This cluster randomized clinical trial investigates whether sending a text message with an invitation to use a digital reproductive decision support tool prior to primary care visits increases reproductive health discussions among female veterans.
Key Points
Question
Does sending veterans a 1-way text message invitation to use MyPath, a patient-facing digital reproductive decision support tool, prior to primary care visits increase discussions about prepregnancy health or contraception?
Findings
In this cluster randomized clinical trial of 349 female veterans aged 18 to 44 years, only half (49.7%) of intervention arm participants sent the text invitation used the tool prior to their primary care visit. Differences in reproductive health discussions between groups were not significant in intent-to-treat analyses.
Meaning
A 1-way text message to deliver a decision support tool in primary care settings, while low cost and scalable, did not significantly increase reproductive health discussions, suggesting that additional strategies are needed to increase engagement with such tools prior to visits.
Abstract
Importance
Discussions about individuals’ reproductive desires and needs for services (eg, prepregnancy care, contraception) are critical but often lacking in primary care.
Objective
To investigate the effectiveness of a 1-way text message invitation to use MyPath, a patient-facing digital reproductive decision support tool, prior to primary care visits in increasing reproductive health discussions at the visit.
Design, Setting, and Participants
This hybrid type 1 study using a cluster randomized clinical trial design was conducted from March 2021 to March 2025 at 12 Veterans Administration health care systems. Participants included primary care practitioners (PCPs), randomized 1:1 to intervention or control, and female veterans aged 18 to 44 years who had an upcoming appointment with a randomized PCP.
Intervention
Veterans in the intervention group were sent a text message invitation to use the decision support tool 1 to 2 days prior to their appointment; veterans in the control group received usual care (no text message).
Main Outcomes and Measures
The primary outcome was self-reported occurrence of a discussion of reproductive health with optimal shared decision-making (SDM) at the visit. Any reproductive health discussion (with or without optimal SDM) was a secondary outcome. The primary analyses followed intent-to-treat (ITT) principles. A prespecified secondary analysis estimated the complier average causal effect to account for nonuse of the decision support tool.
Results
Data from 66 PCPs (mean [SD] age, 47.7 [10.6] years; 61 [92.4%] female) and 349 veterans (mean [SD] age, 35.3 [5.9] years) were analyzed, with 34 PCPs and 190 veterans in the intervention group and 32 PCPs and 159 veterans in the control group. Although all but 1 of the intervention participants were sent a text invitation, only half (83 of 167 [49.7%]) used the decision support tool. Differences between the intervention and control groups in the occurrence of reproductive health discussion with optimal SDM (29 of 168 [17.3%] vs 20 of 139 [14.4%]) or the occurrence of any reproductive health discussion (119 of 168 [70.8%] vs 93 of 142 [65.5%]) were not significant in ITT analyses (adjusted odds ratio, 1.46 [95% CI, 0.70-3.06] and 1.37 [95% CI, 0.79-2.38], respectively). In secondary analyses estimating the intervention effect among compliers, use of the decision support tool was associated with greater odds of any reproductive health discussion (adjusted odds ratio, 3.46 [95% CI, 1.23-9.76]).
Conclusions and Relevance
In this cluster randomized clinical trial of a text invitation to use a decision support tool prior to PCP visits, there were no significant differences in reproductive health discussions between the intervention and usual care groups in ITT analyses. Given higher than expected nonuse of the decision support tool, intervention modifications and implementation strategies to promote greater use while maintaining scalability and sustainment in primary care settings should be explored.
Trial Registration
ClinicalTrials.gov Identifier: NCT04584294
Introduction
National guidelines recommend that clinicians routinely engage their patients in conversations about reproductive desires and needs for reproductive health services, including contraceptive care and care to support healthy pregnancies.1 These services are particularly critical for female veterans, who face elevated risks of adverse pregnancy outcomes compared with nonveterans due to a higher burden of chronic medical and mental health conditions, sexual trauma histories, and intimate partner violence.2,3,4,5 In the Veterans Health Administration (VA), primary care practitioners (PCPs) deliver first-line reproductive health screenings and care, with gynecologist referral available either within the VA or in the community for specialized care.6 However, discussions about reproductive health needs are frequently absent in VA primary care encounters,7 and those that do occur often lack key elements of patient-centered care, including assessment of individual preferences and values.8,9
Clinician-facing interventions, such as electronic health record (EHR) reminders, demonstrate limited success in improving reproductive health services in primary care.10,11,12 Patient-facing decision support tools, in addition to improving patients’ knowledge and decision quality, can improve clinician-patient communication in primary care through prompting conversations and increasing patient engagement in shared decision-making (SDM), a 2-way communication process that incorporates patients’ values and preferences.13 We developed the MyPath patient-facing digital reproductive decision support tool with the goal of empowering veterans to initiate discussions about reproductive health needs and to communicate effectively with clinicians about their preferences and values related to reproductive decisions. In a pilot test among veterans interested in receiving information about prepregnancy health or contraception, the use of the decision support tool on a tablet in clinic waiting rooms prior to scheduled primary care visits was associated with increased discussions about reproductive health needs.14 However, the use of the tool in clinic waiting rooms disrupted clinic flow, and veterans preferred to use it at home with more time to engage with the content.14
Text messaging has been extensively tested as a communication strategy to facilitate health care interactions,15,16 and the vast majority of VA users currently receive 1-way text appointment reminders through the VEText system.17 To address challenges identified in the pilot study, we investigated using this system to invite veterans to use the decision support tool on their smartphone 1 to 2 days before scheduled visits. Initial testing in primary care clinics at VA Puget Sound in 2019 demonstrated the acceptability and feasibility of these standalone messages containing a link to access the decision support tool.18 We designed a cluster randomized clinical trial19 to further evaluate the effectiveness of text message delivery of the decision support tool in increasing discussions about reproductive needs in primary care.
