Background Both the US Centres for Disease Control and Prevention’s WISQARS system and the Healthcare Cost and Utilisation Program (HCUP) provide sample-based national estimates of injury survivors treated in hospital emergency departments and released (ED cases) or admitted. The WISQARS sample is much smaller but also more quickly available and more fully cause-coded than the HCUP data. For admissions, HCUP weights are quite accurate, computed from a near-census. We compared estimated medical and work loss costs from the two systems.

Methods We applied an updated version of essentially the same published injury cost model to both datasets. The HCUP model combined HCUP data on hospital payments per case with estimates by International Classification of Diseases, 9^th Edition-Clinical Modification diagnosis of associated professional fees and lifetime medical costs post-discharge. Work loss costs were estimated from length of stay if admitted and data by diagnosis group on mean days of work loss and the probability of permanent work-related disability. We used 2010 HCUP incidence data to collapse the HCUP costs into WISQARS diagnosis codes.

Results Both data sets estimated 27 million ED cases annually in 2012/13, with annual costs estimated at $165 billion using HCUP data and $167 billion using WISQARS data. For admitted injuries, WISQARS estimates of 3 million cases costing $290 billion exceeded HCUP “gold standard” estimates of 2.7 million (11.5% fewer) and $245 billion (19% less). WISQARS reported admitted injury trends by cause that HCUP suggested were spurious.

Conclusions WISQARS provides adequate injury incidence and cost estimates for surveillance. Its sample is too thin to support trend analyses by cause or for admissions and its admissions estimates should be replaced or reweighted with HCUP data as they become available. For admitted injuries, the sample of 66 hospitals underlying the WISQARS estimates warrants expansion.