HCBE Faculty Articles

Variance in U.S. Hospital Charges in Neonatal Care on the Modal DRG=603: Data-Driven Attributions to Diagnoses, Treatments, Severity of Patient and Characteristics of Hospital


Anne Fiedler0000-0003-4421-3414

Document Type


Publication Title

Journal of Academy of Business and Economics



Publication Date



Purpose Hospital charges in healthcare services in the U.S. has always been mysterious to the general public including many stakeholders such as the healthcare regulators and insurers. Even though the federal and state governments and insurance firms collect a lot of data, the aggregated databases withhold the identity of specific hospitals due to arcane regulations that prevent disclosure of providers and patients in the databases. The best one can do is to conduct statistical analyses that reveal the sources of variance in hospital charges and suggest areas for improvement for future reforms in hospital care in specific areas of patient care. Method This research uses the AHRQ’‘s (Agency for Healthcare Research and Quality) HCUP’‘s (Hospital Cot and Utilization Project) NIS (National Inpatient Sample) database. The web site for the NIS databases is https://www.ahrq.gov/data/hcup/index.html. We use the 2016 data on neonatal care for one particular DRG code 603 which is the most frequent DRG code for neonatal hospitalizations. Hospital charges are highly contingent on the nature and severity of the disease, number of diagnoses, number of treatments, length of stay and a whole series of diagnostic and treatment decisions made by well-qualified doctors or teams of doctors who make clinical decisions that are highly context- and patient- specific. Hence, it is very important to focus on one DRG code to understand the variance in hospital charges in terms of these factors. We use regression analyses on the 83,743 (DRG=603) cases in 2016 sampled in the NIS 2016 database to decompose the variance in hospital charges in terms of factors that contribute to the charges. Results Our results show that the primary determinants of hospital charges, based on patient hospitalization episodes, are length of stay, number of diagnoses and number of treatments which are patient-specific factors, and the size, type and location of the hospital. The interaction terms in our regression analyses reveal more interesting findings in that hospital factors (size, type and location) interact positively with patient factors to increase hospital charges. Future research should explore if different clinical protocols are the cause of such positive interaction effects. Conclusion These results, and the use of regression analysis with interaction terms, highlight both the success in understanding variance in hospital charges and those hospital characteristics that contribute to increases in hospital charges.







First Page


Last Page


This document is currently not available here.

Peer Reviewed

Find in your library