Clusters of Markers Identify High and Low Prevalence of Adolescent Pregnancy in the US

Study Objective: To identify markers that, when clustered, are associated with the highest and lowest pregnancy prevalence among US adolescent females. Design: Secondary database analysis of the National Survey of Family Growth. Participants: Adolescent females 15-19 years old. Setting: United States. Methods: Bivariate analyses were performed of the 2002 and 2006-08 cycles to identify markers associated with pregnancy in females 15-19 years old. Recursive partitioning analysis (RPA) was done to identify markers which, when combined, identify adolescent females with the highest and lowest pregnancy prevalence. Results: Pregnancy prevalence in 2,528 adolescent females was 14%. In RPA, females who ever HIV tested had a 48% pregnancy prevalence vs 4% in those never tested. The highest pregnancy prevalence (78%) was associated with the combination of ever HIV tested, never lived away from parents before 18 years old, age >17 years old, and age of sexual debut ≤14 years old. Adolescent females with a combination of ever HIV tested, never lived away from parents before 18 years old, age >17 years old, age of sexual debut at 15-16 years old, and the adolescent's father not having a high-school diploma had a pregnancy prevalence of 75%. Conclusions: Two clusters of markers yield adolescent females with a very high adolescent pregnancy prevalence, at 75%-78%. Pregnancy prevention efforts might be especially effective when targeting combinations of HIV-testing history, never living away from home before 18 years old, adolescent age, age of sexual debut, and paternal educational attainment.