Outreach and Inreach Strategies for Colorectal Cancer Screening Among Latinos at a Federally Qualified Health Center: A Randomized Controlled Trial, 2015–2018

Abstract

Objectives. To compare usual care, inreach consisting of one-on-one education, mailed outreach offering a fecal immunochemical test (FIT), and a combination of outreach and inreach for promoting colorectal cancer (CRC) screening.

Methods. We conducted a 4-arm randomized controlled trial from 2015 to 2018 at a US federally qualified health center near the California–Mexico border primarily serving low-income Hispanics/Latinos. A total of 673 individuals aged 50 to 75 years not up to date with screening were assigned to 1 of the 4 intervention groups. The primary outcome was CRC screening through 6 months follow-up.

Results. A total of 671 patients were included in intention-to-screen analyses. Their mean age was 59.9 years, 48.9% were male, and 86.3% were primarily Spanish-speaking. Screening was 27.5% for usual care (95% confidence interval [CI] = 0.21, 0.34), 52.7% for inreach (95% CI = 0.45, 0.60), 77.2% for outreach (95% CI = 0.71, 0.83), and 78.9% for combination of inreach and outreach (95% CI = 0.73, 0.85; P < .001 for all comparisons except P = .793 for outreach vs combination).

Conclusions. Among individuals at high risk for noncompletion, inreach with one-on-one education nearly doubled, and outreach offering mailed FIT alone or in combination with inreach nearly tripled screening compared with usual care. Mailed FIT outreach was superior to inreach for promoting screening.

Colorectal cancer (CRC) is the second leading cause of cancer death in the United States. CRC accounts for more than 145 000 new cancer diagnoses and more than 51 000 deaths annually.¹ Screening can reduce CRC incidence and mortality,^2–4 but it is underutilized in the United States. The overall screening rate is 62%, and it is even lower among underserved populations such as racial/ethnic minorities, those with low income and education, or those who are underinsured or uninsured.⁵ At 47%, Hispanics/Latinos (hereafter referred to as Latinos) have among the lowest screening rates and are at increased risk for higher stage at diagnosis.^5,6 Because Latinos are the largest and one of the fastest-growing ethnic populations in the United States,⁷ strategies for increasing CRC screening in this group are needed.

Evidence-based strategies for promoting CRC screening include one-on-one education, patient reminders, provider education, provider reminders, offering screening at time of annual influenza vaccination, patient navigation addressing barriers to screening, and mailed outreach offering screening.^8–17 Multicomponent interventions including more than 1 of these strategies appear to be superior to single-component interventions. Interventions including mailed outreach or patient navigation appear to be associated with the highest absolute increases in screening rates compared with usual care, with absolute increases in screening estimated to be 22% and 18% for outreach and navigation, respectively.¹⁶ Overall, it appears that multiple strategies, in particular those including mailed outreach with or without navigation, are ripe for implementation in primary care settings.¹⁸

Despite availability of evidence-based strategies, a key challenge for health care organizations is selecting what strategies to implement, a choice that should be informed by available resources for delivery. Strategy-specific resources required may vary widely. For example, resources required for an inreach intervention consisting of one-on-one education and navigation include space to deliver education, trained personnel to deliver education and navigation for each patient not up to date with screening, and time to track and follow up with each patient. By contrast, resources required for a mailed outreach intervention with offers to complete screening include a system for identifying individuals not up to date across the health care population such as a comprehensive data query of electronic medical records (EMRs), generating and tracking screening offers, and personnel to deliver invitations, provide telephone reminders, and navigation. Most strategies have not been compared head to head with other strategies with respect to effectiveness for promoting screening, making selection of strategies based on comparative effectiveness a challenge. To our knowledge, head-to-head effectiveness trials in federally qualified health centers (FQHCs) of strategies requiring distinctly different resources for implementation have not been completed.

Our aim was to conduct a 4-arm pragmatic randomized trial to compare usual care, inreach consisting of one-on-one screening education, mailed outreach offering a fecal immunochemical test (FIT), and combination of outreach and inreach for promoting CRC screening completion. Our primary outcome was screening completion within 6 months; we examined screening completion by gender and primary language in stratified analyses. To focus on a population at highest risk for adverse CRC outcomes that has been traditionally understudied, the study was set at an FQHC serving low-income Latinos near the US–Mexico border. As such, the study serves to fill evidence gaps on optimal strategies to promote screening among populations with low rates of screening, as well as on how in-clinic inreach interventions directly compare with system-level outreach interventions.

