Morbidity and mortality in elderly patients admitted to hospital are high. Prognostic information collected at
hospital admission is useful to define care objectives for older people.
to compare the ability of the Mini Nutritional assessment (MNA) and the Karnofsky Performance Score (KS) to
predict mortality in hospitalized geriatric patients.
A prospective 9 -months follow-up study was designed. One hundred thirteen subjects aged 65 years or more,
admitted consecutively over 6 months to a Medicine guard of a Military Hospital in Santiago, Chile, were included. Upon
admission, patients underwent a medical history and examination. MNA and KS were measured. Mortality data during the
9 months after inclusion in the study were obtained from the Chilean population records at the “Registro Civil de Chile”.
Thirty-six patients (32%) died during the follow-up period. According to ROC curves the cut off point for MNA
was 17 and 30 for KS (area under ROC curve=0.79 and 0.81). The Kaplan –Meier curves showed that survival rate in the
group with KS ≤ 30 was 31% and 80% in those with >30. For MNA ≤ 17, survival rate was 47% and 72% with >17. In
multiple step-wise logistic regression analysis, KS and MNA were independent predictors of mortality. However when
mortality OR for MNA were adjusted for KS, the former lost its predictive capacity.
KS ≤ 30 in geriatric patients is a better predictor of mortality than MNA and it can be recommended as a
routine assessment in geriatric hospitalized patients. However MNA is useful to decide the use of nutritional support
measures that could facilitate prognosis of a given patient.