Interpreting quality improvement data with time-series analyses.

Abstract

In quality improvement efforts, the data are frequently a series of measurements taken over time. A collection of statistical methods, commonly referred to as time-series analysis, provides a simple and understandable method for interpreting this longitudinal data. In this article, we present a time-series analysis of data on the quality of prenatal care at a mid-sized public hospital. We will demonstrate some simple tests that alert us to the potential value of using more sophisticated tests of association such as regression. Using regression, we show how to confirm a visual impression of an improvement. The analytical approach we present here is useful with many types of process or outcome data from health care quality improvement efforts.