[Hospital standardised mortality ratios based on data from administrative registries. A pilot project].

INTRODUCTION

Hospital standardised mortality ratios (HSMR) are widely used in quality improvement campaigns. No data exist on whether HSMR can be computed based on Danish administrative registries. We therefore used data from Danish registries to compute HSMRs.

MATERIALS AND METHODS

By linking hospital discharge registries with the Danish Civil Registration System, we identified 77 primary discharge diagnoses that accounted for 80% of all deaths within 30 days of admission. We calculated overall death rates stratified by the 77 primary discharge diagnoses, age, gender, and type of admission and used these to compute the expected number of deaths. HSMR for each hospital was calculated as the ratio of observed to expected number of deaths.

RESULTS

Pneumonia, non-specified was the diagnosis that accounted for most deaths within 30 days after admission. The crude mortality rate varied from 5.7% to 6.3%. HSMR varied little--from 95 and 98 in Hospitals B and D to 102 and 103 in Hospitals A and C, respectively.

CONCLUSION

We found that it was possible to use data from Danish administrative registries to compute HSMR and that HSMR varied little between hospitals with comparable case-mixes.