METHODS

This randomized trial was conducted at an FQHC in the southern region of San Diego, California, from 2015 to 2018. Within this region, Latinos experience cancer disparities in terms of poor access to care, screening,¹⁹ time to treatment, and survival.²⁰ The FQHC has 14 primary care clinic sites and serves more than 97 000 registered patients annually, most of whom are low-income and Spanish-speaking Latinos. In 2015, the CRC screening rate at this FQHC for patients aged 51 through 74 years was 42.9%.²¹

Study Sample and Description

To identify the study sample, we queried the FQHC’s EMRs to identify individuals aged 50 to 75 years, not up to date with CRC screening, who had self-identified as Latino, and had at least 1 primary care visit in the preceding year, served by a single primary care clinic. Medical charts were reviewed to confirm CRC screening status. Not up to date was defined based on absence of fecal occult blood test (FOBT) or FIT in the past 12 months, flexible sigmoidoscopy in the past 5 years, or colonoscopy in the past 10 years.² Those with a family history of CRC, personal history of inflammatory bowel disease, or personal history of CRC or colorectal polyps; who were under active treatment of cancer; or who were without a scheduled primary care visit within 2 months of eligibility assessment documented within the EMR were also excluded.

For individuals meeting inclusion criteria after EMR review, the study biostatistician generated a randomization schedule for group assignment by using online computer software. Randomization was stratified by gender, and men were oversampled with goal of achieving a 50:50 ratio of men to women. Researchers were blinded to study assignment until data collection was complete. To study the intervention effects in a “real-world” unbiased context, we obtained a waiver of consent. In lieu of a written informed consent, we provided a 1-page bilingual informational “research fact sheet” that outlined the goals of the study, potential loss of confidentiality associated with research, and an option to “opt out” of the research study.

For the primary outcome of screening completion within 6 months, we estimated the sample size by considering 5 statistical comparisons: usual care (UC) versus inreach, UC versus outreach, UC versus inreach plus outreach, inreach plus outreach versus outreach alone, and inreach plus outreach versus inreach alone, with a type I error α = 0.01 for each comparison (Bonferroni correction for 5 comparisons) and power of greater than or equal to 0.8.

In planning the study, we assumed the following rates of screening on follow-up among individuals not up to date at baseline: 10% for UC, 25% for inreach and outreach, and 45% for inreach plus outreach combined. The 10% rate for UC and 25% estimate for outreach were conservative assumptions based on our previous work.²² The 25% rate for inreach was a conservative estimate based on observations that inreach interventions have boosted screening completion (among patients referred for colonoscopy) by 41% to 78%. The 45% rate for both interventions combined was considered the minimal clinically significant difference that would justify implementing 2 different interventions for increasing screening. Based on these assumptions, we a priori planned to assign 162 individuals to each intervention group (n = 648). We chose 6 months as the follow-up time because of practical constraints regarding the time available for the research project; adjudicating at 6 months has been used in multiple other studies of CRC screening interventions.^23–26

Interventions

We randomly assigned eligible participants to (1) UC opportunistic offers for screening (n = 167), (2) inreach with in-clinic education including an offer to complete screening with FIT or colonoscopy (n = 168), (3) outreach with invitation to complete an enclosed FIT (n = 171), or (4) combination of outreach and inreach (n = 167).

During the study period, UC at the study site did not include clinical reminders, incentives, or other triggers promoting CRC screening. UC recommendations were at the discretion of the providers and included offering FIT, sigmoidoscopy, and colonoscopy; colonoscopy and sigmoidoscopy required separate clinic visits for completion outside the FQHC. Inreach consisted of a 10-minute scripted session with a bilingual “flipchart” and an offer to complete screening at the time of a UC visit. Mailed outreach components included a culturally tailored postcard primer 2 weeks after randomization and 2 weeks before a subsequent mailed package, which contained a cover letter from a primary care doctor, low-literacy instructions, and a self-addressed stamped envelope with the enclosed FIT to do at home. Outreach and inreach were delivered by bilingual, bicultural trained patient navigators and included telephone reminders for FIT completion. Intervention materials were developed based on the chronic care model relevant for cancer prevention in primary care settings (e.g., use of care teams, decision support, proactive care coordination, promotion of community resources, and clinical information services)²⁷ and were developed and pretested through formative work before implementation in this trial, as previously reported.²⁸

