CLINICAL STUDIES

Routine Use of Anaerobic Blood Cultures:Are They Still Indicated?

Eduardo Ortiz, MD, MPH, Merle A. Sande, MD

PURPOSE: To determine the number of patients with bacte-remia and fungemia and to evaluate the utility of routine anaer-obic blood cultures as part of the work-up for suspected bacte-remia.SUBJECTS AND METHODS: Retrospective review of micro-biology data followed by selective chart review at a university-affiliated Veterans Affairs Medical Center. We determined thenumber of bacterial blood cultures drawn from January 1, 1994,to December 31, 1996, and the number of anaerobic, aerobic,and fungal isolates. Chart reviews were then performed on allpatients with a positive anaerobic result.RESULTS: There were 6,891 sets of blood cultures processedthrough the laboratory, yielding 1,626 patients with positiveresults. Anaerobic isolates were recovered from 36 patients(2.2%) in 48 bottles. Aerobic isolates were recovered from 1550

patients (95.3%), and fungal isolates were recovered from 40patients (2.5%). Seven patients (0.4%) had true anaerobic bac-teremia. All seven patients had an obvious source of anaerobicinfection that was known or suspected before the cultures weredrawn. Antibiotic changes were made in four of these patientsafter the positive anaerobic results were known. Antibioticchanges led to clinical improvement in one patient.CONCLUSIONS: Routine use of anaerobic blood culturesrarely results in clinically important diagnostic or therapeuticbenefits, based on the low incidence of anaerobic bacteremia inpatients who are not at increased risk. Anaerobic blood culturesshould be selectively ordered in patients at risk for anaerobicinfections. Am J Med. 2000;108:445– 447. q2000 by ExcerptaMedica, Inc.

During the past two decades, there has been a de-crease in the incidence of anaerobic bacteremia(1– 6). Despite this decrease, most clinicians rou-

tinely order anaerobic blood cultures in the work-up ofpatients with suspected bacteremia. However, the valueof routine anaerobic blood cultures in guiding clinicaldecisions or improving outcomes is unclear. To evaluatethis issue, we analyzed data on all anaerobic isolates in ourinstitution from 1994 to 1996.

MATERIAL AND METHODS

DesignWe conducted a retrospective review of all blood culturesprocessed through the microbiology laboratory at the SaltLake City Veterans Affairs Medical Center, followed by

retrospective chart reviews on all patients who had a pos-itive anaerobic result.

Blood Culture and Data CollectionThe medical center used the BACTEC NR-660 blood cul-ture system (Becton Dickinson, Towson, Maryland).Each bottle contained 40 mL of brain heart infusionbroth. Blood cultures were drawn by house staff, medicalstudents, or nursing staff and delivered to the laboratoryfor processing and identification by standard techniques.Hospital policy is to collect 20 mL of blood per culture setand inoculate 10 mL into each bottle. A set consists of oneaerobic and one anaerobic bottle.

We determined the number of bacterial blood culturesdrawn from January 1, 1994, to December 31, 1996, andthe number of anaerobic, aerobic, and fungal isolates. Forpatients with a positive anaerobic result, we performedretrospective chart reviews. Positive anaerobic cultureswere segregated into pathogens and contaminants. Anisolate was considered a contaminant if it was a com-monly recognized contaminant, such as Propionbacte-rium acnes (7,8), and there was documentation in themedical chart by either the laboratory, infectious disease,or clinical services stating that the organism was a con-taminant. All charts were reviewed to identify whetherrisk factors (9,10) for anaerobic bacteremia were evidentbefore the blood cultures were drawn. We also assessedthe treatment of patients with a positive result.

From the Department of Medicine, Veterans Affairs Medical Centerand University of Utah Health Sciences Center, Salt Lake City, Utah.

This work was presented in part at the 21st Annual Meeting of theSociety of General Internal Medicine, Chicago, Illinois, April 23–25,1998.

Requests for reprints and correspondence should be addressed toEduardo Ortiz, MD, MPH, Department of Medicine, Health ServicesResearch and Development Unit, Veterans Affairs San Diego Health-care System and University of California, San Diego, 3350 La Jolla Vil-lage Drive, 111N-1, San Diego, California 92161.

