Cellulitis: Do Not Get Blood Cultures

You are treating a patient with left lower leg cellulitis. The nurse is going to establish IV access, draw blood work, and give analgesia and antibiotics. Before walking into the room, the nurse asks, “Do you need me to grab a set of blood cultures?” Additionally the hospitalist had asked you to order a “set of cultures” on your most recent cellulitis admission. Should you proceed?

Background

Cellulitis is one of the ten most common complaints in the ambulatory care setting which includes the emergency department (1-3). Patients are subclassifed into either complex or simple cellulitis depending upon their comorbidity.

• Complicated cellulitis is defined as the presence of an immune-compromised status (HIV/AIDS, in active chemotherapy, status post organ transplantation), diabetes, or peripheral vascular insufficiency. Due to the extensive comorbidity, it typically requires treatment in the inpatient setting and blood cultures have been routinely recommended.
• Uncomplicated (simple) cellulitis is most often treated with oral antibiotics that cover for Staphylococcus and Streptococcus species based on local antibiograms.

Utility of Blood Cultures in Uncomplicated Cellulitis

In 2005, Mills et al (4) performed a search of the best available evidence on blood cultures in patients with cellulitis. Five articles were identified:

Authors	Positive cultures	Contaminated cultures
Perl et al	11/553 (2%)	20/553 (3.6%)
Kulthanan et al	20/150 (17.2%)
Lutomski et al	4/25 (16%)	4/25 (16%)
Ho et al	1/130 (0.77%)	0/130 (0%)
Hook et al	2/13 (4%)	No mention

The authors conclude:

“On the basis of the evidence available, blood cultures do not significantly alter treatment or aid in diagnosing the microbial organism in acute adult cellulitis in normal immunocompetent hosts. Therefore, it would be within the standard of care not to obtain blood cultures in immunocompetent patients who present with apparently uncomplicated cellulitis.”

Utility of Blood Cultures in Complicated Cellulitis

In a retrospective chart review by Paolo et al (5) in 2013, patients were classified by the authors as having complicated or uncomplicated cellulitis. All of the study participants had blood cultures drawn and a comparison was made between the two groups to determine the utility of cultures in this setting. The results were:

Cellulitis Type	Positive blood cultures	Contaminated blood cultures
Complicated	29/314 (9%)	13/314 (4%)
Uncomplicated	17/325 (5%)	10/325 (3%)

The authors stated, “A clinically significant change in management (a change in the class of antibiotic) was found in 6 of 314 cases vs. 4 of 325 controls (p =0.578; OR=1.5525; 95% CI 0.434–5.5541)… This group of clinically significant change in management was about 2% of the entire cohort and most would have been changed to a narrowed antibiotic. “

Patient #	Initial Antibiotic	Second Antibiotic	Blood Culture	Comorbidity
1	Keflex	Zosyn, Vancomycin	Cornebacterium	Diabetes
2	None	Augmentin	Stapylococcus saccharolyticus	Diabetes
3	Vancomycin	Penicillin G	Group B Strep	Diabetes
4	Zosyn, Flagyl	Cephalexin	Streptococcus salivarius	Chemotherapy
5	Clindamycin	Oxacillin	Group B Strep	Chemotherapy
6	Keflex	Linezolid	MRSA	Asplenia

Additionally, out of their entire cohort, only 7 cultures were shown to have gram-negative bacteria. Due to the study time period (2005-2009), MRSA was not as prevalent in their community and is likely more common presently.

Conclusion

In both uncomplicated and complicated cellulitis, blood cultures have a low yield of becoming positive and when they are found to be non-contaminated, they are unlikely to significantly change management. The cases in which non-skin flora grow in the blood, the history from the patient usually has given the provider some cause to suspect bacteria other than routine skin flora.

 

References

Expert Peer Review

March 15, 2014
It is important to utilize diagnostic tests with the greatest likelihood of influencing the management of the presenting patient.  Each particular test that we use will demonstrate differential clinical utility based upon their underlying performance characteristics particularly in the form of sensitivity and specificity.  The combination of the intrinsic testing parameters develops the diagnostic threshold of the test; the point of clinical possibility in which the test should be appropriately applied.  An example is depicted below:

In this scenario, we are looking at the changes in pre-test probability (to posterior probability) of a test given a presumptively high specificity and low sensitivity. As is known from basic EBM, specificity is the needed parameter to rule in disease whereas sensitivity rules out disease. This is evidenced by the steep slope of the maroon lines indicating a potentially relevant change in probability to discover disease whereas the green lines have a softer slope demonstrating the poor performance of the test to change what we initially thought given negative results. The combination of these attributes forms a diagnostic threshold (the horizontal dashed line)—the pre-test probability at which testing should or should not be performed. If we define a patient as likely to have disease (say 40% likely) and we are using a test with the parameters depicted above then our results are ultimately useless and we should not have tested in the first place. A positive and negative result still ends up as likely to have disease rendering the test moot.

Rates of positive blood cultures (pre-test probability) in cellulitis are estimated to be approximately 2% (1). In addition the rate of contamination (false positives) are either equal to or exceed the actual yield making the test a difficult one to recommend. In order for the test to be efficacious given a presumed 2% pre-test probability the testing parameters for clinical relevance would have to be astronomically good—unfortunately they are far from useful. A best estimate demonstrates a 4% sensitivity and an equally poor specificity making it difficult if not impossible to budge the 2% we started with prior to encountering the patient.
Our study (2) sought to determine if these same testing parameters held true in the complicated cellulitis group, a mostly unstudied and expertly defined subcomponent of all patients with cellulitis. As was to be expected the rates of contamination equaled the rates of positive yield for the entire retrospective cohort rendering the test as primarily useless. It is important to note however that two findings had high prediction of positive blood cultures—fever and diabetes. Fever is a parameter that may in fact not predict simple cellulitis but underlying bacteremia from the initial infection. Speculatively this may hold true not just for complicated cellulitis but may be true across all manifestations of disease—this study was not designed to answer this question. Given the retrospective nature of the study it can neither endorse nor exclude the discovered association between fever and positive cultures and therefore the totality of the clinical picture in these complicated patients should be taken into account prior to the ordering of cultures. The yield was also higher in diabetics rather than the rest of the cohort but as with the rest of this study this rarely resulted in a change of management. The argument has been made that bacteremia is a distinct disease from cellulitis that is clinically relevant though it may not result in change in antimicrobial coverage from empiric management. While this may be true it is not clear that the blood culture as opposed to the clinical parameters of the patient would matter in terms of the care for these individuals.

Bottom Line

Our study found little clinically relevant information gained by the addition of blood cultures to complicated cellulitis in a retrospective study. Further prospective studies should be undertaken to confirm these findings and determine if a subgroup of these patients (i.e. fever and diabetes) would benefit from cultures.