Delayed Recognition of a Positive Blood Culture
A 58-year-old woman with metastatic breast cancer recently treated with immunosuppressive therapy presented to the hospital for evaluation of diarrhea. As part of the infectious workup performed in the emergency department (ED), blood cultures and stool studies were sent to the laboratory. The following morning, her stool test returned positive for Clostridium difficile (C. difficile), a potent bacteria that can cause serious infections, especially in immunocompromised hosts. At that time, one out of the two blood cultures drawn in the ED was positive for gram-positive rods, but the speciation (type of bacteria) and sensitivity profile (which suggests which antibiotics are best to use) were not yet available. The medical team ordered appropriate antibiotic treatment for C. difficile but did not initiate treatment for the preliminarily positive blood culture.
After improvement in her symptoms, the patient was discharged 2 days after admission with instructions to complete the course of antibiotics for C. difficile and to follow up with her outpatient providers. On the day of discharge, the blood culture that was positive for gram-positive rods had now speciated and was growing Listeria monocytogenes, a bacterium that can cause life-threatening infection if not treated appropriately. Moreover, a diagnosis of Listeria monocytogenes must be reported to the local department of public health because of its risk of contagion. Unaware that the patient required treatment for listeriosis, the team discharged her.
A later review revealed that the hospital microbiology laboratory was aware of the microbiology result and, per its protocol, had notified the patient's bedside nurse of the positive blood culture. However, this result was not communicated to the primary medical team prior to the patient's discharge. Some team members who saw the positive result in the medical record erroneously thought that Listeria monocytogenes might be a contaminant (a bacteria that can be viewed as a false positive not requiring treatment). Although the positive blood culture was sent to the medical attending's inbox in the electronic health record, he did not see the result until well after the patient was discharged. When the mistake was recognized, the team called the patient and she was immediately readmitted to the medical service. The positive blood culture was reported to the city's department of public health. She received further workup including additional blood cultures and a lumbar puncture, which remained negative. Intravenous antibiotics were initiated for treatment of listeriosis. She was eventually discharged and did well.
by Sarah Doernberg, MD, MAS
Before addressing the issues related to communication, interpretation, and management of positive blood culture results described in this case, it is worth reviewing the indications for ordering a blood culture in the first place. In patients who might have an indication for a blood culture (generally, those with signs of infection such as fever and leukocytosis), providers might consider several strategies, including withholding testing and treatment altogether, withholding treatment while awaiting testing results, empirical treatment while awaiting testing results, or treating without further testing.(1) Determining the best approach involves weighing the probability of disease in the patient, the risks and benefits of treatment, and the diagnostic test performance characteristics. More information does not always lead to better outcomes, and overdiagnosis has proven detrimental to patients, often because of overtreatment.(2-4)
The above construct is true for many tests in medicine, and blood cultures are no exception. Numerous studies demonstrate that false-positive cultures are associated with increased length of stay, cost, and antibiotic use.(5,6) The Society for Healthcare Epidemiology of America advises against ordering blood cultures in patients without signs or symptoms of infection in their Choosing Wisely recommendations.(7) Conversely, inadequate treatment of true positive blood cultures, including listeria, results in higher mortality.(8)
With this in mind, when should blood cultures be drawn? In general, blood cultures are lower yield (2%–13%) in ambulatory patients, those with community-onset fever, community-acquired pneumonia, and cellulitis.(9) On the other hand, the probability of a true-positive culture is higher (38%–69%) in patients with severe sepsis, shock, or suspected bacterial meningitis.(9) Individual signs and symptoms are largely unhelpful, though fever and shaking chills can be important predictors of a positive culture.(9) Clinical gestalt is superior to individual indicators, and multivariable scores may be even more sensitive but require additional study.(9) Unfortunately, most studies of diagnostic accuracy for prediction of bacteremia have excluded immunocompromised patients, limiting extrapolation to the patient described in this case.