Manuscript submitted April 19, 1999, and accepted in revised formSeptember 10, 1999.

q2000 by Excerpta Medica, Inc. 0002-9343/00/$–see front matter 445All rights reserved. PII S0002-9343(99)00410-6

RESULTS

There were 6,891 sets of blood cultures processedthrough the laboratory, yielding 1,626 patients with pos-itive results. Anaerobic isolates were recovered from 36patients (2.2%) in 48 bottles. Aerobic isolates were recov-ered from 1,550 patients (95.3%), and fungal isolateswere recovered from 40 patients (2.5%). Seven patients(0.4%) had true anaerobic bacteremia, 28 patients (1.7%)had isolates determined to be contaminants, and 1 pa-tient’s chart was unavailable for review.

The distribution of anaerobic pathogens was Bacte-roides fragilis in 3 patients, Bacteroides thetaiotamicron in1 patient, Clostridium clostridiiforme in 1 patient, Clos-tridium septicum in 1 patient, and Fusobacterium nuclea-tum in 1 patient. The distribution of contaminants wasPropionbacterium acnes in 22 patients, Peptostreptococcusin 2 patients, Clostridium perfringens in 2 patients, andLactobacillus in 2 patients.

All 7 patients with anaerobic bacteremia had an obvi-ous source of anaerobic infection before the blood cul-tures were drawn, including 2 patients with necrotic dia-betic foot ulcers, 1 patient with an epidural abscess, 1patient with a bowel resection, 1 patient with a rupturedappendix, 1 patient with liver disease, ascites, and perito-nitis, and 1 patient with lymphocytic leukemia and neu-tropenic enterocolitis.

Antibiotic changes were made in 4 of the 7 patientsafter the positive anaerobic results were known. Anaero-bic coverage was added in 2 patients, and antibiotics werediscontinued in 2 patients. Antibiotic changes probablyled to an improvement in the condition of 1 patient.

DISCUSSION

During the 1960s and 1970s, improvements in laboratorytechniques resulted in the increased detection of anaero-bic bacteria (4,5). However, the incidence of anaerobicbacteremia appears to have peaked in the 1970s, with asubsequent decrease in incidence (1– 6). Our results areconsistent with these findings, as only 0.4% of patientswith a positive blood culture had true anaerobic bactere-mia.

Other studies during the past decade have also demon-strated low anaerobic recovery rates (4 – 6,11). Sharp (4)reviewed 8,688 blood cultures and identified 18 culturesyielding anaerobic isolates. Only 9 cultures were clinicallyimportant. Lombardi and Engleberg (5) reviewed 31,758blood cultures at the University of Michigan Hospitalsand 6,594 blood cultures at the Ann Arbor Veterans Af-fairs Medical Center, which yielded a total of 63 isolatesfrom 40 patients with clinically important anaerobes.Bannister and Woods (6) analyzed 12,289 sets of bloodcultures and identified 1,306 sets of positive blood cul-

tures from 808 patients, of whom 15 had true anaerobicbacteremia. Peraino et al (11) analyzed 7,397 blood cul-tures and found only 16 patients with anaerobic bactere-mia that was of clinical importance. Many other investi-gators (2,8,12–15) have also demonstrated anaerobic re-covery rates of less than 5% of positive blood cultures;however, differences in their study designs and reportingmake direct comparisons with our data difficult.

In our study, all 7 patients with anaerobic bacteremiahad an obvious source of infection before the blood cul-tures were drawn. Other investigators have demonstratedsimilar findings in that 88% to 98% of their patients hadan identifiable source of, or risk factors for, anaerobicinfection (3,5,11). These risk factors include oral, gastro-intestinal, gynecologic, or obstetric disease, surgery, ortrauma; malignancy; use of cytotoxic drugs or corticoste-roids; hematologic disorders; immunosuppression; sple-nectomy; aspiration; diabetes; vascular insufficiency; tis-sue obstruction, stasis, anoxia, or destruction; human oranimal bites; burns; and foreign bodies (9,10).

