Abstract
Introduction
Perinatal mental health conditions are common yet frequently undertreated. Perinatal collaborative care models improve access to evidence‐based care but face implementation challenges related to workflow efficiency and scalability. Digital innovations may enhance care delivery and sustainability within perinatal collaborative care models. This study evaluated whether a technology‐enabled service, designed to support and streamline care manager workflow in a perinatal collaborative care model, improves depression and anxiety symptoms, patient satisfaction, and engagement compared with the collaborative care model without digital support tools.
Methods
This randomized controlled trial was conducted from October 2022 through January 2024 within a perinatal collaborative care program serving five obstetric clinics in Chicago, Illinois. Participants included 75 pregnant or postpartum individuals enrolled in collaborative care with symptoms of depression (Patient Health Questionnaire‐9 [PHQ‐9] ≥ 5) or anxiety (Generalized Anxiety Disorder‐7 [GAD‐7] ≥ 5). Participants were randomized either to a technology‐enabled service incorporating a cognitive behavioral therapy mobile application with text‐based coaching and a care manager–facing dashboard or to usual perinatal collaborative care without digital tools. Depression and anxiety symptoms were assessed biweekly for 12 weeks using the PHQ‐9 and GAD‐7. Score results were analyzed using generalized linear mixed models. Additional outcomes included symptom response (≥50% reduction from baseline), symptom remission (PHQ‐9 and GAD‐7 < 5), stepped care utilization, patient satisfaction, and engagement.
Results
A total of 75 individuals were enrolled with 38 randomized to the technology‐enabled service. Depression (β = −0.24; 95% CI, −0.45 to −0.03) and anxiety (β = −0.24; 95% CI, −0.46 to −0.03) symptoms improved over time in both groups. There were no significant differences between the technology‐enabled service and usual care groups for depression (β = −0.06; 95% CI, −0.06 to 0.17) or anxiety (β = −0.06; 95% CI, −0.30 to 0.19). Response and remission rates did not differ between groups for depression (43% vs. 53%, p = 0.4; 46% vs. 32%, p = 0.2) or anxiety (35% vs. 55%, p = 0.08; 32% vs. 32%, p = 0.9). Stepped care occurred in 21% of participants in the technology‐enabled service group and 20% in the usual care group (p > 0.9). Engagement at the study midpoint was higher in the technology‐enabled service group, while satisfaction was comparable across groups.
Conclusions
Although symptom improvement was similar between groups, the technology‐enabled service improved patient engagement within perinatal collaborative care. These findings suggest that while digital tools may not directly benefit clinical care, they may enhance operational efficiency and support the scalability and sustainability of perinatal collaborative care models.
Trial Registration
ClinicalTrials.gov identifier: NCT05525689
Keywords: anxiety, collaborative care model, depression, digital health, perinatal mental health, technology‐enabled service
1. INTRODUCTION
Perinatal mental health conditions are common [1, 2] and associated with adverse maternal and child outcomes. Un‐ or undertreated perinatal mental health conditions lead to impaired quality of life, chronic mental health conditions, and, in its most extreme form, suicide, which remains a leading cause of mortality in the first year after birth [3, 4]. Un‐ or undertreated perinatal mental health conditions also have been associated with adverse neurodevelopmental consequences in offspring with effects particularly pronounced in socioeconomically disadvantaged populations [5]. Therefore, adequate treatment of perinatal mental health conditions has critical public health implications.
People with perinatal mental health conditions are often not diagnosed or treated. While screening and initiation of or referral for treatment are recommended [6, 7], obstetric clinicians report inadequate training on pharmacotherapy and limited resources for referral [8]. Patient‐perceived stigma and access barriers obstruct treatment participation, and many current obstetric treatment models limit care to one clinical visit at 6 weeks postpartum. For all of these reasons, sustained contact for monitoring response to treatment is uncommon. Ultimately, only 50% of people with perinatal mental health conditions accept mental health care referrals and <10% remain in treatment [9].
One solution to these gaps in care is the collaborative care model, which integrates mental health into primary care. Mental health benefits are achieved through adherence to two core principles: population‐based case review and measurement‐based treatment to target (or stepped care) [10]. A care manager serves as the cornerstone of the collaborative care model and facilitates initial treatment planning, brief behavioral care, longitudinal symptom monitoring, and implementation of specialist‐informed stepped care recommendations. Collaborative care models can be effective and cost‐effective approaches to improve mental health outcomes in primary care [11, 12, 13, 14, 15, 16]. However, obstetric care is distinct from primary care in that pregnant and postpartum people have newly competing priorities as they introduce the demands of pregnancy and newborn care [17]. Obstetric clinicians often see their scope of care as more narrowly defined by somatic obstetric concerns. The perceived stigma of being judged as a bad parent limits some patients’ willingness to seek mental health care [18]. Finally, the magnitude of need [2] overwhelms existing infrastructures.
While the collaborative care model is efficacious for perinatal mental health care [19, 20], scant information is available regarding implementation. Attempts at implementation without adherence to the core principles of the model have resulted in a lack of depression symptom improvement in the primary care setting [21, 22]. Even with care managers, the clinical prevalence of perinatal mental health conditions leads to programmatic growth that quickly supersedes capacity. Care managers have difficulty carrying out the basic collaborative care model tasks due to the overwhelming number of patients, the failure of systems to provide the depression monitoring data, difficulties in communication, and the large number of small tasks they are required to complete. Thus, pragmatic implementation of perinatal collaborative care models requires significant optimization of care manager workflow.
