An Educational Review of the Statistical Issues in Analysing UtilityData for Cost-Utility Analysis


The aim of cost-utility analysis is to support decision making inhealthcare by providing a standardised mechanism for comparing resourceuse and health outcomes across programmes of work. The focus of thispaper is the denominator of the cost-utility analysis, specificallythe methodology and statistical challenges associated with calculatingQALYs from patient-level data collected as part of a trial. We providea brief description of the most common questionnaire used to calculatepatient level utility scores, the EQ-5D, followed by a discussionof other ways to calculate patient level utility scores alongsidea trial including other generic measures of health-related qualityof life and condition- and population-specific questionnaires. Detailis provided on how to calculate the mean QALYs per patient, includingdiscounting, adjusting for baseline differences in utility scoresand a discussion of the implications of different methods for handlingmissing data. The methods are demonstrated using data from a trial.As the methods chosen can systematically change the results of theanalysis, it is important that standardised methods such as patient-levelanalysis are adhered to as best as possible. Regardless, researchersneed to ensure that they are sufficiently transparent about the methodsthey use so as to provide the best possible information to aid inhealthcare decision making.

Gianluca Baio
Gianluca Baio
Professor of Statistics and Health Economics