Measures

Following UC protocols, all normal FIT results were communicated to patients via mail, and shared with the primary care provider; abnormal FIT results were communicated to the patient one on one in a primary care provider visit. After the abnormal result was shared with the patient, a referral for colonoscopy was provided. The study team followed up with the participant to ensure that the colonoscopy was completed.

The primary outcome was screening completion within 6 months of randomization, defined as completion of a FIT, colonoscopy, or sigmoidoscopy, ascertained through manual review of EMRs. In secondary analyses, we examined completion stratified by gender and by preferred primary language. We conducted stratification by gender because previous work has suggested that Latino men have lower screening rates than women,²⁹ and impact of interventions for promoting screening has been understudied among Latino men.³⁰ Stratification by primary language was conducted because we postulated that primary language could serve as a proxy for acculturation and other factors that might have an impact on responsiveness to the interventions under study.^31–33

Additional outcomes of interest included intervention group–specific number with abnormal FIT and proportion with abnormal FIT who had colonoscopy. Assessment of receipt of allocated intervention was based on absence of returned mail for outreach, receipt of in-clinic education for inreach, and completion of a clinic visit for UC, assessed through intervention tracking notes and EMR review for intervention groups and UC, respectively.

Analysis

We used descriptive statistics to assess differences in baseline characteristics across groups. We estimated CRC screening rates for each intervention, and 95% confidence intervals (CIs) based on normal approximation or binomial exact test, as appropriate, were provided. For our primary intention-to-screen analysis, we used the Fisher exact test for an overall comparison of the CRC screening rates across groups as well as 5 pairwise comparisons of interest listed previously, using a P level of less than .05 for overall test and a P level of .01 for each pairwise comparison with delineate statistical significance. In the secondary analyses, we conducted an overall comparison of CRC screening rate across groups stratified by gender and language. We computed the number needed to achieve 1 additional screening completion by taking the inverse of the absolute difference in screening completion percentage between each active intervention group and UC. Formal statistical comparisons of number with abnormal FIT and diagnosis of adenoma and CRC were not planned or performed because of the expected and observed small number of these outcomes.

RESULTS

The study flow is depicted in Figure 1. Among 3183 individuals assessed for study eligibility, 2510 were excluded resulting in 673 participants who were randomized to 1 of the 4 intervention groups. Reasons for exclusion included being up to date with screening (colonoscopy in past 10 years n = 1749; sigmoidoscopy in past 5 years n = 5; FIT or FOBT in past year n = 201), absence of upcoming clinic visit within 2 months of eligibility (n = 553), and active treatment of cancer (n = 2). Two individuals opted out of study data collection after randomization, making the final intention-to-screen analytic sample 671 (UC n = 167; inreach n = 167; outreach n = 171; combination n = 166). The mean age was 59.9 years, 48.9% were male, 86.3% were primarily Spanish-speaking, 66.9% were insured by Medicaid, and 7.6% were uninsured. We observed no statistically significant differences in baseline characteristics across groups (Table 1).

FIGURE 1—

Study Flow and Outreach and Inreach Strategies for Colorectal Cancer Screening Among Latinos at a Federally Qualified Health Center: San Diego, CA, 2015–2018

Note. CRC = colorectal cancer; FIT = fecal immunochemical test; FOBT = fecal occult blood test. Study flow, including number assessed for eligibility, excluded, randomized, and analyzed, is depicted.