Technology‐enabled services (TESs) may offer a solution for effective perinatal collaborative care model delivery. TES can (a) include dashboards/interfaces to automate care manager tasks and streamline care manager workflows for population‐based case review; (b) connect the care manager to patients, facilitating communication about stepped care; and (c) directly deliver digital treatments to people who are unable to engage in person due to logistical barriers common in the perinatal period. One solution for direct delivery of digital treatments is the use of smartphone apps and text messaging. Smartphones are broadly available among reproductive‐aged people and are owned by 94% of Americans under age 30 and 89% ages 31–49 [23, 24, 25]. As such, a TES can support adherence to the core tenants of the perinatal collaborative care model, but this novel application of utilizing a TES has not been previously studied. Thus, our objective was to develop and evaluate a TES to support patients and care managers within a perinatal collaborative care model with the unique focus of improving care manager workflow.
2. METHODS
2.1. Trial design
This individual patient‐level randomized controlled trial was approved by the Northwestern University Institutional Review Board prior to study initiation. All participants provided written informed consent prior to randomization. An independent data and safety monitoring committee monitored the trial. Aside from utilization of the TES, participants were managed according to standard clinical care within the perinatal collaborative care model.
2.2. Patient selection
Individuals who were English‐speaking, pregnant or within three months postpartum, at least 18 years of age, referred to the COMPASS (Collaborative Care Model for Perinatal Mental Health Support Services) program, and who own a smartphone and used it within the past week were considered eligible for this study. Exclusion criteria included experiencing a pregnancy loss or suicidality with the presence of a plan. Participants were enrolled between June 2022 and April 2024.
COMPASS is a collaborative care model integrated across five outpatient obstetric clinics affiliated with Northwestern Medicine [26]. These practices include prenatal and postpartum care provided by obstetrician‐gynecologists, maternal‐fetal medicine specialists, nurse practitioners, and certified nurse midwives. All patients receiving prenatal care in these practices are introduced to COMPASS through informational materials provided during their initial prenatal visit. Referrals to COMPASS can be made either by the obstetric clinician or through self‐referral, for mental health concerns arising during pregnancy or within 12 months postpartum.
At COMPASS enrollment, patients complete initial screening assessments, including the Patient Health Questionnaire‐9 (PHQ‐9) [27] for depression and the Generalized Anxiety Disorder‐7 (GAD‐7) [28] for anxiety. Referred participants meet with a care manager, a licensed clinical social worker, to co‐create an individualized care plan. This plan can include psychotherapy, medication management, or both, with the decision informed by symptom severity using a shared decision‐making framework. All enrolled COMPASS participants are monitored in a patient registry, with PHQ‐9 and GAD‐7 surveys sent every 2 weeks while symptomatic and then spaced to every month once in remission. The care managers use a stepped care approach to treatment based on symptom data from the registry with the goal of symptom remission.
COMPASS care managers referred participants to the study team if they had consented to participate in the collaborative care model, their most recent (within 4 weeks) PHQ‐9 or GAD‐7 scores were greater than 4, and they were either pregnant or within 3 months postpartum. Individuals were excluded if they were less than 18 years of age, did not speak English, did not own a smartphone, or had not used a smartphone in the last week. Individuals were also excluded if they experienced a pregnancy loss.
2.3. Treatment allocation
Once eligible participants signed informed consent, they completed a baseline assessment and were randomized in a 1:1 ratio. A computer‐generated randomization sequence using variable block sizes stratified by pregnancy status was created by the study statistician and uploaded to Research Electronic Data Capture (REDCap) [29]. Study staff obtained each participant's randomization assignment from REDCap after confirmation of eligibility was entered.
2.4. Trial intervention
The TES was designed based on Adaptive Health's IntelliCare platform [30, 31, 32]. Prior to trial initiation, via user‐centered design approaches, the intervention was optimized to adapt content to be relevant to the perinatal population, enhance participant engagement in the collaborative care model, and optimize care manager workflows. Specifically, the TES that the intervention group used in this trial consisted of two components: (1) a patient‐facing mobile application (app) designed for the assessment and self‐management of depression and anxiety, and (2) a care manager–facing dashboard to facilitate clinical workflows. The patient‐facing app provided tools for both assessment and self‐management. Participants were prompted to complete symptom assessments within the app, including assessments of their depressive and anxiety symptoms over the previous 2 weeks. For self‐management, the app offered modules based on evidence‐based behavioral strategies, such as scheduling positive activities, cognitive restructuring, and goal setting. Each module addressed a specific treatment goal and could be completed in a brief, user‐friendly format, often requiring only seconds of engagement. Modules included instructional content and interactive components, such as activity monitoring and thought records, to promote behavior change. All participants were required to download the app onto their personal smartphones and were encouraged to use the self‐management modules throughout the study, with guidance from their care manager as needed.
The care manager dashboard was designed to facilitate efficient clinical decision‐making and patient management. This dashboard provided real‐time information on patient engagement with the IntelliCare modules, allowing care managers to monitor which self‐management tools were being utilized by each participant. Additionally, the dashboard displayed mental health assessment scores (i.e., PHQ‐9 and GAD‐7), enabling care managers to track symptom severity and changes over time. A TES section allowed care managers to document patient interactions, while a built‐in text messaging interface supported direct communication with participants, enhancing personalized care and timely interventions based on individual patient needs.
Individuals randomized to treatment as usual received access to a static website that included psychoeducational content specific to perinatal depression. For both groups, beyond the use of the TES, the remainder of care provision was at the discretion of the care manager and the participant under the supervision of an Attending Reproductive Psychiatrist as per the collaborative care model.
2.5. Care manager training and fidelity monitoring
The care managers included two licensed clinical social workers, both trained in collaborative care models, perinatal mental health, and evidence‐based therapy. Care managers were trained by study staff on the functionalities of the IntelliCare mobile app and dashboard, as well as protocols for engaging with participants via the dashboard (e.g., sample messages for various clinical scenarios). To promote engagement for those randomized to the intervention, care managers were asked to send supportive text messages twice weekly within the TES. These text messages either provided positive reinforcement of activities on IntelliCare or offered support to participants with low engagement. The care managers also suggested specific app features or content areas for participants to explore.
Throughout the trial, both care managers participated in weekly supervision with a PhD‐level psychologist. Supervision consisted of discussing participant progress, handling questions of risk, and discussing fidelity to the care manager protocol. Fidelity was measured using the internet‐delivered cognitive behaviour therapy (iCBT) coaching fidelity scale [1, 33]. Each week, the supervisor selected 1 week of text messages between a participant and a care manager, and rated four coaching behaviors on a scale of 1 (inadequate) to 4 (excellent). The supervisor also monitored basic coaching behaviors including sending a message at least one time per week, responding to participant questions within 48 h, and appropriately documenting situations involving risk. An iCBT score of at least 12/16 on scale 1 and 2/3 on scale 2 was considered passing and a score of 15/16 on scale 1 and 3/3 on scale 2 was considered ideal.
2.6. Trial outcomes
All participants in the trial were followed for 12 weeks. The primary outcome was perinatal depression symptoms, measured using the PHQ‐9 [27]. The principal secondary outcome was perinatal anxiety symptoms, measured using the GAD‐7 [28]. Both the PHQ‐9 and GAD‐7 were completed via self‐report. For all participants, screens were sent at baseline, 6 weeks, and 12 weeks. In addition, these screens were sent in REDCap and reviewed by the care managers every 2 weeks while participants were symptomatic (i.e., either GAD‐7 or PHQ‐9 ≥ 5) and every 4 weeks once remission (i.e., both GAD‐7 and PHQ‐9 < 5) was achieved. For individuals randomized to the intervention group, the screens were collected through the digital service platform. For those randomized to usual care, screens were sent to participants via REDCap.
Additional patient experience outcomes included satisfaction and engagement with care. Satisfaction was measured using the Satisfaction Index—Mental Health (SIMH) [34]. Engagement in care was measured using the TWente Engagement with Ehealth Technologies Scale (TWEETS) [35]. Each of these was assessed at Week 6 and Week 12. Additional process outcomes included the prevalence of execution of stepped care, defined as an adjustment in care plans for those who continued to report moderate or higher symptoms (i.e., either GAD‐7 or PHQ‐9 ≥ 10) despite the enactment of an initial treatment plan.
2.7. Coaching fidelity and application usage
For those randomized to the TES, fidelity to coaching was measured using the iCBT coaching fidelity scale and via completed coaching activities. The percentage of scores that were either passing or ideal were estimated. Similarly, the median application usage by participants (measured in days of use) was aggregated and median scores and interquartile ranges (IQRs) were reported. Usage in the first 6 weeks of the trial was compared to usage in the last 6 weeks.
2.8. Trial data analysis
All analyses were conducted following the intention‐to‐treat (ITT) principle. Descriptive statistics were calculated for participant characteristics at baseline and compared between intervention groups using Mann–Whitney U TES for continuous variables and chi‐squared or Fishers’ exact tests for categorical variables. Missing data were addressed using multiple imputation methods to manage potential non‐ignorable mechanisms, assuming data were missing at random.
The primary outcome, depression symptoms as measured by the PHQ‐9, was analyzed using generalized linear mixed models (GLMMs) to assess changes over time between treatment groups. The model included fixed effects for time, treatment group, and their interaction. A random intercept was introduced to account for within‐subject correlation over repeated measures. Anxiety symptoms, as measured by the GAD‐7, were analyzed using similar mixed models.
In addition to examining the overall treatment effects, we conducted moderation analyses to evaluate whether the effect of the intervention varied by participant characteristics, including race, ethnicity, age, health status, pregnancy status at enrollment (i.e., pregnant vs. postpartum), and mobile phone competence. These analyses were performed using GLMMs, incorporating interaction terms between treatment assignment and each potential moderator.
We hypothesized that individuals with moderate symptoms may represent a distinct population in terms of treatment needs and responsiveness. In addition, moderate symptom severity is often a threshold for initiating more intensive interventions, and intervention effects can vary based on initial symptom severity. Accordingly, analyses were also performed for the subgroup of individuals who were referred with moderate or worse depression (i.e., PHQ‐9 > 9) or anxiety (i.e., GAD‐7 > 9) symptoms.
The SIMH and TWEETS scores were analyzed as discrete outcomes at 6 and 12 weeks. Separate analyses were conducted for each time point using the Welch two‐sample t‐test to compare scores across intervention groups. The proportion of participants receiving stepped care was compared between groups using Fisher's exact test. Statistical significance was set at a two‐sided alpha level of 0.05, and all analyses were performed using R software version 4.4.2 (R Foundation for Statistical Computing).
2.9. Sample size
We calculated that a sample size of 75 would provide 80% power to detect an effect size of 0.65, corresponding to a difference of 3.26 points in PHQ‐9 scores between the TES and usual care groups, assuming a standard deviation of 5.
3. RESULTS
3.1. Study participants
A total of 113 participants were assessed for eligibility by the study team for potential inclusion. Of these individuals, 90 participants consented and completed an eligibility screen, and 75 were eligible for participation (Figure 1). These 75 individuals were randomized, with 38 people assigned to the intervention group and 37 people assigned to usual care. Two participants randomized to the intervention did not receive it as intended, and one of these participants was unable to be reached to start treatment or to complete follow‐up surveys. Characteristics of participants are described in Table 1.
FIGURE 1.
CONSORT diagram.
TABLE 1.
Baseline participant characteristics.
| Control | Intervention | |
|---|---|---|
| n = 37 | n = 38 | |
| Age | 33.2 (29.6, 36.7) | 32.0 (28.8, 35.5) |
| Pregnancy status at enrollment | ||
| Pregnant | 18 (48.6%) | 19 (50.0%) |
| Postpartum | 19 (51.4%) | 19 (50.0%) |
| Race | ||
| White | 20 (54.1%) | 13 (34.2%) |
| Black | 5 (13.5%) | 17 (44.7%) |
| Asian | 3 (8.1%) | 3 (7.9%) |
| More than one race | 3 (8.1%) | 3 (7.9%) |
| Decline to report | 6 (16.2%) | 2 (5.3%) |
| Hispanic | ||
| Yes | 11 (29.7%) | 9 (23.7%) |
| No | 26 (70.3%) | 28 (73.7%) |
| Don't know | 0 (0%) | 1 (2.6%) |
| Middle Eastern or North African | ||
| Yes | 2 (5.4%) | 0 (0%) |
| No | 35 (94.6%) | 37 (97.4%) |
| Don't know | 0 (0%) | 1 (2.6%) |
| General health | ||
| Poor | 1 (2.7%) | 0 (0%) |
| Fair | 6 (16.2%) | 8 (21.1%) |
| Good | 15 (40.5%) | 16 (42.1%) |
| Very good | 14 (37.8%) | 13 (34.2%) |
| Excellent | 1 (2.7%) | 1 (2.6%) |
| Engaged in psychotherapy | 31 (83.8%) | 23 (60.5%) |
| Using psychiatric medications | 18 (48.6%) | 13 (34.2%) |
| Insurance (n = 74) | ||
| Private | 26 (70.3%) | 18 (48.6%) |
| Public | 11 (29.7%) | 19 (51.4%) |
| Education | ||
| Some high school or less | 0 (0%) | 1 (2.6%) |
| High school graduate/GED | 4 (10.8%) | 8 (21.1%) |
| Associate degree | 1 (2.7%) | 1 (2.6%) |
| Some college | 11 (29.7%) | 11 (28.9%) |
| Bachelor's degree | 7 (18.9%) | 9 (23.7%) |
| Post‐bachelor's education | 14 (37.8%) | 8 (21.1%) |
| Employed | ||
| Yes | 28 (75.7%) | 24 (63.2%) |
| No | 9 (24.3%) | 12 (31.6%) |
| Prefer not to answer | 0 (0%) | 2 (5.3%) |
| Income | ||
| Less than $10,000 | 0 (0%) | 9 (23.7%) |
| $10,000–$19,999 | 3 (8.1%) | 3 (7.9%) |
| $20,000–$39,999 | 4 (10.8%) | 4 (10.5%) |
| $40,000–$59,999 | 3 (8.1%) | 5 (13.2%) |
| $60,000–$99,999 | 4 (10.8%) | 4 (10.5%) |
| $100,000+ | 19 (51.4%) | 12 (31.6%) |
| Don't know | 1 (2.7%) | 1 (2.6%) |
| Prefer not to answer | 3 (8.1%) | 0 (0%) |
| Marital status | ||
| Divorced | 0 (0%) | 1 (2.6%) |
| Domestic partnership | 1 (2.7%) | 1 (2.6%) |
| Living with a partner | 6 (16.0%) | 6 (15.8%) |
| Married | 25 (67.6%) | 17 (44.7%) |
| Separated | 0 (0%) | 1 (2.6%) |
| Single; never married | 5 (13.5%) | 12 (31.6%) |
| Household | ||
| 1 | 0 (0%) | 2 (5.3%) |
| 2 | 10 (27.0%) | 10 (26.3%) |
| 3 | 18 (48.6%) | 10 (26.3%) |
| 4+ | 9 (24.3%) | 16 (42.1%) |
| 5 | 2 (5.4%) | 5 (13%) |
| 7 | 1 (2.7%) | 1 (2.6%) |
| 8 | 1 (2.7%) | 0 (0%) |
Note: Data presented as median (interquartile range) or n (%).
3.2. Coaching fidelity and application usage
For the care managers, 32.3% of measured encounters achieved a passing iCBT score and 55.4% of measured encounters achieved an ideal iCBT score. For the 35 participants randomized to the TES who utilized the application, use occurred for a median of 31 days (IQR: 24, 40). Usage in the first 6 weeks of the trial was 20 days (IQR: 16, 26) compared to 11 days (IQR 8, 13) in the last 6 weeks.
3.3. Primary outcome
Improvements in depression symptoms were observed in both groups over time (β = −0.24; 95% CI, −0.45 to −0.03; p = 0.025) (Table 2). However, no significant differences were identified between the TES and the usual treatment groups in depression symptom improvement as measured by the PHQ‐9 (β = −0.06; 95% CI, −0.30 to 0.17; p = 0.6) (Figure 2A). Similarly, response (53% vs. 43%, p = 0.4) and remission (32% vs. 46%, p = 0.2) TES for depression showed no significant differences between the TES and usual care groups.
TABLE 2.
Primary and secondary outcomes.
| Controln = 37 | Interventionn = 38 | Relative risk/beta coefficient | p value | |
|---|---|---|---|---|
| Primary outcome | ||||
| Depression symptom trajectories | −0.06 (−0.30 to 0.17) | 0.6 | ||
| Depression symptom response | 16 (43%) | 20 (53%) | 0.4 | |
| Depression symptom remission | 17 (46%) | 12 (32%) | 0.2 | |
| Prespecified secondary outcomes | ||||
| Anxiety symptom trajectories | −0.06 (−0.30 to 0.19) | 0.7 | ||
| Anxiety symptom response | 13 (35%) | 21 (55%) | 0.08 | |
| Anxiety symptom remission | 12 (32%) | 12 (32%) | >0.9 | |
| Satisfaction index—mental health | ||||
| Week 6 (n = 59) | 62 (50, 67) | 58 (47, 68) | >0.9 | |
| Week 12 (n = 60) | 60 (52, 66) | 62 (47, 68) | 0.6 | |
| TWente Engagement with Ehealth Technologies Scale | ||||
| Week 6 (n = 63) | 26 (22, 32) | 38 (30, 43) | 0.007 | |
| Week 12 (n = 62) | 27 (24, 36) | 35 (28, 42) | 0.2 | |
| Stepped Care | 2 (20%) | 4 (21%) | >0.9 |
FIGURE 2.
Individuals randomized to the control group are shown in pink and those randomized to the technology‐enabled service (TES) group in teal. Depression symptoms were measured using the Patient Health Questionnaire‐9 (PHQ‐9), and anxiety symptoms were measured using the Generalized Anxiety Disorder‐7 (GAD‐7). Individual participant trajectories are displayed as semi‐transparent lines, with overlaid smoothed curves representing group‐level trends; shaded areas indicate 95% confidence intervals. Depression (A) and anxiety (B) symptom trajectories stratified by randomization arm.
3.4. Prespecified secondary outcomes
Improvements in anxiety symptoms were observed in both groups over time (β = −0.24; 95% CI, −0.46 to −0.03; p = 0.029) (Table 2). However, no significant differences were identified between the TES and the usual treatment groups in anxiety symptom improvement (β = −0.06; 95% CI, −0.30 to 0.19; p = 0.7) (Figure 2B). Similarly, response (55% vs. 35%, p = 0.08) and remission (32% vs. 32%, p > 0.9) TES for anxiety showed no significant differences between the TES and the usual care groups.
Satisfaction levels were high in both groups, with no significant differences in participant‐reported satisfaction between the TES and usual care at 6 weeks [58 (47–68) vs. 62 (50–67), p > 0.9] and at 12 weeks [62 (47–68) vs. 60 (52–66), p = 0.6] (Table 2). Engagement in perinatal mental health care was higher in the TES group at mid‐point evaluation [38 (30–43) vs. 26 (22–32), p = 0.007], although this difference did not remain statistically significant at 12 weeks [35 (28–42) vs. 27 (24–36), p = 0.2]. Stepped care was administered to 21% of TES participants compared to 20% in the usual care group (p > 0.9).
3.5. Prespecified subgroup analyses
Results indicated that the effect of TES on depressive and anxiety symptoms was consistent across these demographic and clinical subgroups, with no significant moderation observed. Detailed results of these moderation analyses are presented in Table 2. Similarly, among participants presenting with moderate depressive symptoms at baseline (PHQ‐9 > 9), significant symptom improvements were observed over time in both the intervention and control groups. In this subgroup, PHQ‐9 scores decreased by an average of 0.46 points per week (95% CI, −0.72 to −0.21; p < 0.001), with no significant differences in the rate of improvement between the TES and enhanced treatment‐as‐usual groups (interaction term: β = −0.06; 95% CI, −0.36 to 0.24; p = 0.7). Estimated mean PHQ‐9 scores at 12 weeks were 6.83 (95% CI, 4.80 to 8.85) for the TES group and 7.92 (95% CI, 5.93–9.91) for the usual care group, with no statistically significant difference between groups (β = 1.09; 95% CI, −1.43 to 3.62; p = 0.446).
For participants with moderate anxiety symptoms (GAD‐7 > 9), a similar pattern was observed. Both groups exhibited significant reductions in anxiety symptoms over time (β = −0.53 per week; 95% CI, −0.86 to −0.20; p = 0.002). However, there was no significant difference in the trajectory of symptom improvement between TES and usual care groups (interaction term: β = −0.08; 95% CI, −0.41 to 0.26; p = 0.7). At 12 weeks, estimated mean GAD‐7 scores were 6.49 (95% CI, 4.66–8.31) for the TES group and 8.44 (95% CI, 6.49–10.39) for the usual care group, with no statistically significant difference (β = 1.95; 95% CI, −0.68 to 4.58; p = 0.151).
4. DISCUSSION
4.1. Summary of key findings
This randomized clinical trial evaluated the impact of a TES, designed to improve care manager workflow, on depression and anxiety symptoms within a perinatal collaborative care model. Both the TES and usual care groups demonstrated significant improvements in depressive and anxiety symptoms over the 12‐week study period; however, no significant differences were detected in symptom trajectories, response rates, or remission rates between the two groups. Importantly, the TES effectively streamlined care manager workflow and produced clinical outcomes comparable to standard care. These findings suggest that, when thoughtfully implemented, TES may serve as a promising strategy to enhance operational efficiency and promote the sustainability of collaborative care models in perinatal mental health.
Although TES users demonstrated greater engagement with mental health services at the study midpoint, this effect did not persist through 12 weeks. These results indicate that while TES may facilitate initial engagement, further refinement is needed to sustain its impact and enhance its clinical effectiveness. Additionally, exploratory analyses revealed no differential effects based on participant race, ethnicity, age, pregnancy status, overall health, or mobile phone competence, suggesting that the TES was broadly acceptable and feasible across diverse perinatal populations.
4.2. Comparison with existing literature
Our results align with prior studies indicating that collaborative care models are effective for managing perinatal mental health conditions through population‐based care and measurement‐guided treatment strategies [19, 36]. One possible explanation for the lack of differences between the groups is that the marginal impact of digital health interventions may be overshadowed by the overall effect of the collaborative care model as supported by the overall improvements seen across both groups. In contrast, a trial of digital mental health integration within collaborative care models in non‐perinatal populations demonstrated additive clinical benefit. For example, the eIMPACT trial showed that a collaborative care model incorporating digital cognitive behavioral therapy resulted in greater improvements in depression and anxiety outcomes compared with usual collaborative care [37, 38]. Differences in population and the intervention's focus on care delivery versus care manager workflow may account for these divergent findings, particularly given the already high‐functioning collaborative care model in the present study.
The absence of a significant added benefit from the TES over usual care in our study underscores persistent challenges in leveraging digital innovations to enhance clinical outcomes. Consistent with other research on digital mental health interventions, maintaining sustainable user engagement among individuals with mental health conditions remains a significant challenge [39]. In this study, the observed decline in app use over time and attenuation of the early engagement advantage further underscore the difficulty of sustaining engagement beyond initial uptake and the need for implementation strategies that promote sustained engagement, including ongoing reinforcement, integration into clinical workflows, and personalization to evolving patient needs.
4.3. Implications for clinical practice and public health
The observed increase in user engagement associated with the TES suggests its potential use as a tool to enhance participation in perinatal mental health care. In practice, this may be particularly beneficial in populations facing barriers to accessing traditional mental health care, such as unmet health‐related social needs or stigma. However, the absence of superior clinical outcomes, there remains a need for iterative refinement of the TES features to better align with the principles of collaborative care. Enhancing TES functionalities, such as incorporating more robust real‐time feedback or integrating synchronous communication with care managers, may help achieve greater clinical impact. The observed high fidelity rates suggest that care managers are able to successfully integrate coaching and a digital intervention with their existing workflow. These data hold promise that TESs, once optimized, can be successfully implemented into collaborative care models.
4.4. Study limitations
Several limitations should be acknowledged. First, the single‐center design may limit the generalizability of findings to other settings. Second, the modest sample size, while sufficient to detect the hypothesized effect size, limits the ability to identify smaller, yet potentially clinically meaningful differences between groups. Accordingly, the absence of statistically significant differences should not be interpreted as evidence of equivalence. In addition, this sample size constrains the ability to robustly evaluate subgroup differences or heterogeneity of treatment effects. Similarly, although randomization is expected to balance baseline characteristics, in a modestly sized trial, chance imbalances may influence observed outcomes and limit the precision of treatment effect estimates. Finally, while participant attrition was minimal and application usage was consistent, these factors can introduce bias, despite the use of intention‐to‐treat analyses and multiple imputation methods. Specifically, more robust and sustained participant engagement with the TES could have led to improvements in mental health outcomes. As noted, this study utilized a specific TES, with limitations in its functionality. Future research could test a TES that utilizes a dashboard with more robust treatment management functionality with enhanced communication tools that allow for more robust communication with patients on all their care needs. In addition, future research could integrate the TES with the electronic health record to support seamless integration of care.
The study was strengthened by its rigorous user‐centered design approach to the TES, engaging people with lived experience and care managers in its development. Further, the rigor of the randomized design with a prespecified analysis plan minimizes the risks of confounding.
5. CONCLUSION AND FUTURE DIRECTIONS
This study demonstrates that the TES, intentionally designed to support and streamline care manager workflow within the perinatal collaborative care model, achieved clinical outcomes for perinatal depression and anxiety comparable to usual care while improving patient engagement. These findings suggest that thoughtfully integrated digital tools may enhance the efficiency and scalability of collaborative care delivery without compromising clinical effectiveness. Future work will report findings from a planned qualitative evaluation of care manager workflow and implementation processes designed to capture the complexity, usability, and real‐world integration of the TES within clinical practice. Future research should focus on refining TES design to better align digital functionality with patient needs as a more personalized approach to digital mental health may unlock the full potential of technology to address the unmet needs in perinatal mental health care.
CONFLICT OF INTEREST STATEMENT
The authors declare no conflicts of interest.
ETHICS STATEMENT
This study was reviewed and approved by the appropriate institutional review board and was conducted in accordance with applicable ethical standards. Approval was obtained prior to initiation of the study.
ACKNOWLEDGMENTS
This study was supported by the National Institute of Mental Health under grant P50MH119029.
DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request. Access to data may be subject to institutional review board approval and data use agreements to protect participant confidentiality.
REFERENCES
- 1. Gaynes, B. N. , Gavin N., Meltzer‐Brody S., Lohr K. N., Swinson T., Gartlehner G., Brody S., and Miller W. C.. 2005. “Perinatal Depression: Prevalence, Screening Accuracy, and Screening Outcomes: Summary.” In AHRQ Evidence Report Summaries, 119, 1998–2005. Rockville (MD): Agency for Healthcare Research and Quality (US); https://www.ncbi.nlm.nih.gov/books/NBK11838/ [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Wisner, K. L. , Sit D. K. Y., McShea M. C., Rizzo D. M., Zoretich R. A., Hughes C. L., Eng H. F., et al. 2013. “Onset Timing, Thoughts of Self‐Harm, and Diagnoses in Postpartum Women With Screen‐Positive Depression Findings.” JAMA Psychiatry 70(5): 490–8. 10.1001/jamapsychiatry.2013.87. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Metz, T. D. , Rovner P., Hoffman M. C., Allshouse A. A., Beckwith K. M., and Binswanger I. A.. 2016. “Maternal Deaths From Suicide and Overdose in Colorado, 2004–2012.” Obstetrics and Gynecology 128(6): 1233–40. 10.1097/aog.0000000000001695. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Shakespeare, J. and Knight M.. 2015. “Maternal Health in Pregnancy: Messages From the 2014 UK Confidential Enquiry Into Maternal Death.” British Journal of General Practice 65(638): 444–5. 10.3399/bjgp15X686377. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Stein, A. , Pearson R. M., Goodman S. H., Rapa E., Rahman A., McCallum M., Howard L. M., and Pariante C. M.. 2014. “Effects of Perinatal Mental Disorders on the Fetus and Child.” Lancet 384(9956): 1800–19. 10.1016/s0140-6736(14)61277-0. [DOI] [PubMed] [Google Scholar]
- 6. 2023. “Treatment and Management of Mental Health Conditions during Pregnancy and Postpartum: ACOG Clinical Practice Guideline No. 5.” Obstetrics and Gynecology 141(6): 1262–88. 10.1097/AOG.0000000000005202. [DOI] [PubMed] [Google Scholar]
- 7. 2023. “Screening and Diagnosis of Mental Health Conditions during Pregnancy and Postpartum: ACOG Clinical Practice Guideline No. 4.” Obstetrics and Gynecology 141(6): 1232–61. 10.1097/AOG.0000000000005200. [DOI] [PubMed] [Google Scholar]
- 8. LaRocco‐Cockburn, A. , Melville J., Bell M., and Katon W.. 2003. “Depression Screening Attitudes and Practices Among Obstetrician‐Gynecologists.” Obstetrics and Gynecology 101(5 Pt 1): 892–8. [DOI] [PubMed] [Google Scholar]
- 9. Smith, M. V. , Shao L., Howell H., Wang H., Poschman K., and Yonkers K. A.. 2009. “Success of Mental Health Referral Among Pregnant and Postpartum Women With Psychiatric Distress.” General Hospital Psychiatry 31(2): 155–62. 10.1016/j.genhosppsych.2008.10.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Huffman, J. C. , Niazi S. K., Rundell J. R., Sharpe M., and Katon W. J.. 2014. “Essential Articles on Collaborative Care Models for the Treatment of Psychiatric Disorders in Medical Settings: A Publication by the Academy of Psychosomatic Medicine Research and Evidence‐Based Practice Committee.” Psychosomatics 55(2): 109–22. 10.1016/j.psym.2013.09.002. [DOI] [PubMed] [Google Scholar]
- 11. Archer, J. , Bower P., Gilbody S., Lovell K., Richards D., Gask L., Dickens C., et al. 2012. “Collaborative Care for Depression and Anxiety Problems.” Cochrane Database of Systematic Reviews 10: CD006525. 10.1002/14651858.CD006525.pub2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Thota, A. B. , Sipe T. A., Byard G. J., Zometa C. S., Hahn R. A., McKnight‐Eily L. R., Chapman D. P., et al. 2012. “Collaborative Care to Improve the Management of Depressive Disorders: A Community Guide Systematic Review and Meta‐Analysis.” American Journal of Preventive Medicine 42(5): 525–38. 10.1016/j.amepre.2012.01.019. [DOI] [PubMed] [Google Scholar]
- 13. Grypma, L. , Haverkamp R., Little S., and Unützer J.. 2006. “Taking an Evidence‐Based Model of Depression Care From Research to Practice: Making Lemonade Out of Depression.” General Hospital Psychiatry 28(2): 101–7. 10.1016/j.genhosppsych.2005.10.008. [DOI] [PubMed] [Google Scholar]
- 14. Reiss‐Brennan, B. , Briot P. C., Savitz L. A., Cannon W., and Staheli R.. 2010. “Cost and Quality Impact of Intermountain's Mental Health Integration Program.” Journal of Healthcare Management 55(2): 97–113; discussion 113–4. [PubMed] [Google Scholar]
- 15. Solberg, L. I. , Glasgow R. E., Unützer J., Jaeckels N., Oftedahl G., Beck A., Maciosek M. V., and Crain A. L.. 2010. “Partnership Research: A Practical Trial Design for Evaluation of a Natural Experiment to Improve Depression Care.” Medical Care 48(7): 576–82. 10.1097/MLR.0b013e3181dbea62. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Zivin, K. , Pfeiffer P. N., Szymanski B. R., Valenstein M., Post E. P., Miller E. M., and McCarthy J. F.. 2010. “Initiation of Primary Care‐Mental Health Integration Programs in the VA Health System: Associations With Psychiatric Diagnoses in Primary Care.” Medical Care 48(9): 843–51. 10.1097/MLR.0b013e3181e5792b. [DOI] [PubMed] [Google Scholar]
- 17. Bowen, A. , Bowen R., Butt P., Rahman K., and Muhajarine N.. 2012. “Patterns of Depression and Treatment in Pregnant and Postpartum Women.” Canadian Journal of Psychiatry 57(3): 161–7. 10.1177/070674371205700305. [DOI] [PubMed] [Google Scholar]
- 18. Button, S. , Thornton A., Lee S., Shakespeare J., and Ayers S.. 2017. “Seeking Help for Perinatal Psychological Distress: A Meta‐Synthesis of Women's Experiences.” British Journal of General Practice 67(663): e692–e99. 10.3399/bjgp17X692549. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Grote, N. K. , Katon W. J., Russo J. E., Lohr M. J., Curran M., Galvin E., and Carson K.. 2015. “Collaborative Care for Perinatal Depression in Socioeconomically Disadvantaged Women: A Randomized Trial.” Depression and Anxiety 32(11): 821–34. 10.1002/da.22405. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Melville, J. L. , Reed S. D., Russo J., Croicu C. A., Ludman E., LaRocco‐Cockburn A., and Katon W.. 2014. “Improving Care for Depression in Obstetrics and Gynecology: A Randomized Controlled Trial.” Obstetrics and Gynecology 123(6): 1237–46. 10.1097/AOG.0000000000000231. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Solberg, L. I. , Crain A. L., Jaeckels N., Ohnsorg K. A., Margolis K. L., Beck A., Whitebird R. R., Rossom R. C., Crabtree B. F., and Van de Ven A. H.. 2013. “The DIAMOND Initiative: Implementing Collaborative Care for Depression in 75 Primary Care Clinics.” Implement Science 8: 135. 10.1186/1748-5908-8-135. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Solberg, L. I. , Crain A. L., Maciosek M. V., Unutzer J., Ohnsorg K. A., Beck A., Rubenstein L., et al. 2015. “A Stepped‐Wedge Evaluation of an Initiative to Spread the Collaborative Care Model for Depression in Primary Care.” Annals of Family Medicine 13(5): 412–20. 10.1370/afm.1842. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Perrin, A. and Duggan M.. 2015. Americans’ Internet Access: 2000–2015. Washington DC: Pew Research Center. [Google Scholar]
