The process of healthy aging can best be described by plotting the trajectory of health-related variables over time. Unfortunately, graphs including data only from survivors may be misleading because they may confuse patterns of mortality with patterns of change in health. Two approaches for creating graphs that account for death in such situations are 1) to incorporate a category or value for death into the longitudinal health variable and 2) to measure time in years before death or some other event. The first approach has been applied to self-rated health (excellent to poor) and the 36-Item Short-Form Health Survey (SF-36). It allows for flexible and interpretable analyses and may be appropriate for other variables as well. The second approach also accounts fully for death, but the questions it can address are limited. Both approaches are useful and should be used at a minimum for supporting analyses in longitudinal studies in which persons die during observation.
1 December 2003 in Journal of Clinical Epidemiology
Little is known about longitudinal associations among measures of depression, mental and physical health, and quality of life (QOL). We followed 982 clinically depressed persons to determine which measures changed and whether the change was synchronous with change in depressive symptoms.
19 May 2010 in The Journals of Gerontology, Series A: Biological Sciences and Medical Sciences
Older adults are often advised to walk more, but randomized trials have not conclusively established the benefits of walking in this age group. Typical analyses based on observational data may have biased results. Here, we propose a “limited-bias,” more interpretable estimate of the health benefits to sedentary healthy older adults of walking more, using longitudinal data from the Cardiovascular Health Study. The number of city blocks walked per week, collected annually, was classified as sedentary (<7 blocks per week), somewhat active, or active (≥28). Analysis was restricted to persons sedentary and healthy in the first 2 years. In Year 3, some became more active (the treatment groups). Self-rated health at Year 5 (follow-up) was regressed on walking at Year 3, with additional covariates from Year 2, when all were sedentary. At follow-up, 83.5% of those active at baseline had excellent, very good, or good self-rated health, as compared with 63.9% of the sedentary, an apparent benefit of 19.6 percentage points. After covariate adjustment, the limited-bias estimate of the benefit was 11.2 percentage points (95% confidence interval 3.7–18.6). Ten different outcome measures showed a benefit, ranging from 5 to 11 percentage points. Estimates from other study designs were smaller, less interpretable, and potentially more biased. In longitudinal studies where walking and health are ascertained at every wave, limited-bias estimates can provide better estimates of the benefits of walking. A surprisingly small increase in walking was associated with meaningful health benefits.