The Reach of Calls and Text Messages for Mailed FIT Outreach in the PROMPT Stepped-Wedge Colorectal Cancer Screening Trial

Abstract

Background

Mailed fecal immunochemical test (FIT) outreach can improve colorectal cancer screening participation. We assessed the reach and effectiveness of adding notifications to mailed FIT programs.

Methods

We conducted secondary analyses of a stepped-wedge evaluation of an enhanced mailed FIT program (n = 15 clinics). Patients were stratified by prior FIT completion. Those with prior FIT were sent a text message (Group 1); those without were randomized 1:1 to receive a text message (Group 2) or live phone call (Group 3). All groups were sent automated phone call reminders. In stratified analysis, we measured reach and effectiveness (FIT completion within 6 months) and assessed patient-level associations using generalized estimating equations.

Results

Patients (n = 16,934; 83% Latino; 72% completed prior FIT) were reached most often by text messages (78%), followed by live phone calls (71%), then automated phone calls (56%). FIT completion was higher in patients with prior FIT completion vs. without (44% (Group 1) vs. 19% (Group 2 + Group 3); p <.01). For patients without prior FIT, effectiveness was higher in those allocated to a live phone call (20% (Group 3) vs. 18% (Group 2) for text message; p = .04) and in those who personally answered the live call (28% vs. 9% no call completed; p < .01).

Conclusion

Text messages reached the most patients, yet effectiveness was highest in those who personally answered the live phone call.

Impact

Despite the broad reach and low cost of text messages, personalized approaches may more successfully boost FIT completion.

INTRODUCTION

Mailed fecal immunochemical test (FIT) outreach has been shown to increase colorectal cancer (CRC) screening participation in diverse healthcare settings, with average absolute improvements of 22–28% over usual care.(1–3) Prior studies have reported greater effectiveness in Latino populations than non-Latino White patients.(4–6) Most mailed FIT outreach programs include advance notifications and reminders and use of such communications are among best practices on mailed FIT outreach recommended by a panel of experts convened by the Centers for Disease Control and Prevention.(7) While these communications are considered a best practice, few studies have evaluated how well they perform at reaching diverse population subgroups.

Prior research conducted by our team evaluated reach of a health-plan-led mailed FIT outreach program.(8) We found that, of 3,386 age-eligible members overdue for CRC screening, 2,615 (77%) were successfully delivered a FIT kit by mail (i.e., reached), which was defined as the FIT kit not being returned as undeliverable by the post office. Reasons for not being reached included not being considered a clinic patient (478; 14%) or having issues with receiving mail (290; 9%). Lower reach was observed among men (vs. women), those with Medicaid insurance (vs. Medicare), and those not having had a primary care visit in the past year (vs. 4+ visits). However, this prior study did not sample an ethnically diverse population, did not ask about preferred language, and all study participants had health-care coverage. Moreover, this prior evaluation focused on reach for the core intervention component (i.e., having been mailed a FIT), rather than reach of advance notifications and reminders provided by phone or text message, or the impacts on FIT completion.

Here we report on results of research examining reach and effectiveness of a program to send advance notifications and reminders for a mailed FIT outreach program. The program was conducted by a large Latino-serving federally qualified health center (FQHC), headquartered in Los Angeles. We report reach for each intervention component (advance notifications, automated, and live reminders) across relevant socio-demographic characteristics (age, sex, ethnicity, preferred language) and health care utilization variables. Our findings can inform improvements to mailed FIT outreach programs to achieve greater equity and improve screening rates in populations who lag in CRC screening completion rates.

MATERIALS and METHODS

Study Setting

The Participatory Research to Advance Colon Cancer Prevention (PROMPT) study setting, rationale, and design have been reported previously.(9) Briefly, the study was conducted at fifteen clinics (clusters) within a large FQHC based in Southern California, using an open cohort stepped-wedge randomized controlled design. The FQHC operates 25 medical clinics and serves more than 270,000 patients annually. The FQHC has conducted mailed fecal test outreach to patients overdue for CRC screening since 2016. In 2018, they served 41,050 patients between the ages of 50 and 75, of whom more than 97% had English (39%) or Spanish (59%) listed as their preferred language. The FQHC uses the InSureONE FIT (Clinical Genomics; Bridgewater, NJ), which requires collection of two specimens from a single stool sample. The FQHC has near-complete capture of FIT events in the electronic health record (EHR) which is facilitated by a direct electronic interface with the reference laboratory (Quest).

