Demographics and Psychosocial Characteristics (Table 1)
440 (68.5%) participants were postpartum, and 202 (31.5%) were pregnant (1st trimester: 37, 2nd: 114, 3rd: 51). The mean (SD) age of postpartum mothers’ infants was 7.07 (3.17) months. The largest number of participants lived in Staten Island (43.3%). 76% of all participants were White, 81.2% were 22-30 years old, and 64.5% had college/graduate degrees. Perceived stress (mean=45.19, SD=7.71), anxiety (mean=93.67, SD=12.35), and depression (mean=13.49, SD=4.36) all reached levels of clinical significance (40, 80, and 12, respectively). 35.3% had suicidal ideation “at least sometimes”, and 7.2% had it “quite often”. Prevalence rates of substance use, including tobacco (20.7%), cannabis (15.0%), and alcohol (38.0%), were high.
Infected Compared to Uninfected Women (Table 2)
Stress and Psychological Functioning: Infected, compared to uninfected, women reported greater levels of anxiety (100.10 vs. 93.40, p=.00003) and depression (17.05 vs. 13.25, p<.00001), but did not differ on perceived stress.
Suicidal Thoughts: The prevalence of suicidal thoughts among infected women, compared to uninfected, showed a 4-fold increase for “quite often” (17.9% vs. 6.5%, AOR=3.97, 95%CI 1.39-11.38, p=.01) and an overall 13-fold increase for “sometimes” (89.7% vs. 31.8%, AOR=13.20, 95%CI 4.46-39.13, p<.0001).
Substance Use: Infected women, compared to uninfected, had a 3-fold increase in alcohol use (72.5% vs. 35.7%, AOR=3.30, 95%CI 1.50-7.25, p=.003), an over 4-fold increase in tobacco use (70.0% vs. 17.4%, AOR=4.54, 95%CI 2.06-10.00, p=.0002), a 7-fold increase in cannabis (65.0% vs. 11.6%, AOR=7.01, 95%CI 3.02-16.29, p<.0001) and heroin use (30.0% vs. 3.0%, AOR=7.09, 95%CI 2.87-17.54, p=.001), and a 10-fold increase in cocaine use (30.0% vs. 2.8%, AOR=10.05, 95%CI 4.05-24.94, p<.0001).
Influence of COVID-19 infection, time (1st, 2nd, 3rd trimesters pregnancy, and postpartum), and the interaction between the two on the level of distress (Table 3)
GEE was used to evaluate the influence of COVID-19 infection, time (1st, 2nd, 3rd trimesters, and postpartum), and interaction between infection status and time on the level of distress at each time point, and an overall difference in trajectory. Results showed a significant time-effect (p<.001), infection-effect (p=.008), and trajectory difference (i.e., interaction) between infected and uninfected women (p<.001). Specifically, among infected women, the scores at 1st, 2nd, 3rd trimesters, and postpartum were 2.36, 2.27, 2.26, and 2.69, respectively, whereas among uninfected women they were 1.70, 1.88, 1.93, and 2.98. The patterns were very similar in trajectory of each subscale.
Longitudinal changes across pregnancy and postnatal period in distress by COVID-19 infection (Figure 1)
We first examined each distress measure as a function of the intercept plus the linear/quadratic effect of time without predictors and covariates (Panel A). We found that a curvilinear model was the best-fit to explain the trajectory of distress (Table 4). After choosing the best-fit model, Model 1 examined the trajectory of distress with infection status. Then, we evaluated the model with infection, SES, and the interaction between the two (Model 2) to determine whether the effect of infection was moderated by SES.
Model 1. With only infection status
There was a significant difference in distress (t=2.62, p=.009) between infected and uninfected women at the intercept (3rd trimester).
Panel B shows the patterns of change in distress over time by infection status. Among infected women, the level of distress increased slightly throughout the study period. Among uninfected women, the level started low but increased throughout pregnancy and exceeded the level of infected women after childbirth (t=-5.58, p<.001).
Model 2. With infection, SES and interaction of the two
Infection status (t-ratio=11.52, p<.001) and SES (t-ratio=4.81, p<.001) predicted a significantly different distress level at the intercept (3rd trimester). The interaction was also significant (t-ratio=-6.64, p<.001), indicating that SES moderated the effect of infection.
Panel C shows the trajectory of distress by infection status and SES. Both infection status and SES had significant effects in predicting the linear change (t-ratio=-6.50, p<.001, t-ratio=-10.14, p<.001, respectively) and curvilinear changes in distress (t-ratio=-3.35, p<.001, t-ratio=-10.11, p<.001, respectively). The interaction was significantly different for both linear (t-ratio=5.86, p<.001) and curvilinear changes (t-ratio=6.02, p<.001). Specifically, SES has a differential impact on the effect of infection on the trajectory of distress. Among uninfected women, those with low SES had a substantially greater level of distress approaching their 3rd trimester, relative to women with high SES, whereas women with high SES had a lower level of distress toward the end of the 3rd trimester, but it continued to increase postpartum.