strict 12-month cut-off may overestimate the proportion of patients with screening underuse (e.g. patients completing FOBT at 13 months), particularly in the absence of comparative effectiveness data to identify the optimal screening interval. While FIT is recommended annually in the US, many centers in Europe continue to perform biennial FIT.
The factor most strongly associated with repeat FOBT screening patterns in our analysis was study site, suggesting system-level factors may play an important role in consistent CRC screening completion. Although all of the study sites were integrated healthcare systems with organized FOBT-based screening programs, there were major differences in screening delivery strategies (Table 3).23 The highest proportion of consistent screening was observed at KPNC, which delivers FITs through several methods including outreach efforts (e.g. mailed FIT kits timed to coincide with patients’ birthdays) and in-reach efforts (e.g. available FIT kits at flu vaccine centers and opportunistically at clinic-visits). Intermediate proportions of consistent screening were observed in KPSC and KPWA. The use of gFOBT at KPWA until December 2011 may have also adversely affected repeat screening rates because gFOBT required dietary restrictions and more samples. The lowest proportions of consistent screening were observed in Parkland safety net health system, which depended solely on opportunistic clinic-based screening with no organized system-wide outreach or in-reach programs at the time.
Table 3:
Organizational Practices for Promoting FOBT-based CRC Screening Completion, by PROSPR healthcare system, at time of the study
| KP Northern California |
KP Southern California |
KP Washington | Parkland- UTSW |
|
|---|---|---|---|---|
| Outreach: Automated reminders sent to patients that are overdue for CRC screening |
X | X | X | |
| Outreach: System-level outreach program with FOBT kits mailed to patients |
X | X | Varied by clinic | |
| Inreach: FOBT kits available for patients to pick up at other preventive care visits, e.g. flu clinics |
X | X | ||
| Inreach: Automated alerts in the electronic medical record to notify providers that patients are overdue for CRC screening |
X | X | X | X |
| FOBT screening can be ordered by providers at time of clinic visit |
X | X | X | X |
CRC – colorectal cancer; FOBT – fecal occult blood test; KP – Kaiser Permanente; UTSW – UT Southwestern
In our analysis, consistent screening increased with older age and decreased with Charlson comorbidity score, black race, and Hispanic ethnicity. The lower proportions of consistent FOBT screening in racial/ethnic minorities may partly explain the observed higher rates of CRC-related mortality in these subgroups.24–26 Further studies are needed to determine if lower consistent screening rates in these subgroups may be related to patient or provider attitudes and/or screening fatigue, lack of provider recommendation, decreased effectiveness of screening outreach, perceived benefit of screening, or other factors. The inverse association between Charlson comorbidity and consistent screening contrasts with prior studies reporting a positive association, which is likely related to differences in study population (e.g. availability of screening outreach at study sites) and analysis (e.g. adjustment for primary care contact). Patient sex was significantly associated with consistent screening, although size of the associations were small and of questionable clinical significance. Data from a CRC screening program in England similarly suggested patient-level disparities, such as sex disparities, observed with one-time testing may diminish when assessing longitudinal CRC screening adherence.27
Notable strengths of the study include its large cohort of nearly 350,000 patients reflecting real-world clinical practice using contemporary stool-based tests in diverse practice settings in the US, comprehensive capture of screening test results, and our ability to control for both patient- and system-level factors. However, our study has limitations that should be considered when interpreting the results. First, it is possible that some patients in our study may have had limited life expectancy and not been eligible for repeat screening. Adjustment for age and Charlson comorbidity index in our multivariable model may not fully account for patients with a short life expectancy due to unmeasured conditions. Second, we cannot exclude the possibility of unmeasured confounders to explain the variation in repeat screening patterns. For example, we included insurance status, but this does not fully capture socioeconomic status, which is also associated with CRC screening.10, 28 Similarly, we used primary care contact as a measure of healthcare contact and inreach opportunities, but this does not account for other clinic visits in which CRC screening may have been discussed. Third, our study focused on patterns of stool-based screening, so patients who underwent colonoscopy or flexible sigmoidoscopy were excluded. Finally, studies with follow-up periods longer than 3 years are needed to better characterize the long-term adherence to stool-based screening strategies and determining how inconsistent screening patterns may affect CRC incidence and mortality.
