Objective of this study is to identify background infant and maternal characteristics that predict child maltreatment (CM) incidence reporting among Neonatal Intensive Care Unit (NICU) graduates by health care providers versus community sentinels with the goal of identifying ways to improve CM risk surveillance. Demographic, medical data including diagnoses and caregiving needs at discharge for infants treated in a NICU during 2005 to 2008 were obtained from the neonatology databases. CM outcome data was obtained from child welfare databases. Latent class analysis procedures were used to identify observable infant and maternal characteristics that define unobserved groups (latent classes) that predict NICU graduates CM incidence reporting among health care providers versus community sentinels. Medical surveillance (reports made by health care providers) accounted for only 37% of the CM reports made to child welfare. Infant health was more predictive of medical surveillance than maternal characteristics suggesting that health providers may assess risk differently than community sentinels. Based on a simple, two latent class model, the latent class with high infant health indicator membership probabilities was a better predictor of health care provider related reports than the class with lower membership probabilities (odds ratio = 2.72; 95% confidence interval [1.76, 4.20]). Health care providers may be keyed more to an infant's medical frailty than to caregiver (maternal) contextual characteristics and thus may miss an opportunity to identify and intervene to prevent CM among children with medical problems. Findings raise the question of whether increased attention to contextual factors can aid or increase early identification of infants at risk of child maltreatment in NICU settings.
Keywords: NICU graduates; child maltreatment; medical surveillance.