OBJECTIVES: To estimate the incidence rate and the risk of institutionalization in convalescent hospitals among community-dwelling patients with mild cognitive impairment (MCI) or Alzheimer’s disease (AD).

METHODS: A retrospective cohort study was conducted using the Health Insurance Review and Assessment database covering the entire Korean population from JUL 01/2019 to NOV 30/2022. The study focused on patients diagnosed with MCI or AD between JAN 01/2020 and JUN 01/2022 with the index date defined as the first diagnosis date. Institutionalization was defined as a continuous enrollment in a convalescent hospital for over six months. We excluded the patients already residing in a convalescent hospital six months before the index date. The primary outcome was the incidence of institutionalization in convalescent hospitals for MCI or AD patients. The incidence rate per 1000 person-years was calculated, and the risk of institutionalization for AD patients compared to MCI was estimated using hazard ratios (HR) with 95% confidence intervals from a Cox proportional hazards model. To control for confounders including age, sex, type of insurance, Charlson comorbidity index, and infectious diseases, inverse propensity weighting was employed.

RESULTS: The study included 40.8% (n=422,934) patients with MCI and 59.2% (n=613,920) patients with AD. The mean age differed between the two groups (71.7±9.4 years for MCI vs. 80.5±7.8 years for AD). The institutionalization rate per 1000 person-years was different between patients with MCI (0.65[0.59-0.71]) and AD (13.8[13.6-14.0]). The risk of institutionalization was significantly higher in patients with AD compared to MCI (unadjusted HR, 16.6[15.1-18.2]; adjusted HR, 9.2[8.8-9.5]).

CONCLUSIONS: This study provides real-world evidence demonstrating a higher incidence rate and increased risk of institutionalization in convalescent hospitals for patients with AD compared to those with MCI using large population-based data. Inverse propensity weighting was employed to adjust for confounding factors commonly present in observational studies, resulting in more robust and reliable findings.