+++
Make the Diagnosis: Bacteremia
+++
Prior Probability of Bacteremia
++
The probability of bacteremia depends on the clinical setting. Two studies have shown a probability of 4.1%-7.0% when assessed among consecutive patients for whom blood cultures were obtained, with true positive:contaminant ratios of 1.6:1 and 0.88:1.1-2 When the clinician has a suspicion that an underlying source may be causing bacteremia, the prior probability should be adjusted to the likelihood of bacteremia for the specific infection. For example, the probability of bacteremia among patients with cellulitis is about 2%, and 13% among patients with community-acquired pneumonia and fever and a requirement for hospitalization, 19%-25% among patients with pyelonephritis, and 53% among those with acute bacterial meningitis.
+++
Population in Whom Bacteremia Should Be Considered
++
Bacteremia is considered when patients have fever, chills, leukocytosis, focal infections, or signs of shock without an obvious cause, or when the physician suspects endocarditis.
+++
Assessing the Likelihood of Bacteremia
++
As a symptom, a patient's report of fever is an inaccurate predictor of bacteremia, with a likelihood ratio (LR) approaching 1.0. An elevated measured temperature ≥ 38.3°C (summary LR, 1.2; 95% CI, 1.0-1.4) or at a threshold of ≥ 38.5°C (summary LR, 1.4; 95% CI, 1.1-2.0) is not much better as an isolated finding. In febrile patients, the presence or absence of chills is more informative, and the presence of shaking chills may be the most useful finding (Table 89-1). Tachycardia and hypotension have been studied at different thresholds, but generally have LRs < 2.0 and, as isolated findings, have low diagnostic accuracy. Likewise, commonly used laboratory tests (such as for leukocytosis and percentage of banded neutrophils) have been studied at multiple different thresholds with LRs that vary across studies, with most studies showing a positive LR < 2.0.
++