Association between severity of illness and mortality from nosocomial infection

Abstract

Background: For the years 1987 through 1992, a study was undertaken to analyze nosocomial infection mortality data and to stratify risk according to severity of underlying illness to compare with published data from the Centers for Disease Control and Prevention. Methods: Nosocomial infections that contributed to or caused death were identified. In addition, during 1990 through 1992, severity of illness was determined as the subjective estimate of the risk of death or lack of risk of death during the current hospital admission before the onset of the nosocomial infection. These groups were named + SIC and − SIC, respectively. Results: It was determined that the data from death certificates in cases of known nosocomial infection were not sufficient to determine whether nosocomial infection contributed to or caused death. There was a 24% increase in cases of nosocomial infection contributing to or causing death when a physician reviewed deaths in patients with nosocomial infections who did not have a diagnosis of nosocomial infection listed on the death certificate. The rates for nosocomial infections contributing to or causing death are as follows: nosocomial pneumonia, 20%; and bloodstream infections, 19%. In patients who died and had severity of illness determination, there was a statistically significant difference in the rates of nosocomial infections contributing to or causing death between −SIC and +SIC groups for both nosocomial pneumonia and bloodstream infections. The rates for bloodstream infections were as follows: −SIC, 5%; and +SIC, 21%. For nosocomial pneumonia, the rates were as follows: −SIC, 13%; and + SIC, 23%. Conclusion: In published reports from the Centers for Disease Control and Prevention, a rate of 13% is given for nosocomial pneumonia and bloodstream infections contributing to or causing death; however, there is no stratification for severity of illness in these reports. The presence of life-threatening illness before the onset of nosocomial pneumonia or bloodstream infection accounts for most deaths among our patients. For valid comparisons, mortality outcome data for nosocomial infections should be stratified for risk according to severity of underlying illness.