Psychobiological Interventions in Pregnancy

Study Design This will be a randomized, parallel group, unblinded, single site superiority trial evaluating the impact of a physical activity intervention on reducing symptoms in people with known or screen-positivity for depression or anxiety during pregnancy. Participants will be randomized to a step count intervention versus usual care. Adherence will be monitored via Actigraph Link GT9X accelerometer watches.

Primary outcome: The primary outcome will be the mean within-person change in EPDS score between Time 2 (randomization) and Time 3 (37 weeks).

Secondary outcomes: Secondary outcomes will include additional psychobiologic measures, pregnancy complications, SMM events, and neonatal outcomes as shown below:

• Maternal Psychologic and Clinical: For psychologic measures, within person change in EPDS anxiety subscale score between Time 2 and Time 3, STAI score between Time 2 and Time 3, as well as within person change in both EPDS and STAI from Time 2 and Time 4 (6 weeks postpartum), will be assessed. Delivery outcomes (e.g. mode of delivery, gestational age at delivery) as well as complications including hypertensive disorders and SMM indicator events will be abstracted from the medical record.
• Maternal Biologic: For biologic measures, within person change in leukocyte telomere length and hair cortisol level from Time 2 and Time 3 will be assessed. Saliva samples for telomeres will be retained until the study is complete, at which point DNA will be extracted from leukocytes and telomere length will be measured using quantitative polymerase chain reaction. All correlated samples will be analyzed on the same 96-well plates. Telomere length results are reported as telomere "T" to single copy gene "S" standardized ratios which can be converted to base pairs using a standard equation. Within-person correlation of repeated continuous measures will be assessed using generalized estimated equations to evaluate longitudinal changes. Hair cortisol will be measured in two 3-inch hair sections, one at Time 2 and one at Time 3.
• Infant: Birthweight, Apgar scores, neonatal ICU admission, and length of stay will be evaluated.

Covariates: Demographic, clinical, and socioeconomic covariates of interest will be collected from participants. These variables will be ascertained via patient report (e.g. highest level of education, annual income, number of jobs worked, self-identified race and ethnicity, engagement in mental health treatments) and via chart abstraction (e.g. medical comorbidities, substance use, body mass index, neonatal birthweight and Apgar score). If covariates are unbalanced between groups, results from Aim 1 will be used to prioritize which covariates should be accounted for in adjusted models.

Analyses. The primary analysis will compare the difference in mean within person EPDS score change between Time 2 and Time 3 in the study across the control and intervention groups. If the mean EPDS differences are normally distributed, analysis will be conducted using independent t-tests. If not, analysis will be conducted using nonparametric tests such as Wilcoxon rank-sum tests. If any demographic or clinical characteristic emerges unbalanced between the randomization groups, an additional analysis will be conducted adjusting for this as a potential confounder as long as assumptions are met for linear regression techniques. Secondary outcomes will be compared using t-tests for normally distributed continuous variables, Wilcoxon rank sum tests for nonparametric continuous variables, and chi square tests for categorical variables. Changes in secondary questionnaire outcomes over time (such as differences in STAI, EPDS) will be compared using the differences in an individual's scores. Mean maternal leukocyte telomere and hair cortisol differences between Time 2 and Time 3 will be compared across groups as continuous variables.

Missing data: Based on our prior longitudinal studies with similar populations, we expect completion rates of >70%, with little item-level missing data. We will conduct intent-to-treat analyses for Aim 2 based on all available data. Missing data will not be imputed for primary outcomes. Given the exploratory nature of the biologic outcomes and small sample size, missing values will also not be imputed. Multiple imputation will be used for missing covariates deemed necessary to adjust for in statistical modeling. Patterns of missingness will be examined as needed.

Power: Using EPDS as the continuous outcome variable compared between independent control and experimental subjects with a 1:1 allocation ratio, standard deviation of 4.89 in the control group, and true difference in mean EPDS scores of 4.00, we will need 25 subjects per group using an independent t-test to reject the null hypothesis that the population means of the groups are equal with power of 0.80 and Type 1 error probability of 0.05. To account for 20% loss to follow up, stratification, and block randomization in blocks of 2 and 4, we will enroll 44 per group for a total N=88 participants.