Abstract
OBJECTIVE: To develop survival prediction tables to inform physicians and patients about survival probabilities after the diagnosis of dementia and to determine whether survival after dementia diagnosis can be predicted with good accuracy. METHODS: We conducted a nationwide registry-linkage study including 829 health centers, i.e., all memory clinics and approximately 75% of primary care facilities, across Sweden. Data including cognitive function from 50,076 people with incident dementia diagnoses >/=65 years of age and registered with the Swedish Dementia Register in 2007 to 2015 were used, with a maximum follow-up of 9.7 years for survival until 2016. Sociodemographic factors, comorbidity burden, medication use, and dates of death were obtained from nationwide registries. Cox proportional hazards regression models were used to create tables depicting 3-year survival probabilities for different risk factor profiles. RESULTS: By August 2016, 20,828 (41.6%) patients in our cohort had died. Median survival time from diagnosis of dementia was 5.1 (interquartile range 2.9-8.0) years for women and 4.3 (interquartile range 2.3-7.0) years for men. Predictors of mortality were higher age, male sex, increased comorbidity burden and lower cognitive function at diagnosis, a diagnosis of non-Alzheimer dementia, living alone, and using more medications. The developed prediction tables yielded c indexes of 0.70 (95% confidence interval [CI] 0.69-0.71) to 0.72 (95% CI 0.71-0.73) and showed good calibration. CONCLUSIONS: Three-year survival after dementia diagnosis can be predicted with good accuracy. The survival prediction tables developed in this study may aid clinicians and patients in shared decision-making and advance care planning.