Intervention Development

In developing the intervention, including the format and content for advance notifications and reminders, we did not strictly apply a behavioral theoretical framework. Instead, we used a validated community engagement process (boot camp translation), with separate sessions conducted in English and Spanish.(10,11) All materials (e.g., text messages, automated and live phone call scripts) were approved by the health center’s marketing department and are provided in Supplemental Table 1. The intervention was pilot tested prior to the main trial.(12) Based on the pilot findings, we designed the main trial to compare the addition of advance notifications and automated call reminders (enhanced mailed FIT outreach) because it achieved the highest FIT completion in the pilot study (35.7%), to live outreach alone (the usual care practice at the health center). Given the low FIT completion among screening-naïve patients in the pilot (17% vs. 45% for patients with prior CRC screening), we conducted an embedded patient-level randomization to live phone call advance notifications or text message advance notifications in patients with no EHR evidence of prior FIT screening. We anticipated that screening-naïve patients would be more likely to respond to an advance notification delivered by live phone call than by text message. Our main-trial findings showed that the enhanced program boosted FIT completion at 6 months, both in intention-to-treat [+2.8%, 95% confidence interval (CI; 0.4–5.2)] and per-protocol [limited to individuals who were reached; +16.9%, 95% CI (12.3–20.3)] analyses.(13)

All procedures and intervention materials were reviewed and approved by the Institutional Review Board of Kaiser Permanente Northwest, with ceding agreements from the FQHC and Oregon Health & Science University. The study was conducted in accordance with recognized ethical guidelines of the Belmont Report, the US Common Rule, and the US Office of Civil Rights. The study obtained a waiver of informed consent and authorization for use of protected health information, given the minimal risk posed to patients. The trial was registered in clinicaltrials.gov (National Clinical Trial (NCT) Identifier NCT03167125).

Patient Eligibility

We used codes from prior research to identify average risk patients who were due for CRC screening and eligible for the mailed FIT intervention.(14) Eligible patients were ages 50–75, overdue for CRC screening based on EHR data, had at least two clinic visits in the prior 24 months, had an address listed in the EHR, and had no known history of colectomy, CRC, or a known life-limiting condition (e.g., on hospice or diagnosis of metastatic cancer).

Study Design

The trial used a stepped-wedge design to test the effectiveness and implementation of the mailed FIT advance notifications and reminders across the 15 clinics within the health system (Consolidated Standards of Reporting Trials (CONSORT), Figure 1).(9) The 15 clinics were randomized to one of two wedges based on two stratification factors: proportion of eligible patients up-to-date with CRC screening recommendations, and proportion of patients who had Spanish listed as their preferred language in the medical record. Both wedges delivered usual care during the baseline period (June 25, 2017 – June 24, 2018). Wedge 1 began the intervention during year 1 (June 25, 2018 – June 24, 2019), while Wedge 2 remained in usual care. Wedge 2 began the intervention in year 2 on June 25, 2019; data were collected through June 24, 2020.

Figure 1.

Consolidated Standards of Reporting Trials (CONSORT) Study Flow Diagram

Figure 1 shows the design of the stepped-wedge trial of advance notification and reminders to a mailed FIT outreach program.

Training to implement live call advance notifications

Centralized clinic outreach staff were trained to deliver live advance notification phone calls during a 4-hour in-person training delivered (in English and Spanish) by the study team. Training addressed CRC and screening, effective messages to promote screening, and phone call tracking. Outreach staff were given phone call scripts that included answers to frequently asked questions (in English and Spanish).