Even in patients at risk for anaerobic bacteremia, it isnot evident that obtaining anaerobic blood cultures leadsto improved outcomes. Only 1 patient in our study ap-peared to benefit from antibiotic changes made as a resultof the positive anaerobic culture. In 3 cases, microbiologyresults were not even known until after the patient haddied or been discharged.

Others have also noted this problem. Murray et al (2)found that anaerobic blood culture results rarely influ-enced the management of patients, either because othercultures were positive, patients were already on appropri-ate antibiotic therapy, or the isolate was not clinicallyimportant. Lombardi and Engleberg (5) noted that posi-tive anaerobic blood cultures rarely influenced the clini-cal management of their patients with anaerobic bactere-mia. Goldstein et al (16) stated, “When performed, theresults of anaerobic identification and susceptibility stud-ies are often returned to the chart long after there is anypotential for clinical relevance.”

Although routine use of anaerobic blood cultures maynot provide much benefit, discontinuation of this prac-tice is unlikely to result in substantial cost savings. Even inpatients not at risk for anaerobic bacteremia, 20 to 30 mLof blood per culture should still be collected, since this isthe optimal volume for detecting bacteremia (17,18). Thecurrent practice of inoculating one vented aerobic andone unvented anaerobic bottle would therefore be re-placed with the inoculation of two vented aerobic bottles.This practice would optimize the recovery of the morecommon bacterial pathogens, which include Staphylo-cocci, Streptococci, Enterobacteriaceae, and other gram-negative bacilli (2,19). Because the cost of the bottles isequal, no savings would be realized. In previous years,indirect savings may have occurred because anaerobicbacteriology was more labor intensive and required more

Use of Anerobic Blood Cultures/ Ortiz and Sande

446 April 15, 2000 THE AMERICAN JOURNAL OF MEDICINEt Volume 108

expertise than other laboratory activities (16). With theuse of automated blood culture systems, this issue nolonger appears to be important.

Our study had some limitations. We did not evaluatethe utility of anaerobic blood cultures for the detection offacultative anaerobes, which include certain species ofStaphylococci, Streptococci, and Enterococci, and membersof the Enterobacteriaceae family. Some (3,19 –21) but notall (2,6,17) investigators have shown that these organismsmay preferentially grow in the anaerobic bottles. Our re-sults are based on retrospective review of microbiologyand chart data, which can introduce selection and infor-mation biases. We did not evaluate the effects, if any, of anegative anaerobic culture.

We conclude that clinically important anaerobic bac-teremia is rare and that it usually occurs in patients whohave a suspected source of, or risk factors for, anaerobicbacteremia. Routine use of anaerobic blood culturesrarely results in clinically relevant diagnostic or therapeu-tic gains. Anaerobic blood cultures should be selectivelyordered in patients at risk for anaerobic infections.Healthcare facilities should assess their incidence of clin-ically important anaerobic bacteremia and use this infor-mation to guide local policies (22). Further evaluation isneeded to determine the feasibility of ordering selectiveanaerobic blood cultures, the utility of anaerobic bloodcultures in detecting facultative anaerobes, and the role ofanaerobic blood cultures in improving clinical outcomes.

ACKNOWLEDGMENTWe wish to thank David Porter, MT (ASCP) for his assistancewith the data collection and Brent James, MD, M.Stat for hissupport.

REFERENCES1. Dorsher CW, Rosenblatt JE, Wilson WR, Ilstrup DM. Anaerobic

bacteremia: decreasing rate over a 15-year period. Rev Infect Dis.1991;13:633– 636.

2. Murray PR, Traynor P, Hopson D. Critical assessment of bloodculture techniques: analysis of recovery of obligate and facultativeanaerobes, strict aerobic bacteria, and fungi in aerobic and anaero-bic culture bottles. J Clin Microbiol. 1992;30:1462–1468.

3. Morris AJ, Wilson ML, Mirrett S, Reller LB. Rationale for selectiveuse of anaerobic blood cultures. J Clin Microbiol. 1993;31:2110 –2113.

