Tempo de permanência na Unidade de Terapia Intensiva e o uso de Redes Bayesianas como ferramenta de gestão

Abstract

The lack of beds in intensive care units (ICU) is a public health problem. The length of stay in these units contributes to reducing bed turnover, which slows down the admission of other critical patients who end up receiving inadequate care, increasing hospital mortality rates. The general objective of this work is to statistically analyze the possible causal relationships using probabilistic inferences between factors related to the length of stay in an intensive care environment through the use of Bayesian Networks for management strategies and support to decision making. A retrospective study was carried out in the general ICU of Hospital Calixto Midlej Filho de Itabuna, with a qualitative and quantitative approach with 49 patients aged between 14 and 92 years. Data collection was performed through the patients' medical records via the hospital's audit department and / or Medical and Statistical File Service (SAME). The variables collected were: gender, age, APACHE II (English: Acute Physiology and Chronic Health Evaluation II), presence of mechanical ventilation, development of infection and length of stay in the unit. The collected data were inserted and tabulated in an Excel® spreadsheet. and used in the manufacture of Bayesian Networks (RB) using GeNIe 2.0 software. The results showed that gender, age, mechanical ventilation and APACHE II classification factors influence the length of stay in the ICU. Therefore, the use of Bayesian methods that, through probabilistic reasoning, have a good performance to work with causes and effects relationships, can be used as a tool to support management for decision making and optimization of the time spent in critical ICU patients.