When should we order blood cultures?
The article: Shapiro NI, Wolfe RE, Wright SB, Moore R, Bates DW. Who Needs a Blood Culture? A Prospectively Derived and Validated Prediction Rule. The Journal of Emergency Medicine. 2008;35(3):255-264. doi:10.1016/j.jemermed.2008.04.001.
PICO: In patients presenting to the emergency department with suspected infection, can a clinical decision rule predict bacteremia and need for blood cultures?
Design and Population:
Prospective cohort study, included all patients who had a blood culture obtained either in the ED or within 3h of admission
Patients were randomly assigned to derivation (2/3) or validation (1/3) sets
The derivation set had a total of 2466 patient visits with 204 culture sets with positive growth for true pathogens (8.3% bacteremia).
The validation set had 1264 patient visits with 101 culture sets growing true pathogens (8.0% bacteremia)
The outcome was true bacteremia
Bacteremia and sepsis are common, with an estimated 751,000 cases of sepsis per year in the United States, or approximately 2000 cases per day.
The general indications for blood culture use are poorly defined, and as a result, overall blood culture yields from hospitalized patients remain remarkably low at 4-8%
The low yield of blood cultures has significant financial costs, wastes health care worker time, and results in unnecessary needle sticks for patients and risk for health care workers.
In addition, cultures that grow contaminants may result in increased hospital charges and unnecessary antibiotic treatment, hospital admission, and resource utilization
When present in the ED, the following factors are associated with an increased risk of bacteremia:
Major Criteria: clinical presentation suggesting endocarditis, temperature > 39.4°C (103° F), indwelling catheter
Minor Criteria: creatinine > 2.0 mg/dL, shaking chills, neutrophil percent > 80, temperature 38.3-39.3°C (101-102.9°F), any hypotension (systolic blood pressure < 90 mm Hg), platelets < 150,000 cells/mm3, white blood cell count > 18,000 cells/mm3, bands > 5%, vomiting, age > 65 years.
Either 1 major or 2 minor criteria were an indication for blood culture.
The sensitivity was 98.0% (95% confidence interval [CI] 96-100%) in the derivation set and 97.0% (95% CI 94-100%) in the validation set. The specificity was 29.0% (95% CI 27-31%) and 28.8% (95% CI 26.2-31.4%) for each respective set. The positive predictive value was 11.1% (95% CI 10-13%) in the derivation set and 10.6% (95% CI 9-13%) in the validation set, whereas the negative predictive value was 99.4% (95% CI 99-100%) and 99.1% (95% CI 98-100%), respectively.
Seven bacteremic patients were missed with the decision rule (4 in the derivation and 3 in the validation set). Upon review of the cases, in only 2 did the blood culture results change treatment–both of those cases technically would have met the clinical rule criteria (one was febrile to 103 at home, the second spiked a fever on 2nd day of hospitalization).
The study was single center. Decision to send blood cultures was left up to the clinicians and nursing staff–it is possible that patients who may have been infected were not included in the patient population. There was also the risk of misclassification bias as it was difficult occasionally to determine whether a bacterial isolate was a “true positive” or “contaminant.”