Behaviorally Informed Text Messaging to Promote Colon Cancer Screening: A Quality Improvement Randomized Clinical Trial.

Introduction
Colorectal cancer (CRC) screening is a proven strategy for early detection and mortality reduction, yet uptake remains suboptimal across the US.1 As of 2023, only 72.6% of adults aged 50 to 75 years were up to date with screening, falling short of national targets.2 Screening rates are even lower among historically underserved populations. At Federally Qualified Health Centers (FQHCs), just 41.2% of patients were up to date with CRC screening as of 2022.2 Individuals from low-income and minoritized groups are less likely to undergo timely screening3,4,5 and more likely to receive a diagnosis of advanced-stage disease.6 At the Family Health Centers at NYU Langone (FHC), an FQHC network in Brooklyn, New York, 59% of eligible adults were up to date with screening in 2024, higher than the FQHC average, but still well below national targets. These gaps underscore the need for scalable interventions to improve screening uptake in underserved populations.
Among available CRC screening modalities, the fecal immunochemical test (FIT) is a widely used, stool-based method recommended annually for average-risk adults.7,8 FIT is effective at reducing cancer incidence and mortality, and its low cost and ease of distribution have led many organized CRC screening programs to adopt it as their primary screening modality.9,10,11 Despite its advantages, FIT completion remains a challenge, often because of logistical or cognitive barriers, such as forgetting to complete the test, uncertainty about instructions, or discomfort around handling stool.12 In response, a growing body of research has explored ways to increase FIT return rates. Interventions such as mailed FIT outreach and patient navigation have demonstrated success across diverse populations,13 but these approaches can be resource-intensive, limiting their feasibility and scalability in safety-net settings.
To improve screening uptake, some programs have adopted low-cost strategies grounded in behavioral economics. These strategies, often referred to as nudges, work by subtly reshaping how choices are presented to make the desired action, such as returning a FIT kit, more intuitive, urgent, or socially desirable.14 Effective examples from prior studies include notifying patients in advance of mailing a FIT kit (to support planning and preparation),15 opting patients into screening by default (to reduce decision inertia),16 and sending reminders signed by a trusted practitioner (providing social pressure).15,16,17,18,19,20,21 Adding a return-by deadline to screening reminders has also been shown to increase FIT completion in randomized studies.20,21 Nudges such as these can be easily embedded within reminder messages, with text messaging in particular offering a low-cost, scalable delivery method that is adaptable to local workflows and easy to refine over time. In prior work within our FQHC network, for example, we found that small adjustments to the framing and timing of text messages were associated with significantly improved vaccination uptake.22
Building on prior evidence, we conducted a randomized test of a behavioral economics–informed text message reminder strategy designed to increase FIT completion at a large FQHC network. The intervention and randomization were embedded in routine clinical care. We hypothesized that this approach would increase FIT return rates compared with standard telephone outreach and offer a scalable model for improving uptake of preventive care.

Methods
We conducted a quality improvement randomized clinical trial at the FHC, a network of 8 FQHC sites in Brooklyn, New York, serving over 100 000 patients annually. This study was determined to be a quality improvement initiative by the NYU Langone Health Quality Improvement Oversight Committee and was not formally supervised by the institutional review board per NYU Langone Health institutional policies. Participant consent was not required. We followed the Standards for Quality Improvement Reporting Excellence (SQUIRE) reporting guidelines.23
Enrollment occurred from April 7 to June 24, 2025, with follow-up through July 15, 2025, to allow 21 days for FIT completion. The primary objective was to evaluate whether a behaviorally informed text message reminder could increase FIT completion compared with standard telephone outreach. We hypothesized that automated outreach incorporating behavioral nudges would improve screening uptake. The trial protocol is shown in Supplement 1.

Participants
Eligible participants were adults (aged ≥18 years) with a new FIT order placed between April 7 and June 24, 2025, who had not opted out of text messaging and listed English, Spanish, or Chinese (Mandarin or Cantonese) as their preferred language. Patients preferring other languages or who opted out of texts were excluded and continued to receive usual care. Patients with FIT orders who were part of the FHC’s Community Medicine Program were also excluded from randomization because that program had an ongoing CRC screening study. Finally, patients who had completed a FIT within the prior 12 months were excluded because, per clinic protocol, they were not eligible to receive nurse-led reminder telephone calls. This was intended to reserve nursing time for patients in need of updated screening. Given technical limitations in identifying these patients in real time, those with recent FITs were manually excluded after randomization: they were either screened out by the callers in the telephone group (thereby not receiving calls) or were manually removed by the study team from the text group after being sent a message.
Eligible participants were randomized 1:1 to receive text message reminders (intervention) or standard nurse-led telephone outreach (control) using a computer-generated sequence at the time of FIT order. Because of the nature of the intervention, neither staff nor participants were blinded.

