Background
Community-acquired bacteremia is a potential source of infection and cause of fever in patients presenting to the emergency department (ED) with suspected sepsis. Ambiguous or false-positive blood culture results may lead to unnecessary testing and overtreatment with substantial implications on antimicrobial stewardship and associated healthcare costs. There is little clinical information available to determine with a high degree of certainty whether a patient may be febrile secondary to bacteremia. The primary aim of this study was to assess the diagnostic utility of fever in patients with suspected sepsis secondary to community-acquired bacteremia.
Methodology
A retrospective review of a consecutive sample of electronic health records of patients presenting to the ED of an academic tertiary care hospital with an annual census of approximately 100,000 visits was performed. Structured demographic and clinical data were summarized. Receiver operating characteristics of fever, defined as an initial recorded temperature ≥38°C, in predicting bacteremia, defined as at least one positive blood culture result, were calculated. The associations between fever and bacteremia and admission and inpatient all-cause mortality rates were analyzed using multivariable logistic regression.
Results
A total of 100,270 pediatric and adult ED visits were screened for eligibility. Of the 10,220 (10.2%) patients who had at least one blood culture result and temperature recorded, 1,175 (11.5%) were febrile and 487 (4.8%) had bacteremia. Febrile patients were more likely to have blood cultures drawn than afebrile patients (34% vs. 10%, p < 0.001), and the median initial temperature was higher in patients with a positive blood culture result (37.1°C vs. 36.9°C, p < 0.001). Fever was not sensitive for bacteremia across pediatric, adult, and geriatric cohorts (25.0%, 21.6%, and 23.4%, respectively), but was more specific in adult and geriatric cohorts (90.2% and 92.1%, respectively) than the pediatric cohort (68.6%). A positive blood culture result was associated with an increased likelihood of admission (adjusted odds ratio (AOR) = 2.85, 95% confidence interval (CI) = 2.14-3.81, p < 0.001) and inpatient all-cause mortality (AOR = 3.67, 95% CI = 2.68-5.04, p < 0.001) while fever increased the likelihood of admission only (AOR = 1.58, 95% CI = 1.36-1.83, p < 0.001).
Conclusions
A strict definition of a fever of ≥38°C is specific but not sensitive for bacteremia. Febrile patients are 1.58 times as likely to be admitted when compared to afebrile patients while adjusting for age, sex, and the presence of a positive blood culture result. Bacteremia is associated with increased admission and mortality rates in patients with suspected sepsis presenting to the ED.