Customize your JAMA Network experience by selecting one or more topics from the list below.
Story From the Front Lines
A 67-year-old woman with a history of hypertension presented to the hospital after 1 week of fever, cough, and dyspnea. Physical examination demonstrated a temperature of 39°C, respiratory rate of 24 breaths/min, and inspiratory crackles auscultated at the left lung base. A chest radiograph showed a left lower lobe infiltrate. Findings from laboratory studies were notable for a blood urea nitrogen level of 21 mg/dL (to convert to millimoles per liter, multiply by 0.357). The patient was admitted to the general medicine inpatient service and started on ceftriaxone and azithromycin therapies for community-acquired pneumonia. During hospital day 1, the patient noted improvement in her symptoms. On attending physician rounds, a plan was made for transition to oral antibiotics on the following day, with consideration for discharge.
Just prior to midnight on hospital day 1, the night coverage resident was called because the patient was found to have a temperature of 38.3°C. At this time she was asymptomatic, and the remainder of her examination findings were normal. The resident instructed the nurse to draw 2 sets of blood cultures. On hospital day 2, the patient noted continued improvement in her symptoms and remained afebrile since the previous night. Blood cultures drawn the night prior resulted as preliminarily positive for gram-positive cocci in clusters from 1 of 2 sets. The team subsequently started vancomycin therapy for possible methicillin-resistant Staphylococcus aureus bacteremia. This regimen was continued awaiting organism identification.
On hospital day 4, the positive blood culture was reported as Staphylococcus epidermidis. Owing to the absence of signs or symptoms consistent with active infection by this organism, the result was concluded to be a contaminant. Vancomycin therapy was discontinued, and antibiotics were narrowed to azithromycin to complete a 7-day course. Following evaluation by physical therapy, the patient was deemed safe for discharge on hospital day 5. The patient described anxiety regarding her diminished physical strength following a 5-day inpatient stay, in addition to concern that the added interventions and length of stay would result in financial hardship.
Blood cultures have been the “gold standard” for detecting bacteremia since their earliest application, when Streptococcus viridans was identified as the cause of malignant endocarditis.1 With approximately 200 000 cases of bacteremia occurring in the United States each year and an associated mortality approaching 50% in certain populations, physicians carry a low index of suspicion for ordering blood cultures. Quiz Ref IDAs a result, a low percentage (4%-7%) of blood cultures are positive.2 Published guidelines do not specifically state when blood cultures should be drawn for most clinical conditions.3
Quiz Ref IDUnfortunately a large portion (approximately half) of positive blood cultures are subsequently determined to be contaminants—skin flora inoculated into culture bottles.4 These false positives lead to inappropriate tests and interventions. Compared with true-negative results, false-positive results are independently correlated with a 39% increase in charged amounts for antibiotics and a 20% increase in total subsequent laboratory charges, in addition to an average increased length of stay of 4.5 days.4 Other potential harms for the patient include adverse effects of the unnecessary antibiotics, as well as concerns for increasing antibiotic resistance patterns in the setting of overuse as exemplified by the emergence of vancomycin-resistant organisms.
The decision to obtain blood cultures for the febrile inpatient should consider the pretest probability of bacteremia, in addition to the likelihood that the results will influence management. Quiz Ref IDA review by Coburn et al2 demonstrated that the pretest probability of bacteremia varies widely based on the clinical condition, ranging from 2% for cellulitis to 69% for septic shock.Quiz Ref IDIn terms of clinical predictors of bacteremia, elevated temperature (for ≥38.3°, likelihood ratio [LR], 1.2) or leukocytosis (LR, 1.4) in isolation do not accurately predict bacteremia. The presence of chills is a more useful predictor with a positive LR of 2.2, increasing to 4.7 for “shaking chills.”Quiz Ref IDThe systemic inflammatory response syndrome (defined as ≥2: temperature <36°C or >38°C, heart rate >90 beats/min, respiratory rate >20 breaths/min or Pco2 <32 mm Hg, and white blood cell count <4000/µL or >12 000/µL or >10% immature neutrophils) is a very sensitive tool for predicting bacteremia, with absence significantly lowering the probability (LR, 0.09). Preexisting antibiotic use also decreases the pretest probability of bacteremia (LR, 0.63). In a retrospective study of 139 inpatients with community-acquired infections or fever, only 1 patient (0.72%) had a new pathogen isolated on blood culture while receiving antibiotic therapy.5
It is common for an immunocompetent inpatient who presented with fever due to an infectious source to have persistent fever during the initial 72 hours of treatment. In response, physicians may reflexively order blood cultures. Our patient’s presenting condition (community-acquired pneumonia: pretest probability of bacteremia approximately 5%-10%), current use of antibiotics, and absence of useful clinical predictors of bacteremia make the use of blood cultures in this setting inappropriate.2 Avoidable diagnostic and therapeutic interventions continue to occur because of reflexive ordering of blood cultures for inpatients with fever, at a high cost.
Corresponding Author: Bryan LeBude, MD, Department of Medicine, Thomas Jefferson University Hospital, 1025 Walnut St, Ste 805, Philadelphia, PA 19107 (firstname.lastname@example.org).
Published Online: August 11, 2014. doi:10.1001/jamainternmed.2014.3687.
Conflict of Interest Disclosures: None reported.
LeBude B, Diemer G. Routine Blood Cultures for the Febrile Inpatient: A Teachable Moment. JAMA Intern Med. 2014;174(10):1546–1547. doi:10.1001/jamainternmed.2014.3687
Create a personal account or sign in to: