A Mobile Application for Monitoring and Management of Depressed Mood in a Vulnerable Pregnant Population

This study was an RCT drawing from a convenience sample. Pregnant women were recruited from March to September 2015 from an urban ambulatory prenatal clinic within an academic medical center serving a predominantly low-income and racial-ethnic minority population. Women at least 18 years old and 32 weeks gestation or less, per electronic health record (EHR), were approached by a research coordinator at a routine prenatal appointment. Enrollment was offered to English-speaking women with depressive symptoms (≥5 on the Patient Health Questionnaire–9 [PHQ-9]) who owned a smartphone with an iOS or Android operating system (8).

Participants were randomly assigned to one of three conditions: a mobile app allowing access to a patient portal (PP), the standard of care available to all patients at the health center, that enabled e-mail-like communication with providers; PP with the addition of an MTA app (Ginger.io; San Francisco, California) that alerted providers when participant mood symptoms worsened, prompting the provider to contact the participant; or the PP app and MTA app with a lottery incentive to encourage MTA app use. Additional description of the apps is available as an online supplement to this brief report. Research staff assisted participants with the app(s) download, provided information packets explaining the assigned mobile app(s), and administered a baseline interview (demographic information, prenatal care engagement, and satisfaction). After eight weeks, care engagement and satisfaction were reassessed. Service delivery data on patient-provider contacts during the eight-week period were extracted from the EHR of each participant. As an exploratory outcome, depressive and anxiety symptoms were assessed after eight weeks within the MTA group. A priori sample size calculations were based on a moderate effect size and a power of .8. This study was approved by the University of Pennsylvania Institutional Review Board, conducted according to the guidelines of the Declaration of Helsinki, and registered on ClinicalTrials.gov.

Patient engagement and care satisfaction questions addressed confidence in ability to manage one’s own health, feeling that prenatal providers understand and respond to one’s unique needs as a patient, feeling connected with providers, ability to communicate with prenatal providers, whether one was left with unanswered questions after prenatal visits, and general satisfaction with prenatal care; items were rated on a Likert scale. To assess service delivery, data on patient-provider interactions were abstracted from the EHR for all encounters dated within the participant’s eight-week study period. Information obtained included medical history, office visit frequency, telephone contact, secure messages sent and received via PP, mental health specialist referrals and visits, and encounters that mentioned mental health or the MTA app. To assess app acceptability, participants in the MTA group were asked eight Likert-scale questions about the perceived acceptability and utility of the mobile app [see online supplement]. As exploratory measures, depressive and anxiety symptoms were assessed after eight weeks in the MTA group by using the PHQ-9 and the Generalized Anxiety Disorder–7 (GAD-7) (9).

Descriptive statistics were generated and examined for normality. Student’s t and chi-square tests assessed group differences in demographic characteristics. Patient engagement, care satisfaction, and service delivery outcomes were treated as continuous variables and were analyzed with analysis of covariance (ANCOVA) by using baseline levels as covariates. Because prenatal visits are more frequent in the later stages of pregnancy, ANCOVAs were adjusted for gestational age. A p value was considered significant at p<.05; values presented were not corrected for multiple comparisons. Analyses were performed with IBM SPSS Statistics for Macintosh, version 23.

The sample included 72 participants (PP control app, N=24; PP plus MTA, N=25; PP plus MTA with lottery, N=23) of 91 screened. One participant did not receive the allocated intervention because of incomplete app download, one participant withdrew, and five participants were lost to follow-up. Randomization created generally equivalent groups; the only significant difference among the three groups in demographic characteristics, health, or psychiatric history was gestational age, which was included as a covariate in ANCOVAs (Table 1). There was no difference in utilization or outcomes for the two MTA groups related to the presence of the lottery; therefore, the groups (with and without lottery) were combined for all additional analyses.

Regarding patient engagement and care satisfaction, as gestational age increased, women in the MTA group rated their ability to manage their own health significantly higher than the control group after eight weeks (F=4.03, df=4 and 49, p=.007; p_interaction=.04). At week 8, controlling for baseline care satisfaction scores, there were no significant differences in care satisfaction between the PP and MTA groups.

In terms of service delivery, compared with the PP group, the MTA group had significantly more telephone encounters with providers that mentioned mental health (F=6.0, df=1 and 55, p=.02), but this group did not have significantly more participants with mental health referrals or significantly better referral adherence than the PP group (Table 1). Within the MTA group, 17 women (41%) received a phone call from a provider that was triggered by an alert from the app. Those women who received a provider contact triggered by the MTA app had a significantly higher rate of referral to a mental health provider compared with women who did not receive an MTA app–triggered contact (t=−2.3, df=15, p=.03). Exploratory analysis of depressive and anxiety symptoms revealed that mean daily mood score was significantly positively correlated with number of calls that a participant received in weeks 1–4 (p<.05 for both comparisons). Participants who received an MTA-triggered call had consistently higher PHQ-9 and GAD-7 scores across the eight weeks compared with those who did not receive an MTA-triggered call. These differences were significant at weeks 1–4 for the PHQ-9 and at Weeks 3–4 for the GAD-7 (p<.05 for both comparisons).

Among participants randomly assigned to the MTA app, the minimum number of days that a participant used the app was 10. The most frequent users (N=3) used the app daily. Mean use frequency was 42.49 (±13.32) days, or 75.9% of days (about 5.3 days per week). In terms of acceptability, after eight weeks, at least half of MTA users said that the app helped them to feel open and honest about their feelings or helped them to manage their mental health. Less than 50% felt that the app helped their prenatal team to understand and respond to their unique needs, or to feel more connected to their care team [see online supplement].

Among MTA users, PHQ-9 (F=7.87, df=2.5 and 47.3, p=.001), GAD-7 (F=6.32, df=2.2 and 42.1, p=.003), and self-reported daily mood scores (F=2.62, df=4.2 and 139.9, p=.03) significantly improved over eight weeks.

The aim of this study was to assess whether an MTA mobile app used in a high-risk obstetric population would improve patient engagement and mental health care delivery, over and above a standard app that connected patients to providers through a PP. Participants had a high rate of mental health concerns, including moderate depressive symptoms and a psychiatric diagnosis in more than half. Participants were largely African American, unmarried, young, relatively early in pregnancy, and receiving Medicaid or Medicare, representing a highly vulnerable population.

After the eight-week trial, women in the third trimester in the MTA group experienced a significant improvement in their perceived ability to manage their own health, whereas those in the PP control group did not. Previous research has suggested that mHealth interventions can help patients to self-manage certain health conditions (10).

Because the MTA app alerted providers when a patient was experiencing worsening mood symptoms, triggering telephone contact from the provider, it follows that the MTA users had more phone contact with their providers regarding mental health. Furthermore, women who received an MTA app-triggered call from a provider had significantly higher mean PHQ-9 and GAD-7 scores early in the study, indicating that the app was correctly identifying women in distress. Finally, women who received an MTA app-triggered telephone contact had significantly more referrals to a mental health provider than users who did not have a triggered contact from a provider. This finding indicates that an app such as MTA is a feasible option to improve mental health service delivery via monitoring at-risk patients between visits; this app also facilitated patient-provider contact when needed rather than relying on the patient to decide to send an electronic message to her provider through a PP.

Although a comparison group was not available, measures of depressive and anxiety symptoms and self-reported daily mood improved over eight weeks among MTA app users. However, patients with depressive symptoms tend to improve over time (11), so we cannot make any causal claims that symptom improvement was due to app use.

This study’s strengths included randomization to the MTA app or a control app and an extended (eight-week) period of app use. Although greater than 90% of women in this clinic had smartphones, less than 10% of all patients seen in this clinic had activated the PP app. Limitations included restricting participation to smartphone users actively seeking prenatal care and unblind status of the research staff. We were also underpowered to adjust our assessment of statistical significance for multiple comparisons. Also, the study was focused on care delivery rather than on clinical outcomes; future studies should assess depressive and anxiety symptoms in both app and control groups to assess the role of these apps in improving these outcomes.

A mobile app that monitored patients between routine obstetric visits with targeted provider contact when needed was well accepted, and it improved mental health service delivery in a high-need obstetric practice. Tools such as this app may help monitor and treat women at high risk of developing psychiatric symptoms during pregnancy.

The authors thank the members of the “Stress in Pregnancy: Improving Results With Interactive Technology (SPIRIT)” group at the University of Pennsylvania for their assistance in carrying out the study; research coordinators Rebecca Henderson and Alicia Lo; Dina Appleby, M.S., and Mary Sammel, Sc.D., for guidance on statistical analysis; the staff of the University of Pennsylvania Helen O. Dickens Center for Women's Health for providing study-related clinical services; Katy Mahraj, M.S.I., and Roy Rosin, M.B.A., from the University of Pennsylvania School of Medicine Center for Healthcare Innovation for technical guidance; and Lucy Edwards and Leanne Kaye at Ginger.io for technical guidance.