Usual care condition: Standard mailed FIT program

Patients identified in clinics during the pre-intervention timepoints (standard mailed FIT program; usual care) were mailed a FIT and, starting six weeks later, were delivered a live reminder call (up to three attempts were made to reach patients) by clinic staff on care teams, as time allowed.

Intervention condition: Enhanced mailed FIT program

The intervention has been described previously, and included patient-level stratification based on prior FIT completion and an embedded patient-level randomization for patients who had never completed a prior FIT (patients who never completed a prior FIT may or may not have been offered a FIT in the past).(9) Patients were categorized by: (1) those who had completed a FIT in the past (Group 1), and (2) those who had never completed a FIT (Groups 2 and 3). Group 1 patients were delivered the following components: (1) an advance notification one-way text message (1–2 days prior to anticipated receipt of the mailed FIT); (2) a mailed FIT test; (3) a series of two automated phone call reminders (sent 2 weeks and 4 weeks following the mailing); and (4) live reminder calls delivered at the discretion of clinic staff on care teams (beginning six weeks following the mailing), consistent with usual care. Patients who never completed FIT were randomized to receive a vendor-delivered advance notification text message (Group 2) on the same schedule as for Group 1, or an advance notification live phone call delivered by centralized outreach staff who made up to three attempts to reach the patient, leaving a voicemail on the final attempt (Group 3). There were 12,138 participants in Group 1 (prior CRC screening, text), 2,548 in Group 2 (no prior CRC screening, text), and 2,248 in Group 3 (no prior CRC screening, live call). The remaining intervention components (mailed FIT and reminders) were delivered on the same schedule as for Group 1 patients. Text messages and automated calls were delivered and tracked by a contracted vendor (Stericycle Communication Solutions). The outgoing phone number for the automated and live outreach was a local number (to encourage the participant to answer the call), and the timing and outcomes of the live call advance notifications were recorded by health education staff using an Excel tracking form.

Definitions of Reach and Effectiveness

We assessed the proportion of eligible patients who were reached by each intervention component (text message, live phone call advance notification; automated phone call reminder). Wae define reach according to the RE-AIM framework as the absolute number, proportion, and representativeness of individuals who are willing to participate in a given intervention.(15) Reach for a text message was defined as the text message having been successfully sent (as we could not confirm whether it was read). Reach for an automated phone call was defined as the call having been answered by the patient or a voice message having been left. Reach data for the text message and automated phone calls were obtained from the vendor. Reach for the live phone calls (left voice mail, patient answered) was documented by outreach staff using the study’s Excel tracking system. We defined effectiveness as completing a FIT within 6 months of the date patients were identified as eligible for the study (i.e., list pull date).

Data Analysis

Our analysis is limited to patients in clinics at the post-intervention timepoints, as these components were not delivered during the usual care phase (Figure 1). We report the demographic and health care utilization patterns of patients in Group 1, Group 2, and Group 3. Using the definitions above, we report reach and effectiveness for each intervention component.

We calculated the association between reach (outcome) and health care utilization and demographic variables (independent variables; race/ethnicity, language preference, household income, etc.) using generalized estimating equations with logit link and binomial distribution. For these analyses, we focused on three questions: First, how were the associations between patient factors and reach for the text message moderated by prior FIT completion? For this analysis, we compared the patient factors-reach associations in those who had ever completed a prior FIT (Group 1) to those who had never completed a prior FIT (Group 2). Second, how were associations between patient factors and reach for the automated phone call moderated by prior FIT completion? For this analysis, we compared the patient factors-reach associations in those who had ever completed a prior FIT (Group 1) and those who had never completed a prior FIT (Group 2 plus Group 3). Finally, in the subgroup of patients who never completed prior FIT (n = 4,796), how were associations between patient factors and reach moderated by outreach modality (text vs. live phone call)? This analysis compared the associations of patient factors and reach by outreach modality (text message, Group 2 vs. live phone call, Group 3). We performed analysis of interactions by expanding the generalized linear mixed models to include our stratification variables (i.e., ever complete FIT, never completed FIT; text message, live call) and the product of the stratification variable and patient characteristic (representing the interaction) as independent variables. We graphed the simple main effects and associated 95% confidence intervals using forest plots and report p values for the interactions.

