OBJECTIVES: Dementia is one of the major causes of disability and decreased quality of life among elder people worldwide. The aim of this study was to develop and validate a prediction model for the risk of developing dementia in elderly people using a Markov approach that accounts for mortality as a competing risk.

METHODS: The Markov model was constructed with three mutually exclusive health states: alive without dementia, dementia and death. Within a 1-year cycle, patients move from one state to another or stay in the current state until death. The individual transition probabilities based on baseline characteristics were calculated using results from previous longitudinal cohort studies for elderly adults. We applied the Markov model to a dataset from the Alzheimer's Disease Neuroimaging Initiative (ADNI) cohort study for validation. The discrimination performance of the Markov model was evaluated using the area under the ROC curve (AUC) and its 95% confidence interval (CI).

RESULTS: The probability that someone of a given age will develop dementia during his or her remaining lifespan was calculated taking account for probability of death that can occur prior to the onset of dementia. The model showed that the probability of developing dementia tended to decrease after a certain point in age, as the probability of natural death grew as they get older. A total of 515 participants from ADNI study were included in the validation analysis. The area under the ROC curve for this model was 0.78 (95% CI: 0.73 to 0.82).

CONCLUSIONS: The model is useful for elderly people to predict their yearly pattern of probability for developing dementia based on their baseline condition, by taking the increasing probability of mortality as they get older, so that they can consider modifying life behavior to change the probability of future event.