4. Sharp SE. Routine anaerobic blood cultures: still appropriate to-day? Clin Microbiol Newsl. 1991;13:179 –181.

5. Lombardi DP, Engleberg NC. Anaerobic bacteremia: incidence, pa-tient characteristics, and clinical significance. Am J Med. 1992;92:53– 60.

6. Bannister ER, Woods GL. Evaluation of routine anaerobic bloodcultures in the BacT/Alert blood culture system. Clin Microbiol In-fect Dis. 1995;104:279 –282.

7. Weinstein MP, Reller LB, Murphy JR, Lichtenstein KA. The clinicalsignificance of 500 episodes of bacteremia and fungemia in adults. I.laboratory and epidemiologic observations. Rev Infect Dis. 1983;5:35–53.

8. Murray PR. Determination of the optimum incubation period ofblood culture broths for the detection of clinically significant sep-ticemia. J Clin Microbiol. 1985;21:481– 485.

9. Finegold SM. Host factors predisposing to anaerobic infections.FEMS Immunol Med Microbiol. 1993;6:159 –163.

10. Brook I. Anaerobic bacterial bacteremia: 12-year experience in twomilitary hospitals. J Infect Dis. 1989;160:1071–1075.

11. Peraino VA, Cross SA, Goldstein EJC. Incidence and clinical signif-icance of anaerobic bacteremia in a community hospital. Clin InfectDis. 1993;16(suppl 4):S288 –S291.

12. Courcol RJ, Durocher AV, Roussel-Delvallez M, Fruchart A, Mar-tin GR. Routine evaluation of BACTEC NR-16A and NR-17A me-dia. J Clin Microbiol. 1988;26:1619 –1622.

13. Reimer LG, Reller LB, Mirrett S. Controlled comparison of a newBecton Dickinson agar slant blood culture system with Roche Septi-Chek for the detection of bacteremia and fungemia. J Clin Micro-biol. 1989;27:2637–2639.

14. Levi MH, Gialanella P, Motyl MR, McKitrick JC. Rapid detection ofpositive blood cultures with the BACTEC NR-660 does not requirefirst-day subculturing. J Clin Microbiol. 1988;26:2262–2265.

15. Weinstein MP, Mirrett S, Reimer LG, Reller LB. Effect of agitationand terminal subcultures on yield and speed of detection of theOxoid Signal blood culture system versus the BACTEC radiometricsystem. J Clin Microbiol. 1989;27:427– 430.

16. Goldstein EJC, Citron DM, Goldman RJ. National hospital surveyof anaerobic culture and susceptibility testing methods: results andrecommendations for improvement. J Clin Microbiol. 1992;30:1529 –1534.

17. Washington JA, Ilstrup DM. Blood cultures: issues and controver-sies. Rev Infect Dis. 1986;8:792– 802.

18. Forbes BA, Granato PA. Processing specimens for bacteria. In:Murray PR, Baron EJ, Pfaller MA, Tenover FC, Yolken RH, eds.Manual of Clinical Microbiology, 6th ed. Washington, DC: ASMPress, 1995:265–281.

19. Cockerill FR, Hughes JG, Vetter EA, et al. Analysis of 281,797 con-secutive blood cultures performed over an eight-year period: trendsin microorganisms isolated and the value of anaerobic culture ofblood. Clin Inf Dis. 1997;24:403– 418.

20. Martin WJ. Routine anaerobic blood cultures: reasons for contin-ued use. Clin Microbiol Newsl. 1992;14:133–134.

21. Hellinger WC, Cawley JJ, Alvarez S, et al. Assessment of routine useof an anaerobic bottle in a three-component high-volume bloodculture system. J Clin Microbiol. 1996;34:2544 –2547.

22. Goldstein EJC. Anaerobic bacteremia. Clin Inf Dis. 1996;23(suppl1):S97–S101.

Use of Anerobic Blood Cultures/ Ortiz and Sande

April 15, 2000 THE AMERICAN JOURNAL OF MEDICINEt Volume 108 447