Interventions
In both groups, practitioners were prompted to order annual FIT screening for adults aged 45 years and older. Practitioners could also order screening for younger patients at their discretion. FIT kits were distributed in person (or mailed after virtual visits) at no cost to the patient, with nurses providing instructions. In the control group, patients with unreturned kits received a single telephone call reminder from a nurse on day 8, per clinic protocol. Call completion varied across sites based on staffing and workload and was not modified for the study. When patients did not answer and voicemail was available, nurses left a message that explicitly reminded patients to return the FIT and included a callback number for questions. Intervention group patients received up to 3 automated text messages on days 2, 5, and 8. Messages were withheld if FIT completion was logged earlier. Each message included the patient’s clinic address and a centralized telephone number for questions or replacement kits. Message content was developed with clinic input and underwent review for health literacy and communication best practices. Messages were sent via NYU Langone Health’s existing text platform and in the patient’s preferred language (English, Spanish, or Chinese).
The messages incorporated several behavioral science elements (Table 1). To encourage timely completion, they included a specific due date (day 9 after the order), leveraging the motivating effect of deadlines.20,21 They also invoked social norms by referencing the patient’s practitioner in the second message (“your provider is waiting to receive…”), fostering a sense of accountability.17,18,19 The messages employed gain-framing by emphasizing that screening may “save your life,” highlighting the potential benefits of action. Finally, repetition was used by sending 3 reminders, reinforcing the importance of the task and mitigating forgetfulness.24

Outcomes
The primary outcome was FIT completion, defined as return and laboratory processing of the test kit within 21 days of the order, as recorded in the electronic health record (EHR). We also assessed completion at 7 and 14 days. Post hoc exploratory subgroup analyses evaluated potential differences in intervention effects by sex, age group, self-reported race and ethnicity (categorized as Asian; Black or African American; Hispanic, Latino, or Spanish; White; or other, defined as non-Hispanic American Indian or Alaska Native, Native Hawaiian or Other Pacific Islander, Middle Eastern or North African, multiracial, and any race or ethnicity not otherwise specified), preferred language, and patient portal engagement. All variables were extracted from the EHR. Data on race and ethnicity were included because of known differences in colorectal cancer screening rates by race and ethnicity.

Sample Size
We estimated that enrolling 720 participants would provide 80% power to detect a 10–percentage point absolute difference in FIT completion between groups (2-sided α = .05). This threshold was selected as the minimum clinically meaningful difference. We decided to enroll twice as many participants to allow for additional analyses of differences in effect among population subgroups.

Statistical Analysis
Analyses followed an intention-to-treat approach among eligible patients. FIT completion at 7, 14, and 21 days was compared between groups using χ2 tests, performed in SAS statistical software 9.4 (SAS Institute). We also fit a logistic regression model with treatment assignment as the independent variable to estimate the overall intervention effect while controlling for race, ethnicity, age, insurance type, sex, and patient portal use. From this model, we calculated average marginal effect (AME) with a 95% CI using robust SEs.
To examine whether the intervention effect varied across patient subgroups, we then fit separate logistic regression models that included an interaction term between treatment assignment and each covariate of interest (race and ethnicity, age, insurance type, sex, and patient portal use). For each model, we evaluated the coefficient for the interaction term and its P value to assess evidence of heterogeneity relative to the reference category. All regression analyses were conducted in R statistical software version 4.4.1 (R Project for Statistical Computing). Statistical significance was defined as 2-sided P < .05.
To estimate the operational impact of replacing telephone calls with text messages, we calculated the average number of calls made per month in the call group and estimated that the calls on average, including completed and attempted calls, required 5 minutes of staff time. To estimate the monthly number of additional completed FITs, we applied the observed AME of text messaging to the average monthly number of new FIT orders across all FHC sites. Using EHR data on the proportion of abnormal FIT results and published estimates of the positive predictive value of a onetime FIT, we projected the number of abnormal tests and CRC cases that would be detected if text messages were used in place of telephone calls.

