Mortality one-year postdischarge from a veterans affairs geriatric evaluation and management unit: Assessing mortality risks

OBJECTIVE: To assess at Geriatric Evaluation and Management Unit (GEM) admi ssion factors that affect mortality 12-months postdischarge and to develop a preliminary risk scoring protocol to guide targeting of GEM care. SETTING: A 24 bed-GEM at a Veterans Affairs (VA) Medical Center. DESIGN: Relative risks(RR) were assessed using prospective data; a risk pro tocol from 1988-1989 data was tested on 1990-1991 patients. SUBJECTS: A total of 283 male patients, aged 60 to 102, discharged over 4 y ears. RESULTS: Age at GEM admission did not correlate with death (r = .14; P = .1 45), but did correlate with risk scores (r = .33, P < .001). The risk proto col had a sensitivity of .67 and specificity of 1.00. High and low risk pat ients had mortalities of 51% versus 20%, a Wilcoxon (Gehan) statistic of 15 .22, df = 1, and P < .001. Differences in mortality ceased about 100 days p ostdischarge. Three univariate RR exceeded 1.00 at a 99% Confidence Interva l (CI): IADL score (RR: 1.12; CI, 1.03-1.21); nursing acuity score (RR: 1.7 8; CI: 1.02-3.11); and a primary diagnosis of pneumonia/sepsis (RR: 3.95; C I, 1.60-9.78). Four RRs exceeded 1.00 at a 90% CI: dementia (RR: 1.78; CI, 1.02-3.09);transfer into the GEM from a medical service (RR: 1.47; CI, 1.02 -2.12); deconditioning/functional decline (RR: 1.67; CI, 1.12-2.48); and us e of a Foley catheter (RR: 2.22; CI, 1.11-4.45), Thirteen other potential r isk factors were found in a multivariate analysis. CONCLUSIONS: The point estimates of risk factors may help clinicians target GEM care, but the development of a useable risk protocol requires addition al work. Causal models may be needed to assess patient conditions related t o successful treatment in GEMs.