Data Availability Statement

The data generated in this study are not publicly available under data use agreements, but are available upon reasonable request from the corresponding author.

RESULTS

Eligible patients at post-intervention timepoints (n=16,934) were mostly ages 50–64 (79%), and 56 percent were female (Table 1). Eighty-three percent had Latino ethnicity and over two-thirds had Spanish listed as their preferred language in the EHR. Half of the sample had a household income at or below the federal poverty level, and 83 percent had a recorded clinic visit in the past year. Seventy-five percent of the sample lived in Los Angeles County; the rest lived in Orange County.

Table 1.

Demographic Characteristics of Patient Population in Post-Intervention Groups (Wedge 1: Years 1, 2 (8 clinics); Wedge 2: Year 2 (7 clinics))

Characteristic	Ever Completed Prior FIT		Never Completed Prior FIT				Total (N = 16,934)
	Text Message (Group 1) (n = 12,138)		Text Message (Group 2) (n = 2,548)		Live Phone Call (Group 3) (n = 2,248)
	N	%	N	%	N	%	N	%
Age Mean (SD)	59.9	(6.0)	57.1	(6.0)	57.4	(6.1)	59.1	(6.1)
50–64	9329	76.9	2207	86.6	1906	84.8	13442	79.4
65–75	2809	23.1	341	13.4	342	15.2	3492	20.6
Female Sex	7047	58.1	1321	51.8	1198	53.3	9566	56.5
Hispanic or Latino Ethnicity 1	10435	86.0	1963	77.0	1749	77.8	14147	83.5
Preferred Language 2
English	3115	25.7	1107	43.4	970	43.1	5192	30.7
Spanish	8840	72.8	1393	54.7	1228	54.6	11461	67.7
Federal Poverty Level
< 100%	6322	52.1	1170	45.9	983	43.7	8475	50.0
100% - 200%	1890	15.6	360	14.1	290	12.9	2540	15.0
>200%	321	2.6	75	2.9	83	3.7	479	2.8
Missing	3605	29.7	943	37.0	892	39.7	5440	32.1
Clinic Visit History
No visit in past year	1854	15.3	516	20.3	450	20.0	2820	16.7
Visit in past year	10284	84.7	2032	79.7	1798	80.0	14114	83.3
County
Orange	2969	24.5	694	27.2	625	27.8	4288	25.3
Los Angeles	9169	75.5	1854	72.8	1623	72.2	12646	74.7

¹data are missing for 179 (101 in Group 1; 38 in Group 2; 40 in Group 3); data are ‘other’ for 80 (49 in Group 1; 20 in Group 2; 11 in Group 3);

²data are missing for 15 (7 in Group 1; 5 in Group 2; 3 in Group 3); data are ‘other’ for 266 (176 in Group 1, 43 in Group 2, and 47 in Group 3)

Overall, FIT completion was higher in patients who had previously completed a FIT versus those who had not (44% vs. 19%; p < .01). Reach varied by intervention component (Table 2). Advance notifications were delivered to more than three-quarters of intended participants: text message advance notifications were successfully delivered to 9,467/12,138 (78%) of Group 1 patients and 2,022/2,548 (79%) of Group 2 patients; live phone calls reached 1,590/2,248 (71%) of Group 3 patients (37% were answered and for 34%, a voice message was left). Patients to whom the text message was successfully delivered had 13–14% higher odds of completing FIT than patients who did not receive the text, a difference that was only significant for the ever-completed FIT group. Compared to patients who were not reached by the phone call, FIT completion was higher for those who personally answered the phone (OR: 3.77, 95% CI: 2.68, 5.30) and for those who were left a voice message (OR: 2.22, 95% CI: 1.52, 3.23). Reminders in the form of automated phone calls reached more than half of all listed phone numbers; 6,190/12,138 (53%) of Group 1 patients (18% were answered, 35% were left a voice message), 2,390/4,796 (52%) of Group 2 and Group 3 combined (12% answered and 40% were left a voice message). Answering the automated call was associated with a more than 2-fold higher odds of completing FIT (OR: 2.23, 95% CI: 2.06, 2.42 in Group 1; OR: 2.25, 95% CI: 1.86, 2.72 in Groups 2 and 3); leaving a voice message was not significantly associated with FIT completion. In adults who had never completed prior FIT testing and who were randomized to receive a text vs. live phone call advance notification (Group 2 vs. Group 3), live phone call outreach was associated with higher FIT completion (20.0% vs. 17.7% for text messaging; p value = 0.04).

