Simple statistical measures for analyzing categorical data

The proportions of patients undergoing a given procedure or experiencing a specific health outcome often are important in studies of healthcare quality. Run charts are a nonstatistical method for evaluating changes in these proportions or rates over time. Simple epidemiological methods for comparing rates between two groups allow statistical comparison of rates between two time periods. The chi-square test measures the statistical significance of differences in rates between two groups. Practical aspects of the use of these statistics and interpretation of their results are discussed and illustrated in this article, with realistic examples originally prepared by the author for a drug usage evaluation study. Readers with appropriate software and hardware may request a simple spreadsheet from the author to calculate many of the analyses described in the article.