From SMS Text Messaging to Screening Completion: Understanding How SMS Responses Influence Cancer Screening Uptake in Federally Qualified Health Center Clinics.

From SMS Text Messaging to Screening Completion: Understanding How SMS Responses Influence Cancer Screening Uptake in Federally Qualified Health Center Clinics - PMC Skip to main content An official website of the United States government Here's how you know Here's how you know Search Log in Primary site navigation Search Log in Search PMC Full-Text Archive Search in PMC Journal List User Guide As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: PMC Disclaimer | PMC Copyright Notice Health Serv Res . 2026 May 28;61(3):e70134. doi: 10.1111/1475-6773.70134 From SMS Text Messaging to Screening Completion: Understanding How SMS Responses Influence Cancer Screening Uptake in Federally Qualified Health Center Clinics Omolola E Adepoju 1, 2, ✉ , Tonghui Xu 2 , Tracy Angelocci 3 Author information Article notes Copyright and License information 1 Tilman J. Fertitta Family College of Medicine, University of Houston, Houston, Texas, USA 2 Humana Integrated Health Systems Sciences Institute, Houston, Texas, USA 3 Lone Star Circle of Care, Georgetown, Texas, USA * Correspondence: Omolola E. Adepoju ( oadepoju@central.uh.edu ) ✉ Corresponding author. Revised 2026 Mar 26; Received 2025 Oct 15; Accepted 2026 May 11; Issue date 2026 Jun. © 2026 The Author(s). Health Services Research published by Wiley Periodicals LLC. This is an open access article under the terms of the http://creativecommons.org/licenses/by-nc-nd/4.0/ License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non‐commercial and no modifications or adaptations are made. PMC Copyright notice PMCID: PMC13238827  PMID: 42206315 ABSTRACT Objective To examine how patient engagement with SMS text reminder scan mediate cancer screening completion among patients at Federally Qualified Health Center clinics (FQHCs). Study Setting and Design Patients overdue for cancer screening were randomly assigned to one of two groups: a 3‐week interactive SMS intervention or a 6‐week theory‐informed SMS intervention. The binary outcome was cancer screening completion after 90 days of SMS intervention. A mediation model was used to assess associations between text‐message response rates and cancer screening completion. Exposure variables included sociodemographic characteristics, type of SMS intervention, type of cancer screening needed, and health status variables. Data Sources and Analytic Sample Data were obtained from two large FQHC networks in Texas and California in 2023. We included 4,344 patients who participated in the 3‐week or 6‐week intervention groups, successfully received all interactive messages, had unique phone numbers, and did not opt out of the intervention. Principal Findings Overall effects were statistically significant, indicating partial mediation. Full mediation was observed for insurance type: compared with Medicaid, private insurance was associated with a 5‐percentage‐point (PP) higher predicted probability of screening completion (Average marginal effect [AME] = 5.0 PP, 95% CI [0.01, 0.09]). Partial mediation was observed for Medicare (AME = 7.0 PP, 95% CI [0.02, 0.13]), uninsurance (AME = 5.0 PP, 95% CI [0.01, 0.09]), patient with diabetes (AME = −4.0 PP, 95% CI [−0.08, −0.01]), non‐primary English language (AME = 8.0 PP, 95% CI [0.05, 0.10], p < 0.001), and large‐city residence (AME = 9.0 PP, 95% CI [0.07, 0.10]), all of which were associated with higher predicted probabilities of screening completion. Conclusions Engagement with SMS reminders was a significant mediator of cancer screening completion, suggesting that increasing message interactivity may improve cancer screening uptake. Theory‐informed interactive messaging may promote greater patient engagement. Keywords: cancer screening, federally qualified health centers, interactive SMS message Callout Box What Is Known on This Topic? ○ Underserved populations such as those receiving care in FQHCs have lower rates of cancer screening completion. ○ Gaps in cancer screening completion persist across sociodemographic groups. ○ SMS text messages are evidence‐based interventions for increasing care continuity in office‐based settings. What This Study Found ○ Patient interaction with SMS messages partially mediated improvements in cancer screening completion in FQHC settings. ○ Patients with private insurance showed that the effect on screening completion mainly operated through patient interaction with the SMS. ○ For uninsured patients, those with Medicare, patients with diabetes, non–English‐speaking patients, and those living in large cities, the effect was only partially explained by SMS interaction. ○ Sending more interactive, theory informed messages increased patient engagement with messages and may help improve patients' knowledge and awareness of cancer risk. 1. Introduction The burden of cancer is increasing in the United States, with breast [ 1 ], cervical [ 2 ], and colorectal (CRC) cancer [ 1 ] being some of the leading causes of cancer‐related deaths. Efforts to boost preventive screenings as a strategy to delay disease progression and reduce mortality rates [ 3 ] have been widely successful, with recent numbers indicating sizable increases in the proportion of US adults who were up to date with their screenings, at 75.7%, 75.2%, 72.2% for breast, cervical, and CRC screening respectively [ 4 ]. While this trend appears promising, uptake rates still lag behind among individuals belonging to lower socioeconomic statuses [ 5 ]. This is especially true for patients served in Federally Qualified Health Center (FQHC) settings, which care for more than 34 million patients annually [ 6 ], many of whom are racial and ethnic minorities, low‐income, and uninsured [ 7 ]. Work by Amboree et al. reports that in FQHC settings, cancer screening rates have remained relatively low, lagging behind national averages: 45.4% for breast cancer, 51% for cervical cancer, and 40.2% for CRC [ 8 ]. FQHCs face multiple challenges in addressing this lag, including limited funding [ 9 ], shortages of patient navigators [ 10 ], and logistical barriers associated with serving high‐need populations who often face structural obstacles to preventive care [ 