Recent trends in Victorian risk‐adjusted in‐hospital mortality: 1999–2009

Abstract

The background of the study is a comparison of risk-adjusted mortality across hospitals from different jurisdictions is now common worldwide. To examine temporal trends in risk-adjusted mortality in Victoria over the last decade. Retrospective cohort study of 6.89 million adult (>14 years) patient episodes from 23 major Victorian public hospitals between 1999 and 2009. The primary outcome was in-hospital death. Three measures were calculated: the crude mortality rate, risk-adjusted mortality rate and standardised mortality ratio (SMR). The Hospital Outcome Prediction Equation (HOPE) was applied to generate estimates of predicted mortality that were used to compute the SMR and risk-adjusted mortality rates. The HOPE model includes 26 exogenous risk factors for which providers have no influence. The model was calibrated using the 2004-2005 data. Temporal mortality trends from 1999-2009 were evaluated using negative binomial regression for crude mortality and SMR estimates and random-intercept hierarchical logistic regression for risk-adjusted mortality. The study population included 84 423 in-hospital deaths (1.2%). Crude mortality risk declined from 1.5% in 2000 to 1.1% in 2005-2009 (incidence rate ratio (IRR): 0.96; 95% confidence interval (CI): 0.95-0.97; P < 0.001). There were 1.39 million episodes in the HOPE calibration cohort. Between 1999 and 2009, the SMR decreased from 1.4 to 0.9 (IRR = 0.91; 95% CI: 0.90-0.97; P < 0.001) and adjusted mortality risk declined from 2.1% to 0.9% (odds ratio = 0.94, 95% CI: 0.94-0.94, P < 0.001). Declining mortality trends were evident in the tertiary, metropolitan and regional peer groups (P < 0.001). Analysis of in-hospital risk-adjusted mortality trends using the HOPE model indicates significant improvement in patient outcomes in the State of Victoria over the past decade.