Latent Class Analysis of Prenatal Substance Exposure and Child Behavioral Outcomes
Sarah E Maylott,Elisabeth Conradt,Monica McGrath,Emily A Knapp,Xiuhong Li,Rashelle Musci,Judy Aschner,Lyndsay A Avalos,Lisa A Croen,Sean Deoni,Karen Derefinko,Amy Elliott,Julie A Hofheimer,Leslie D Leve,Juliette C Madan,Maxwell Mansolf,Liza B Murrison,Jenae M Neiderhiser,Sally Ozonoff,Jonathan Posner,Amy Salisbury,Sheela Sathyanarayana,Julie B Schweitzer,Carl Seashore,Meagan E Stabler,Leslie W Young,Steven J Ondersma,Barry Lester,program collaborators for Environmental influences on Child Health Outcomes (ECHO),P B Smith,K L Newby,L P Jacobson,C B Parker,R Gershon,D Cella,F P Perera,J B Herbstman,M R Karagas,J Ganiban,A Ferrara,L A Croen,N Paneth,J M Kerver,D M Ruden,S Deoni,V A D'Sa,C Blair,J M Neiderhiser,K Z LeWinn,N R Bush,D Dabelea,M O'Shea,R Fry,L D Leve,A Mason,J B Stanford,A Giardino,C A Porucznik,C Karr,S Sathyanarayana,J Gern,A J Elliott,I Hertz-Picciotto,J B Schweitzer,D Bennett
DOI: https://doi.org/10.1016/j.jpeds.2023.113468
Abstract:Objectives: To predict behavioral disruptions in middle childhood, we identified latent classes of prenatal substance use. Study design: As part of the Environmental influences on Child Health Outcomes Program, we harmonized prenatal substance use data and child behavior outcomes from 2195 women and their 6- to 11-year-old children across 10 cohorts in the US and used latent class-adjusted regression models to predict parent-rated child behavior. Results: Three latent classes fit the data: low use (90.5%; n = 1986), primarily using no substances; licit use (6.6%; n = 145), mainly using nicotine with a moderate likelihood of using alcohol and marijuana; and illicit use (2.9%; n = 64), predominantly using illicit substances along with a moderate likelihood of using licit substances. Children exposed to primarily licit substances in utero had greater levels of externalizing behavior than children exposed to low or no substances (P = .001, d = .64). Children exposed to illicit substances in utero showed small but significant elevations in internalizing behavior than children exposed to low or no substances (P < .001, d = .16). Conclusions: The differences in prenatal polysubstance use may increase risk for specific childhood problem behaviors; however, child outcomes appeared comparably adverse for both licit and illicit polysubstance exposure. We highlight the need for similar multicohort, large-scale studies to examine childhood outcomes based on prenatal substance use profiles.