TABLE 1—

Baseline Characteristics and Outreach and Inreach Strategies for Colorectal Cancer Screening Among Latinos at a Federally Qualified Health Center: San Diego, CA, 2015–2018

Baseline Characteristics	Total Sample (n = 671), Mean ±SD or % (No.)	Study Group				P
		Inreach (n = 167), Mean ±SD or % (No.)	Outreach (n = 171), Mean ±SD or % (No.)	Combination (n = 166), Mean ±SD or % (No.)	Usual Care (n = 167), Mean ±SD or % (No.)
Age, y	59.9 ±6.6	59.4 ±6.5	60.0 ±7.1	60.0 ±6.5	60.1 ±6.3	.7
Gender						.99
Female	51.1 (343)	50.9 (85)	50.3 (86)	51.8 (86)	51.5 (86)
Male	48.9 (328)	49.1 (82)	49.7 (85)	48.2 (80)	48.5 (81)
Insurance						.54
Yes	92.4 (620)	92.2 (154)	93.6 (160)	89.8 (149)	94.0 (157)
No	7.6 (51)	7.8 (13)	6.4 (11)	10.2 (17)	6.0 (10)
Insurance type						.09
Medi-Cal managed care	66.9 (449)	72.5 (121)	69.0 (118)	62.0 (103)	64.1 (107)
Medicare	19.5 (131)	16.8 (28)	20.5 (35)	21.7 (36)	19.2 (32)
Private insurance	6.0 (40)	3.0 (5)	4.0 (7)	6.0 (10)	10.8 (18)
Uninsured	7.6 (51)	7.8 (13)	6.4 (11)	10.2 (17)	6.0 (10)
Preferred language						.42
English	13.4 (90)	15.6 (26)	13.5 (23)	13.9 (23)	10.8 (18)
Spanish	86.3 (579)	83.2 (139)	86.6 (148)	86.1 (143)	89.2 (149)
Other	0.3 (2)	1.2 (2)	. . .	. . .	. . .

Primary Results

On intention-to-screen analysis, CRC screening completion was 27.5% for UC (95% CI = 0.21, 0.34), 52.7% for inreach (95% CI = 0.45, 0.60), 77.2% for outreach (95% CI = 0.71, 0.83), and 78.9% for combination (95% CI = 0.73, 0.85; P < .001 for all 5 between-group pairwise comparisons except P = .793 for outreach vs combination; Figure 2). The number needed to achieve 1 additional screen compared with UC was 1.9 for inreach, 1.3 for outreach, and 1.3 for combination; number needed to screen was 4 for outreach compared with inreach.

FIGURE 2—

Screening Completion at 6-Month Follow-Up, Intention-to-Screen Analysis, Outreach and Inreach Strategies for Colorectal Cancer Screening Among Latinos at a Federally Qualified Health Center: San Diego, CA, 2015–2018

Note. CI = confidence interval. Colorectal cancer (CRC) screening completion was defined as fecal immunochemical test, colonoscopy, or sigmoidoscopy 6 months from randomization as measured in the electronic medical record. On intent-to-screen analysis (n = 671), CRC screening completion was 27.5% for usual care, 52.7% for inreach, 77.2% for outreach, and 78.9% for combination of inreach and outreach. Five between-group pairwise comparisons were calculated, and the P values were adjusted for 5 prespecified comparisons by Bonferroni correction. They were all significantly different (P < .001) except for outreach vs combination (P = .793).

Screening by Gender and Language

Among men, CRC screening completion was 25.9% for UC (95% CI = 0.16, 0.35), 53.7% for inreach (95% CI = 0.43, 0.64), 67.1% for outreach (95% CI = 0.57, 0.77), and 72.5% for combination (95% CI = 0.63, 0.82; P < .001 for overall comparison). Among women, CRC screening completion was 29.1% for UC (95% CI = 0.19, 0.39), 51.8% for inreach (95% CI = 0.41, 0.62), 87.2% for outreach (95% CI = 0.80, 0.94), and 84.9% for combination (95% CI = 0.77, 0.92; P < .001 for overall comparisons; Figure 3).

FIGURE 3—

Screening Completion at 6-Month Follow-Up, Intention-to-Screen Analysis, Stratified by Primary Language and Gender, Outreach and Inreach Strategies for Colorectal Cancer Screening Among Latinos at a Federally Qualified Health Center: San Diego, CA, 2015–2018

Note. Colorectal cancer (CRC) screening completion was defined as fecal immunochemical test, colonoscopy, or sigmoidoscopy 6 months from randomization as measured in the electronic medical record. CRC screening was significantly different across intervention groups in stratified analyses (P < .001).