In summary, we found only about half of all patients with index FOBT screening underwent consistent FOBT screening and 10% did not repeat screening over the subsequent 3-year period. However, there was very large variation in receipt of consistent screening across healthcare systems, which could be in part related to differences in the organization of screening programs. The proportion of patients with consistent screening ranged from 1.0% at Parkland, which depends solely on opportunistic visit-based screening, to 54.3% at KPNC, which uses a combination of outreach and in-reach efforts to promote CRC screening completion. The variation in FOBT screening patterns suggests the importance of system-level factors including screening outreach and in-reach for increasing repeat screening rates, but evidence-based interventions are needed to improve CRC screening adherence in all settings.
Supplementary Material
WHAT THIS PAPER ADDS.
What is already known on this subject
Simulation models have demonstrated annual stool-based CRC testing can achieve similar life-years gained as colonoscopy performed every 10 years; however, its effectiveness depends on annual testing. Little is known about different patterns of repeat stool-based screening in clinical practice and how screening patterns vary across different healthcare settings.
What this study adds
Our study suggests adherence with annual FOBT screening is suboptimal, with less than half of patients undergoing consistent screening over a 3-year period; however we found very large variation across healthcare delivery systems. Settings with more organized screening programs performed better than those with opportunistic screening, but evidence-based interventions are needed to improve CRC screening adherence in all settings.
Acknowledgments
Financial support: This study was conducted as part of the NCI-funded consortium Population-Based Research Optimizing Screening through Personalized Regimens (PROSPR) with support from NIH/NCI grants U54CA163308, U54CA163262, and U54CA163261. Support also comes from AHRQ Grant R24 HS022418 (AGS, CSS, JT, EAH) and NIH/NCI Cancer Center Support Grant P30 CA142543 (EAH, CSS, JT). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Footnotes
Conflicts of Interests: None of the authors have any conflicts of interest
Transparency Declaration: Dr. Singal affirms the manuscript is an honest, accurate, and transparent account of the study being reported, no important aspects have been omitted, and that any discrepancies from the study as planned have been explained.
REFERENCES
- 1.World Cancer Report 2014. Lyon, France: International Agency for Research on Cancer, WHO, 2014. [Google Scholar]
- 2.Elmunzer BJ, Singal AG, Sussman JB, et al. Comparing the effectiveness of competing tests for reducing colorectal cancer mortality: a network meta-analysis. Gastrointest Endosc 2015;81:700–709.e3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Kronborg O, Fenger C, Olsen J, et al. Randomised study of screening for colorectal cancer with faecal-occult-blood test. Lancet 1996;348:1467–71. [DOI] [PubMed] [Google Scholar]
- 4.Mandel JS, Bond JH, Church TR, et al. Reducing mortality from colorectal cancer by screening for fecal occult blood. Minnesota Colon Cancer Control Study. N Engl J Med 1993;328:1365–71. [DOI] [PubMed] [Google Scholar]
- 5.Hardcastle JD, Chamberlain JO, Robinson MH, et al. Randomised controlled trial of faecal-occult-blood screening for colorectal cancer. Lancet 1996;348:1472–7. [DOI] [PubMed] [Google Scholar]
- 6.Zauber AG, Lansdorp-Vogelaar I, Knudsen AB, et al. Evaluating test strategies for colorectal cancer screening: a decision analysis for the U.S. Preventive Services Task Force. Ann Intern Med 2008;149:659–69. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Screening for colorectal cancer: U.S. Preventive Services Task Force recommendation statement. Ann Intern Med 2008;149:627–37. [DOI] [PubMed] [Google Scholar]
- 8.Klabunde CN, Cronin KA, Breen N, et al. Trends in colorectal cancer test use among vulnerable populations in the United States. Cancer Epidemiol Biomarkers Prev 2011;20:1611–21. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Meester RG, Doubeni CA, Zauber AG, et al. Public health impact of achieving 80% colorectal cancer screening rates in the United States by 2018. Cancer 2015;121:2281–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Sabatino SA, White MC, Thompson TD, et al. Cancer screening test use - United States, 2013. MMWR Morb Mortal Wkly Rep 2015;64:464–8. [PMC free article] [PubMed] [Google Scholar]
