Menu
Menu
Menu
Menu
Background: Among HIV-infected women, perinatal depression compromises clinical, maternal, and child health outcomes. Antiretroviral therapy (ART) is associated with lower depression symptom severity but the uniformity of effect through pregnancy and postpartum periods is unknown.
Uganda is an important setting in which to investigate depression over perinatal periods given a total fertility rate of 6.2 births per woman41 and HIV prevalence of 7.2%,20,42 higher among women (8.2%) than among men (6.1%). In addition, 70% of treatment-eligible Ugandans access ART.20,28 ART coverage among women with pregnancy is expected to increase with Uganda’s adoption of “Option B+” to prevent perinatal HIV transmission.43,44 Mbarara District is a primarily rural setting located approximately 265 km southwest of Kampala. Adult HIV prevalence in the district is estimated at 10%45 and is higher among women compared with men. The Mbarara University HIV clinic is located within the Mbarara Regional Referral Hospital and offers comprehensive HIV care services, including ART, at no cost to patients.46 Study participants were enrolled in the Uganda AIDS Rural Treatment Outcomes (UARTO) cohort study, initiated in July 2005 with the primary objective of determining predictors of virologic failure and antiretroviral resistance. Participants were recruited from HIV-infected, treatment-naive adults initiating ART at the Mbarara University HIV clinic. Patients who were at least 18 years old and living within 60 km of the clinic were eligible to enroll. At the time of this analysis, 447 women were enrolled in UARTO. Loss-to-follow-up (participants for whom we were unable to confirm vital status after ≥180 days without cohort follow-up) was 2% at 1 year and 5% at 2 years after enrollment. Participants were seen at baseline (ie, at ART initiation) and quarterly for bloodwork and to complete standardized interviewer-administered questionnaires detailing sociodemographics, mental and physical health, sexual risk behaviour, and partner dynamics. Incident pregnancies were assessed by self-report. Interviews were administered in the dominant local language (Runyankole). This analysis includes data from female participants (18–49 years) enrolled from June 2005 followed through December 2012. Depression symptom severity was measured using a modified version of the Hopkins Symptom Checklist (HSCL)-15 for depression.47 Based on previous studies using HSCL in Uganda, we included a 16th item, “Feeling like I don’t care about my health.”48 Each symptom was scored on a 4-item Likert scale ranging from 1 (not at all) to 4 (extremely) and the total depression severity score was calculated as the mean of the 16 items, with higher scores indicating greater depression symptom severity. We also assessed a dichotomous measure of “probable depression” defined as an HSCL score >1.75, a commonly used threshold for a positive screen of depression.47–50 The depression subscale of HSCL has been used to assess depression in general population samples and among people living with HIV in sub-Saharan African countries.47–49,51–54 The HSCL has been further shown to have good reliability55 and construct validity56 among people living with HIV in Uganda specifically. Although the HSCL was not designed to specifically screen for perinatal depression, it and other generic depression scales are commonly used to measure depression among antenatal women.54,57–59 The most reliable, valid, and frequently used measure of perinatal depression in sub-Saharan Africa is the Edinburgh Postnatal Depression Scale54; however, the reliability and validity of the Edinburgh Postnatal Depression Scale outside of perinatal periods is unknown. The primary predictor variable is a 3-level variable indicating pregnancy status, which we classified as being pregnant, up to 1 year postpartum or neither pregnant nor post-partum (“non–pregnancy-related”). Women who reported sterilization (tubal ligation or hysterectomy) at baseline were excluded from the analysis. If the procedure was reported during follow-up, women were censored upon reported date of the procedure. Periods of pregnancy were defined based on self-report at baseline and over the follow-up period and included both first and subsequent pregnancies. Pregnancy start was defined as the visit date when pregnancy was first reported and pregnancy end was the subsequent date at which women reported no longer being pregnant. For the few women (n = 8) who reported a live birth outcome and had a computed period of pregnancy of 11 months, live birth date was used to back-calculate a start date to account for an estimated 9-month gestation. The postpartum period was defined as the period from the end of pregnancy until 12 months after any pregnancy outcome.60,61 All other follow-up times were assessed as non–pregnancy-related. At study enrollment, women reported whether they were or had been pregnant in the previous 12 months but did not report dates of pre-enrollment pregnancy or postpartum status. Thus, no women were classified as postpartum at study entry. We examined the association between depression and pregnancy status, adjusting for baseline and time-updated covariates considered potential confounders. Baseline variables included sociodemographic characteristics (including age, education, employment, household income, and marital status), reproductive history (including parity), and clinical history (including time since HIV diagnosis, AIDS defining illnesses, CD4 cell count at enrollment, and body mass index). Time-updated variables were measured quarterly and included age, time on HIV treatment, CD4 cell count, HIV viral load <400 copies per milliliter (ie, viral suppression), the Medical Outcomes Study HIV Health Survey Physical Health Summary score (scored on a 0–100 scale, where a higher score indicates better health),62,63 and sexual activity in the previous 3 months. Descriptive statistics were used to characterize baseline distributions of study variables. Baseline differences between women with and without incident pregnancy were compared using Wilcoxon rank sum test for continuous variables and Pearson χ2 test for categorical variables. Although our regression models are based on time-varying pregnancy-related status, comparison of fixed categories of pregnancy are provided to inform understanding of differences between women who exclusively contributed to non-pregnant periods and those who contributed to all 3 periods of follow-up. We calculated the mean (SD) HSCL score at baseline and then compared mean HSCL scores across pregnant, postpartum, and non–pregnancy-related periods using analysis of variance. We plotted mean HSCL scores (with residual standard error bars) across the 3 periods by time on ART. We fit a multivariable linear regression model to the data, using generalized estimating equations to estimate model parameters. We used an autoregressive working correlation structure. Pregnancy and the postpartum period were investigated as independent predictors of depression symptom severity. After testing normality assumptions and collinearity, variables with a significant association with depression in bivariate analyses (at P < 0.20) were considered for the full model to obtain the relative contribution of each covariate. Model selection was achieved by minimizing the Akaike Information Criterion while maintaining P-values for covariates below 0.20.64 All statistical tests were 2-sided and were considered statistically significant at α = 0.05. Using a binary variable for “probable depression,” we also fit a multivariable GEE logistic regression model. The same model selection procedures and covariates were used. Data were analyzed with SAS version 9.3 (SAS Institute Inc., Cary, NC).65 All participants provided voluntary, written informed consent at study enrollment. Ethical approval for all study procedures was obtained from the Institutional Review Committee, Mbarara University of Science and Technology; the Partners Human Research Committee, Massachusetts General Hospital; the Committee on Human Research, University of California at San Francisco; and the Research Ethics Board of Simon Fraser University. Consistent with national guidelines, we received clearance for the study from the Uganda National Council for Science and Technology and from the Research Secretariat in the Office of the President.