Methods
Study Design
We conducted a hybrid type 1 effectiveness-implementation study, which tests effectiveness of an intervention while gathering data on implementation,19 using a cluster randomized clinical trial design with randomization at the PCP level. A description of the trial design was previously published,20 and the full trial protocol is provided in Supplement 1. The study protocol was approved by the VA Central Institutional Review Board. Verbal informed consent was obtained for participation, as enrollment was conducted via telephone. We followed the Consolidated Standards of Reporting Trials (CONSORT) reporting guideline.21 A woman veteran engagement group provided advice on research objectives, study materials, and interpretation of results.
Study Participants and Recruitment
We recruited VA PCPs and veterans from primary care clinics at 12 VA health care systems across the US from March 2021 through March 2025 (eFigure 1 in Supplement 2). Primary care for female veterans is delivered by designated women’s health PCPs (WH-PCPs) who have received specialized women’s health training.6 WH-PCPs may practice in general primary care clinics or in Women’s Health Clinics (WHCs) with separate entrances and waiting rooms.7,22 PCPs were eligible for recruitment if they were designated WH-PCPs regardless of practice setting. They were excluded if they were in training or planned to leave the VA in the next 18 months. With facilitation from the VA Women’s Health Practice–Based Research Network,23 study staff conducted initial outreach to PCPs via email and Microsoft Teams message and then screened and enrolled PCPs via telephone or videoconference.
Veterans 18 to 44 years of age were eligible if they had an upcoming primary care appointment with an enrolled study PCP, had no evidence of hysterectomy or bilateral oophorectomy, and were interested in receiving information about or discussing contraception or prepregnancy health. They were ineligible for enrollment if they were currently pregnant, reported previously using the decision support tool, or had opted out of VA text messaging. We used EHR data to identify potentially eligible female veterans assigned to study PCPs and mailed study information to potential participants. On a weekly basis, we identified potential participants who had upcoming appointments with study PCPs and contacted them by email, telephone, or both to assess eligibility.
Randomization and Data Collection Procedures
Enrolled PCPs were randomized to the intervention or usual care group with a 1:1 allocation ratio. Randomization was stratified by VA health care system and by clinic type, categorized as WHCs vs general primary care clinics, given data linking care received in WHCs to improved reproductive health outcomes.7,22 Veterans were automatically assigned to the study group of the clinician they were scheduled to visit. Staff involved in veteran recruitment and outcome assessment remained blinded to the PCPs’ group throughout the study. Veterans completed a baseline survey administered via telephone at enrollment. Study outcomes were captured in a postvisit telephone survey administered by study staff. Starting in 2023, veterans were given the option to complete the postvisit survey online. Veterans were compensated $25 and $50 for completing the baseline and postvisit surveys, respectively.
Intervention Group
Veterans in the intervention group were sent a single, 1-way text message from a 5-digit number inviting them to use the decision support tool 1 to 2 business days prior to their scheduled primary care appointment. The theoretical basis, development, and pilot testing of the decision support tool have been previously published.14,24 The tool integrates a reproductive goals assessment module and information about fertility and prepregnancy health with a previously tested evidence-based contraceptive decision support module (My Birth Control).14,24 As the intervention was designed to be patient-facing with minimal requirements for PCPs, study PCPs were not advised of patient enrollment. The principal investigator (L.S.C.) delivered a 30-minute overview of the tool to intervention PCPs after enrollment.
Control Group
PCPs in the control group received no orientation or resources but were informed about the purpose and aims of the study. Veterans in the control group received usual care and did not receive a text message.
Text Messaging Receipt and Decision Support Tool Use
We obtained self-reported data from intervention participants on whether they saw the text and used the decision support tool. Participants who saw the text but did not use the tool were asked about reasons for nonuse. Responses were typed when the survey was self-administered online or were transcribed by study staff (S.S.M., Z.H.P., A.Y., and C.G.M.) when surveys were administered via telephone.
Outcomes and Measures
The primary outcome was a composite measure based on veteran self-report of whether a discussion about reproductive health needs occurred at the visit and, if so, whether the conversation included optimal SDM. First, whether a reproductive health discussion occurred at the visit was measured as a yes response to 1 or more the following survey items: “During your visit, did you and your provider discuss… (1) your thoughts about pregnancy or preventing pregnancy? (2) your health in relation to a future pregnancy? (3) preventing pregnancy or birth control?” (eTable in Supplement 2). If they reported discussing any of these topics at their appointment, SDM was measured using the validated 3-item CollaboRATE scale scored from 0 to 12: “Thinking about the conversation you had with your provider about pregnancy and/or birth control… (1) How much effort was made to help you understand your health issues or choices? (2) How much effort was made to listen to the things that matter most to you? (3) How much effort was made to include what matters to you in choosing what to do next?” (eTable in Supplement 2).25 Optimal SDM was defined as a top-box CollaboRATE score (12 of 12 vs all other scores) as is recommended in the literature based on demonstrated ceiling effects with this and other patient experience measures.26 Veterans’ self-report of the occurrence of any reproductive health discussion at the appointment regardless of SDM scores was a secondary outcome.
For PCP participants, baseline characteristics were collected by self-report. For veteran participants, baseline characteristics obtained by self-report included sociodemographic and health characteristics. Self-identified race and ethnicity were captured as social characteristics given their established association with poor experiences of reproductive health counseling and care.8,27,28 Race categories were based on US Census Bureau categories and included American Indian or Alaska Native, Asian, Black or African American, Native Hawaiian or Pacific Islander, White, equally identified with more than 1 category, no racial identity, and a write-in category. Baseline veteran characteristics obtained from EHR data included age at study enrollment, history of prior appointments with study PCP, and history of any positive screen for military sexual trauma.
Sample Size
Initial sample size and power calculations were based on our primary outcome and relied on formulas specified for use in cluster randomized trials accounting for intervention nonuse and attrition.29 Calculations assumed 30 PCP clusters and baseline reproductive health discussions with SDM prevalence of 50% in the usual care group, 30% decision support tool nonuse in the intervention group,18 and 5% attrition between enrollment and the postvisit survey. Assuming a κ coefficient of 0.01, a sample of 456 veterans yielded 80% power in 2-sided tests with a type I error rate of 5% to detect an increase of 20 percentage points in discussions, estimated based on the observed difference in the pilot study.14 Due to lower than expected cluster sizes and challenges with recruiting veterans, we changed the trial protocol to continue PCP recruitment throughout the study to increase our number of clusters.
Statistical Analysis
For the primary intent-to-treat (ITT) analyses, we used multilevel logistic regression models to test whether the odds of the primary and secondary outcomes were significantly different in the intervention vs control group postvisit, including fixed effects for VA health care system and clinic type and a random effect for PCPs. Adjustment for selected veteran characteristics was prespecified in the protocol. Due to a smaller realized sample size and to avoid overparameterization, the decision was made to further limit adjustment to characteristics that were most prognostic of the outcome based on the literature (age, race, and socioeconomic status captured by educational achievement and health insurance status).30,31 Analyses were prespecified to impute missing outcome and baseline variables using Multiple Imputation by Chained Equations in R, version 4.5.0 (R Project for Statistical Computing) with missing values imputed separately for each treatment group. The analysis was performed with 50 imputations using outcome and baseline variables prespecified for adjustment.32 Statistical significance was assessed using a 2-sided α level of .05.
In a prespecified secondary analysis, we estimated the complier average causal effect (CACE) of the intervention, which is the causal effect of the decision support tool restricted to a subpopulation of patients who would comply with the treatment assigned.33 We used a principal stratification framework and structural equation modeling approach implemented in Mplus, version 8.11 (Muthén & Muthén), and steps included (1) a multilevel logistic (or normal) regression model for outcomes adjusting for complier type and the same ITT analyses adjustment variables; and (2) a multilevel logistic regression model for binary complier type adjusting for the ITT analyses adjustment variables as well as additional variables prespecified to estimate complier type (decision support tool intervention receipt, sexual orientation, housing status, and orientation toward pregnancy).
Results
Study Participants
A total of 74 PCPs were enrolled and randomized; at least 1 veteran patient was enrolled for 66 PCPs over the study time frame (Figure 1). A total of 1297 veterans were screened for eligibility, and 391 were initially enrolled (Figure 2). Veterans were withdrawn prior to outcome assessment per the study protocol if they did not attend their visit (16 intervention, 18 control), learned that they were pregnant between enrollment and the postvisit survey (0 intervention, 1 control), or decided to withdraw (3 intervention, 4 control).20 This approach resulted in 349 veterans included in analyses. Overall, 34 PCPs and 190 veterans were included in the intervention group and 32 PCPs and 159 veterans in the control group. The trial was stopped before achieving recruitment goals due to budget and staffing constraints. Under the original sample size calculation assumptions, the achieved sample size was estimated to provide 70% power to detect the prespecified primary outcome effect size for α level .05.
Figure 1. Study Flow Diagram for Primary Care Practitioners (PCPs) .
Figure 2. Study Flow Diagram for Veterans .
PCP indicates primary care practitioner; VA, Veterans Administration.
aOne veteran missing data for primary outcome.
bThree veterans missing data for primary outcome.
Table 1 presents the baseline characteristics of 66 included PCPs (mean [SD] age at enrollment, 47.7 [10.6] years; 61 [92.4%] female, 5 [7.6%] male) and 349 included veterans (mean [SD] age at enrollment, 35.3 [5.9] years). PCPs self-reported race as Asian (18 [27.3%]), Black or African American (11 [16.7%]), White (30 [45.5%]), and other (6 [9.0%]). Veterans self-reported race as American Indian or Alaska Native (8 [2.3%]), Asian (11 [3.2%]), Black or African American (141 [40.4%]), White (140 [40.1%]), equally identified with 1 or more category (19 [5.4%]), and other (29 [8.3%]). Intervention and control group PCPs were similar across most characteristics; however, a greater percentage of intervention vs control PCPs were nurse practitioners (11 [32.4%] vs 5 [15.6%]) and a lower percentage were physicians (21 [61.8%] vs 24 [75.0%]). The mean PCP cluster size was 5.3 veterans (range, 1-28). Among veteran participants, a greater percentage of the intervention group compared with the control group identified as Black or African American (84 [44.2%] vs 57 [35.8%]) and a smaller percentage identified as White (67 [35.3%] vs 73 [45.9%]). A smaller percentage of veterans in the intervention group compared with the control group reported the VA as their only source of health insurance coverage (94 [49.5%] vs 94 [59.1%]) or had a bachelor’s degree or higher (91 [47.9%] vs 87 [54.7%]). A smaller percentage of the intervention vs control group had previously visited the study PCP (135 [71.1%] vs 136 [85.5%]).
Table 1. Baseline Characteristics by Study Group.
| Characteristic | Participants, No. (%) | ||
|---|---|---|---|
| Total | Intervention | Control | |
| VA primary care practitioners | |||
| No. | 66 | 34 | 32 |
| Age at enrollment, mean (SD), y | 47.7 (10.6) | 47.7 (11.6) | 47.7 (9.6) |
| Sex | |||
| Female | 61 (92.4) | 32 (94.1) | 29 (90.6) |
| Male | 5 (7.6) | 2 (5.9) | 3 (9.4) |
| Hispanic ethnicity | 6 (9.1) | 4 (11.8) | 2 (6.2) |
| Race | |||
| Asian | 18 (27.3) | 7 (20.6) | 11 (34.4) |
| Black or African American | 11 (16.7) | 6 (17.6) | 5 (15.6) |
| White | 30 (45.5) | 17 (50.0) | 13 (40.6) |
| Othera | 6 (9.0) | 4 (11.8) | 2 (6.2) |
| Missing | 1 (1.5) | 0 | 1 (3.1) |
| Clinician type | |||
| Physician | 45 (68.2) | 21 (61.8) | 24 (75.0) |
| Nurse practitioner | 16 (24.2) | 11 (32.4) | 5 (15.6) |
| Physician assistant | 5 (7.6) | 2 (5.9) | 3 (9.4) |
| Practices in a women’s health clinic | 17 (25.8) | 9 (26.5) | 8 (25.0) |
| Veterans | |||
| No. | 349 | 190 | 159 |
| Age at enrollment, mean (SD), yb | 35.3 (5.9) | 35.5 (5.9) | 35.1 (5.9) |
| Previously visited study clinicianb | 271 (77.7) | 135 (71.1) | 136 (85.5) |
| Hispanic ethnicity | 71 (20.3) | 40 (21.1) | 31 (19.5) |
| Race | |||
| American Indian or Alaska Native | 8 (2.3) | 8 (4.2) | 0 |
| Asian | 11 (3.2) | 3 (1.6) | 8 (5.0) |
| Black or African American | 141 (40.4) | 84 (44.2) | 57 (35.8) |
| White | 140 (40.1) | 67 (35.3) | 73 (45.9) |
| Identified equally with >1 category | 19 (5.4) | 12 (6.3) | 7 (4.4) |
| Otherc | 29 (8.3) | 15 (7.9) | 14 (8.8) |
| Missing | 1 (0.3) | 1 (0.5) | 0 |
| Straight sexual orientation | 269 (77.1) | 149 (78.4) | 120 (75.5) |
| Highest achieved educational level | |||
| High school diploma or GED | 28 (8.0) | 17 (8.9) | 11 (6.9) |
| Some college or associate’s degree | 142 (40.7) | 81 (42.6) | 61 (38.4) |
| Bachelor’s degree or higher | 178 (51.0) | 91 (47.9) | 87 (54.7) |
| Missing | 1 (0.3) | 1 (0.5) | 0 |
| Health insurance only through VA benefits | 188 (53.9) | 94 (49.5) | 94 (59.1) |
| Unstable housingd | 39 (11.2) | 16 (8.4) | 23 (14.5) |
| Has ≥1 chronic medical conditionse | 241 (69.1) | 131 (68.9) | 110 (69.2) |
| Has ≥1 mental health conditionsf | 298 (85.4) | 158 (83.2) | 140 (88.1) |
| History of military sexual traumab | 149 (42.7) | 78 (41.1) | 71 (44.7) |
| Sexually active with a male partner last 12 mo | 264 (75.6) | 147 (77.4) | 117 (73.6) |
| Orientation toward pregnancy: wants to become pregnant… | |||
| Now | 40 (11.5) | 21 (11.1) | 19 (11.9) |
| Not trying, but would be okay with it | 50 (14.3) | 23 (12.1) | 27 (17.0) |
| Not now, but in the future | 101 (28.9) | 55 (28.9) | 46 (28.9) |
| Never | 105 (30.1) | 60 (31.6) | 45 (28.3) |
| Not sure | 53 (15.2) | 31 (16.3) | 22 (13.8) |
Abbreviations: GED, General Educational Development; VA, Veterans Administration.
Other race for clinicians includes self-identified (write-in) other racial identity categories such as Latino or Middle Eastern, and identified equally with more than 1 racial identity categories.
Indicates this characteristic was obtained from VA medical records data; all other characteristics are self-reported by participants.
Other race for veterans includes Native Hawaiian or Pacific Islander, self-identified (write-in) categories such as Latino, and no racial identity category.
Unstable housing defined as report of not having a stable place to stay during the past 3 months or being worried about not having a stable place to stay in the next 3 months.
Chronic medical condition(s) defined as responding yes to the following question, “Do you have any chronic medical conditions, such as diabetes, high blood pressure, migraines, asthma, or others?”
Mental health conditions defined as responding yes to the following question, “Do you have any mental health conditions, such as depression, anxiety, bipolar disorder, posttraumatic stress disorder, or others?”
Text Message Receipt and Decision Support Tool Use
Among intervention participants who completed postvisit survey questions (n = 168), study team records indicated that a decision support tool text message was sent to all but 1 (167 of 168) (Table 2). Among intervention participants who were sent the text, 49.7% used the tool (83 of 167). Nonuse of the tool was driven by participants who reported that they did not remember receiving the text message at all (46 of 167 [27.5%]) and by participants who reported that they did receive the text message but did not use the decision support tool (38 of 121 [31.4%]). The most common reasons for tool nonuse among the latter group are summarized in Table 2.
Table 2. Text Message Delivery and Receipt, Decision Support Tool Use, and Reasons for Nonuse Among Intervention Group Participants.
| Intervention participant outcome | Participants, No. (%) | Example quotations |
|---|---|---|
| Delivery of text to intervention participantsa | ||
| No. of intervention participants | 168 | NA |
| Study team sent text to participant | 167 (99.4) | NA |
| Study team did not send text to participant | 1 (0.6) | NA |
| Self-reported text receipt among participants sent a text | ||
| No. sent a text | 167 | NA |
| Participant recognized receipt of text | 121 (72.4) | NA |
| Participant did not recognize receipt of text | 46 (27.5) | NA |
| Self-reported use of the decision support tool among participants who recognized receipt of text | ||
| No. who recognized receipt of text | 121 | NA |
| Participant used the tool | 83 (68.6) | NA |
| Participant did not use the tool | 38 (31.4) | NA |
| Self-reported reasons for nonuse of tool among participants who recognized receipt of text | ||
| No. did not use the decision support tool | 38 | NA |
| Too busy or forgot to go back to use it later | 16 (42.1) | “I received the text during my work hours and wasn’t able to respond, and I eventually lost track of it in my messages.” |
| Did not fully read text invitation or did not understand what it was for | 8 (21.1) | “I wasn’t sure what it was for, so I didn’t click on the link as I sometimes get random messages.” |
| Reproductive health was not the main reason for this appointment | 2 (5.3) | “I didn’t need more information about [reproductive health] because my appointment was not to talk about that.” |
| Technology problems using the decision support tool | 2 (5.3) | “Still trying to figure it out; only half of it loaded.” |
| No reason given or reason was unclear | 10 (26.3) | NA |
Abbreviation: NA, not applicable.
The intervention participant group (n = 168) excludes 21 participants who did not complete the postvisit survey and 1 participant with missing responses for questions regarding receipt and use of the decision support tool.
Effectiveness Outcomes
The majority of veterans reported discussing at least 1 reproductive health topic at their primary care visit (119 of 168 [70.8%] in the intervention group, 61 of 82 [74.4%] in intervention group participants who used the decision support tool, and 93 of 142 [65.5%] in the control group) (Table 3). Reports of reproductive health discussions that included optimal SDM using the top-box CollaboRATE score were uncommon in all groups (# of # [17.3%] in the intervention group, # of # [17.1%] in the decision support tool users, and # of # [14.4%] in the control group). The distribution of CollaboRATE scores is included in eFigure 2 in Supplement 2.
Table 3. Reproductive Health Discussion by Study Group.
| Outcome | Veterans, No./total No. (%) | Adjusted odds ratio (95% CI) | |||
|---|---|---|---|---|---|
| Intervention (n = 169)a | Decision support tool users (n = 83)a | Control (n = 142)a | ITT modelb | CACE modelb | |
| Discussion with shared decision-makingc,d | 29/168 (17.3) | 14/82 (17.1) | 20/139 (14.4) | 1.46 (0.70-3.06) | No convergence |
| Occurrence of any discussiond | 119/168 (70.8) | 61/82 (74.4) | 93/142 (65.5) | 1.37 (0.79-2.38) | 3.46 (1.23-9.76) |
Abbreviations: CACE, complier average causal effect; ITT, intent-to-treat.
Frequencies and proportions exclude participants with missing outcomes.
