Failure to Re-identify and Manage Known Dementia Associated With Hospital Readmission and LOS: A Retrospective Cohort Study

The extent to which known dementia is re-identified and managed during hospitalization and the impact of this on patient outcomes warrant investigation. Among 7,919 patients with an index admission coded with dementia between 1 July 2006 to 30 June 2015 in a regional health district in Australia, we examined total length of stay (LOS) and Difference from Expected LOS (using mixed linear regression and quantile regression), and 30-day readmission (using a generalized linear mixed model). In models adjusted for age, complexity, specialty, and discharge destination, uncoded stays had significantly shorter total LOS (β = -2.9, 95% CI: -3.5, -2.3) and Difference from Expected LOS (β = -5.2, 95% CI: -5.8, -4.6). Quantile regression results confirmed these findings but were smaller in effect size. People without dementia coded were 1.2 times more likely to be re-readmitted (95% CI: 1.1, 1.3) than when coded. Consistent re-identification and active management of known dementia is needed to improve long-term patient and health system outcomes.