Study Population and Design
This retrospective cohort study assessed the outcome of ADHD in children born at all KPSC hospitals during January 1, 2011–December 31, 2014. The outcome of autism spectrum disorder has been evaluated in a similar cohort of Tdap-vaccinated pregnant women. KPSC is an integrated health-care delivery organization that provides health care in 15 hospitals and about 234 medical offices to approximately 4.6 million members who are broadly representative of the Southern California population. KPSC uses electronic medical records (EMRs) to integrate medical information, such as diagnoses and immunizations. All recommended immunizations are free to KPSC members regardless of copayment status. This study was approved by the KPSC Institutional Review Board, which waived the requirement for informed consent.
Eligibility was restricted to pregnant women without assisted conceptions (i.e., in vitro fertilization) who had a documented birth of a live singleton infant no later than 45 weeks' gestation. Children's medical records were linked to those of their biological mothers using unique identifiers. Mother-child pairs were included if women had had continuous KPSC membership at least since the beginning of the 27th week of pregnancy (allowing for a ≤31-day gap) and infants were members for at least 90 continuous days after turning 3 years of age. Eligibility criteria reflected exposure opportunity, as the minimum gestational age recommended for vaccination is 27 weeks' gestation. In addition, requiring membership after age 3 years reflects the minimum preschool age, when children would have the opportunity to be identified if behavioral manifestations consistent with ADHD were present. In addition, we only included children with at least 1 outpatient visit during the observation period, to assure that everyone in the cohort had an opportunity to be seen by the KPSC health system. Finally, pairs were excluded if children were diagnosed with chromosomal or congenital anomalies.
Maternal Tdap Vaccination
Maternal vaccination captured in the EMR was defined as receipt of Tdap vaccine anytime during pregnancy (yes, no) up to the day prior to the delivery date. The unvaccinated group was defined as women who were not vaccinated with Tdap during the observed pregnancy period.
ADHD was defined by 2 or more clinical diagnoses on different dates or any 1 diagnosis with 2 or more dispensed ADHD medications after the child turned 3 years of age, recorded in the EMR during January 1, 2014–December 31, 2018 (International Classification of Diseases, Ninth Revision (ICD-9), codes 314.0, 314.1, 314.2, 314.8, and 314.9; International Classification of Diseases, Tenth Revision (ICD-10), codes F90.0, F90.1, F90.2, F90.8, and F90.9; ADHD medications: amphetamine aspartate, amphetamine sulfate, dextroamphetamine aspartate, dextroamphetamine sulfate, and methylphenidate hydrochloride). Initial diagnoses documented in the EMR were made by experts in the field of childhood neurodevelopmental disorders (i.e., child/adolescent psychiatrists, psychologists, and developmental/behavioral pediatricians) who used stringent ADHD diagnostic criteria based on the Diagnostic and Statistical Manual of Mental Disorders, Fourth or Fifth Edition, requiring patients to be symptomatic for at least 6 months with symptoms affecting the quality of functioning in at least 2 settings (i.e., home and school).[14,24] ADHD cases were refuted if there was evidence of a subsequent diagnosis of autism spectrum disorder (ICD-9 codes 299.0, 299.8, and 299.9; ICD-10 codes F84.0, F84.5, F84.8, and F84.9). However, children diagnosed with autism spectrum disorder first and subsequently diagnosed with ADHD were considered cases. This ADHD diagnosis record case identification approach was previously validated in this same setting.[25,26]
Information on maternal and child characteristics was obtained from the EMR. Data on maternal characteristics at delivery included: maternal age (≤25, 26–30, 31–34, or ≥35 years), race/ethnicity (non-Hispanic White, non-Hispanic Black, Hispanic, non-Hispanic Asian/Pacific Islander, other/multiple/unknown), primary language (English, other), need for an interpreter (yes, no), educational attainment (high school or less, some college, bachelor's degree, master's degree or more, unknown), history of an ADHD diagnosis, and medical center of delivery. Additional pregnancy information included use of Medicaid insurance (yes, no), parity (0, 1, or ≥2), time of prenatal care initiation (≤3 months' gestation, >3 months' gestation, no care/missing), receipt of influenza vaccine during pregnancy (yes, no), and medical and obstetrical complications (pregestational hypertension, preeclampsia/eclampsia, pregestational or gestational diabetes, placenta previa/placental abruption). Child data included sex (male or female), birth year (2011–2014), birth season (winter, spring, summer, fall), gestational age at birth (<37 weeks, ≥37 weeks), and birth weight (g).
The distributions of maternal and child characteristics by prenatal Tdap vaccination status were determined, and tests were performed for differences in characteristics (either χ 2 test or Student's t test). In addition, the distributions of follow-up characteristics, diagnosis prevalence by birth year, age at ADHD diagnosis, and the incidence rate of ADHD diagnosis by Tdap vaccination status were calculated. Children were followed from the start of enrollment (after 3 years of age) to ADHD diagnosis, the end of membership in the health plan, or the end of the study follow-up period (December 31, 2018), whichever came first.
Cox proportional hazards regression was used to estimate unadjusted and adjusted hazard ratios and their 95% confidence intervals to evaluate the magnitude of association between maternal Tdap vaccination and ADHD diagnosis in children. To evaluate potential variations in the magnitude of associations, we also conducted stratified analyses by child's birth year and by assessing results among women giving birth to their first child (nulliparous women). We used propensity score analyses with inverse probability of treatment weighting (IPTW) to adjust for potential confounding. First, we used a logistic regression model including variables associated with vaccination a priori[27,28] or variables from bivariate results (P < 0.05) that could reasonably affect Tdap receipt and were not collinear with each other (i.e., influenza vaccination and birth season) to estimate the probability of Tdap vaccination: child's sex, birth year, gestational age at birth (<37 weeks or ≥37 weeks), maternal age, race/ethnicity, education, history of ADHD diagnosis, medical center of delivery, Medicaid insurance, parity, start of prenatal care, an emergency department visit during pregnancy, and influenza vaccination during pregnancy. Second, the weight for each mother was calculated as the inverse of her predicted probability of Tdap exposure, and it was normalized by dividing by the mean weight of each exposure group. Standardized difference scores were used to assess whether balance of covariates between the comparison groups was achieved. Unlike P values, for which the magnitude is highly related to the sample size, the standardized difference is a unified approach to quantifying the magnitude of differences between groups regardless of sample size; an absolute value less than 0.10 is considered a negligible difference.[29,30] To consider the potential importance of adjusting for factors associated with the outcome, we performed a sensitivity analysis including maternal self-reported smoking, self-reported alcohol drinking, and prepregnancy body mass index (weight (kg)/height (m)2) in the propensity score model.
Am J Epidemiol. 2020;189(10):1163-1172. © 2020 Oxford University Press