Results

Study Participants
Between April 7 and June 24, 2025, 1275 eligible patients with new FIT orders were randomized: 649 to the text group and 626 to the call group (Figure 1). No one was lost to follow-up. Follow-up for the primary outcome concluded on July 15, 2025. Baseline characteristics were well-balanced between groups (Table 2). Participants were predominantly Hispanic (329 participants [50.7%] in text vs 317 participants [50.6%] in call), female (418 participants [64.4%] vs 398 participants [63.6%]), and Spanish-speaking (407 participants [62.7%] vs 372 participants [59.4%]). The mean (SD) age was also similar across groups (56.4 [9.3] vs 56.7 [9.6] years).
Of 457 call group participants eligible for reminders (excluding patients who returned their FIT before day 7), nurses successfully reached 209 (45.7%), left voicemails for 67 (14.7%), and made unsuccessful attempts to contact 20 (4.4%). No call data were available for 161 participants (35.2%). As intended, no call attempts were documented in the text group. Among participants assigned to the text group, 94.5% of messages (1452 of 1537 messages) were successfully delivered.

Primary Outcome
FIT completion rates were consistently higher in the text group across all follow-up intervals, with statistically significant differences at days 14 and 21 (Figure 2). At day 7, 212 of 649 participants (32.7%) in the text group had completed FIT screening compared with 173 of 626 (27.6%) in the call group (absolute difference, 5.0 percentage points; 95% CI, −0.0 to 10.1 percentage points; P = .051). By day 14, completion increased to 352 of 649 participants (54.2%) in the text group vs 252 of 626 participants (40.3%) in the call group (absolute difference, 14.0 percentage points; 95% CI, 8.6 to 19.4 percentage points; P < .001). At day 21, 382 of 649 participants (58.9%) in the text group had completed screening compared with 312 of 626 participants (49.8%) in the call group (absolute difference, 9.0 percentage points; 95% CI, 3.6 to 14.5 percentage points; P = .001). In multivariable logistic regression, assignment to the text group was associated with higher odds of FIT completion at 21 days (odds ratio, 1.58; 95% CI, 1.25 to 2.00; P < .001), corresponding to a 10.4–percentage point AME (95% CI, 5.2 to 15.7 percentage points). Full regression results are available in the eTable in Supplement 2.

Exploratory Subgroup Analyses
Only the age interaction term was statistically significant. Those aged 18 to 44 years had the lowest odds of returning the FIT vs those older than 64 years (Table 3).

Estimated Operational and Clinical Impact
Replacing telephone calls with text messages for FIT outreach offers measurable clinical and operational benefits. Given the call volume and observed effect, we estimated savings of 18 staff hours per month, return of 48 additional completed FITs within 21 days per month, of which an average of 2.2 would be abnormal, and detection of 1 to 2 additional CRC cases per year.