11 ]. These disparities influence treatment and mortality rates, with greater delays in diagnostic evaluations, suboptimal treatment [ 12 , 13 ], and ultimately higher cancer deaths for breast, cervical, and CRC [ 14 ]. There is evidence suggesting that Short Message Service (SMS) text reminders are effective in improving patient appointment compliance [ 15 ]. A review by Uy et al. found that text messaging interventions were associated with moderate increases in breast and cervical cancer screening rates and smaller, but still positive, effects on CRC screening [ 16 ]. SMS reminder services have been proposed to be particularly beneficial in FQHC settings for reaching out to eligible patients and initiating coordinated efforts to get them screened [ 6 ]. Further evidence has found SMS reminder services to potentially reduce disparities in cancer screenings in high‐risk settings [ 17 ], making them an ideal complementary intervention in increasing screening rates among FQHC populations. In optimizing the effectiveness of SMS reminders, adjusting variables such as reminder rate and timing have been explored [ 18 ] with mixed results. Moreover, personalized, tailored reminders may increase effectiveness, particularly theory‐informed messages, by targeting attitudes, norms, and perceived control related to adherence to screening recommendations [ 19 ]. Use of theory‐informed messaging allows reminders to be tailored to emphasize positive outcomes [ 20 ] and include persuasive language [ 21 ] that may encourage screening uptake. This tailoring has great potential in FQHC settings, as culturally tailored SMS interventions [ 22 ] have previously shown success in diverse populations. Despite growing interest in SMS reminders, there remains limited understanding [ 23 ] of the direct and indirect factors that influence cancer screening uptake, particularly how patients engage with SMS reminders and how such engagement translates into behavioral outcomes within community health center settings. Exploring the direct and indirect contributions of SMS reminders is key to establishing effective bi‐directional communication in underserved settings such as FQHCs, which have historically faced challenges in engaging patients in preventive care [ 24 ]. In this study, we evaluated the impact of interactive SMS reminder services on cancer screening uptake across multiple FQHC clinics. Using a mediation model, we examined how patient engagement with SMS reminders influenced screening behavior. Studies of this nature are critical for understanding how communication strategies like SMS reminders can be optimized to improve preventive care uptake, particularly in populations facing structural barriers to access. 2. Methods 2.1. Data Sample The study sample consisted of active adult patients (defined as having had at least one visit (of any type) within the prior 12 months) and were overdue for breast, cervical, or colorectal cancer screening at the time of the intervention data pull. Because we aimed to examine how patient responses to SMS messages impacted the completion of cancer screening, we excluded patients who had previously opted out of messaging, did not have a valid mobile phone number, or had a non‐unique phone number. Patients were from two FQHC networks: (a) FQHC 1—large FQHC with 56 clinic sites in Texas, and (b) FQHC 2—large FQHC with 7 clinic sites (2 full‐time sites and 5 satellite locations) in California. Together, both FQHCs serve over 200,000 patients annually, providing pediatric, family medicine, behavioral health, and obstetrics and gynecology services to individuals primarily from underserved and medically vulnerable populations. In 2023, both FQHCs implemented an SMS/text messaging reminder initiative to increase cancer screening gaps closure rates. While the intervention included patients who met the eligibility criteria, this study focused on examining the effect of SMS engagement on cancer screening completion. Therefore, patients who did not receive any SMS messages were excluded from the analytic sample, as they had no opportunity to engage with or respond to SMS outreach. Figure 1 shows the analytic sample waterfall. A total of 22,861 patients were overdue for a breast, cervical, or CRC screening test. After excluding those who previously opted out of messaging, did not have unique telephone numbers, or whose invitational SMS message did not deliver, we retained 8946 eligible subjects. The FQHC then worked with academic partners to randomly assign eligible patients to one of two study arms: (i) 3‐week SMS overdue and reminder messages (total of 3 SMS messages; one per week) and (ii) 6‐week SMS messages that are behavior theory‐informed (total of 18 SMS messages, three per week). FIGURE 1. Open in a new tab Analytic sample waterfall showing study arms. 2.2. Theory‐Informed Messages The theory‐informed SMS messages were designed to be bidirectional allowing the recipient to respond—to either opt out of the message program, answer a question, or respond to a survey. Since all patients had already been identified as overdue for screening, reminder SMS messages did not explicitly state that patients were overdue, but focused on behavioral outcomes to expect in response to the reminder SMS message. For example, patients received the following message: Bum, behind, bottom. Whatever you call it, if you are embarrassed to talk about the tests that screen for colon cancer, you are not alone. Check out this video: https://msg.care/8SnkK . Call ###‐###‐#### to schedule a test today. It's hard waiting for your Pap smear results to come back. Many women have thoughts of cancer on their mind. Don't fear the results! An abnormal Pap usually means a pre‐cancerous condition that can be treated, long before cancer develops. You are more than halfway through this program. Type NO to stop now. A Black woman's lifetime risk of breast cancer is 12%. Black women are diagnosed with breast cancer less often than white women but are 40% more likely to die from it. It is important to talk to your doctor about your personal risk. Type “Y” to schedule a test today. The FQHC and SMS messaging service maintained a record of all delivered messages and participant responses. De‐identified data on those who completed the final step were provided to the academic team for analysis, ensuring that patients who completed the program received all text messages, and had unique, unshared mobile phone numbers. 