Among those identifying English as their primary language, CRC screening completion was 27.8% for UC (95% CI = 0.10, 0.53), 38.5% for inreach (95% CI = 0.20, 0.59), 69.6% for outreach (95% CI = 0.47, 0.87), and 65.2% for combination (95% CI = 0.43, 0.84; P = .014 for overall comparisons). Among those identifying Spanish as their primary language, CRC screening completion was 27.5% for UC (95% CI = 0.20, 0.35), 56.1% for inreach (95% CI = 0.48, 0.64), 78.4% for outreach (95% CI = 0.72, 0.85), and 81.1% for combination (95% CI = 0.75, 0.88; P < .001 for overall comparisons; Figure 3).

Screening Follow-Up

Among 671 individuals who were randomized, 397 completed a screening test: 312 with FIT, 1 with sigmoidoscopy, and 84 with colonoscopy. Distribution of test type by each group is depicted in Appendix A (available as a supplement to the online version of this article at http://www.ajph.org). For UC, 13.2% completed FIT, 13.8% completed colonoscopy, and 0.6% completed sigmoidoscopy. For inreach, 34.7% completed FIT, 18.0% completed colonoscopy, and 0% completed sigmoidoscopy. For outreach, 66.7% completed FIT, 10.5% completed colonoscopy, and 0% completed sigmoidoscopy, and for combination of outreach and inreach, 71.1% completed FIT, 7.8% completed colonoscopy, and 0% completed sigmoidoscopy. Of the 312 patients who returned a FIT test, 21 patients had an abnormal FIT (5 from UC, 0 from inreach, 6 from outreach, and 10 from combination), and 14 (66.7%) completed a colonoscopy (4 from UC, 2 from outreach, and 8 from combination). Reasons for colonoscopy noncompletion were medical comorbidities precluding colonoscopy for 2 individuals (1 UC, 1 combination), and unknown for 5 individuals (0 UC, 4 outreach, 1 combination).

DISCUSSION

In a 4-arm pragmatic randomized trial for increasing CRC screening in a FQHC clinic system serving low-income Latinos, we found inreach with one-on-one education nearly doubled (53%), and outreach offering mailed FIT alone (77%) or in combination with inreach (79%) nearly tripled, screening compared with UC (28%). Mailed outreach was superior to inreach with one-on-one education for promoting screening completion. Results were qualitatively similar among men and women, but screening participation across each intervention group was higher among individuals who preferred Spanish versus English as their primary language.

Superiority of mailed outreach and of inreach compared with UC offers for CRC screening has been consistently demonstrated through randomized controlled trials.^8,16 Randomized controlled trials of mailed outreach and patient education have often included additional components such as reminders¹⁶ to complete screening, components that were included in both of our outreach and inreach strategies, confirming that our strategies were generally similar to those used in previous work. Both mailed outreach and patient education have been shown to specifically increase screening among underserved populations, including Latinos.^{9,16,17,22,34–40} Thus, our results showing superiority of mailed outreach compared with UC and superiority of inreach including patient education compared with UC among Latinos confirm previous findings and extend current knowledge by demonstrating that mailed outreach (including a mailed primer, invitation to complete screening, and a live telephone reminder) is superior to another established strategy, one-on-one education (with a live telephone reminder).

Our results fill a major gap in current literature because we conducted the first head-to-head trial comparing inreach and outreach—2 logistically very different screening promotion strategies—and found that mailed outreach was substantially superior to inreach. Furthermore, we found that the combination of both, while superior to UC, was not superior to mailed outreach alone for screening completion. For leaders in cancer control, public health, and health care organizations such as FQHCs, these results suggest that a mailed outreach strategy, because of its ability to reach patients outside of UC visits, may be a first-line choice for screening promotion over implementation of one-on-one education as a strategy, particularly in resource-strapped settings. The choice between implementing mailed outreach versus one-on-one education has significant implications, as the resources required to implement mailed outreach are distinct from resources required to deliver one-on-one education.

