Factors associated with length of stay in hospital for men and women aged 85 and over: a quantile regression approach

Objectives: Explore characteristics of hospital use for adults aged 85 and over in their last year of life and examine factors associated with cumulative overnight length of stay (LOS). Data source/study setting: NSW 45 and Up Study linked with hospital data. Study design: Longitudinal cohort study. Methods: Quantile regression models were performed for men and women (N = 3145)to examine heterogeneity in predictors of overnight hospital admissions. Coefficients were estimated at the 25th, 50th, 75th and 90th percentiles of the LOS distribution. Principal findings: 86% had at least one hospitalisation in their last year of life, with 60% dying in hospital. For men, first admission for organ failure was associated with a 26 day increase at the 90th LOS percentile, and a 0.22 day increase at the 10th percentile compared to men with cancer. Women admitted with influenza had decreased LOS of 20.5 days at the 75th percentile and 6 to 8 fewer days at the lower percentiles compared to those women with cancer. Conclusions: Poor health behaviours were a major driver of highest LOS among older men, pointing to opportunities to achieve health care savings through prevention. For older women, influenza was associated with shorter LOS, which could be an indicator of the high and rapid mortality rates at older ages, and may be easily prevented. Other factors associated with LOS among women, included where they lived before they were admitted, and discharge destination.