Predictive mathematical model of mortality for community acquired pneumonia

Abstract

Background: pneumonia acquired in the community constitutes an important problem worldwide. It is the fourth cause of death at our country. Prognoses index are helpful to detect the patients with high risk of death, however these prognoses have low sensibility and specificity. Objective: to propose a predictive mathematical mortality model for acquired pneumonia disease in the community. Methods: a longitudinal analytic study with two groups was conducted. Mann Withney's test was made to obtain predictive variables of mortality. Pearson correlation was calculated with meaningful variables. Then, a mathematical model was elaborated and was taken to ROC curve to find the curve area as well as coordinates. Sensibility and specificity were calculated. Results: age from 79± 11 years with 50 % accounting for females. Global mortality was 35%. The variables with different behavior between the two groups were systolic blood pressure (x2=0,001) and diastolic blood pressure (x2=0,001), Creatinine was (x2=0, 03), respiratory frequency rate (x2=0, 01), pressure of oxygen (x2=0,036) just like the numbers of hemoglobin, and finally the level of potassium (K) (x2=0,004) with marked difference between groups. Conclusions: this mathematical model is a useful tool to evaluate the patient's state because it helps to give a more precise diagnostic. DeCS: PNEUMONIA; MATHEMATICAL MODELS; SENSITIVITY AND SPECIFICITY; CAUSALITY; LONGITUDINAL STUDIES.