Models used multiple imputation and included prespecified adjustment variables (veteran age, race, educational attainment, and health insurance status).
Shared decision-making defined as a top score on all items of the 3-item CollaboRATE scale.
Occurrence of any discussion measured as a yes response to 1 or more of 3 survey items.
ITT analyses showed no difference by study group in the occurrence of reproductive health discussion with optimal SDM (adjusted odds ratio [AOR], 1.46 [95% CI, 0.70-3.06]) or the occurrence of any reproductive health discussion (AOR, 1.37 [95% CI, 0.79-2.38]) (Table 3). In analyses estimating the CACE that accounted for decision support tool nonuse, the users had higher odds of reporting any reproductive health discussion compared with controls (AOR, 3.46 [95% CI, 1.23-9.76]). Due to small cell size, the CACE estimation for discussions with optimal SDM did not converge.
Discussion
We evaluated the effect of a single, 1-way text message invitation to use a decision support tool prior to primary care visits on self-reported occurrence of reproductive health discussions at the visit. Our primary ITT analyses found no significant differences between groups in reproductive health discussion with SDM or any reproductive health discussion at visits, with lower than expected engagement with the tool.
The findings in our primary analyses may be related to several factors. First, the study was underpowered due to recruitment challenges, including the COVID-19 pandemic and a lower proportion of veterans meeting eligibility criteria than anticipated, resulting in a lower than planned sample size. Second, only half of intervention participants used the decision support tool prior to their visit, which was lower than expected and further reduced statistical power. Prior research on text message use in health care interventions suggests that engagement with text messages may be higher with greater frequency of text messaging, personalization of texts, and clarity around the sender of the text message.15,16,34 Our inability to personalize text message content due to VA data security and privacy requirements, reliance on 5-digit number to deliver the text message through the VA text program, and lack of automation capacity that could have facilitated higher-frequency texting all likely contributed to suboptimal text engagement.
Occurrence of reproductive health discussions with optimal SDM was low in both the intervention and control groups, with no effect of the intervention. This contrasts with previous work demonstrating that use of the contraceptive decision support component (My Birth Control) of the decision support tool prior to scheduled visits for contraceptive care resulted in an increased proportion reporting a top-box score on the Interpersonal Quality of Family Planning (IQFP) scale.24 We used CollaboRATE rather than the IQFP to enable us to capture counseling quality in the broader scope of reproductive health discussions covered by our studied decision support tool, including prepregnancy health. However, use of a generic SDM measure rather than one tailored for reproductive health is a plausible explanation for the discrepancy. Another explanation may be the different clinical settings, as My Birth Control was tested in family planning and safety net clinics, in which clinicians may have had greater comfort with reproductive health counseling than PCPs in the VA. Future areas for investigation include development of validated measures to capture counseling quality in reproductive health discussions beyond contraception and strategies to build VA PCPs’ reproductive health counseling skills.
Based on anticipated less-than-perfect decision support tool use in the trial, we prespecified a secondary analysis to estimate the CACE of the intervention, which isolated the effect of the decision support tool among those who would use it if assigned and addressed the dilution that imperfect use introduced into ITT analyses. Our finding of greater odds of any reproductive health discussion among the decision support tool users compared with their control group counterparts is consistent with prior pilot work by members of our team14 as well as a growing number of studies demonstrating increased preventive health discussions following use of patient-facing tools prior to PCP visits,13 including colon cancer screening35,36 and end-of-life goals of care.37 Additional research is warranted to investigate adaptions and implementation strategies, such as using text platforms allowing automation and personalization; implementing systems to complement texts, such as secure messages from patient portals; or leveraging existing interactions with health care staff, which could increase text engagement and allow for better assessment of the decision support tool impact.
Strengths and Limitations
The strengths of the study include the pragmatic design, the low cost and scalability of delivering the intervention, and the geographic and sociodemographic diversity of the sample. Several limitations deserve mention. As discussed, the original sample size was not achieved due to recruitment challenges. In addition, higher than expected nonuse of the decision support tool reduced statistical power in the ITT analysis and limited our ability to evaluate the effectiveness of tool use. While estimation of the CACE preserves the benefits of randomization without the selection bias inherent in naive per-protocol analyses,33 our CACE findings were limited by the lower statistical power and resulting imprecision of the estimates. Finally, due to the nature of any health care professional–level cluster randomization, we observed imbalances in veteran characteristics by study group; although we adjusted for a limited set of key variables, residual unmeasured confounding between groups may have remained.
Conclusions
In this cluster randomized clinical trial of a 1-way text message invitation to use a decision support tool prior to VA primary care visits, we did not find significant differences in the occurrence of reproductive health discussions with or without SDM, with lower than expected use of the tool. This study contributes to the literature on the limitations of 1-way text messaging as a delivery strategy for patient-facing web-based tools. Additional research could identify modifications and implementation strategies that prompt greater use of the decision support tool while maintaining scalability and sustainment in primary care settings.
Trial Protocol
eFigure 1. Map of MyPath Trial Recruitment Sites (n=12)
eTable. Survey Items Used to Define Reproductive Discussion and Shared Decision Making Outcomes
eFigure 2. Distribution of 3-Item CollaboRATE Scores Among Trial Participants Who Had a Reproductive Health Discussion
Data Sharing Statement
References
- 1.Romer SE, Blum J, Borrero S, et al. Providing quality family planning services in the United States: recommendations of the U.S. Office of Population Affairs (revised 2024). Am J Prev Med. 2024;67(6S):S41-S86. doi: 10.1016/j.amepre.2024.09.007 [DOI] [PubMed] [Google Scholar]
- 2.Kimerling R, Gima K, Smith MW, Street A, Frayne S. The Veterans Health Administration and military sexual trauma. Am J Public Health. 2007;97(12):2160-2166. doi: 10.2105/AJPH.2006.092999 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Kimerling R, Iverson KM, Dichter ME, Rodriguez AL, Wong A, Pavao J. Prevalence of intimate partner violence among women veterans who utilize Veterans Health Administration primary care. J Gen Intern Med. 2016;31(8):888-894. doi: 10.1007/s11606-016-3701-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Eisen SV, Schultz MR, Vogt D, et al. Mental and physical health status and alcohol and drug use following return from deployment to Iraq or Afghanistan. Am J Public Health. 2012;102(Suppl 1)(suppl 1):S66-S73. doi: 10.2105/AJPH.2011.300609 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Lehavot K. Coping strategies and health in a national sample of sexual minority women. Am J Orthopsychiatry. 2012;82(4):494-504. doi: 10.1111/j.1939-0025.2012.01178.x [DOI] [PubMed] [Google Scholar]
- 6.Handbook VHA. 1330.01: Health Care Services for Women Veterans (May 21, 2010). Accessed August 8, 2026. https://www.womenvetsusa.org/cmsfiles/docs/vha-handbook-1330-01-health-care-services-for-women-veterans-21-may-10.pdf?1786182297
- 7.Quinn DA, Sileanu FE, Borrero S, Callegari LS. Reproductive health services: a missed opportunity in VA primary care? Contraception. 2023;120:109957. doi: 10.1016/j.contraception.2023.109957 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Callegari LS, Tartaglione EV, Magnusson SL, et al. Understanding women veterans’ family planning counseling experiences and preferences to inform patient-centered care. Womens Health Issues. 2019;29(3):283-289. doi: 10.1016/j.whi.2019.03.002 [DOI] [PubMed] [Google Scholar]
- 9.Callegari LS, Mahorter SS, Benson SK, Zhao X, Schwarz EB, Borrero S. Perceived contraceptive counseling quality among veterans using VA primary care: data from the ECUUN study. J Gen Intern Med. 2022;37(suppl 3):698-705. doi: 10.1007/s11606-022-07586-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Srinivasulu S, Shah SD, Schechter CB, Prine L, Rubin SE. Effectiveness of clinical decision support to enhance delivery of family planning services in primary care settings. Contraception. 2020;101(3):199-204. doi: 10.1016/j.contraception.2019.11.002 [DOI] [PubMed] [Google Scholar]
- 11.Schwarz EB, Parisi SM, Handler SM, et al. Clinical decision support to promote safe prescribing to women of reproductive age: a cluster-randomized trial. J Gen Intern Med. 2012;27(7):831-838. doi: 10.1007/s11606-012-1991-y [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Miller MK, Mollen C, Behr K, et al. Development of a novel computerized clinical decision support system to improve adolescent sexual health care provision. Acad Emerg Med. 2019;26(4):420-433. doi: 10.1111/acem.13570 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Stacey D, Lewis KB, Smith M, et al. Decision aids for people facing health treatment or screening decisions. Cochrane Database Syst Rev. 2024;1(1):CD001431. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Callegari LS, Nelson KM, Arterburn DE, et al. Development and pilot testing of a patient-centered web-based reproductive decision support tool for primary care. J Gen Intern Med. 2021;36(10):2989-2999. doi: 10.1007/s11606-020-06506-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Suffoletto B. Deceptively simple yet profoundly impactful: text messaging interventions to support health. J Med Internet Res. 2024;26:e58726. doi: 10.2196/58726 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Dobson R, Whittaker R, Abroms LC, et al. Don’t Forget the humble text message: 25 years of text messaging in health. J Med Internet Res. 2024;26:e59888. doi: 10.2196/59888 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.VEText enhances access and improves experience for veterans. VA News. March 7, 2019. Accessed August 8, 2026. https://news.va.gov/57288/vetext-enhances-access-improves-experience-veterans/