Table 2.

Reach and Effectiveness of Intervention Components, by Prior FIT Completion and Study Condition

Outreach communications	Ever Completed Prior FIT				Never Completed Prior FIT
	Reach	Effectiveness			Reach	Effectiveness
	N (Col %)	Completed FIT		Odds Ratio, 95% CI a	N (Col %)	Completed FIT		Odds Ratio, 95% CI a
		N	%			N	%
Text message advance notifications b	Group 1: N = 12,138				Group 2 (allocated to text message): N = 2,548
Delivered	9467 (78.0)	4255	44.9	1.15 (1.04, 1.27)	2011 (78.9)	367	18.2	1.17 (0.89, 1.54)
Not delivered	2671 (22.0)	1103	41.3	ref	537 (21.1)	85	15.8	ref
Live phone call advance notification c					Group 3 (allocated to live phone call): N = 2,248
Answered in-person	--	--	--	--	823 (36.6)	228	27.7	3.77 (2.68, 5.30)
Left voicemail	--	--	--	--	767 (34.1)	139	18.1	2.22 (1.52, 3.23)
No call needed					240 (10.7)	44	18.3	2.20 (1.55,3.13)
Not completed	--	--	--	--	418 (18.6)	39	9.3	ref
Mailed FIT	12,138 (100)	5358	44.1	--	4796 (100)	902	18.8	--
Eligible for reminders d	11,671 (96.2)	--	--	--	4498 (93.8)	--	--	--
Automated phone call reminders e					Groups 2 and 3: N = 4,796
Answered in-person	2105 (18.0)	1263	60.0	2.23 (2.06, 2.42)	559 (12.4)	166	29.7	2.25 (1.86, 2.72)
Answered by machine	4085 (35.0)	1627	39.8	0.99 (0.92, 1.08)	1831 (40.7)	340	18.6	1.24 (1.00, 1.53)
Not completed	5045 (43.2)	2009	39.8	ref	2000 (44.5)	307	15.4	ref

^aOdds ratio based on generalized estimating equations, accounting for clinic.

^b525 patients had ‘do not text’ in their medical record (467 who completed prior FIT, 58 who never completed prior FIT); these patients were included in ‘not delivered’ text

^cNo call was needed for 240 patients allocated to the live call primer (these patients were excluded or ineligible)

^dExcludes participants for whom no call was needed and who had ‘do not text’ in their medical record

^eNo call was needed for 544 patients allocated to the automated call reminder (436 who completed prior FIT; 108 who never completed prior FIT)

Associations and Interactions of Reach and Patient Characteristics

How were the associations between patient factors and reach for the text message moderated by prior FIT completion?

In analysis limited to those allocated to receive the text message advance notification (Group 1 and Group 2), reach was associated with age and language preference, with higher reach observed among those who were younger (50–64 vs. 65–75) and had English as their preferred language (vs. Spanish: Figure 2). Having a clinic visit in the past year was associated with reach, but only for those who had never completed a prior FIT (OR: 1.75, 95%CI: 1.30, 2.35; p for interaction = <.05). The association of reach and ethnicity was moderated by prior FIT completion (p <.01); with Latino adults having higher odds of being reached if they ever completed prior FIT, and a lower odds if they never completed prior FIT.

Figure 2.