2.3. Mediation Model This study uses Structural Equation Modeling (SEM) for mediation analysis, a theory‐guided analytic framework. SEM employs a conceptual model, path diagram, and a system of linked regression‐style equations to capture complex and dynamic relationships among observed and unobserved variables [ 25 ]. A key strength of SEM is its ability to simultaneously estimate direct, indirect, and total effects in one model, which improves statistical efficiency and reduces bias. In general, a mediation model includes: (a) the direct path which represents the direct effect (referred to as c ′ ) of predictors (X) on outcome (Y) controlling for the mediator (M); (b) the effect of X on M (referred to as a ); (c) the effect of M on Y after accounting for X (referred to as b), (d) the indirect pathway from X to Y through M, quantified as the product of the a and b paths ( a × b ) and (e) the total effect path, which represents the total effect (referred to as c = a × b + c ′ ) [ 26 ]. 2.4. Measures The outcome measure was a binary indicator of whether the participant completed cervical, CRC, or breast cancer screening within 90 days of completing the intervention. For patients eligible for more than one screening test (less than 5% of the analytic sample), each test was retained in the analytic sample, with the corresponding outcome marked. Screening completion was coded as 0 = No and 1 = Yes. The ordinal mediator variable was defined as an ordinal measure of engagement, based on the distribution of SMS response rates, categorized as follows: 0% (no engagement), 1%–49% (low), 50%–75% (moderate), and 76%–100% (high). Exposure variables included demographic characteristics: race/ethnicity (Non‐Hispanic White, Hispanic, Black/African American, Asian/others), language (English, Spanish/others), and insurance type (uninsured, Medicaid, Medicare, private). Additional categorical variables included intervention cohort (three‐message vs. six‐message per week), cancer screening type (cervical, breast, CRC), and the presence of medical conditions such as heart disease, diabetes mellitus, chronic lower respiratory disease, overweight/obesity, and depression or mood disorders. Using patients' ZIP code information, three city‐ or county‐level variables were merged into the analytic data: (a) the proportion of residents who reported a routine checkup with a doctor in the prior year, obtained from the City Health Dashboard data [ 27 ], (b) 2025 county‐level population‐to‐primary care physician (PCP) ratio, sourced from County Health Rankings & Roadmaps data [ 28 ], and (c) the city population size, obtained from the US Census Bureau QuickFacts (2023) [ 29 ]. 2.5. Statistical Data Analysis During data cleaning, observations meeting the following criteria were removed using listwise deletion: (a) missing ZIP code information and (b) missing outcome variable. The final analytic sample included 4344 participants: 2227 in the 3‐week message group, and 2117 in the 6‐week message group. Descriptive statistics, including frequencies and proportions, were used to summarize patient sociodemographic characteristics, health status indicators, and SMS message response rates. Additionally, average marginal effects (AMEs) were computed to support probability‐scale interpretation of the total and direct effects. AMEs were calculated as differences in predicted probabilities at different values of each predictor, averaged across the sample. For path a, the X → mediator pathway (where the mediator is an ordinal scale), the AME was defined as the average change in the mediator's expected value on its original 1–4 scale for a given change in X; therefore, its interpretation differs from probability‐based AMEs. Given the complexity of the model, which includes an ordinal mediator and a binary outcome, we focus on the core mediation analyses: the total effect, the direct effect, and X → mediator effect. This approach allows us to test the hypothesized pathways without introducing transformations that could obscure the model structure. Therefore, for the indirect models of each variable and the overall effects across all paths in the model, we report only the probit coefficients. Partial mediation is defined as a statistically significant direct, indirect, and total effect. Full mediation is defined as a statistically significant indirect and total effect, with a non‐significant direct effect. Mediation analyses were conducted using path regression models with the Weighted Least Squares Mean and Variance Adjusted (WLSMV) estimator, implemented in the Lavaan package in R version 4.4.2 through RStudio. Statistical significance was assessed at p < 0.05, and results are reported as unstandardized regression coefficients and AMEs, along with standard errors (SEs) and 95% confidence intervals (CIs). 3. Results 3.1. Statistical Description Table 1 presents the descriptive statistics of the independent and mediator variables. Of the 4344 participants, 51.3% were assigned to the 3‐week intervention group and 48.7% to the 6‐week group. 58.5% were aged 25–54 years, 23.0% were aged 55–64 years, and 18.5% were aged 65 years or older. A majority reported English as their primary language (57.6%). With respect to race, 19.7% identified as Non‐Hispanic White, 54.2% as Hispanic, 11.1% as Black or African American, and 15% as Asian or others. Nearly half of the participants were uninsured (45.5%), followed by 23.5% with private insurance, 19.4% with Medicare, and 11.6% with Medicaid coverage. TABLE 1. Statistical descriptions of patient characteristics. Variable Frequency (%) Intervention type 3‐week message 2227 (51.3) 6‐week message 2117 (48.7) Age group 25–54 2539 (58.5) 55–64 1001 (23.0) 65+ 804 (18.5) Primary language English 2503 (57.6) Others 1841 (42.4) Diabetes Yes 1284 (29.6) No 3060 (70.4) Race/ethnicity Non‐Hispanic White 889 (19.7) Hispanic 2444 (54.2) Black or African American 498 (11.1) Asian/others 674 (15.0) Insurance type Medicare 844 (19.4) Medicaid 506 (11.6) Private 1019 (23.5) Uninsurance 1975 (45.5) Heart disease Yes 2104 (48.4) No 2240 (51.6) Hypertension Yes 2268 (52.2) No 