Falls and long-term survival among older adults residing in care homes

Abstract

Objectives To assess the association between having suffered a fall in the month prior to interview and long-term overall survival in nursing-home residents. Methods Retrospective cohort study conducting an overall survival follow-up of 689 representative nursing-home residents from Madrid, Spain. Residents lived in three types of facilities: public, subsidized and private and its information was collected by interviewing the residents, caregivers and/or facility physicians. Residents contributed to follow-up time from their baseline interviews until death or being censored at the end of the 5-year follow-up period. The association between suffering a fall during the month prior to interview and long-term overall survival was analyzed using Cox proportional hazards models. To adjust for potential confounders we used progressive adjusted models. We then repeated the analyses with severity of the fall (no fall, non-severe, severe) as the main independent variable. Results After a 2408 person-year follow-up (median 4.5 years), 372 participants had died. In fullyadjusted models, residents who had suffered any kind of fall in the previous month showed virtually the same survival rates compared to non-fallers (hazard ratio (HR) = 1.03; 95% CI = 0.75-1.40). There was a weak graded relationship between increased fall severity and survival rates for the non-severe fall group (HR = 0.92; 95% CI = 0.58-1.45) and the severe fall group (HR = 1.36; 95% CI = 0.73-2.53) compared with residents who had not suffered any kind of fall. The hazard ratios for severe falls were higher in men, residents with less comorbidity, fewer medications, and those functionally independent.

Towards Clinical Prediction with Transparency: An Explainable AI Approach to Survival Modelling in Residential Aged Care

“…Cox proportional hazards regression has been used extensively to analyse survival in community-dwelling older people and those residing in aged care facilities [18, 19, 20, 21, 22, 23, 24, 25, 26]. This method, along with other similar standard statistical methods, are predominately used to identify key predictors of mortality during the development of prognostic indices.…”