Author Affiliations: Division of Geriatrics, University of California, San Francisco (Drs Smith and Lee) (firstname.lastname@example.org); and Department of Medicine, Scripps Mercy Hospital, San Diego, California (Dr Yourman).
In Reply: In response to Dr Stern, we agree that estimating life expectancy from life tables has utility.1 Furthermore, life expectancy may be more interpretable to clinicians and patients than mortality risk.2 Using multiple sources (both prognostic models and life tables) may improve the accuracy of the prognostic estimate. As described by Walter and Covinsky,3 life expectancy estimates generated by US census data are 1 method of obtaining this information. However, clinicians must first decide if their patient falls in the upper 25th percentile, middle 50th percentile, or lowest 25th percentile of health for their age group. Prognostic indices may assist with this determination. Keeler et al4 refined the life table method to incorporate mobility-disability and activities of daily living. Unlike the prognostic models in our study, however, this method has not been subject to validation. The prognostic model and life table strategies should be tested and compared in diverse clinical settings to evaluate their accuracy, acceptability, and impact on clinician behavior and patient outcomes.
Smith AK, Yourman LC, Lee SJ. Prognostic Models for Older Adults—Reply. JAMA. 2012;307(18):1911-1912. doi:10.1001/jama.2012.3615