2076 (47.8) Overweight/obese Yes 3579 (82.4) No 765 (17.6) Chronic lower respiratory Yes 302 (7.00) No 4042 (93.00) Depression or mood disorders Yes 1326 (30.5) No 3018 (69.5) Intervention cohort Cervical screening 1298 (29.9) Breast cancer screening 747 (17.2) Colorectal screening 2299 (52.9) 2023 size of the city Small town/city 598 (13.8) Large city ( ≥ 150,000) 3746 (86.2) 2022 city‐level routine checkup of doctor Less than the average 1173 (27.0) Higher than the average 3171 (73.0) 2025 county level PCP ratio Less than the average 1104 (25.4) Higher than the average 3240 (74.6) Response message rate (mediator) 0.00% 3004 (69.1) 1.00%–49.00% 247 (5.7) 50.00%–75.00% 563 (13.0) 76.00%–100% 530 (12.2) Total sample size 4344 Open in a new tab Abbreviation: PCP, primary care physician. Regarding health status, 29.6% of participants had diabetes, 48.4% had heart disease, 52.2% had hypertension, 82.4% were classified as overweight/obese, 7% had chronic lower respiratory disease, and 30.5% reported depression or mood disorders. With regard to cancer screening type, 29.9% were in the cervical cancer cohort, 17.2% in the breast cancer cohort, and 52.9% in the colorectal cancer cohort. Most participants resided in large cities (86.2%), with 73% living in areas reporting above‐average rates of routine doctor checkups and 74.6% in counties with above‐average population to PCP ratios. Engagement, measured by SMS response rates, showed that 69.1% of participants did not respond to any messages, 5.7% responded to 1%–49% of messages, 13% to 50%–75%, and 12.2% to 76%–100%. Figure 2 shows the frequency of responses to SMS messages and whether screening was completed. Across all three cancer cohorts, 1263 patients (29.1%) completed the screening within 90 days: 442 cervical, 232 breast, and 589 colorectal. In the cervical cancer cohort, 405 participants (34.1%) responded to at least one SMS message. In the breast cancer cohort, 218 participants (38.9%) responded to at least one message. In the colorectal cancer cohort, 558 participants (34.1%) responded to at least one SMS message. FIGURE 2. Open in a new tab Patient Response to SMS reminders and Subsequent Completion of Breast, Cervical, and Colorectal Screening Test. 3.2. Overall Path Effect Results Figure 3 presents the results of the partial mediation model. We report unstandardized probit coefficients for the overall direct effect, which represents the combined influence of multiple pathways. The overall direct effect remained a significant predictor of the outcome after adjusting for the mediator. Specifically, the direct probit coefficient was 2.18 (95% CI: 1.30–3.05, p < 0.01), indicating a strong direct association. The intervention and demographic characteristics were also associated with the mediator, message response rate, with a coefficient of 0.86 (95% CI: 0.05–1.68, p = 0.04), suggesting that these factors improve patient engagement with SMS reminders. For the mediator to outcome path, we report a coefficient of 0.23 (95% CI: 0.18–0.27, p < 0.001), indicating that higher message response rates were associated with a higher likelihood of cancer screening completion. FIGURE 3. Open in a new tab Overall pathway of mediation model. Furthermore, a statistically significant indirect effect was observed through message response rate (coefficient = 0.19, 95% CI: 0.06–0.38, p = 0.04), indicating that the intervention and demographic characteristics may also influence screening completion through increased message engagement. The total effect, combining both direct and indirect pathways, was 2.37 (95% CI: 1.48–3.26, p < 0.01), indicating a strong overall association with screening completion. Tables A1 and A2 present the estimated coefficients from the mediation model. 3.3. Model 1 (Path c) AMEs for the total effect model are reported in Table 2 . Compared with individuals aged ≥ 65 years, those aged 25–54 had a 8 percentage‐point higher predicted probability of completing their cancer screening test (AME = 0.08, 95% CI [0.04, 0.12], p < 0.001). Non‐primary English speakers had a 8 percentage‐point higher predicted probability (AME = 0.08, 95% CI [0.05, 0.10], p < 0.001), compared to those who identified English as their primary langauage. Patients with diabetes had a 4 percentage‐point lower predicted probability (AME = −0.04, 95% CI [−0.08, −0.01], p = 0.01), compared to those without diabetes. Compared to patients with Medicaid, patients with Medicare had a 7 percentage‐point higher predicted probability (AME = 0.07, 95% CI [0.02, 0.13], p = 0.01), patients with private insurance had a 5 percentage‐point higher predicted probability (AME = 0.05, 95% CI [0.01, 0.09], p = 0.04), and patients without insurance had a 5 percentage‐point higher predicted probability (AME = 0.05, 95% CI [0.01, 0.09], p = 0.02). Patients with obesity/overweight had a 4 percentage‐point higher predicted probability (AME = 0.04, 95% CI [0.00, 0.07], p = 0.04), whereas patients with depressive disorder had a 7 percentage‐point lower predicted probability (AME = −0.07, 95% CI [−0.10, −0.04], p < 0.001). Compared with the CRC screening group, cervical cancer screening (AME = 0.09, 95% CI [0.05, 0.12], p < 0.001) and breast cancer screening (AME = 0.10, 95% CI [0.05, 0.15], p < 0.001) were each associated with 10 percentage‐point higher predicted probability of completion. In addition, patients living in large city areas had a 9 percentage‐point higher predicted probability of completing screening (AME = 0.09, 95% CI [0.07, 0.10], p < 0.001), and those living in cities with higher routine checkup rates had a 10 percentage‐point higher predicted probability (AME = 0.10, 95% CI [0.02, 0.15], p = 0.003). TABLE 2. The AME results of direct, mediator, and total effects model results. Model 1: Path c = c ′ + ab (total effects) Model 2: Path a (X → Mediator effects) Model 3: Path c ′ (Direct effects) AME 95% [CIs] p AME 95% [CIs] p AME 95% [CIs] p Intervention cohort 3 week REF — — REF — — REF — — 6 week 0.02 (0.01) −0.01, 0.05 0.18 0.43 (0.03) 0.38, 0.49 < 001 −0.02 (0.01) −0.04, 0.01 0.24 Age group 25–54 0.08 (0.02) 0.04, 0.12 < 001 0.09 (0.05) −0.01, 0.19 0.07 0.07 (0.02) 0.03, 0.11 < 001 55–64 0.04 (0.02) −0.01, 0.09 0.09 0.15 (0.06) 0.05, 0.27 < 001 0.02 (0.02) −0.02, 0.07 0.28 65+ REF — — REF — — REF — — Primary language English REF — — REF — — REF — — Others 0.08(0.01) 0.05, 0.10 < 001 0.09 (0.04) 0.02, 0.15 0.02 0.07 (0.13) 0.04, 0.09 < 001 Diabetes Yes −0.04(0.02) −0.08, −0.01 0.01 −0.10(0.03) −0.16, −0.03 0.005 −0.04 (0.02) −0.07, −0.01 0.03 No REF — — REF — — REF — — Race/ethnicity Non‐Hispanic White REF — — REF — — — — Hispanic −0.02 (0.06) −0.07, 0.02 0.32 −0.12 (0.05) −0.25, −0.03 < 001 REF −0.06, 0.03 0.65 Black/African American 0.01 (0.03) −0.04, 0.07 0.77 −0.19 (0.06) −0.31, −0.08 < 001 −0.01 (0.02) −0.03, 0.09 0.36 Asian/Others 0.03 (0.03) −0.03, 0.08 0.33 −0.23 (0.05) −0.33, −0.12 < 001 0.03 (0.03) −0.01, 0.11 0.07 Insurance type Medicare 0.07 (0.03) 0.02, 0.13 0.006 0.19 (0.07) 0.04, 0.31 0.03 0.06 (0.03) 0.01, 0.12 0.05 Medicaid REF — — REF — — REF — — Private 0.05 (0.02) 0.01, 0.09 0.04 0.21 (0.05) 0.11, 0.31 < 001 0.03 (0.02) −0.01, 0.07 0.19 Uninsurance 0.05 (0.02) 0.01, 0.09 0.02 0.11 (0.05) 0.02, 0.19 0.01 0.04 (0.02) 0.00, 0.09 0.04 Heart disease Yes −0.04 (0.04) −0.13, 0.03 0.22 −0.14 (0.08) −0.34, 0.01 0.08 −0.03 (0.04) −0.11, 0.04 0.38 No REF — — REF — — REF — — Hypertension Yes −0.03 (0.04) −0.09, 0.05 0.47 0.10 (0.08) −0.05, 0.28 0.19 −0.03 (0.03) −0.10, 0.03 0.34 No REF — — REF — — REF — — Obesity Yes 0.04 (0.02) 0.00, 0.07 0.04 0.02 (0.05) −0.08, 0.11 0.71 0.04 (0.02) 0.00, 0.08 0.04 No REF — — REF — — REF — — Chronic lower respiratory Yes 0.01 (0.03) −0.05, 0.06 0.84 −0.08 (0.06) −0.19, 0.04 0.16 0.01 (0.03) −0.05, 0.07 0.65 No REF — — REF — — REF — — Depression Yes −0.07 (0.02) −0.10, −0.04 < 001 −0.07 (0.04) −0.15, 0.01 0.09 −0.07 (0.02) −0.10, −0.03 < 001 No REF — — REF — — REF — — Intervention cohort Cervical screening 0.09 (0.02) 0.05, 0.12 < 001 −0.08 (0.04) −0.14, 0.03 0.08 0.10 (0.02) 0.05, 0.13 < 001 Breast screening 0.10 (0.02) 0.05, 0.15 < 001 0.01 (0.05) −0.08, 0.14 0.72 0.10 (0.02) 0.05, 0.14 < 001 Colorectal screening REF — — REF — — REF — — 2023 Size of the city a Small town/city REF — — REF — — REF — — Large city ( ≥ 150,000) 0.09 (0.01) 0.07, 0.10 < 001 −0.23 (0.11) −0.51, −0.08 < 001 0.10 (0.01) 0.08, 0.12 < 001 2022 City routine checkup of doctor b Less than the average REF — — REF — — REF — — Higher than the average 0.10 (0.03) 0.02, 0.15 0.003 0.07 (0.11) 0.22, 0.25 0.63 0.09 (0.03) 0.00, 0.14 0.004 2025 County PCP ratio c Less than the average REF — — REF — — REF — — Higher than the average 0.03 (0.05) 0.11, 0.10 0.52 0.21 (0.05) 0.02, 0.25 0.04 0.01 (0.05) 0.14, 0.07 0.90 Open in a new tab Note: The value inside the parentheses is the standard error. Abbreviations: AME, average marginal effect; CI, confidence interval; p , p ‐value; PCP, primary care physician; REF, reference group. a The city population is based on the population estimate as of April 1, 2020 (V2023) from the U.S. Census quickfacts. b The average county level is calculated by summing the routine checkup of doctor rates of all sample cities and dividing by the number of cities in the sample. The data is based on the City Health Dashboard website. c The average county level is calculated by summing the PCP ratio of all sample cities and dividing by the number of cities in the sample. The PCP ratio is sourced from County Health ranking & Roadmaps website. 3.4. Model 2 (Path a) Compared with the 3‐message group, the 6‐message group had an average 0.43‐point increase in message engagement level (AME = 0.43, 95% CI [0.38, 0.49], p < 0.001). In practical terms, this provides evidence that the intervention increased the engagement level of message responses. Patients aged 55–64 had an average 0.15‐point increase in message engagement level (AME = 0.15, 95% CI [0.05, 0.27], p < 0.001), compared to those 65+. Non‐primary English speakers showed, on average, a 0.09‐point higher message engagement level (AME = 0.09, 95% CI [0.02, 0.15], p = 0.02), compared to those who identified English as their primary language. Patients with diabetes had an average 0.10‐point decrease in message engagement level (AME = −0.10, 95% CI [−0.16, −0.03], p = 0.01). Hispanic (AME = −0.12, 95% CI [−0.25, −0.03], p < 0.001), Black or African American (AME = −0.19, 95% CI [−0.31, −0.08], p < 0.001), and Asian or other race groups (AME = −0.13, 95% CI [−0.33, −0.12], p < 0.001) exhibited lower average levels of message engagement, compared to their non‐Hispanic white counterparts. However, compared to patients with Medicaid, patients with Medicare had an average 0.19‐point increase in message engagement level (AME = 0.19, 95% CI [0.04, 0.31], p = 0.03), patients with private insurance had an average 0.21‐point increase in message engagement level (AME = 0.21, 95% CI [0.11, 0.31], p < 0.001), and patients without insurance had an average 0.11‐point increase in message engagement level (AME = 0.11, 95% CI [0.02, 0.19], p = 0.01). In addition, patients living in large city areas had an average 0.23‐point decrease in message engagement level (AME = −0.23, 95% CI [−0.51, −0.08], p < 0.001). In contrast, patients living in counties with a high PCP ratio had an average 0.21 point increase in message engagement (AME = 0.21, 95% CI [0.02, 0.25], p = 0.04). 3.5. Model 3 (Path c') and Indirect Effect (Path a × b ) For the direct effect model, we reported only variables whose statistical significance changed relative to the total‐effect model. Private insurance group was not statistically significant in the direct effect model (AME = 0.03, 95% CI [−0.01, 0.07], p = 0.19). Table A2 ( Supporting Information ) presents the results of the indirect effects. The results indicate that the intervention cohort, age group, language, diabetes status, race/ethnicity, insurance type, city size, and routine physician checkups were statistically significant (all p < 0.05). After adjusting for the mediator in the total effect model, private insurance became statistically significant. 