- 11.Fenton JJ, Elmore JG, Buist DS, et al. Longitudinal adherence with fecal occult blood test screening in community practice. Ann Fam Med 2010;8:397–401. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Janda M, Hughes KL, Auster JF, et al. Repeat participation in colorectal cancer screening utilizing fecal occult blood testing: a community-based project in a rural setting. J Gastroenterol Hepatol 2010;25:1661–7. [DOI] [PubMed] [Google Scholar]
- 13.Liss DT, Petit-Homme A, Feinglass J, et al. Adherence to repeat fecal occult blood testing in an urban community health center network. J Community Health 2013;38:829–33. [DOI] [PubMed] [Google Scholar]
- 14.Parente F, Boemo C, Ardizzoia A, et al. Outcomes and cost evaluation of the first two rounds of a colorectal cancer screening program based on immunochemical fecal occult blood test in northern Italy. Endoscopy 2013;45:27–34. [DOI] [PubMed] [Google Scholar]
- 15.van Rossum LG, van Rijn AF, Laheij RJ, et al. Random comparison of guaiac and immunochemical fecal occult blood tests for colorectal cancer in a screening population. Gastroenterology 2008;135:82–90. [DOI] [PubMed] [Google Scholar]
- 16.Oort FA, Terhaar Sive Droste JS, Van Der Hulst RWM, et al. Colonoscopy-controlled intra-individual comparison to screen relevant neoplasia: faecal immunochemical test vs. guaiac-based fecal occult blood test. Alimentary Pharmacology & Therapeutics 2009; 31(3): 432–439. [DOI] [PubMed] [Google Scholar]
- 17.Beaber EF, Kim JJ, Schapira MM, et al. Unifying screening processes within the PROSPR consortium: a conceptual model for breast, cervical, and colorectal cancer screening. J Natl Cancer Inst 2015;107:djv120. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Tiro JA, Kamineni A, Levin TR, et al. The CRC screening process in community settings: A conceptual model for the Population-based Research Optimizing Screening through Personalized Regimens Consortium. Cancer Epidemiol Biomarkers Prev 2014. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Bibbins-Domingo K, Grossman DC, Curry SJ, et al. Screening for Colorectal Cancer: US Preventive Services Task Force Recommendation Statement. Jama 2016;315:2564–75. [DOI] [PubMed] [Google Scholar]
- 20.Martin J, Halm EA, Tiro JA, et al. Reasons for Lack of Diagnostic Colonoscopy After Positive Result on Fecal Immunochemical Test in a Safety-Net Health System. Am J Med 2016. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Crotta S, Segnan N, Paganin S, et al. High rate of advanced adenoma detection in 4 rounds of colorectal cancer screening with the fecal immunochemical test. Clin Gastroenterol Hepatol 2012;10:633–8. [DOI] [PubMed] [Google Scholar]
- 22.Jensen CD, Corley DA, Quinn VP, et al. Fecal Immunochemical Test Program Performance Over 4 Rounds of Annual Screening: A Retrospective Cohort Study. Ann Intern Med 2016;164:456–63. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Chubak J, Garcia MP, Burnett-Hartman AN, et al. Time to Colonoscopy after Positive Fecal Blood Test in Four U.S. Health Care Systems. Cancer Epidemiol Biomarkers Prev 2016;25:344–50. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Chien C, Morimoto LM, Tom J, et al. Differences in colorectal carcinoma stage and survival by race and ethnicity. Cancer 2005;104:629–39. [DOI] [PubMed] [Google Scholar]
- 25.DeSantis C, Naishadham D, Jemal A. Cancer statistics for African Americans, 2013. CA Cancer J Clin 2013;63:151–66. [DOI] [PubMed] [Google Scholar]
- 26.Gomez SL, O’Malley CD, Stroup A, et al. Longitudinal, population-based study of racial/ethnic differences in colorectal cancer survival: impact of neighborhood socioeconomic status, treatment and comorbidity. BMC Cancer 2007;7:193. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Lo SH, Halloran S, Snowball J, et al. Colorectal cancer screening uptake over three biennial invitation rounds in the English bowel cancer screening programme. Gut 2015;64:282–91. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Goebel M, Singal AG, Nodora J, et al. How can we boost colorectal and hepatocellular cancer screening among underserved populations? Curr Gastroenterol Rep 2015;17:22. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Zeger SL, Liang KY, Albert PS. Models for longitudinal data: a generlized estimating equation approach. Biometrics 1988; 44: 1049–60. [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.