Interpretation of positive blood cultures should take into account both pretest probability and Gram stain characteristics. Approximately 40% of all positive blood cultures adjudicated by infectious diseases physicians, and just over 80% of blood cultures with gram-positive rods, are contaminants.(10,11) The most commonly isolated gram-positive rods—Corynebacterium, Propionobacterium, and Bacillus species—are almost always contaminants.(10,11) Other factors, including higher number of positive cultures and faster time to culture positivity, may suggest that the gram-positive rod is a true pathogen.(12) Whether a positive blood culture should be treated empirically depends on a patient's comorbidities, clinical stability, pretest probability for bacteremia, and the organism's morphology. Gram-negative rods are rarely contaminants and should be treated, while treatment of gram-positive cocci and gram-positive rods before speciation is less straightforward.(10)
In this case, given the clinical improvement, alternative diagnosis, and statistical likelihood that a gram-positive rod is a contaminant, if the culture had not speciated on the day of discharge, withholding antibiotics and discharging the patient would have been reasonable, even in the setting of immunosuppression. With concurrent Clostridium difficile infection, the harms of empiric antibiotic therapy likely outweigh benefits in stable patients.(13) Yet, there is also a possibility that this patient's diarrhea could have been entirely explained by listeriosis, with the positive C. difficile test—especially if detected by polymerase chain reaction—attributable to colonization with a toxigenic strain, rather than true infection.(14)
It's also crucial to consider best practices for follow-up of blood cultures pending at the time of discharge. An estimated 11% of inpatient microbiology tests result after discharge and approximately 2% of those require follow-up or a change in therapy.(15) Since one-third of providers are unaware that tests pending at discharge were even ordered and more than 60% are unaware of the results, putting systems in place to notify providers and patients of actionable results is crucial for patient safety.(16) The standard of care for reporting of positive blood cultures is under one hour.(17) Reporting critical values, including positive blood cultures, to ordering clinicians with confirmation is mandated by accreditation bodies, but even with mandates in place, results can fall through the cracks.(18) Novel approaches to improve timely and successful notification include automated notifications, call centers, and required manual review by infectious disease physicians.(18-20) At one institution, in addition to standard notification processes (e.g., a phone call from microbiology), an automated email notification was sent to discharging physicians if a pending culture turned positive after discharge and no antibiotic had been prescribed to the patient. While these efforts resulted in a 16% increase in documented follow-up, follow-up still only occurred 28% of the time.(20)
In this case, the reporting of the positive result from the microbiology laboratory was appropriate as nurses are considered suitable recipients for critical laboratory values.(18) Yet, gaps in communication among team members contributed to the delay in treating the patient. First, the nurse who received the initial notification from the microbiology laboratory may not have communicated the critical value to the treating team promptly. One could speculate causes for delay, including high clinical workload, competing demands, issues with hierarchy, and failure to recognize the urgency. In addition, team members had a medical knowledge gap; they did not recognize that listeria growing from blood cultures should be considered a true pathogen requiring immediate antibiotic treatment.(8) Developing a culture supportive of inquiry may have facilitated escalation of this result to the attending physician in real time, rather than leaving him to discover it in his inbox later. Finally, consultation with infectious disease experts would have been appropriate to discuss the significance of listeria bacteremia and appropriate treatment.
Notification of public health authorities allows for monitoring, control, and prevention of communicable diseases. Laws mandating notification of local health departments vary by municipality, but generally both the attending physician and the laboratory are required to report. Notification occurs at both the state and national levels, permitting identification of foodborne and other outbreaks as well as widespread surveillance for trends. Listeria is a reportable disease at both the state and national level.
A model notification system would address the gaps identified in this case. Initially, notification from the microbiology laboratory should go straight to the ordering or attending provider, which might obviate multiple handoffs of information with attendant degradation of information or misunderstanding of urgency. Automated calls could offload the work from the microbiology technicians. If the attending physician is not the one notified, a best-practice alert in the medical record for critical values might help to draw attention to the results, though much has been written about alert fatigue and development of new alerts requires careful consideration before implementation.(21) Lastly, the addition of clinical decision support regarding the significance of listeria bacteremia might help team members ensure appropriate treatment. Such support could suggest consulting infectious disease for treatment recommendations and contacting local public health authorities for reporting. Although systems approaches will help decrease errors in communicating and appropriately treating positive blood culture results, ensuring a work environment that respects high-value care and one that facilitates open dialogue is essential for preventing these types of events.
- Blood cultures have a high rate of contamination, especially if a gram-positive rod is identified.
- Before sending blood cultures, the pretest probability of bacteremia should be considered, and the test should not be sent if there is a low likelihood of true bacteremia.
- Discharging a patient with pending microbiology tests may be appropriate in certain situations, but systems should be in place to identify critical untreated results and notify the correct providers so that prompt action can be taken.
- Infectious disease consultation should be sought if there is uncertainty about how to interpret microbiology test results, including positive blood cultures.
- Clinicians should be aware of the need for mandated reporting of certain communicable diseases to public health authorities.
