(1) We examined the hypothesis that in utero/peripartum antiretroviral (IPA) exposure may affect the likelihood of developmental disorders—i.e., attention deficit and hyperactivity disorder (ADHD), autism spectrum disorder (ASD), and functional impairment (FI). (2) Children and their primary caregivers were enrolled and followed for 12 months. The sample included 250 children perinatally HIV-infected (CPHIV), 250 children HIV-exposed and uninfected (CHEU) of women living with HIV, and 250 children HIV unexposed and uninfected (CHUU) at 6–18 years of age. CHEU’s IPA exposure-type was established via medical records and categorized as no IPA, single-dose nevirapine with/without zidovudine (SdNVP ± AZT), SdNVP + AZT + Lamivudine (3TC), or combination ART (cART). Developmental disorders were assessed at months 0, 6, and 12 per caregiver response to standardized questions from the third edition of Behavioral Assessment System for Children. Multivariable repeated measures linear regression models estimated standardized mean differences (SMDs) with 95% confidence intervals (95% CI) according to the IPA exposure type relative to CHUU with adjustment for the dyad’s sociodemographic and psychosocial factors. (3) Relative to the CHUU, outcomes were similar for CPHIV/CHEU with cART, SdNVP ± AZT, and no anti-retroviral drug exposure in the peripartum period. For CHEU relative to CHUU, SdNVP + AZT + 3TC exposure was associated with lower resiliency (SMD = −0.26, 95% CI: −0.49, −0.51), and elevated scores on ADHD (SMD = 0.41, 95% CI: 0.12, 0.70), ASD (SMD = 0.40, 95% CI: 0.19, 0.61), and EBD (SMD = 0.32, 95% CI: 0.08, 0.56) probability and functional impairment (SMD = 0.39, 95% CI: 0.18, 0.61) index scores. With the exception of ADHD, the adverse association between SdNVP + AZT + 3TC and outcomes were replicated for CPHIV vs. CHUU. (4) The results provided reassuring evidence that cART exposure in the peripartum period is unlikely to be adversely associated with developmental disorder probability scores in late childhood and adolescent years. However, the peripartum SdNVP + AZT + 3TC exposure associated elevation in developmental disorder probability and functional limitation at 6–18 years of life is a concern.
Ugandan children aged 6–18 years and their primary adult caregivers were enrolled on a first-come, first-served basis as part of two prospective cohort studies implemented between 16 March 2017 and 30 June 2021. By design, an equal number n = 250 of CPHIV, CHEU, and CHUU were recruited. CPHIV were enrolled from among patients cared for in a community health center in urban Kampala, Uganda. CHEU were identified from the early infant diagnosis (EID) registers at the same clinic. CHUU were recruited by referrals using the social networks of the child-caregiver dyads already enrolled. Current HIV status for both the CHEU and CHUU was ascertained using an HIV-rapid diagnosis test at enrolment. Each child-caregiver pair was followed for 12 months or until loss to follow-up with study related assessments at enrolment, 6, and 12 months. Eligible children were between 6 and 18 years old at enrolment with a consenting adult (i.e., age ≥ 18 years) caregiver who had served in the caregiving role for at least six months before study enrolment. To ensure medical record documentation of child ART exposure status during pregnancy, their mother’s HIV status, mother’s participation in the PMTCT program, and the child’s HIV status at birth, only children born in a hospital or medical setting were enrolled. In addition to maternal ART exposure type, the child’s birth weight and prematurity status were abstracted from the delivery records. Children not born in a clinical setting and child-caregiver dyads without antenatal, delivery, or care records were ineligible because peripartum ART exposure and the HIV status of their pregnancy could not be ascertained. Protocol for their respective cohort studies were reviewed by professionals from the research ethics boards of Michigan State University (IRB Protocol numbers: 16-828 and 205), Makerere University College of Health Sciences, School of Medicine (Protocol REC REF numbers: 2017-017 and 2018-099), and the Uganda National Council for Science and Technology (Protocol #s: SS4378 and HS 2466). All caregivers gave written informed consent, and children provided assent or consent for study participation. Five separate caregiver reported indices were scored following instructions provided in the manual of the Behavioral Assessment System for Children, Third Edition (BASC-3) [25]. Each was analyzed as a continuous outcome variable (z-scores) that quantified the children’s propensity to develop behavioral or emotional functioning disorders, be functionally impaired, or exhibit a resilient profile relative to age-matched Ugandan CHUU. The BASC-3 items were forward and back translated for the study setting with adaptations made to preserve meaning in the local context as previously described [26]. Cronbach’s alpha was calculated as measure of reliability for each tool, and the values for respective outcomes ranged from a low of 0.81 to a high of 0.94, demonstrating that the items included within the outcome measures had good to excellent consistency in measuring underlying constructs. Snacks were given to all participants prior to each interview to reduce the distracting impact of hunger/fatigue on their responses. Autism Spectrum Disorder (ASD) Probability Index: Defined per response to 13 (if child < 11 years) or 18 (if child ≥ 11 years) questionnaire items where caregivers rated the severity of withdrawal, atypical behaviors, impaired communication, leadership aversion, and poor social skills in their children. The ASD probability index score is positively correlated with behavioral ratings in children with clinically diagnosed autism and indicates a greater propensity of exhibiting unusual behaviors, including problems developing/maintaining peer relationships. Attention Deficit or Hyperactivity Disorder (ADHD) Probability Index: Defined per summation of 11(if child < 11 years) or 9 (if child ≥ 11 years) questionnaire items that positively correlate with behavioral ratings for children clinically diagnosed with ADHD. Higher scores are suggestive of difficulty in academic settings or completion of tasks requiring attention due to inability to sustain focus, plan tasks, make decisions, or moderate activity level. Emotional Behavioral Disorder (EBD) Probability Index: Defined per summation of 30 (if child < 11 