Our study population was drawn from the Childhood Autism Perinatal Study, a large case–control study examining pre-, peri-, and neonatal risk factors for ASDs among the membership of Kaiser Permanente of Northern California (KPNC; Croen et al. 2005
). KPNC is a group model integrated health plan that provides care for over 3.2 million northern California residents. The KPNC membership represents approximately 30% of the insured population in the region and is demographically similar to the residents of the counties served by KPNC, except that the very poor and very wealthy are underrepresented (Krieger 1992
). Cases and controls were identified from the cohort of infants born at a KPNC facility between January 1995 and June 1999 who remained KPNC members for at least 2 years following birth (n
Cases were defined as children with at least one diagnosis of an ASD, including autism (International Classification of Diseases 1999
; ICD-9-CM code 299.0) and Asperger’s Disorder or Pervasive Developmental Disorder Not Otherwise Specified (ICD-9-CM code 299.8) recorded anytime between January 1995 and December 2002 in KPNC outpatient clinical databases (n
420). Children with fragile X syndrome (n
2), tuberous sclerosis (n
0), or neurofibromatosis (n
0) were excluded. One control per case was randomly selected from the cohort of KPNC births without an ASD diagnosis, frequency matched to cases on sex, birth year, and delivery hospital. Since we were interested in examining characteristics of the mother in relation to autism risk in the offspring, we randomly sampled one child for inclusion for each woman who contributed two children to the original study sample (13 case mothers, 5 control mothers). Only children whose mothers were KPNC members with pharmacy benefits from at least 1 month before conception through the end of the pregnancy were included (291 cases, 284 controls). Women with pharmacy benefits during this time period were very likely to have purchased all of their prescription medications at a KPNC pharmacy.
Information on maternal exposure to B2AR agonists was ascertained from the KPNC pharmacy database, which records all dispensed prescriptions at KP pharmacies, and abstracted from prenatal medical records using a standardized form. All inpatient and outpatient prescriptions were identified in the 30 days before conception through the end of the pregnancy with the study child. B2AR agonists included: Advair®, albuterol, Combivent®, Foradil®, Maxair®, metaproterenol, Ritodrine®, Serevent®, terbutaline, Tornalate®, and Xopenex® (Table ). Only two B2AR agonists have been used for tocolysis in the USA, ritodrine and terbutaline. Terbutaline is the generic for Brethine, Bricanyl, and Brethaire, all of which were searched in the pharmacy database along with Ritodrine but only prescriptions for terbutaline were found. Since generic terbutaline is inexpensive and comes in a variety of forms for administration, it was the B2AR agonist in use for tocolysis during the time period of this study. Inhalers, injections, and pills were the only routes of administration included. Topical, ophthalmic, and otic routes were excluded.
B2AR agonists and mimics included in analyses, listed by indication
Four time periods of exposure were defined: preconception (the 30 days prior to the last menstrual period (LMP)), first trimester (in the 90 days post-LMP), second trimester (90–180 days post-LMP), and third trimester (180 days post-LMP to date of delivery). To account for possible inaccuracies in LMP date, the preconception period was included as a proxy for very early pregnancy exposure.
The date the prescription was dispensed and the number of days supplied were used to determine exposure status during each time period. Exposure during a given time period was assumed if a prescription was dispensed during the time period or the days supplied overlapped some portion of the time period. Duration of exposure was derived by summing the total number of days during pregnancy for which a B2AR agonist was prescribed. Maternal conditions that were indications for B2AR agonists during pregnancy were identified from inpatient and outpatient databases and abstracted from prenatal records. Clinical indicators included asthma, preterm labor, and conditions that predispose to preterm labor (bicornuate uterus, cervical incompetence, multiple fibroids, hypertension, diabetes, anemia, bacterial vaginosis, pyelonephritis, asymptomatic bacteriuria, chorioamnionitis, urinary tract infection, and sexually transmitted diseases). Information on several maternal and infant characteristics was obtained from health plan and vital statistics databases.
Characteristics of cases and controls were compared using contingency tables for categorical variables and comparisons of means for continuous variables. Unconditional logistic regression analysis was conducted to estimate relative risks of ASD associated with maternal use of B2AR agonists during pregnancy. Separate models were run for all B2AR agonists combined, terbutaline alone, and albuterol alone, the two most frequently used B2AR agonists. For each exposure definition, we examined time period of exposure as well as duration of exposure. We defined exposure duration categories based on the distribution of days of exposure among the control group. For terbutaline, most use was for 1 or 2 days, reflecting acute exposure, or for 10 or more days, reflecting maintenance tocolysis. We also examined exposure to terbutaline for ≥2 weeks, the duration found to be associated with ASD concordance among dizygotic twins in a previous study. For albuterol, duration was dichotomized as >21 days vs. ≤21 days, the cut-point representing the median length of exposure among control mothers. Women with no prescriptions or days supplied for any B2AR agonists for the entire period from 30 days before conception through the end of pregnancy were considered unexposed for all analyses. Maternal and infant characteristics associated with maternal exposure to B2AR agonists or case–control status were treated as possible confounders and included as covariates in multivariable models as were the matching variables. All covariates were entered simultaneously into regression models and all statistical tests were two-tailed. Study procedures were approved by the KPNC Northern California Institutional Review Board and the California State Committee for the Protection of Human Subjects.