Purpose: Over two-thirds of U.S. infant homicide is attributable to inflicted neurotrauma (IN). Diagnosis of IN is challenging and misdiagnosis leads to increased morbidity and mortality. The purpose of this study is to estimate the cost-effectiveness of cranial computed tomography (CT) to screen for occult IN.

Methods: We used decision analysis to compare immediate CT or discharge home from an emergency department in an asymptomatic 5-week-old with either 1) a history of an apparent life-threatening event or 2) unexplained scalp bruising. Research supports a prevalence of IN in such infants of 2.4% and 24% respectively. We examined these clinical scenarios from three perspectives: a case-finding model to estimate costs per case found, a Markov model from a payer perspective and a Markov model from a societal perspective. Health states modeled were no injury, undiagnosed IN, mild IN, severe IN and death. Infants were entered into the model with no IN or mild IN according to prevalence. Probabilities of IN-related disability or death were linked to IN detection. We used available literature to estimate model parameters associated with medical, legal and welfare costs and probabilities. Informed assumptions were made regarding quality of life in IN survivors. The model terminated at death or at 52 weeks of age to limit prediction of long-term costs and utilities of child abuse, which are poorly understood.

Results: Case finding was inexpensive for both high and low prevalence scenarios ($835 and $10,471 per case respectively). From a payer perspective, screening under conditions of high prevalence saves money ($1008) and improves outcomes (0.009 QALYs); for conditions of low prevalence, screening is expensive ($136,676/QALY) but sensitive to variation in CT cost and IN utilities. From a societal perspective, however, screening is uniformly expensive ($325,636/QALY for high prevalence and $560,419/QALY for low prevalence) due to high costs of child protection following case detection.

Conclusion: From a payer perspective, CT screening for inflicted neurotrauma can be cost saving. From a societal perspective, screening appears expensive due to high costs of caring for survivors. This time-limited model cannot account for future benefits of case detection. Our models emphasize the challenge of selecting appropriate time horizons in pediatric cost-effectiveness analysis and the importance of improved understanding of costs and outcomes of child abuse.