Use of Logistic Regression for Developing Risk-adjusted Quality Indicators in Nursing

Abstract

Summary Objectives: A low rate of newly developed pressure ulcers is considered as an important quality indicator in nursing. However, the result of a hospital depends not only on the quality of care but on the risk profile of its patients as well. Therefore, based on multiple logistic regression models we describe a method for calculating risk-adjusted quality indicators in nursing. Method: Based on data of 1,009,989 patients from 1747 hospitals in 2009, we developed two multiple logistic regression models to identify and to weigh a possible joint influence of several risk factors on newly developed pressure ulcers. In a further step, we calculated risk-adjusted rates. Results: Factors remaining in the regression models were “micro-movements on admission”, “diabetes mellitus”, “age” and “days on intensive care unit”. Based on the corresponding regression coefficients and the logistic function, the expected rate of newly developed pressure ulcers was calculated for every hospital. Fi nally, expected rates and observed rates both were used to calculate risk-adjusted rates. Conclusion: The simultaneous consideration of relevant risk factors by means of risk- adjusted quality indicators ensures a fair comparison of hospitals.