4. Discussion In this study, we examined interactions with SMS reminder services and cancer screening uptake in FQHCs using a mediation model. Overall paths were statistically significant, consistent with partial mediation, suggesting that patient interaction with SMS messages improves cancer screening completion. Patients with private insurance showed evidence consistent with full mediation, indicating that the relationship between insurance type and screening completion appeared to operate through patient interaction with the SMS messages. In contrast, partial mediation was observed among uninsured individuals, Medicare beneficiaries, non‐English speakers, patients with diabetes, and patients living in large cities, suggesting that SMS interactions explained part but not all of the differences in screening completion for these groups. In addition, among these groups, only patients with diabetes showed a lower predicted probability of completing cancer screening. 4.1. Response Rate Mediates Cancer Screening Completion Patients with private insurance had a higher predicted probability of completing their cancer screening, aligning with prior research. Work by Halpern et al. found that patients with private insurance were more likely to receive an early cancer diagnosis and hypothesized this was due to greater engagement with preventive care, including responsiveness to text messages and participation in screening examinations [ 30 ]. Our partial mediation findings are note worthy, considering that uninsured patients and those enrolled in Medicare had higher predicted probabilities of screening completion than Medicaid enrollees. This unexpected finding may reflect FQHC no‐cost screening programs that actively engage uninsured patients through targeted outreach and follow‐up. We posit that uninsured individuals receiving care at FQHCs may be more motivated to take advantage of these no‐cost screenings, compared to Medicaid patients who may delay care particularly if they are navigating multiple social challenges [ 31 ]. Non‐primary English speakers had a higher predicted probability of screening completion. Although work has long suggested language barriers contribute to lower screening rates [ 32 ], review by Genoff et al. found that interventions, specifically those tailored to cultural preferences, can significantly improve screening rates [ 33 ]. Additionally, because our intervention provided messages in Spanish, this may have helped non‐primary English speakers to complete the screening tests. For patients with diabetes who had a lower probability of screening completion, our finding aligns with current literature, which reports lower screening rates for cervical [ 34 ], breast [ 34 ] and CRC [ 35 ] patients with diabetes. Finally, patients living in large cities had a higher predicted probability of screening completion, aligning with a review by Walji et al. that reported screening uptake for all three cancers being generally lower in rural areas [ 36 ]. 4.2. Engagement With SMS Messages We identified several factors influencing patients' engagement with SMS reminder messages. First, participatns in the arm receiving theory‐informed messages (6‐week arm) showed significant higher levels of message engagement. This suggests that adjusting the reminder frequency might be a strategy to increase engagement, as shown in other studies to reduce missed appointments and same‐day cancellations [ 37 ] for clinical visits. Second, older patients showed lower levels of message engagement—this aligns with prior work suggesting older adults are less likely to engage in technological interventions [ 38 ]. Third, patients with diabetes had a lower message engagement, which mirrors work reporting worsened engagement for an SMS diabetic intervention among those with uncontrolled HbA1c [ 39 ]. Fourth, patients from racial and ethnic minority groups had lower average levels of message engagement, consistent with prior studies showing lower interaction among these subpopulations [ 40 ]. However, patients whose primary language was not English had higher average message engagement, likely due to the availability of both English‐ and Spanish‐language messages. Compared to Medicaid enrollees, those with private insurance, or covered by Medicare had a higher level of message engagement, aligning with other studies that report heightened engagement among commercially insured patients [ 23 ]. Patients living in large cities had lower levels of message engagement [ 41 ], whereas those residing in counties with higher population to PCP ratios had high levels of message engagement, which may reflect better healthcare access and stronger patient–provider connectivity. In summary, these findings suggest that SMS‐based outreach can be integrated as a core component of preventive care delivery, particularly for FQHC populations. Policies that support investment in digital infrastructure, patient‐centered messaging platforms, and analytics to monitor engagement could enable FQHCs to move beyond one‐size‐fits‐all reminders toward adaptive, data‐driven communication strategies. Because message response and engagement mediate screening completion, payment models and quality metrics (e.g., in value‐based care and Health Resources and Services Administration performance measures) should incentivize not only the delivery of reminders but also the optimization of their design and tailoring to patient characteristics. 4.3. Limitations First, key sociodemographic variables, such as income level and citizenship status, were not included in the analysis, which may limit understanding of differential impacts across diverse populations. Additionally, participant engagement with the intervention was limited to SMS response, leaving gaps in other potential engagement metrics. Finally, while the study was conducted within a FQHC setting, findings may not be fully generalizable to other healthcare environments with different patient demographics, resource constraints, or organizational structures. Future research should explore the adaptability of these approaches across varied clinical and community contexts. 