A three-year study of positive blood cultures, with emphasis on prognosis.

A study of 37,156 blood cultures over a 3-year period yielded 1,972 positive blood culture episodes, of which 63% were of clinical significance, 26% represented contamination, 7% represented transient bacteremia, and 3% were of indeterminate significance. Mortality curves were calculated for clinically significant bacteremia according to etiologic organism and source. Several curves with different shapes were demonstrated. Quantitative differences were noted between cases of bacteremia arising from different sources and caused by different organisms. No mortality was associated with bacteremia of bone and joint origin. Mortality from bacteremic pneumococcal pneumonia, bacteremia with Escherichia coli of urinary tract origin, endocarditis, and beta-hemolytic streptococcal bacteremia showed an early plateau effect, with a drop before day 20. Cases of bacteremia from intravascular sources other than endocarditis were associated with no initial mortality, but mortality rose progressively after day 5. Bacteremia related to most organisms and sources was associated with mortality that continued until at least day 20.