METHODS TO ACCOUNT FOR FOLLOW-UP TIME DIFFERENCES WHEN CALCULATING QALYS FROM RANDOMISED CONTROLLED TRIALS
Author(s)
Alshreef A1, Spackman E2
1University of Sheffield, Sheffield, UK, 2University of York, Heslington, York, UK
OBJECTIVES: Methods for calculating quality-adjusted life years (QALY) are well developed and employed in economic evaluations alongside randomised controlled trials (RTCs). However, patient follow-up does not always occur at the same time and the effect of these time differences in collecting EQ-5D data on QALY calculations has not been tested. Since time is a key part in calculating the QALY, it might be important to take these time differences into account. The objective of this study was to assess different methods for estimating the differences in follow-up time when calculating QALYs from EQ-5D data collected alongside RCTs. METHODS: Alternative approaches were considered for estimating QALYs include: i) using the proposed time points for EQ-5D responses as stated in the trial protocol; ii) using the average actual follow-up time; and iii) using the individual patient-level time. The last two approaches were estimated over the actual trial follow-up time; and truncated at the protocol time horizon as alternative methods. Five methods were identified. These methods were illustrated using a hypothetical example. They were further illustrated by application with empirical analyses of RCT data. A seemingly unrelated regression model was fit for estimating QALYs and costs for comparing these methods. RESULTS: Different methods generated different estimates for QALYs using the trial data. However, the magnitude of differences is relatively small ranging from -0.9% to 11.8% (n=752) when different methods were compared with the conventional method. CONCLUSIONS: The effect of follow-up time differences in calculating QALYs using the RCT data is relatively small. However, differences could matter when the estimated ICERs are very close to the cost-effectiveness threshold. The most appropriate method is using individual patient-level timing data truncated at the proposed trial follow-up time horizon, as it uses the most data available without biasing the results due to total follow-up time differences.
Conference/Value in Health Info
2015-11, ISPOR Europe 2015, Milan, Italy
Value in Health, Vol. 18, No. 7 (November 2015)
Code
PRM166
Topic
Methodological & Statistical Research
Topic Subcategory
PRO & Related Methods
Disease
Multiple Diseases