Forest plot of associations between socio-demographic characteristics and reach, stratified by prior FIT completion, among adults who were sent a text message advance notification

Figure 2 shows the associations between socio-demographic characteristics and reach, stratified by prior FIT completion, among adults who were sent a text message advance notification. This analysis includes 14,686 individuals (Group 1: n = 12,138 who ever completed prior FIT; Group 2: n = 2,548 who never completed prior FIT).

Figure 2 footnotes: Reference categories were female; not Latino/Hispanic, English-language preference, <100% federal poverty level, no clinic visits in past year, and resident of Orange County.

How were the associations between patient factors and reach for the automated phone call moderated by prior FIT completion?

In analysis of those allocated to receive the automated call reminder (Group 1 vs. Group 2 and Group 3), reach was higher among those who had a clinic visit in the past year, irrespective of prior FIT completion history (Figure 3). Among those who had ever completed a prior FIT, reach was lower in those whose household income was more than 200% of the federal poverty level (vs. <100% FPL) and among males (vs. females). Age was associated with reach, but only for those who had never completed a prior FIT (OR: 1.19, 95%CI: 1.03, 1.37; p for interaction = .02).

Figure 3.

Forest plot of associations between socio-demographic characteristics and reach, stratified by prior FIT completion, among adults who were sent an automated phone call reminder

Figure 3 shows the associations between socio-demographic characteristics and reach, stratified by prior FIT completion, among adults who were sent an automated phone call reminder. This analysis includes 16,934 individuals (Group 1: n = 12,138 who ever completed prior FIT; Groups 2 and 3: n = 4,796 who never completed prior FIT).

Figure 3 footnotes: Reference categories were female; not Latino/Hispanic, English-language preference, <100% federal poverty level, no clinic visits in past year, and resident of Orange County.

In the subgroup of patients who never completed prior FIT (n = 4,796), how were the associations between patient factors and reach moderated by outreach modality (text vs. live phone call)?

In a subgroup analysis of adults who had never completed prior FIT testing and who were randomized to receive a text vs. live phone call advance notification (Group 2 vs. Group 3), reach was higher in the text message, compared to the live phone call, group (79% vs. 71%; Figure 4). Age was associated with reach, but only for those who were allocated to the text message (OR: .73, 95%CI: .60, .87). Compared to females, males had lower odds of being reached by the live call (OR: .78, 95%CI: .66, .93), but not for the text message. For this subgroup, we observed lower odds of being reached for the text message among Spanish-preferring adults (OR: .76; 95%CI: .64, .89) and higher odds of being reached by the live phone call among Hispanic adults (OR: 1.23; 95%CI: 1.06, 1.43; p value for interaction = <.01). A similar pattern was observed for ethnicity (p for interaction= .02). Household income was associated with reach, but only for those allocated to receive a text message (OR: .77; 95%CI: .64, .89, comparing 100%< 200% vs. <100%; p value for interaction = .01). Compared to those with no prior-year clinic visit, those with a clinic visit in the past year were more likely to be reached by either text message (OR: 1.67, 95%CI: 1.26, 2.21) or live phone call (OR: 1.28; 95%CI: 1.05, 1.57).

Figure 4.

Forest plot of associations between socio-demographic characteristics and reach, stratified by outreach modality (text, live call), among adults who had not completed a prior FIT

Figure 4 shows the associations between socio-demographic characteristics and reach, stratified by outreach modality (text, live call), among adults who had not completed a prior FIT. This analysis includes 4,796 individuals (Group 2: n = 2,548 text message advance notification; Group 3: n = 2,248 live phone call advance notification).

Figure 4 footnotes: Reference categories were female; not Latino/Hispanic, English-language preference, <100% federal poverty level, no clinic visits in past year, and resident of Orange County.

