An assessment of failure to rescue derived from routine NHS data as a nursing sensitive patient safety indicator (report to Policy Research Programme)

Objectives: This study aims to assess the potential for deriving 2 mortality based failure to rescue indicators and a proxy measure, based on exceptionally long length of stay, from English hospital administrative data by exploring change in coding practice over time and measuring associations between failure to rescue and factors which would suggest indicators derived from these data are valid.Design: Cross sectional observational study of routinely collected administrative data.Setting: 146 general acute hospital trusts in England.Participants: Discharge data from 66,100,672 surgical admissions (1997 to 2009).Results: Median percentage of surgical admissions with at least one secondary diagnosis recorded increased from 26% in 1997/8 to 40% in 2008/9. The failure to rescue rate for a hospital appears to be relatively stable over time: inter-year correlations between 2007/8 and 2008/9 were r=0.92 to r=0.94. No failure to rescue indicator was significantly correlated with average number of secondary diagnoses coded per hospital. Regression analyses showed that failure to rescue was significantly associated (p<0.05) with several hospital characteristics previously associated with quality including staffing levels. Higher medical staffing (doctors + nurses) per bed and more doctors relative to the number of nurses were associated with lower failure to rescue. Conclusion: Coding practice has improved, and failure to rescue can be derived from English administrative data. The suggestion that it is particularly sensitive to nursing is not clearly supported. Although the patient population is more homogenous than for other mortality measures, risk adjustment is still required