The effects of audits and fines on upcoding in neonatology

Upcoding is a common type of fraud in healthcare. However, how audit policies need to be designed to cope with upcoding is not well understood. We provide causal evidence on the effect of random audits with different probabilities and financial consequences. Using a controlled laboratory experiment, we mimic the decision situation of obstetrics staff members to report birth weights of neonatal infants. Subjects' payments in the experiment depend on their reported birth weights and follow the German non-linear diagnosis-related group remuneration for neonatal care. Our results show that audits with low detection probabilities only reduce fraudulent birth-weight reporting, when they are coupled with fines for fraudulent reporting. For audit policies with fines, increasing the probability of an audit only effectively enhances honest reporting, when switching from detectable to less gainful undetectable upcoding is not feasible. Implications for audit policies are discussed.