[Nosocomial infection and multiple causes of death].

OBJECTIVE The purpose of this study was to evaluate the possibility for collecting data on nosocomial infections by means of death certificates. METHODS The medical charts of children who died after 48 hours of their admission to a pediatric hospital were revised to get information about the causes of death. Death certificates were also revised to verify whether they were properly filled out. The death certificates were redrafted according to the data obtained from patients' medical charts. Nosocomial infections and procedures related to them were codified using the Supplementary External Causes of Injury and Poisoning codes of the International Disease Classification-9th Revision. RESULTS The causes mentioned on medical death certificates compared with revised causes obtained after revision of patients' charts showed an agreement of 69.9%, with a Kappa coefficient of 0.65. Only one case of nosocomial infection was found on original death certificates, whereas the redrafted certificates revealed 88.9% of nosocomial infections in the study population up to fifteen days before death. CONCLUSION Death certificates are not a good source of information on nosocomial infection because they are often filled out inadequately. The solution to this is to qualify physicians so that they can identify and record nosocomial infections on the patient's medical chart and complete death certificates accordingly.