A pressure ulcer audit and feedback project across multi-hospital settings in the Netherlands.

Abstract

OBJECTIVE: To examine whether participating in a pressure ulcer prevalence survey and receiving feedback results in an improvement in quality of care. DESIGN: Cross-sectional studies from 1998 to 2002 were compared over time. SETTING: Sixty-two acute care hospitals in the Netherlands. STUDY PARTICIPANTS: Patients hospitalized at the moment of the surveys. INTERVENTIONS: Each hospital was given hospital-specific performance data and national aggregate data, and peer comparisons to improve the quality of care. MAIN OUTCOME MEASURES: The case-mix-adjusted prevalence of pressure ulcers of grade >or=2, the percentage of high-risk patients receiving adequate prevention, and the total number of enabling conditions present were compared between successive surveys using multi-level analysis, in order to estimate a linear trend model and trend differences for each hospital. RESULTS: The case-mix-adjusted prevalence of pressure ulcers decreased over the 5-year period, while the percentage of patients receiving adequate prevention and the total number of enabling conditions present increased. The total number of enabling conditions had a significant effect on the decrease in case-mix-adjusted prevalence: more enabling conditions led to a lower case-mix-adjusted prevalence (chi(2) = 125; degrees of freedom = 1; P < 0.00). The percentage of patients receiving adequate prevention also had an effect on the change in case-mix-adjusted prevalence, with a higher percentage leading to a lower case-mix-adjusted prevalence. This effect, however, was not significant. CONCLUSIONS: Monitoring prevalence and giving feedback results in an improvement in quality of care in terms of pressure ulcer prevention. It is very important to continue conducting surveys to avoid attention moving away from this topic, which may in turn lead to a deterioration in the quality of pressure ulcer care. Further research to find the most effective feedback approach is needed.