DISCUSSION

Our evaluation found that advance notifications delivered via live phone call or text message reached over 70%, and that automated phone call reminders reached over 50%, of provided phone numbers in the PROMPT enhanced mailed FIT program. While text messages reached the most patients, FIT completion was highest when the call was answered and there was opportunity for live interaction with clinic outreach staff trained to answer frequently asked questions. Personalized approaches may be more successful at boosting completion of mailed FIT tests in populations that traditionally have low CRC screening rates, such as Latinos and Spanish-speaking adults. Our findings fill an important research gap and can inform efforts to tailor outreach to achieve equitable reach and effectiveness of multi-component mailed FIT outreach programs.

Our findings suggest mailed FIT outreach, when implemented in a large safety-net health system, successfully engaged patients outside of primary care visits. Observational studies support the effectiveness of mailed outreach. At Kaiser Permanente Northern California, for example, an organized CRC screening program with mailed outreach doubled the proportion of adults up-to-date with screening (from 40% to over 80%), which led to tandem decreases in CRC incidence and mortality, of 26% and 52%, respectively.(16) The potential to substantially reduce CRC incidence and mortality underscored the importance of enhancements to mailed FIT outreach programs.

Our study also found that FIT completion was highest when the live call was answered and a discussion occurred between the patient and clinic outreach staff. Similarly, in a mailed FIT outreach study conducted in a safety net health system that targeted patients who had completed prior FIT, Lee and colleagues reported higher 1-year FIT completion among patients who were spoken with during and advance notification phone call, than among those who were left a voicemail or could not be reached (80% vs. 69% vs. 50%; p < .01).(17) Importantly, our study, along with others, has highlighted that individuals who had previously completed FIT had a significantly higher likelihood of completing a subsequent FIT compared to those who had not. In our study, we observed FIT completion rates of 44% versus 19% for these two groups (p < .01). Likewise, Somsouk and colleagues reported FIT completion rates of 72% versus 36% for these groups (p < .001).(18) Murphy and colleagues found a 50% higher probability of CRC screening in individuals with a history of FIT completion in comparison to those without (odds ratio: 1.47; 95% CI: 1.33, 1.63).(4) Our findings contribute to this body of literature by suggesting that improvements can be realized even when live phone calls are selectively employed for individuals with a lower likelihood of completing FIT (i.e., those without prior FIT completion).

Prior evaluations of mailed FIT programs have reported variation in reach across modalities. Our reach for text messages (78%) was at the top of the range of reach reported in previous studies (published from 2012 through 2018) conducted in community health centers (51 – 78%),(5,19–22) likely reflecting recent increases in acceptance and use of text messaging in the 50 and older age group. At 71%, our reach for advance notification live phone calls was higher than in several prior evaluations (37–48%)(19–21) and was lower than the 81% previously reported by our team(5) and the 91% reach reported by Lee and colleagues;(17) however, Lee and colleagues’ inclusion criteria was limited to participants who had completed a FIT in the prior year (a group that is more contactable). Our observed reach of automated reminder phone calls (53%) was lower than in prior studies (65 – 88%),(5,21–23) perhaps reflecting lower use of call functions on cell phones (in favor of text messaging), or more prevalent screening of calls from unfamiliar phone numbers than in previous years. Even though our evaluation was limited to patients who had at least two clinic visits in the past two years, our finding that patients with a clinic visit in the past year were more likely to be reached for many components likely reflects the greater accuracy of contact information in the EHR for more recently seen patients. Thus, our findings underscore the importance of maintaining updated patient contact information for optimal program delivery, especially in FQHCs where patient visits are generally used as a proxy for clinic enrollment.

Latino patients and patients with a documented Spanish language preference had lower odds of being reached by text message, but higher odds of being reached by a live phone call, than non-Latino patients and those with English language preference. Reach for automated phone calls was associated with Spanish language preference, but not with ethnicity. These findings suggest that live phone calls may be useful to raise CRC screening rates among populations whose CRC screening participation has traditionally lagged. A known drawback of live phone call outreach is that it is resource intensive; prior research suggests that labor costs for live phone call outreach costs is about $6.59 - $9.08 (converted from Canadian dollars to US dollars) per person (when up to two attempts are made),(24) which is higher than costs for text messaging found in our study ($0.13 per text). Given the resource demands of live phone call outreach, future research should explore the use of video text messages, text-delivered fotonovelas, or other asynchronous, resource-efficient approaches to deliver health reminders and education.(25,26)

