[Value of C-reactive protein for detecting bacteremia in febrile patients].

OBJECTIVES

Mortality among bacteremic adults ranges between 14 and 35%, and there is no biological clue to identify such patients a priori. As C-reactive protein (CRP) blood level rises in children during bacteremia, we investigated its accuracy to identify bacteremic patients among febrile adults who were admitted to our hospital either for study or treatment.

METHODS

Patients older than 14 entering the emergency room with objective axilar temperature above 37.5 degrees C and admitted either for study or treatment were elligible for enrollment. After initial examination, samples were obtained for blood cultures and CRP measurement. Follow-up during hospitalization was assessed. All variables were related with one another by bivariant statistical methods performed with Sigma Horus hardware. After bivariant study we used the program BMDP Statistical Software (1991) to perform the multivariate study in its discriminant analysis.

RESULTS

One hundred seventy-four patients entered the study with an average age of 58.9 years; 47.7% were over 65 years of age; 88% of febrile syndromes were of infectious origin and among them, bacteremia ranged up to 17.2%. Values of CRP were significantly related with the following variables: "age": patients younger than 45 had average CRP concentrations of 9.5 mg%, age over 45 had 17.4 mg% (Mann Whitney U (MW) p < 0.001); "clinical situation on admission": for poor, intermediate and stabilized situations average CRP levels were 14.7, 18.5 and 12.4 mg% respectively (MW p < 0.05); "time from fever onset"; "clinical outcome": for death average CRP was 23.01 mg% and for cure 14.6 mg% (MW, p < 0.05), in that way probability of death was 21%, 9% and 3% for CRP concentrations higher than 20, between 10 and 20 and under 10 mg% respectively. There was no link between CRP serum levels and sex, severity of chronic illness, cause of fever, localization of infection and existence of bacteremia. The multivariate study in its discriminate mode was of no use to identify bacteremic patients.

CONCLUSION

According to our findings CRP determination can not be used alone or associated to identify patients with positive blood cultures among the febrile ones. CRP blood determination should not be considered before 24 hours from disease onset as this time is required for CRP to stabilize. Patients with CRP levels higher than 20 mg% on admission have a higher risk of dying during hospitalization.