J Am Med Dir Assoc. 2021 Dec 14:S1525-8610(21)01007-0. doi: 10.1016/j.jamda.2021.11.020. Online ahead of print.
OBJECTIVES: To determine which nursing home (NH) resident-level admission characteristics are associated with potentially preventable emergency department (PPED) transfers.
DESIGN: We conducted a population-level retrospective cohort study on NH resident data collected using the Resident Assessment Instrument-Minimum Data Set Version 2.0 and linked to the National Ambulatory Care Reporting System for ED transfers.
SETTING: We used all NH resident admission assessments from January 1, 2017, to December 31, 2018, in Ontario.
PARTICIPANTS: The cohort included the admission assessment of 56,433 NH residents.
METHODS: PPED transfers were defined based on the International Classification of Disease, Version 10 (Canadian) We used logistic regression with 10-fold cross-validation and computed average marginal effects to identify the association between resident characteristics at NH admission and PPED transfers within 92 days after admission.
RESULTS: Overall, 6.2% of residents had at least 1 PPED transfer within 92 days of NH admission. After adjustment, variables that had a prevalence of 10% or more that were associated with a 1% or more absolute increase in the risk of a PPED transfer included polypharmacy [of cohort (OC) 84.4%, risk difference (RD) 2.0%], congestive heart failure (OC 29.0%, RD 3.0%), and renal failure (OC 11.6%, RD 1.2%). Female sex (OC 63.2%, RD -1.3%), a do not hospitalize directive (OC 24.4%, RD -2.6%), change in mood (OC 66.9%, RD -1.2%), and Alzheimer's or dementia (OC 62.1%, RD -1.2%) were more than 10% prevalent and associated with a 1% or more absolute decrease in the risk of a PPED.
CONCLUSIONS AND IMPLICATIONS: Though many routinely collected resident characteristics were associated with a PPED transfer, the absence of sufficiently discriminating characteristics suggests that emergency department visits by NH residents are multifactorial and difficult to predict. Future studies should assess the clinical utility of risk factor identification to prevent transfers.