Changes in nutritional status during the hospital stay: A predictor of long-term survival

The objectives of this prospective observational study were to assess wheth er: 1) midarm circumference (MAC), previously shown to predict in-hospital mortality, maintains its prognostic implication after discharge; 2) in-hosp ital changes in aspecific indicators of the health status are predictors of long-term survival. The study population consisted of 249 patients from th e general community [mean age 80+/-7 (70-99) years], consecutively discharg ed from geriatric and medical wards of an acute care hospital. Changes in h ealth status during hospitalization were recorded (dynamic or delta variabl es) and health-related variables were collected at discharge (discharge var iables). The relationship of both sets of variables to survival over a S-ye ar period was assessed by Cox's proportional hazards regression analysis. T he discriminatory efficacy of predictive models was estimated by the Hanley and McNeil method. Survival curves were drawn with the patients alternativ ely grouped according to the presence or absence of each of the predictive variables. Serum albumin<3.5 g/dL (hazard rate=0.57, 95% confidence limits= 0.33-0.96) and dependency in at least one ADL (h.r.=0.87, c.l.=0.79-0.98) w ere found to be associated with increased mortality, and delta MAC (h.r.=1. 03, c.l.=1.01-1.05), i.e., there was a positive change or no change in MAC from admission to discharge, with increased survival. A slightly weaker pre dictive model was obtained using only discharge variables. However, Hanley and McNeil's analysis showed that both models were far from achieving the o ptimal discrimination of high from lour risk subjects. Effects on survival of individual variables varied in magnitude and dependency on time. We conc luded that measuring in-hospital changes in nutritional status might improv e prediction of long-term survival. Attempts should be made to identify var iables having the strongest prognostic implications, and to tailor dynamic assessment to the needs of selected categories of patients. (C) 1998, Editr ice Kurtis.