To the Editor.
—We are concerned with many methodological weaknesses in the study by Dr Ware and colleagues,1 which cast grave doubts on the conclusions stated in the article, especially since the results run counter to most other published literature on this subject.The nonrandom sample analyzed in the study was derived based on high attrition rates, participation refusal, death, and self-selection and, in total, amounted to less than 30% of the intended sample. Looking at small subpopulations within this nonrepresentative sample does not improve external validity. Further, the complex sampling design was never accounted for in the calculation of standard errors used to determine statistical significance.We question the use of zero in measuring change in Medical Outcomes Study (MOS) 36-Item Short-Form Health Survey (SF-36) scores for members who died during the course of the 4-year follow-up period for the linear comparisons. Based on this methodology, a self-reported change in health