Study Data: The Honolulu Heart Program (HHP) is a prospective study of heart disease and stroke involving a cohort of Japanese-American men born between 1900 and 1919. Clinical and demographic information were collected during three midlife examinations (EX) in 1965 (EX1), 1968 through 1970 (EX2), and 1971 through 1974 (EX3). As an extension of the HHP, the Honolulu Asia Aging Study (HAAS) was started in 1991 to study dementia prevalence, incidence and risk factors. At the first HAAS exam (examination 4), 3,734 members of the HHP cohort (80 percent of survivors) participated. Subsequent follow-up examinations for dementia were conducted in 1994 (EX5), 1996-1997 (EX6) and 1999-2000 (EX7) using a multistep procedure. The 100-point Cognitive Abilities Screening Instrument (CASI) was used to screen the entire sample. The CASI is a cross-culturally validated test of global cognitive function designed for use in comparative studies of dementia in the United States and Japan. In EX 4, CASI score and age were used to identify a subgroup of participants for dementia evaluation. In follow-up examinations, subjects with either CASI scores falling below education-adjusted cut-points (77 for participants with low education and 79 for high education) or with absolute drops of at least 9 points were included in a subgroup for specific dementia examination. Subjects who met the criteria for dementia underwent neuro-imaging and blood tests for diagnosis of dementia subtype. A consensus diagnosis was reached by the study neurologist and at least two other physicians with expertise in geriatrics and dementia. We used the death data recorded until June 20, 2002 .? ? Among the 3,734 participants in EX 4, 226 people were diagnosed with dementia, and they were called prevalent cases. There were 135, 112 and 52 incident dementia cases diagnosed at EX 5, 6, and 7, respectively.? ? Study 1: Estimate age-specific dementia rate and the impact of dementia on survival.? ? Summary: Early work discussed the age-specific incidence of dementia. Previous approaches either treated deaths simply as censored or did not include any covariates, however. We proposed an illness-death model for survival data with interval-censoring and left-truncation. This approach can include risk factors related to dementia onset and death, and allow semi-Markov models where survival after dementia could depend on dementia onset age. This model can be used to estimate the cumulative risk of developing dementia in the presence of competing risks of death. We applied this method to dementia data to estimate the age-specific and cumulative risks of incident dementia and to examine the effect of major risk factors on dementia onset and death.? ? Based on the analysis, we found that having dementia and early onset of dementia both significantly increase mortality rate. Having dementia increases the mortality hazard by 7.9-fold. ? ? Study 2: Joint modeling for cognitive trajectory and risk of dementia in the presence of death? ? Summary: We used a Bayesian change point model to fit the trajectory of cognitive function for demented people as well as that for normal aged people. In real life, aging people are subject to two competing events, e.g., dementia and death without dementia. The majority of people do not develop dementia. In this article, we used a mixture model to fit the survival data with competing risks consisting of dementia onset time after the change point of cognitive function for demented subjects and death time for non-demented subjects. Cognitive trajectories and the survival process were modeled jointly and parameters were estimated using the Markov chain Monte Carlo method. Using data from the HAAS, we showed the trajectories of cognitive function and the effect of several major covariates.? ? Based on the analysis, we found that for healthy aging subjects, CASI scores decrease gradually from 90 at age 70 to about 60 at age 95. For demented subjects, the change points for accelerated decline are age 72 and 84 for subjects with lower and higher education levels, respectively. For higher educated subjects, acceleration of cognitive decline occurs later but with a faster rate than for lower educated subjects. This finding is consistent with the theory of cognitive reserve. At age 95, all demented subjects have similar cognitive function regardless of education level.
| Yu, Binbing (2010) A Bayesian MCMC approach to survival analysis with doubly-censored data. Comput Stat Data Anal 54:1921-1929 |
