Using marginal structural models to identify the cardiovascular adverse effects of second generation antipsychotics in children and adolescents

Abstract

In pediatric patients, we examined the association between exposure to five second generation antipsychotics (SGAs) and incidentcardiovascular events (arrhythmic or ischemic/myocardial) over time using marginal structural models (MSM), while controllingfor salient comorbid conditions and co-prescribed psychotropic medications. A retrospective cohort, longitudinal/ observationalstudy design was used to evaluate Medicaid medical and pharmacy claims in 4140 children and adolescents prescribed SGAsfrom South Carolina USA’s Medicaid program covering outpatient and inpatient medical services and medication prescriptionsbetween January, 1996 and December, 2005. Exposure to multiple SGAs (Risk Ratio [RR]=2.37; 95% CI=1.17-4.83), coprescribedpsychostimulants (RR=1.37; CI=1.03-1.81), and comorbid hypertension (RR=2.23; CI=1.28-3.89) were associatedwith a significantly increased risk of arrhythmias compared to those not exposed, whereas exposure to co-prescribed serotoninnorepinephrine reuptake inhibitor/heterocyclic compounds was associated with a significantly decreased risk of arrhythmias(RR=0.59; CI=0.35-0.99). The risk of incident ischemic/myocardial events was significantly associated with the co-prescription ofmood stabilizers (RR=1.68; CI=1.06-2.68) or selective serotonin reuptake inhibitors (RR=1.91; CI=1.18-3.09), and the presenceof comorbid hypertension (RR=3.97; CI=1.96-8.07) and obesity (RR=2.21; CI=1.34-3.67). MSM analyses comparing multipletreatments while controlling for confounding variables in an observational, longitudinal data set provide important, differentialestimates of outcome, when randomized, controlled trials estimating low-incidence outcomes such as cardiovascular adverseevents in large pediatric patient populations are not feasible.