Commentary on "Disability Trajectories at the End of Life: A Countdown Model" The Problems With Time-to-Death as a Predictor of Disability.

Abstract

ThE study of disability trajectories at the end of life is both interesting and important. Everyone dies, and most adults live well into advanced ages in which they experience some level of physical decline before death. This period of decline is relevant to us as individuals, obviously, but also to society, which ultimately bears much of the financial burden of care via Medicare and Medicaid expenditures. Thus, it is important to understand the disability process in late life. In particular, concepts like “terminal drop” and “terminal decline” are of interest because, if we could recognize who is likely to experience which type of pattern, we might be able to develop interventions to improve quality of death, and we might be able to better predict end-of-life expenditures. The recent paper by Wolf et al. is intriguing for these reasons. In brief, the authors use time-to-death (TTD) as a predictor of disability trajectories. The key goal is to determine the number and shape of disability trajectories that immediately precede death, ultimately estimating the prevalence of those who experience terminal drop, terminal decline, and other patterns. Although I think the methodology is interesting and certainly meticulous, there are two key problems with studying disability trajectories as a function of TTD that should make one cautious when employing such an approach. First, treating TTD as a predictor of disability patterns that precede it confuses the direction of causality. Second, the approach suggested by the authors cannot yield any more useful information than a model that does not confuse the direction of causality can.