Prediction Model for Mortality May Inform Dementia Treatment in Older Adults

The development of a prognosis prediction model may help clinicians predict mortality risk among older adults with dementia.

A prognosis prediction model showed accuracy in predicting mortality in older adults with dementia, according to study findings published in the journal JAMA Internal Medicine.

In 2022, 6.5 million Americans aged 65 years and older are living with Alzheimer disease (AD) and dementia. The leading cause of morbidity and mortality worldwide is dementia. The best-known mortality predictor for this patient population was developed for use in nursing homes, however, approximately 70% of individuals with dementia live in the community.

To develop a tool for use in the community setting, researchers from the University of California, San Francisco sourced data for this analysis from the Health and Retirement Study (HRS), which was conducted between 1998 and 2016, and the National Health and Aging Trends Study (NHATS), which was conducted between 2011 and 2019. Candidate predictors for mortality dementia were obtained from the literature. The predictive model was formulated with data from HRS (n=4267) and validated with data from NHATS (n=2404).

The HRS and NHATS cohorts comprised 69.4% and 59.3% women, aged mean 82.2 years and not reported, 75.1% and 45.4% were White, 12.1% and 32.8% were Black, and 37.7% and 25.2% lived alone, respectively.

The mortality risk estimates may help guide discussions regarding treatment decisions and advance care planning.

The primary outcome was time to all-cause death.

The final significant predictors for mortality included age category, gender, body mass index (BMI) category, smoking status, activities of daily living (ADL) dependency count, Instrumental ADL difficulty count, vigorous physical activity, diabetes, heart disease, non-skin cancer, and lung disease (all P £.007).

In the derivation cohort, the prediction model had an area under the curve (AUC) for predicting mortality of 0.76 (95%, CI; 0.75-0.76) overall, 0.73 (95% CI, 0.70-0.75) at year 1, 0.75 (95% CI, 0.73-0.77) at year 5, and 0.84 (95% CI, 0.82-0.85) at year 10. In the validation cohort, the model had an AUC for predicting mortality of 0.73 (95% CI, 0.70-0.76) at year 1 and 0.74 (95% CI, 0.71-0.76) at year 5.

The researchers noted that the mortality prediction model showed good discrimination and good calibration when predicting risk from 1 to 10 years in older adults with dementia. “The mortality risk estimates may help guide discussions regarding treatment decisions and advance care planning,” the researchers concluded.

The major limitations of this study were that data about disease etiology and severity were not available.


Deardorff WJ, Barnes DE, Jeon SY, et al. Development and external validation of a mortality prediction model for community-dwelling older adults with dementia. JAMA Intern Med. Published online September 26, 2022. doi:10.1001/jamainternmed.2022.4326