A unique contribution of this study was our findings from our embedded randomization that sought to answer the question: “do screening-naïve patients need more intensive interventions (i.e., live phone call advance notification) to complete a FIT test?” We observed that the live phone call, when answered, increased FIT completion among patients who had never completed a prior FIT (27.7% vs. 18.2% for those who received text messages), suggesting that live phone call outreach may warrant the additional cost. This finding is particularly important considering recent changes to the US Preventive Services Task Force recommendations, calling for CRC screening to begin at age 45, rather than 50, resulting in an influx of 20 million adults who are now screening-eligible, most of whom have never had prior screening. Moreover, these recommendation changes coupled with suspension of screening services and routine care visits enacted during the COVID-19 pandemic have created an unprecedented demand for colonoscopy services, which has spurred the use of at-home stool-based tests.(27,28) Notably, these findings can inform program planners about how best to engage populations with lagging CRC screening rates.

Our findings can inform the design of mailed FIT outreach programs. First, information about the recency of prior clinic visits may be further leveraged in mailed FIT outreach programs, given that having more, or more recent, clinic visits was directly associated with FIT completion in our study and in others.(4,29) Moreover, recency of prior clinic visit is typically easy to obtain from clinic EHRs. Programs that mail FITs on a quarterly basis to patients with a visit in the past 3 months, for example, might reach more patients than programs that mail FITs to patients who haven’t been recently seen. Second, our expected finding that higher FIT completion was achieved in the groups that personally answered automated or live phone calls might underscore the importance of delivering phone calls during hours when patients are most likely to be reached (usually evenings and weekends). Relatively low FIT completion among screening-naïve patients, even with phone call or text message prompts, might underscore the need for additional education, community awareness approaches, or navigation to improve uptake in this subgroup.(1) Identifying approaches that can successfully boost CRC screening in screening-naïve individuals is increasingly important given the recent changes to US Preventive Services Task Force guidelines that broadened the screening eligibility age to include adults ages 45–49. Future research might test optimal timing of phone calls or test an approach that sends a text message immediately prior to a live phone call to determine whether this multi-modal sequenced approach boosts frequency of personal conversations. A similar sequenced approach has been successfully used by our team to engage patients in a combined text messaging and digital story (fotonovela) program.(26) Finally, future research might apply theory-based methods to design and evaluate clinic-based CRC screening interventions, especially those focused on younger, screening-naïve populations.

Strengths and limitations.

Our study had several strengths, including its large and diverse sample, randomized design, and the near-complete capture of FIT events and demographic characteristics in the EHR. The delivery of live phone call advance notifications by bilingual (English and Spanish) clinic staff optimized the consistency of intervention delivery among those whose preferred language was English or Spanish. In addition to its strengths, our study had several limitations. First, the content of messages delivered via live call and text message naturally differed; thus, observed differences in effectiveness could be attributed to a combination of content and modality. Some data limitations made it difficult to determine whether a patient received the advance notification or reminder: automated phone messages can be delivered to unintended recipients (through message interception, or cell phone theft or loss), and we could not be certain that patients read the text messages or listened to voice messages. For our assessment of reach for text messages and automated phone calls, we were limited to the data available from the vendor and had no way of assessing magnitude of any misclassifications. Additionally, our findings were from a single large FQHC in California and may have limited generalizability to FQHCs in other regions. Finally, our study included adults aged 50–75; thus, we were unable to assess reach in the younger population for whom screening is now recommended (ages 45–49). Understanding reach of notifications and reminders in this population is an important topic for future research.

Conclusion

We observed high reach for live phone calls and text messages notifications and moderate reach for automated phone call reminders used as part of the PROMPT enhanced mailed FIT program. The live phone call advance notifications reached a higher proportion, and text message advance notifications reached a lower proportion, of Latinos and Spanish-speaking adults and led to higher FIT completion rates among those with no prior FIT testing.