35PREDICTING MORTALITY IN THE ELDERLY BEYOND HSMRS

    loading  Checking for direct PDF access through Ovid

Abstract

Introduction: Hospital standardised mortality ratios (HSMRs) are used as a proxy measure of quality of care. The reliance on mortality ratios has been criticised for several reasons: the risk adjustment process only adjusts for variables that can be measured. There is no ‘coding’ for frailty, despite the fact that this contributes to poorer outcomes. HSMRs vary by 60% across UK hospitals, indicating that it is unlikely this difference can be solely attributed to variations in quality of care. Variations in the ‘coding depth’ may be due to different coding practices, this is open to potential abuse by upgrading risk assessments. A focus on HSMRs may result in clinicians practicing more aggressively, resulting in excess morbidity and more expensive care.

Innovation: We hypothesised that a complementary method could be to express mortality as the number of years added or subtracted to patients' age to give the observed mortality in that population. This method would allow comparison of mortality ratios between different teams.

Evaluation: In a geratology unit over a 4-year period 2,467 patients of whom 1,528 were female were discharged with 428 deaths (241: female, 187: male). The patients' average age was 84.9 years (IQR: 81–90).

The number of deaths that would be expected in a population with this age if they were to experience the average UK mortality was calculated (ONS interim life tables 2007–09) (162 females and 99 males). By adding 5 and 6 years to the age of every female and male to produce an age shifted theoretical population the predicted no deaths are 444 deaths (254 females and 190 males). This matches closely the observed number of deaths in the study population.

Conclusions: These results suggest that this may be used to complement HSMRs and give a richer picture to the expected mortality. We suggest that it needs to be validated by further research.

Related Topics

    loading  Loading Related Articles