Using Latent Variable Modeling for Discrete Time Survival Analysis: Examining the Links of Depression to Mortality |
| |
Authors: | Tenko Raykov Anna Zajacova Philip B Gorelick George A Marcoulides |
| |
Institution: | 1. Michigan State University;2. University of Western Ontario;3. Mercy Health Hauenstein Neurosciences at Saint Mary’s and Michigan State University;4. University of California, Santa Barbara |
| |
Abstract: | Using a latent variable modeling approach to discrete time survival analysis, the dynamics of the relationships of depression and body mass index to mortality are examined with data from the multiwave, nationally representative Health and Retirement Study. A set of medical and demographic variables are employed as time-invariant covariates along with lag-1 depression scores and body mass indexes as time-varying covariates for mortality within an up to 2-year follow-up interval. The results indicate marked links of immediately prior depression levels, as well as notable relations of the body mass indexes, to within-wave mortality in middle-aged and older adults. The approach highlights the benefits of using latent variable modeling for survival analysis, and its findings represent potentially important relationships of clinical and theoretical relevance. |
| |
Keywords: | body mass index depression discrete time survival analysis latent variable modeling mortality time-invariant covariate time-varying covariate |
|