Sarah Doernberg, MD, MAS Assistant Clinical Professor Medical Director, Adult Antimicrobial Stewardship Department of Medicine, Division of Infectious Diseases University of California, San Francisco
1. Pauker SG, Kassirer JP. The threshold approach to clinical decision making. N Engl J Med. 1980;302:1109-1117. [go to PubMed]
2. Sisson JC, Schoomaker EB, Ross JC. Clinical decision analysis. The hazard of using additional data. JAMA. 1976;236:1259-1263. [go to PubMed]
3. Morgan DJ, Wright SM, Dhruva S. Update on medical overuse. JAMA Intern Med. 2015;175:120-124. [go to PubMed]
4. Morgan DJ, Dhruva SS, Wright SM, Korenstein D. 2016 Update on medical overuse: a systematic review. JAMA Intern Med. 2016;176:1687-1692. [go to PubMed]
5. Alahmadi YM, Aldeyab MA, McElnay JC, et al. Clinical and economic impact of contaminated blood cultures within the hospital setting. J Hosp Infect. 2011;77:233-236. [go to PubMed]
6. Bates DW, Goldman L, Lee TH. Contaminant blood cultures and resource utilization. The true consequences of false-positive results. JAMA. 1991;265:365-369. [go to PubMed]
7. Choosing Wisely. Don't perform urinalysis, urine culture, blood culture or C. difficile testing unless patients have signs or symptoms of infection. Tests can be falsely positive leading to over diagnosis and overtreatment. Society for Healthcare Epidemiology of America; October 1, 2015. [Available at]
8. Thønnings S, Knudsen JD, Schønheyder HC, et al; Danish Collaborative Bacteraemia Network (DACOBAN). Antibiotic treatment and mortality in patients with Listeria monocytogenes meningitis or bacteraemia. Clin Microbiol Infect. 2016;22:725-730. [go to PubMed]
9. Coburn B, Morris AM, Tomlinson G, Detsky AS. Does this adult patient with suspected bacteremia require blood cultures? JAMA. 2012;308:502-511. [go to PubMed]
10. Pien BC, Sundaram P, Raoof N, et al. The clinical and prognostic importance of positive blood cultures in adults. Am J Med. 2010;123:819-828. [go to PubMed]
11. Weinstein MP, Towns ML, Quartey SM, et al. The clinical significance of positive blood cultures in the 1990s: a prospective comprehensive evaluation of the microbiology, epidemiology, and outcome of bacteremia and fungemia in adults. Clin Infect Dis. 1997;24:584-602. [go to PubMed]
12. Mushtaq A, Chen DJ, Strand GJ, et al. Clinical significance of coryneform Gram-positive rods from blood identified by MALDI-TOF mass spectrometry and their susceptibility profiles—a retrospective chart review. Diagn Microbiol Infect Dis. 2016;85:372-376. [go to PubMed]
13. Pop-Vicas A, Shaban E, Letourneau C, Pechie A. Empirical antimicrobial prescriptions in patients with Clostridium difficile infection at hospital admission and impact on clinical outcome. Infect Control Hosp Epidemiol. 2012;33:1101-1106. [go to PubMed]
14. Polage CR, Gyorke CE, Kennedy MA, et al. Overdiagnosis of Clostridium difficile infection in the molecular test era. JAMA Intern Med. 2015;175:1792-1801. [go to PubMed]
15. El-Kareh R, Roy C, Brodsky G, Perencevich M, Poon EG. Incidence and predictors of microbiology results returning postdischarge and requiring follow-up. J Hosp Med. 2011;6:291-296. [go to PubMed]
16. Roy CL, Poon EG, Karson AS, et al. Patient safety concerns arising from test results that return after hospital discharge. Ann Intern Med. 2005;143:121-128. [go to PubMed]
17. Schifman RB, Meier FA, Souers RJ. Timeliness and accuracy of reporting preliminary blood culture results: a College of American Pathologists Q-probes study of 65 institutions. Arch Pathol Lab Med. 2015;139:621-626. [go to PubMed]
18. Liebow EB, Derzon JH, Fontanesi J, et al. Effectiveness of automated notification and customer service call centers for timely and accurate reporting of critical values: a laboratory medicine best practices systematic review and meta-analysis. Clin Biochem. 2012;45:979-987. [go to PubMed]
19. Wilson JW, Marshall WF, Estes LL. Detecting delayed microbiology results after hospital discharge: improving patient safety through an automated medical informatics tool. Mayo Clin Proc. 2011;86:1181-1185. [go to PubMed]
20. El-Kareh R, Roy C, Williams DH, Poon EG. Impact of automated alerts on follow-up of post-discharge microbiology results: a cluster randomized controlled trial. J Gen Intern Med. 2012;27:1243-1250. [go to PubMed]
21. Ranji SR, Rennke S, Wachter RM. Computerised provider order entry combined with clinical decision support systems to improve medication safety: a narrative review. BMJ Qual Saf. 2014;23:773-780. [go to PubMed]