5. Conclusion In conclusion, patient interactions with SMS messages partially influenced cancer screening completion at FQHCs. Our findings suggest that sending interactive messages can help increase screening uptake, and that patients who receive more messages may be more likely to read and engage with them. Importantly, interactive messaging does not incur substantially higher costs than traditional SMS messaging. Even when patients do not complete cancer screening, interactive messages provide information related to cancer risk, which may help increase patients' knowledge and awareness. This suggests that policies supporting the use of interactive messages rather than traditional messages in FQHC settings. Funding The authors have nothing to report. Conflicts of Interest The authors declare no conflicts of interest. Supporting information Table A1 . The coefficient results of direct effect, X to mediator effect, and total effect. Table A2 . Coefficient of indirect effect model. HESR-61-e70134-s001.docx (25.5KB, docx) Data Availability Statement Data available on request from the authors. The data that support the findings of this study are available from the corresponding author upon reasonable request. References 1. Siegel R. L., Miller K. D., Wagle N. S., and Jemal A., “Cancer Statistics, 2023,” CA: A Cancer Journal for Clinicians 73, no. 1 (2023): 17–48, 10.3322/caac.21763. [ DOI ] [ PubMed ] [ Google Scholar ] 2. Bray F., Laversanne M., Sung H., et al., “Global Cancer Statistics 2022: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries,” CA: A Cancer Journal for Clinicians 74, no. 3 (2024): 229–263, 10.3322/caac.21834. [ DOI ] [ PubMed ] [ Google Scholar ] 3. Moleyar‐Narayana P., Leslie S. W., and Ranganathan S., Cancer Screening (StatPearls Publishing LLC, 2025). [ PubMed ] [ Google Scholar ] 4. Sabatino S. A., Thompson T. D., White M. C., et al., “Up‐to‐Date Breast, Cervical, and Colorectal Cancer Screening Test Use in the United States, 2021,” Preventing Chronic Disease 20 (2023): E94. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 5. Sabatino S. A., White M. C., Thompson T. D., and Klabunde C. N., “Cancer Screening Test Use ‐ United States, 2013,” MMWR. Morbidity and Mortality Weekly Report 64, no. 17 (2015): 464–468. [ PMC free article ] [ PubMed ] [ Google Scholar ] 6. Spencer J. C. and Pignone M. P., “Cancer Screening Through Federally Qualified Health Centers,” JAMA Internal Medicine 184, no. 6 (2024): 679–680, 10.1001/jamainternmed.2024.0702. [ DOI ] [ PubMed ] [ Google Scholar ] 7. Shin P., Rosenbaum S. J., and Paradise J., “Community Health Centers: The Challenge of Growing to Meet the Need for Primary Care in Medically Underserved Communities,” (2012). 8. Amboree T. L., Montealegre J. R., Parker S. L., et al., “National Breast, Cervical, and Colorectal Cancer Screening Use in Federally Qualified Health Centers,” JAMA Internal Medicine 184, no. 6 (2024): 671–679, 10.1001/jamainternmed.2024.0693. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 9. Rosenblatt R. A., Andrilla C. H., Curtin T., and Hart L. G., “Shortages of Medical Personnel at Community Health Centers: Implications for Planned Expansion,” JAMA 295, no. 9 (2006): 1042–1049, 10.1001/jama.295.9.1042. [ DOI ] [ PubMed ] [ Google Scholar ] 10. Varkey A. B., Manwell L. B., Williams E. S., et al., “Separate and Unequal: Clinics Where Minority and Nonminority Patients Receive Primary Care,” Archives of Internal Medicine 3 (2009): 243–250. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 11. Chuang E., Pourat N., Chen X., et al., “Organizational Factors Associated With Disparities in Cervical and Colorectal Cancer Screening Rates in Community Health Centers,” Journal of Health Care for the Poor and Underserved 30, no. 1 (2019): 161–181. [ DOI ] [ PubMed ] [ Google Scholar ] 12. Elmore J. G., Nakano C. Y., Linden H. M., Reisch L. M., Ayanian J. Z., and Larson E. B., “Racial Inequities in the Timing of Breast Cancer Detection, Diagnosis, and Initiation of Treatment,” Medical Care 43, no. 2 (2005): 141–148, 10.1097/00005650-200502000-00007. [ DOI ] [ PubMed ] [ Google Scholar ] 13. Gwyn K., Bondy M. L., Cohen D. S., et al., “Racial Differences in Diagnosis, Treatment, and Clinical Delays in a Population‐Based Study of Patients With Newly Diagnosed Breast Carcinoma,” Cancer 100, no. 8 (2004): 1595–1604, 10.1002/cncr.20169. [ DOI ] [ PubMed ] [ Google Scholar ] 14. Jemal A., Siegel R., Ward E., Hao Y., Xu J., and Thun M. J., “Cancer Statistics, 2009,” CA: A Cancer Journal for Clinicians 59, no. 4 (2009): 225–249, 10.3322/caac.20006. [ DOI ] [ PubMed ] [ Google Scholar ] 15. Schwebel F. J. and Larimer M. E., “Using Text Message Reminders in Health Care Services: A Narrative Literature Review,” Internet Interventions 13 (2018): 82–104, 10.1016/j.invent.2018.06.002. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 16. Uy C., Lopez J., Trinh‐Shevrin C., Kwon S. C., Sherman S. E., and Liang P. S., “Text Messaging Interventions on Cancer Screening Rates: A Systematic Review,” Journal of Medical Internet Research 19, no. 8 (2017): e296, 10.2196/jmir.7893. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 17. Muller C. J., Robinson R. F., Smith J. J., et al., “Text Message Reminders Increased Colorectal Cancer Screening in a Randomized Trial With Alaska Native and American Indian People,” Cancer 123, no. 8 (2017): 1382–1389, 10.1002/cncr.30499. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 18. Boksmati N., Butler‐Henderson K., Anderson K., and Sahama T., “The Effectiveness of SMS Reminders on Appointment Attendance: A Meta‐Analysis,” Journal of Medical Systems 40, no. 4 (2016): 90, 10.1007/s10916-016-0452-2. [ DOI ] [ PubMed ] [ Google Scholar ] 19. Ajzen I., “The Theory of Planned Behavior,” Organizational Behavior and Human Decision Processes 50, no. 2 (1991): 179–211, 10.1016/0749-5978(91)90020-T. [ DOI ] [ Google Scholar ] 20. Guenther L., Gaertner M., and Zeitz J., “Framing as a Concept for Health Communication: A Systematic Review,” Health Communication 36, no. 7 (2021): 891–899, 10.1080/10410236.2020.1723048. [ DOI ] [ PubMed ] [ Google Scholar ] 21. Groden P., Capellini A., Levine E., et al., “The Success of Behavioral Economics in Improving Patient Retention Within an Intensive Primary Care Practice,” BMC Family Practice 22, no. 1 (2021): 253, 10.1186/s12875-021-01593-8. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 22. Bonful H. A., Addo‐Lartey A. A., Sefenu R. S., et al., “Developing a Culturally Tailored Short Message Service (SMS) Intervention for Improving the Uptake of Cervical Cancer Screening Among Ghanaian Women in Urban Communities,” BMC Women's Health 22, no. 1 (2022): 154, 10.1186/s12905-022-01719-9. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 23. Levitz C. E., Kuo E., Guo M., et al., “Using Text Messages and Fotonovelas to Increase Return of Home‐Mailed Colorectal Cancer Screening Tests: Mixed Methods Evaluation,” JMIR Cancer 9 (2023): e39645, 10.2196/39645. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 24. Huguet N., Hodes T., Holderness H., Bailey S. R., JE D. V., and Marino M., “Community Health Centers' Performance in Cancer Screening and Prevention,” American Journal of Preventive Medicine 62, no. 2 (2022): e97–e106, 10.1016/j.amepre.2021.07.007. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 25. Gunzler D., Chen T., Wu P., and Zhang H., “Introduction to Mediation Analysis With Structural Equation Modeling,” Shanghai Archives of Psychiatry 25, no. 6 (2013): 390–394, 10.3969/j.issn.1002-0829.2013.06.009. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 26. MacKinnon D. P., Introduction to Statistical Mediation Analysis (Routledge, 2012). [ Google Scholar ] 27. “City Health Dashboard,” accessed May 23, 2025, https://www.cityhealthdashboard.com/ . 28. “County Health Ranking & Roadmaps Data,” accessed May 23, 2025, https://www.countyhealthrankings.org/ . 29. “U.S. Census Bureau QuickFacts: United States,” accessed May 23, 2025, https://www.census.gov/quickfacts/fact/table/US/PST045224 . 30. Halpern M. T., Ward E. M., Pavluck A. L., Schrag N. M., Bian J., and Chen A. Y., “Association of Insurance Status and Ethnicity With Cancer Stage at Diagnosis for 12 Cancer Sites: A Retrospective Analysis,” Lancet Oncology 9, no. 3 (2008): 222–231. [ DOI ] [ PubMed ] [ Google Scholar ] 31. McIntyre A., Sommers B. D., Aboulafia G., et al., “Coverage and Access Changes During Medicaid Unwinding,” JAMA Health Forum 5, no. 6 (2024): e242193, 10.1001/jamahealthforum.2024.2193. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 32. Toraih E. A., Hussein M. H., Malik M. S., Malik A. N., Kandil E., and Fawzy M. S., “Unraveling the Link Between Language Barriers and Cancer Risk,” Cancer Causes & Control 36, no. 4 (2025): 399–407, 10.1007/s10552-024-01946-5. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 33. Genoff M. C., Zaballa A., Gany F., et al., “Navigating Language Barriers: A Systematic Review of Patient Navigators' Impact on Cancer Screening for Limited English Proficient Patients,” Journal of General Internal Medicine 31, no. 4 (2016): 426–434, 10.1007/s11606-015-3572-3. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 34. Bhatia D., Sutradhar R., Austin P. C., et al., “Periodic Screening for Breast and Cervical Cancer in Women With Diabetes: A Population‐Based Cohort Study,” Cancer Causes & Control 33, no. 2 (2022): 249–259, 10.1007/s10552-021-01517-y. [ DOI ] [ PubMed ] [ Google Scholar ] 35. Wilkinson J. E. and Culpepper L., “Associations Between Colorectal Cancer Screening and Glycemic Control in People With Diabetes, Boston, Massachusetts, 2005–2010,” Preventing Chronic Disease 8, no. 4 (2011): A82. [ PMC free article ] [ PubMed ] [ Google Scholar ] 36. Walji L. T., Murchie P., Lip G., Speirs V., and Iversen L., “Exploring the Influence of Rural Residence on Uptake of Organized Cancer Screening ‐ A Systematic Review of International Literature,” Cancer Epidemiology 74 (2021): 101995, 10.1016/j.canep.2021.101995. [ DOI ] [ PubMed ] [ Google Scholar ] 37. Ulloa‐Pérez E., Blasi P. R., Westbrook E. O., Lozano P., Coleman K. F., and Coley R. Y., “Pragmatic Randomized Study of Targeted Text Message Reminders to Reduce Missed Clinic Visits,” Permanente Journal 26, no. 1 (2022): 64–72, 10.7812/tpp/21.078. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 38. Van Acker J., Maenhout L., and Compernolle S., “Older Adults' User Engagement With Mobile Health: A Systematic Review of Qualitative and Mixed‐Methods Studies,” Innovation in Aging 7, no. 2 (2023): igad007. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 39. Nelson L. A., Spieker A., Greevy R., LeStourgeon L. M., Wallston K. A., and Mayberry L. S., “User Engagement Among Diverse Adults in a 12‐Month Text Message‐Delivered Diabetes Support Intervention: Results From a Randomized Controlled Trial,” JMIR mHealth and uHealth 8, no. 7 (2020): e17534, 10.2196/17534. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 40. Conroy M., Kamaraju S., Powell M., et al., “Racial and Ethnic Differences in the Use of Electronic Medical Record Messaging Among Patients With Breast Cancer: A Quality Improvement Study,” Clinical Breast Cancer 23, no. 7 (2023): e434–e440. [ DOI ] [ PubMed ] [ Google Scholar ] 41. Iqbal I., Thompson L., and Wilson P., “Patient Satisfaction With General Practice in Urban and Rural Areas of Scotland,” Rural and Remote Health 21, no. 4 (2021): 1–10. [ DOI ] [ PubMed ] [ Google Scholar ] Associated Data This section collects any data citations, data availability statements, or supplementary materials included in this article. Supplementary Materials Table A1 . The coefficient results of direct effect, X to mediator effect, and total effect. Table A2 . Coefficient of indirect effect model. HESR-61-e70134-s001.docx (25.5KB, docx) Data Availability Statement Data available on request from the authors. The data that support the findings of this study are available from the corresponding author upon reasonable request. Articles from Health Services Research are provided here courtesy of Health Research & Educational Trust ACTIONS View on publisher site PDF (1.5 MB) Cite Collections Permalink RESOURCES Similar articles Cited by other articles Links to NCBI Databases Follow NCBI NCBI on X (formerly known as Twitter) NCBI on Facebook NCBI on LinkedIn NCBI on GitHub NCBI RSS feed Connect with NLM NLM on X (formerly known as Twitter) NLM on Facebook NLM on YouTube National Library of Medicine 8600 Rockville Pike Bethesda, MD 20894 Web Policies FOIA HHS Vulnerability Disclosure Help Accessibility Careers NLM NIH HHS USA.gov Back to Top

Related documents

• Behaviorally Informed Text Messaging to Promote Colon Cancer Screening: A Quality Improvement Randomized Clinical Trial.#136001
• Access My Breast Cancer Care App#812497
• Effects of an ICT-platform for Assessment and Management of Patient-reported Symptoms During Treatment for Breast Cancer#882676
• Randomized Phase II Study to Evaluate Efficacy of Web-based Patient-reported Outcome (PRO) Application in Solid Cancer Patients#838692
• Mobile health architecture for health-care information delivery: case study: a breast cancer awareness mobile application (PINK RIBBON)#956491