[A prognostic multifactorial study of sepsis in an internal medicine service]

All cases of sepsis attended at the Service of Internal Medicine from 1985 to 1989, both inclusive, were retrospectively analyzed by the statistical study of several clinical, epidemiological, bacteriological and laboratory factors. The parameters acting as prognosis factors were analyzed using Cox's method of logistic regression, in order to obtain more reliable information on the multifactorial spectrum determining "death due to sepsis". The unifactorial analysis (UFA) suggested that the following factors were associated to an statistically significant increase in the mortality due to sepsis: age above 70; male sex; presence of shock upon hospitalization; nosocomial etiology; high levels of urea, creatinine and LDH; proteinuria; family of the causal germ (gram-positive coccus) and underlying pathology. The relative risks (RR) or "odd ratios" of creatinine and LDH were, respectively, 2.8 and 2.9 in the UFA; 3 and 3.3 in the multivariant analysis (MVA). In the older patients developing sepsis within their communities, RR were respectively 2.7 and 1.98 in the UFA and 1.1 and 2.7 in the MFA. The results of the univariant and multivariant analysis showing the relative risk (odd ratio) associated to each of these variables with statistical significance, demonstrate that the mortality rate due to sepsis increases with age, nosocomial etiology and elevation of creatinine and lactodehydrogenase (LDH) levels. We stress the great importance of the LDH as a prognostic factor of sepsis.