The results also emphasize the need for future additional studies comparing proven CRC screening strategies head to head, rather than only against UC, given that there are now multiple strategies that have been confirmed to work in multiple trials. Primary care CRC screening strategies previously shown superior to UC that could be considered for head-to-head comparison include offering FIT at the time of annual influenza vaccination,¹⁴ patient navigation,⁹ and provider reminders.¹⁷ Choice of whether to compare these strategies individually or to include easier-to-implement strategies (such as provider reminders through EMRs) as a baseline strategy for both comparison groups must be based on assessments of costs and feasibility.

Our results also highlight several additional opportunities for promoting screening. Latino men have among the lowest rates of screening compared with other groups nationally (50% vs 63%)⁶ and have generally been understudied. We purposely oversampled Latino men and found that mailed outreach and inreach with education can substantially increase screening completion for this group, suggesting that either intervention should be considered for implementation in settings with substantial male Latino populations. We also found our interventions were more effective among individuals identifying Spanish as their primary language. Some research has shown that Spanish-speaking Latinos are more likely to receive cancer screenings than English-speaking Latinos,³² while other studies have shown that English-speaking Latinos are more likely to screen.^31,33 Future research is needed to understand the role that language and acculturation play in intervention effectiveness in FQHC settings.

Limitations

The study setting was a clinic in the southernmost region of California, adjacent to the Mexico border, and is 60% Latino (as compared with 34% for San Diego County).⁴¹ Cultural beliefs and practices of this population may limit generalizability of these study results to some other health settings.

We did not study implementation of interventions across multiple clinics. Coronado and colleagues conducted a large cluster randomized trial comparing intent to implement mailed outreach versus UC across 26 FQHCs in Oregon and California.^39,42 While their results showed that intervention clinics had higher rates of screening completion compared with UC clinics (14% vs 10%), the absolute difference was much smaller than the average 28% absolute difference observed by a meta-analysis of multiple mailed outreach randomized controlled trials.³⁶ Coronado’s study underscores that success of mailed outreach may differ when implemented widely and may depend heavily on the approach to implementation taken. Implementation challenges may also apply to patient education and were not a focus of this work.

We assessed a package of interventions for mailed outreach (mailed primer, invitation with included FIT kit, and live telephone reminders) and inreach (one-on-one education using a flip chart, FIT distribution or referral to colonoscopy, and live telephone reminders). As such, our study was not designed to address questions such as whether mailed FIT could have been equally effective with use of text messages or automated phone primers or reminders, or in absence of such primers or reminders. Outreach was shown to be superior to inreach, but many individuals assigned inreach alone or in combination did not receive in-clinic education. Indeed, in per-protocol analyses, screening completion was 88.4% (76 of 86) for the patients assigned inreach who completed a clinic visit and the allocated education and 78.9% (131 of 166) for patients assigned combination who both received mailed outreach and completed a clinic visit and in-clinic education.

Our intent-to-screen analysis excluded 2 individuals who opted out of the study, including further interventions and data collection regarding screening completion. In a posthoc analysis, inclusion of these individuals in the intent-to-screen analysis, assuming both as non–screening completers, resulted in qualitatively similar results for screening completion: 27.5% for UC, 52.4% for inreach, 77.2% for outreach, 78.4% for combination (P < .001 for all 5 pairwise comparisons except for combination vs outreach). Cost-effectiveness analysis was not conducted, though previous work has suggested that mailed outreach can be highly cost-effective.^43–45 Imbalance in randomization across providers could have led to differences in screening completion, as UC was a component of all intervention arms. Provider characteristics were not collected to allow for assessment of such imbalance, though randomization procedures should have prevented imbalance in provider characteristics and other unmeasured potential confounders.

Strengths of this study included its focus on Latinos (including men) who have been traditionally understudied, head-to-head comparison of 2 proven strategies for screening completion, and use of a waiver of written informed consent to maximize participation by the target population selected for interventions.

Public Health Implications

In the first head-to-head comparison of previously proven strategies for CRC screening, we found inreach with one-on-one education nearly doubled, and outreach offering mailed FIT alone or in combination with inreach nearly tripled, screening compared with UC. Mailed outreach was superior to inreach for promoting screening completion, making it potentially ripe for widespread implementation. Studies of implementation strategies to optimize the impact of mailed outreach in UC practice, as well as of head-to-head comparisons of other strategies for promoting CRC screening, are needed to help further optimize uptake and impact of CRC screening.