Discussion
In this quality improvement randomized trial embedded within a large FQHC network, a behaviorally informed, automated text message strategy was associated with a 9.0–percentage point increase in FIT completion by 21 days compared with standard telephone outreach. Although telephone calls are commonly used for preventive care reminders, our findings suggest that automated, but thoughtfully designed, messaging can outperform live outreach in a clinical safety-net setting.
Text messaging likely succeeded through greater reach and, potentially, through text message content. Many patients in the call group were never reached, reflecting the practical limitations of manual outreach in busy clinical settings. In contrast, text messages can be delivered automatically and consistently. Although live communication may be more motivating under ideal conditions,25 our findings illustrate a trade-off between effectiveness and reliable delivery in clinical practice. Moreover, patients in the text group could receive up to 3 reminders, compared with a single telephone call in the control group, further amplifying differences in reach. More resource-intensive interventions, such as mailed FIT kits, patient navigation, or education, can yield larger effects, but these require substantial investment.13,26 Our results demonstrate that even a simple, automated reminder can still meaningfully boost screening uptake when more intensive efforts are not feasible.
Other randomized studies of text messaging for CRC screening have reported similar improvements, although most assessed longer follow-up intervals. In a trial among veterans mailed FIT kits, automated text reminders increased completion from 28% to 38% at 90 days (an increase of 10 percentage points).27 In a Spain-based study, text reminders 14 days after FIT pick-up increased return rates from 53.7% to 64.2% at 30 days (an increase of 10.5 percentage points).28 Our study adds evidence that effects can be observed earlier, within 3 weeks of the test order.
Behavioral design features may also have contributed to this program’s effectiveness. We incorporated several evidence-based nudges aimed at addressing common barriers to completing screening.14 One was the use of 3 separate reminder messages. Prior research on preventive care has shown that multiple reminders are more effective than a single reminder. Huf et al24 tested a bundled intervention including a prealert text, opt-out mailed FIT kit, and 3 follow-up text reminders, achieving a 17.7–percentage point higher return rate compared with only a single reminder. In vaccination, a megastudy of 19 influenza vaccine reminder strategies found that multimessage approaches outperformed single reminders, with the best 2-message strategy increasing uptake by 2.9 percentage points.29 When implementing multimessage strategies, however, it is important to note that repeated outreach can lead to message fatigue, highlighting the need to balance message frequency and efficiency.30 The messages also included a return-by deadline, which prior studies suggest can support planning and follow-through.31,32,33 Robb et al,21 for example, found that adding a 2-week suggested deadline to mailed FITs increased returns by 2 percentage points.
We also incorporated gain-framed wording (“this test could save your life”) and a social norms cue (“your provider is waiting”). Prior research shows that these types of message features can influence behavior, but effects are generally small and variable across studies.34,35 In CRC specifically, Hagoel et al36 reported that interrogative framing increased screening completion by 3 to 6 percentage points, and Muller et al37 found that culturally tailored messages improved colonoscopy uptake by 3.3 percentage points. Practitioner-linked cues have also proven effective: in a megastudy of influenza vaccination reminders, messages describing a shot as “reserved for you” increased uptake by 4.6 percentage points.38 Taken together, these findings suggest that while the impact of any single wording change may be small and inconsistent, such cues can serve as useful additions to broader outreach strategies.
Importantly, this study was not designed to disentangle the independent contributions of message content, delivery modality, and contact frequency. Isolating content effects would require an attention control, such as outreach matched in modality and frequency, which was beyond the scope of this study but may be investigated in a future study. Emerging automation technologies may also shift the trade-offs observed in this study. For example, artificial intelligence–assisted telephone outreach could enable repeated, standardized contact by telephone without increasing staff burden. Whether such approaches can achieve engagement comparable to that of text messaging, particularly as many patients might not answer calls from unknown numbers, remains an open question.
In our exploratory subgroup analysis, we found no statistically significant effect modification by race, ethnicity, insurance status, sex, or patient portal engagement. However, the effect of the intervention did appear smaller among participants aged 18 to 44 years. Although this suggests that this age bracket may be less responsive to text-based outreach, perhaps because screening recommendations were only recently extended to those younger than 50 years, further research is needed to confirm this finding and determine whether age-tailored messaging could enhance impact. The lack of differential effects by patient portal use also suggests that text message reminders can benefit patients regardless of prior digital engagement. We can also infer that older adults and publicly insured patients, groups often perceived as less digitally engaged, can still meaningfully benefit from thoughtfully designed digital outreach.39,40 Importantly, many subgroups had relatively small sample sizes, and the study may have been underpowered to detect modest differences in effect across groups.
Methodologically, this project demonstrates the feasibility of embedding rapid randomized evaluations into routine care. By integrating randomization into clinic workflows and using EHR data for outcome tracking, we were able to conduct a low-cost, minimally disruptive study that produced timely, actionable insights for clinic leadership. This rapid-cycle approach, adapted from A/B testing models in the technology industry, offers a scalable model for other health systems seeking to test and improve outreach strategies in clinical practice settings.41

Limitations
This study has limitations that should be mentioned. This intervention was implemented across 1 clinic network in 1 geographic area and so these findings may not be generalizable to other contexts, although the methods used may be applicable elsewhere. Second, although behavioral strategies can mitigate cognitive barriers to screening, they do not address structural obstacles, such as unstable housing, transportation barriers, or competing priorities, that disproportionately affect FQHC populations and may persistently limit screening participation.

Conclusions
In this quality improvement randomized clinical trial, a behaviorally informed, automated text message strategy improved FIT completion in a diverse primary care setting. Although simple and low-cost, the intervention produced clinically meaningful improvements in preventive care uptake with minimal operational burden. On the basis of these findings, the text message intervention was adopted across the FQHC network for all eligible patients with new FIT orders. Future work should focus on optimizing message content and frequency to prevent message fatigue, identifying which behavioral components are most effective, and ensuring these strategies remain impactful across diverse patient populations and settings.