Construction of an instrument for the prognostic evaluation of elderly persons in intensive care units

Abstract

Abstract Objective: To create an instrument for the prognostic evaluation of elderly patients hospitalized in an intensive care unit. Methods: A cohort study, with prospective data collection, which included elderly persons aged 60 years or older, was carried out in the city of João Pessoa, in the state of Paraíba, Brazil. Data collection was performed using an instrument created from a pilot study and the Katz Index. Poisson’s regression was used for data analysis. This technique estimates relative risk, retaining variables with p≤0.10 in the instrument, and ensures biological plausibility. The classification of risk of death was performed using quartile analysis, confirmed by the Receiver Operator Characteristic (ROC) curve. Results: 205 elderly patients with an average age of 74.6 years and a 59% risk of mortality, were included. Of the total sample and based on the scores, 16.6% of elderly persons had a low risk of developing death, 23.9% were at moderate risk, 40% had a high risk, and 19.5% exhibited a very high risk of death. The positive predictive value of the instrument was 77% and the negative value was 67.5%, with a concordance index of 0.78. The cutoff score of the instrument was 9 points or over. The sensitivity was 77.7% and the specificity was 66.7%. Conclusions: The instrument developed may be useful in the identification of elderly people with risk factors who require increased care. The instrument described can therefore be applied in Brazilian intensive care units.