Can we update the Summary Hospital Mortality Index (SHMI) to make a useful measure of the quality of hospital care? An observational study.

Abstract

To advance methods for the estimation of hospital performance based upon mortality ratios. Observational study estimating trust performance in a year derived according to comparative standards from a 3-year period, accounting for patient-level case-mix and overdispersion (unexplained variability). 23 363 630 admissions to the English National Health Service (NHS) by NHS Trust. Number of SDs (QUality and Outcomes Research Unit Measure, QUORUM banding) and comparative odds of hospital mortality difference from mean performance by trust compared for 2010/2011, 2008/2009 and 2009/2010, accounting for patient-level case-mix. The model was highly predictive of mortality (C statistic=0.93), and well calibrated by risk stratum. There was substantial overdispersion. No trusts were more than 3 SDs above the mean, and only one trust was more than 2 SDs above the mean for 2010/2011. QUORUM is highly predictive of patient mortality in hospital or up to 30 days after admission. However, like the Summary Hospital Mortality Indicator (SHMI), QUORUM is subjected to considerable remaining legitimate but unexplained variation. It is unlikely that measures like QUORUM and SHMI will be useful beyond identifying a very small number of trusts as potential outliers.