- 18.Callegari LS, Benson SK, Bondzie JA, et al. Leveraging text messaging to disseminate a web-based family planning decision support tool in primary care. J Gen Intern Med. 2021;36(suppl 1):S45. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Curran GM, Bauer M, Mittman B, Pyne JM, Stetler C. Effectiveness-implementation hybrid designs: combining elements of clinical effectiveness and implementation research to enhance public health impact. Med Care. 2012;50(3):217-226. doi: 10.1097/MLR.0b013e3182408812 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Callegari LS, Benson SK, Mahorter SS, et al. Evaluating the MyPath web-based reproductive decision support tool in VA primary care: Protocol for a pragmatic cluster randomized trial. Contemp Clin Trials. 2022;122:106940. doi: 10.1016/j.cct.2022.106940 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Hopewell S, Chan AW, Collins GS, et al. CONSORT 2025 Statement: updated guideline for reporting randomized trials. JAMA. 2025;333(22):1998-2005. doi: 10.1001/jama.2025.4347 [DOI] [PubMed] [Google Scholar]
- 22.Koenig AF, Borrero S, Zhao X, Callegari L, Mor MK, Sonalkar S. Factors associated with long-acting reversible contraception use among women veterans in the ECUUN study. Contraception. 2019;100(3):234-240. doi: 10.1016/j.contraception.2019.05.010 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Frayne SM, Carney DV, Bastian L, et al. The VA Women’s Health Practice-Based Research Network: amplifying women veterans’ voices in VA research. J Gen Intern Med. 2013;28(Suppl 2)(suppl 2):S504-S509. doi: 10.1007/s11606-013-2476-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Dehlendorf C, Fitzpatrick J, Fox E, et al. Cluster randomized trial of a patient-centered contraceptive decision support tool, My Birth Control. Am J Obstet Gynecol. 2019;220:565e1-565e12. [DOI] [PubMed] [Google Scholar]
- 25.Elwyn G, Barr PJ, Grande SW, Thompson R, Walsh T, Ozanne EM. Developing CollaboRATE: a fast and frugal patient-reported measure of shared decision making in clinical encounters. Patient Educ Couns. 2013;93(1):102-107. doi: 10.1016/j.pec.2013.05.009 [DOI] [PubMed] [Google Scholar]
- 26.Forcino RC, Barr PJ, O’Malley AJ, et al. Using CollaboRATE, a brief patient-reported measure of shared decision making: results from three clinical settings in the United States. Health Expect. 2018;21(1):82-89. doi: 10.1111/hex.12588 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Gomez AM, Wapman M. Under (implicit) pressure: young Black and Latina women’s perceptions of contraceptive care. Contraception. 2017;96(4):221-226. doi: 10.1016/j.contraception.2017.07.007 [DOI] [PubMed] [Google Scholar]
- 28.Higgins JA, Kramer RD, Ryder KM. Provider bias in long-acting reversible contraception (LARC) promotion and removal: perceptions of young adult women. Am J Public Health. 2016;106(11):1932-1937. doi: 10.2105/AJPH.2016.303393 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Friedman LM, Furberg CD, DeMets DL, Reboussin DM, Granger CB. Fundamentals of Clinical Trials. Springer International Publishing; 2015. doi: 10.1007/978-3-319-18539-2 [DOI] [Google Scholar]
- 30.Kavanaugh ML, Pliskin E, Hussain R. Associations between unfulfilled contraceptive preferences due to cost and low-income patients’ access to and experiences of contraceptive care in the United States, 2015-2019. Contracept X. 2022;4:100076. doi: 10.1016/j.conx.2022.100076 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Welti K, Manlove J, Finocharo J, Faccio B, Kim L. Women’s experiences with person-centered family planning care: differences by sociodemographic characteristics. Contracept X. 2022;4:100081. doi: 10.1016/j.conx.2022.100081 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.von Hippel PT. How many imputations do you need? a two-stage calculation using a quadratic rule. Sociol Methods Res. 2020;49(3):699-718. doi: 10.1177/0049124117747303 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Little R, Yau L. Statistical techniques for analyzing data from prevention trials: treatment of no-shows using Rubin’s causal model. Psychol Methods. 1998;3(2):147-159. doi: 10.1037/1082-989X.3.2.147 [DOI] [Google Scholar]
- 34.Ziegenfuss JY, Asche SE, Dehmer SP, Reams M, Grossman ES, Solberg LI. The impact of text messaging to promote patient reported outcome measures (PROMs) completion in orthopedic practice: findings from a randomized controlled study. Am J Med Qual. 2024;39(2):78-85. doi: 10.1097/JMQ.0000000000000174 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Lewis CL, Kistler CE, Dalton AF, et al. A decision aid to promote appropriate colorectal cancer screening among older adults: a randomized controlled trial. Med Decis Making. 2018;38(5):614-624. doi: 10.1177/0272989X18773713 [DOI] [PubMed] [Google Scholar]
- 36.Miller DP Jr, Denizard-Thompson N, Weaver KE, et al. Effect of a digital health intervention on receipt of colorectal cancer screening in vulnerable patients: a randomized controlled trial. Ann Intern Med. 2018;168(8):550-557. doi: 10.7326/M17-2315 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Sudore RL, Schillinger D, Katen MT, et al. Engaging diverse English- and Spanish-speaking older adults in advance care planning: the PREPARE randomized clinical trial. JAMA Intern Med. 2018;178(12):1616-1625. doi: 10.1001/jamainternmed.2018.4657 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Trial Protocol
eFigure 1. Map of MyPath Trial Recruitment Sites (n=12)
eTable. Survey Items Used to Define Reproductive Discussion and Shared Decision Making Outcomes
eFigure 2. Distribution of 3-Item CollaboRATE Scores Among Trial Participants Who Had a Reproductive Health Discussion
Data Sharing Statement
