Hospital mortality under surgical care

The 'hospital standardised mortality ratio' (HSMR) has been used in England since 1999 to measure NHS hospital performance. Large variations in reported HSMR between English hospitals have recently led to heavy criticism of their use as a surrogate measure of hospital performance. This paper aims to review the mortality data for a consultant general surgeon contributed by his NHS trust over a 3-year period as part of the trust's HSMR calculation and evaluate the accuracy of coding the diagnoses and covariates for case mix adjustment.

The Dr Foster Intelligence database was interrogated to extract the NHS trust's HSMR benchmark data on inpatient mortality for the surgeon from 1 April 2006 to 31 March 2009 and compared to the hospital notes.

30 patients were identified of whom 12 had no evidence of being managed by the surgeon. This represents a potential 40% inaccuracy rate in designating consultant responsibility. The remaining 18 patients could be separated into 'operative' (11 patients) and 'non-operative' (7 patients) groups. Only 27% in the operative group and 43% of the non-operative mortality group respectively had a Charlson co-morbidity index recorded despite 94% of the cases having significant co-morbidities

Highlighting crude and inaccurate clinician-specific mortality data when only 1-5% of deaths under surgical care may be associated with avoidable adverse events seems potentially irresponsible.