To Register      SMDM Homepage

Monday, 18 October 2004 - 4:15 PM

This presentation is part of: Oral Concurrent Session B - Methodological Advances

QUALITY-ADJUSTED YEARS OF LIFE GAINED FROM IMPLANTABLE DEFIBRILLATOR: COMPARISON OF A NEW METHOD WITH TRADITIONAL ANALYSES IN CLINICAL TRIALS

Katia Noyes, PhD, MPH1, Andrew W. Dick, PhD1, Hongwei Zhao, PhD2, Ethan Corona, BA1, W. Jackson Hall, PhD2, Jack Zwanziger, PhD3, and Arthur J. Moss, MD4. (1) University of Rochester School of Medicine, Department of Community and Preventive Medicine, Rochester, NY, (2) University of Rochester School of Medicine, Biostatistics and Computational Biology, Rochester, NY, (3) University of Illinois at Chicago School of Public Health, Health Policy and Administration, Chicago, IL, (4) University of Rochester School of Medicine, Cardiology, Rochester, NY

Purpose: Traditionally results of clinical trials are analyzed using either change in health-related quality of life (HRQOL) from baseline that includes survivors only, not accounting for data censoring, and does not consider intermittent outcomes, or the Kaplan-Meier survival approach that accounts for the weakness of the previous method but does not recognize differences in the HRQOL of survivors. Here we examine a novel approach that incorporates the strengths of the traditional methods, use it to examine HRQOL during the additional years of life saved, and compare our results with the results from traditional methods.

Methods: We used the original data from the MADIT II study that examined mortality in patients with implantable cardiac defibrillators (ICD) (n=1089). To estimate changes in HRQOL, we used a general health preference instrument, Health Utility Index 3 (HUI3). First, we developed a multivariate fixed-effects model to impute the missing HUI3 data assuming missing at random mechanism. Then, using the imputed data, we calculated the total HRQOL gain over 3 years as the area under the normalized HRQOL profile over time, for each treatment arm. We also estimated the difference in HRQOL gains between treatments, as the area between the two treatment profiles. We bootstrapped all standard errors. These results were compared to the traditional estimates of changes in HRQOL from baseline and to results of survival analysis.

Results: Over the 3 years, subjects in the ICD arm experienced a better survival (&Delta Life Years Saved (ICD-control) = .122 years, p = .03) and had a similar change in the HUI3 score from baseline (ICD: -.056 (SE .032), control: -.039 (SE .035), &Delta (ICD-control) = -.017, p = .73) conditional on survival. When adjusted for survival and censoring, use of ICD resulted in a loss of -0.257 (SE 0.050) QALYs, while subjects in the control arm lost on average -0.311 (SE 0.011) QALYs, &Delta HRQOL (ICD-control) = .054, p = .48.

Conclusions: Despite prolonging survival of severely ill cardiac patients, ICD implantation was associated with a similar change in HRQOL over 3 years compared with patients undergoing medical treatment. Using quality-adjusted survival analysis in the trials with high differential mortality permits evaluation of the intervention effect on all domains of patient health.


See more of Oral Concurrent Session B - Methodological Advances
See more of The 26th Annual Meeting of the Society for Medical Decision Making (October 17-20, 2004)