We studied 45 patients, aged 50 years and older, with acute renal fail
ure during their admission. We assessed the chance oi death using a lo
gistic regression stepwise model, with 16 variables. The mean age of t
he group was 68,4 +/- 10 years; 13 patients were hemodialised and the
global mortality was 37,8 %. Mortality was not increased in hemodialys
ed patients. The statistical model used was: Omega = 0.271 - 0.197 (d
ays of admission) + 7.139 (mechanical ventilation). In this model the
chance of death is computed as: p = e(Omega)/(7 + e(Omega)) It predic
ted death with a a 94, 1 % sensitivity, 92,9 specificity and 93,3 % pr
edictive value. Our model was compared with 2 other models (Corwin and
Liano), previously described. We conclude that it is feasible to pred
ict mortality in elderly people with acute renal failure using a stati
stical model.