Context Clinicians order blood cultures liberally among patients in whom bacteremia is suspected, though a small proportion of blood cultures yield true-positive results. Ordering blood cultures inappropriately may be both wasteful and harmful.
Objective To review the accuracy of easily obtained clinical and laboratory findings to inform the decision to obtain blood cultures in suspected bacteremia.
Data Sources and Study Selection A MEDLINE and EMBASE search (inception to April 2012) yielded 35 studies that met inclusion criteria for evaluating the accuracy of clinical variables for bacteremia in adult immunocompetent patients, representing 4566 bacteremia and 25 946 negative blood culture episodes.
Data Extraction Data were extracted to determine the prevalence and likelihood ratios (LRs) of findings for bacteremia.
Data Synthesis The pretest probability of bacteremia varies depending on the clinical context, from low (eg, cellulitis: 2%) to high (eg, septic shock: 69%). Elevated temperatures alone do not accurately predict bacteremia (for ≥38°C [>100.3°F], LR, 1.9 [95% CI, 1.4-2.4]; for ≥38.5°C [>101.2°F], LR, 1.4 [95% CI, 1.1-2.0]), nor does isolated leukocytosis (LR, <1.7). The severity of chills graded on an ordinal scale (shaking chills, LR, 4.7; 95% CI, 3.0-7.2) may be more useful. Both the systemic inflammatory response syndrome (SIRS) and a multivariable decision rule with major and minor criteria are sensitive (but not specific) predictors of bacteremia (SIRS, negative LR, 0.09 [95% CI, 0.03-0.26]; decision rule, negative LR, 0.08 [95% CI, 0.04-0.17]).
Conclusions Blood cultures should not be ordered for adult patients with isolated fever or leukocytosis without considering the pretest probability. SIRS and the decision rule may be helpful in identifying patients who do not need blood cultures. These conclusions do not apply to immunocompromised patients or when endocarditis is suspected.