years) or 20 (if child ≥ 11 years) items that positively correlate with behavioral scores obtained in children identified within educational settings as having emotional/behavioral disturbance or disability. Higher scores suggest greater likelihood of exhibiting disruptive, atypical, and antisocial behaviors or high levels of negative emotions like anger, sadness, and pessimism—behavioral traits that contribute to strain in peer and adult relationships. Functional Impairment (FI): This index was measured using a total of 44 items that capture the level of difficulty children have with engaging in successful or appropriate behaviors during interactions with others, in the execution of age-appropriate tasks, in the regulation of mood, and success with school-related tasks. Elevated scores suggest difficulty responding appropriately in everyday social interactions in a variety of settings. Resiliency Indices: This index was defined using nine items gauging the overall capacity to engage in adaptive behaviors like coping with change, recovering from setbacks, and problem solving in the face of adversity. High scores suggest greater likelihood of exhibiting a resilient trajectory in the short-term and correlate with positive mental health. As previously described [26], IPA exposure was established from medical records: namely, the mother’s ART treatment card and antenatal or early-infant diagnosis registers for CPHIV and CHEU. CHEU and CPHIV were exposed to one of four IPA types: (1) no IPA, (2) intrapartum prophylactic single-dose nevirapine (SdNVP) ± zidovudine (AZT), (3) intrapartum prophylactic SdNVP + AZT plus Lamivudine (3TC), i.e., SdNVP + AZT + 3TC, and (4) combination ART (cART, including at least two antiretroviral drug classes), and these children were compared to CHUU (reference). Current ART Regimen: The current ART regimen of mothers in the CPHIV and CHEU subsamples was abstracted from their medical records at enrolment into the study. Their current ART regimen was in four categories: namely, (1) NRTI (Abacavir or TDF inclusive), (2) NNRTI (EFV and NVP), (3) protease inhibitors (Ritonavir, Atazanivar, Lopinavir/Kaletra), and (4) no current ART/unknown. Sociodemographic Factors: Biological sex was defined as male vs. female; chronologic age and formal education completed (in years) and developmental stage—i.e., preadolescent vs. adolescent (<11 vs. ≥11 years)—were defined for caregivers and dependent children. Caregiving Context: The severity of acute caregiver psychosocial stress over the past month was assessed using the perceived stress scale which ranges from 0 (no stress) to 40 (highest stress) [27,28]. Lifetime stress—i.e., sum of occurrence for 13 adverse experiences over the life course—was assessed using the stressful life events questionnaire [29]. Their subjective social standing (lowest = 1 to highest = 10) was assessed using the MacArthur scale of subjective social standing [30]. Their functioning within the caregiving role (lowest = 1 to highest = 120) was measured using an adapted version of the Barkin index of maternal functioning scale [31,32]. Caregiver depressive and anxiety symptoms were measured using 15 and 10 items, respectively, from the Hopkins Symptom Checklist-25 [33,34]. As in our prior studies, social support was measured as the summed score of eight questions in which caregivers expressed agreement or not with statements about their ability to access wanted emotional, monetary, and physical support resources [35]. This secondary analysis included four IPA exposure groups that were compared with CHUU as the primary determinant; hence we estimated the minimum detectable effect size with 80% power using a two-sided test of significance at alpha = 0.05. The minimum detectable effect size varied according to the available number of children within IPA groups in the cohort. Specifically, compared to the 251 CHUU, this study has 80% power with 95% confidence to detect effect sizes of (a) ≥0.3 among the 250 children without IPA exposure, (b) ≥0.4 among the 96 children exposed to SdNVP ± AZT, and (c) ≥0.5 in all IPA categories as each includes ≥75 children per group. Analysis of variance for continuous measures and chi-square tests for categorical variables were used to summarize differences according to the type of early-life ART exposure and, for CPHIV, according to the developmental stage. Multivariable repeated measures linear regression models were implemented using the generalized estimating equations approach to quantify IPA-related standardized mean differences (SMD) in age standardized outcome measures over 12 months’ follow-up via SAS PROC GENMOD. In all models, confounders such as caregivers’ age, sex, socioeconomic status, depression, and functioning in caregiving role stress were adjusted for based on subject matter knowledge. The random effect of the caregiver was included in all models to account for nesting of children within households. Time was entered as a class variable to model potentially non-linear patterns, and interactions between time and IPA-type were examined to assess potential variation in respective outcomes over 12 months’ observation according to IPA type. In the absence of IPA by time interaction, time averaged associations between repeated assessments of disorders over 12 months were presented. Since outcomes were standardized by age, standardized mean differences (SMD) were estimated—a measure comparable to Cohen’s effect size and thus permitting the determination of clinical importance. Per prior precedent based on studies of quality of life, |SMD| of ≥0.33 were deemed clinically important for child growth which has wide-ranging lifelong implications [36]. Hence, |SMD| < 0.33, 0.33 ≤ |SMD| < 0.50 and |SMD| ≥ 0.50 were considered to be of small to modest, moderate, and large clinical importance, respectively. To explore the perinatal developmental stage as a potential moderator of IPA associated variations in respective outcomes over 12 months, IPA by developmental stage interaction was introduced in multivariable models. Potential heterogeneity was indicated when p-value for the interaction was <0.1. In that case, the results for IPA-related SMD in developmental disorders were presented separately for preadolescent vs. adolescent children. In the absence of heterogeneity, the results were presented for the overall sample. All analyses were performed with SAS version 9.4 (SAS Institute, Inc., Cary, NC, USA). All hypothesis tests were two-sided at alpha = 0.05.