- 24. Smith, A. 2015. U.S. Smartphone Use in 2015 . Washington DC: Pew Research Center. http://www.pewinternet.org/2015/04/01/us‐smartphone‐use‐in‐2015/. [Google Scholar]
- 25. Poushter, J. 2016. Smartphone Ownership and Internet Usage Continues to Climb in Emerging Economies . Washington DC: Pew Research Center. [Google Scholar]
- 26. Miller, E. S. , Grobman W. A., Ciolino J. D., Zumpf K., Sakowicz A., Gollan J., and Wisner K. L.. 2021. “Increased Depression Screening and Treatment Recommendations After Implementation of a Perinatal Collaborative Care Program.” Psychiatric Services 72(11): 1268–75. 10.1176/appi.ps.202000563. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Kroenke, K. , Spitzer R. L., and Williams J. B.. 2001. “The PHQ‐9: Validity of a Brief Depression Severity Measure.” Journal of General Internal Medicine 16(9): 606–13. 10.1046/j.1525-1497.2001.016009606.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Spitzer, R. L. , Kroenke K., Williams J. B., and Löwe B.. 2006. “A Brief Measure for Assessing Generalized Anxiety Disorder: The GAD‐7.” Archives of Internal Medicine 166(10): 1092–7. 10.1001/archinte.166.10.1092. [DOI] [PubMed] [Google Scholar]
- 29. Harris, P. A. , Taylor R., Thielke R., Payne J., Gonzalez N., and Conde J. G.. 2009. “Research Electronic Data Capture (REDCap)—A Metadata‐Driven Methodology and Workflow Process for Providing Translational Research Informatics Support.” Journal of Biomedical Informatics 42(2): 377–81. 10.1016/j.jbi.2008.08.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Graham, A. K. , Greene C. J., Kwasny M. J., Kaiser S. M., Lieponis P., Powell T., and Mohr D. C.. 2020. “Coached Mobile App Platform for the Treatment of Depression and Anxiety among Primary Care Patients: A Randomized Clinical Trial.” JAMA Psychiatry 77(9): 906–14. 10.1001/jamapsychiatry.2020.1011. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Mohr, D. C. , Schueller S. M., Tomasino K. N., Kaiser S. M., Alam N., Karr C., Vergara J. L., Gray E. L., Kwasny M. J., and Lattie E. G.. 2019. “Comparison of the Effects of Coaching and Receipt of App Recommendations on Depression, Anxiety, and Engagement in the IntelliCare Platform: Factorial Randomized Controlled Trial.” Journal of Medical Internet Research 21(8): e13609. 10.2196/13609. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Mohr, D. C. , Tomasino K. N., Lattie E. G., Palac H. L., Kwasny M. J., Weingardt K., Karr C. J., et al. 2017. “IntelliCare: An Eclectic, Skills‐Based App Suite for the Treatment of Depression and Anxiety.” Journal of Medical Internet Research 19(1): e10. 10.2196/jmir.6645. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Hadjistavropoulos, H. D. , Schneider L. H., Klassen K., Dear B. F., and Titov N.. 2018. “Development and Evaluation of a Scale Assessing Therapist Fidelity to Guidelines for Delivering Therapist‐Assisted Internet‐Delivered Cognitive Behaviour Therapy.” Cognitive Behaviour Therapy 47(6): 447–61. 10.1080/16506073.2018.1457079. [DOI] [PubMed] [Google Scholar]
- 34. Nabati, L. , Shea N., McBride L., Gavin C., and Bauer M. S.. 1998. “Adaptation of a Simple Patient Satisfaction Instrument to Mental Health: Psychometric Properties.” Psychiatry Research 77(1): 51–6. 10.1016/s0165-1781(97)00122-4. [DOI] [PubMed] [Google Scholar]
- 35. Kelders, S. M. , Kip H., and Greeff J.. 2020. “Psychometric Evaluation of the TWente Engagement With Ehealth Technologies Scale (TWEETS): Evaluation Study.” Journal of Medical Internet Research 22(10): e17757. 10.2196/17757. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Miller, E. S. , Wisner K. L., Gollan J., Hamade S., Gossett D. R., and Grobman W. A.. 2019. “Screening and Treatment After Implementation of a Universal Perinatal Depression Screening Program.” Obstetrics and Gynecology 134(2): 303–9. 10.1097/aog.0000000000003369. [DOI] [PubMed] [Google Scholar]
- 37. Stewart, J. C. , Patel J. S., Polanka B. M., Gao S., Nurnberger J. I., MacDonald K. L., Gupta S. K., et al. 2023. “Effect of Modernized Collaborative Care for Depression on Depressive Symptoms and Cardiovascular Disease Risk Biomarkers: EIMPACT Randomized Controlled Trial.” Brain, Behavior, and Immunity 112: 18–28. 10.1016/j.bbi.2023.05.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Williams, M. K. , Shell A. L., Wu W., MacDonald K. L., Callahan C. M., Nurnberger J. I., Crawford C. A., Schuiling M. D., and Stewart J. C.. 2026. “Longer‐Term Effects of Modernized Collaborative Care for Depression on Multiple Mental Health Factors (eIMPACT): A Randomized Controlled Trial.” Journal of Affective Disorders 392: 120272. 10.1016/j.jad.2025.120272. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39. Borghouts, J. , Eikey E., Mark G., De Leon C., Schueller S. M., Schneider M., Stadnick N., Zheng K., Mukamel D., and Sorkin D. H.. 2021. “Barriers to and Facilitators of User Engagement With Digital Mental Health Interventions: Systematic Review.” Journal of Medical Internet Research 23(3): e24387. 10.2196/24387. [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.
Data Availability Statement
The data that support the findings of this study are available from the corresponding author upon reasonable request. Access to data may be subject to institutional review board approval and data use agreements to protect participant confidentiality.