The volume of blood drawn in adult patients is the most important single factor governing the sensitivity of blood cultures [5-16], and 10 to 20 mL of blood per draw has been recommended [5, 13, 17-20]. This recommendation is based on studies showing that the concentration of microorganisms in one half of adult bloodstream infections is 1 colony-forming unit (CFU) per milliliter [10, 21-25] and less than 0.1 CFU per milliliter in 20% of cases [10].

We recently discovered that 15% of adult blood-culture specimens at the University of Wisconsin Hospital and Clinics were being collected in 3.5-mL pediatric blood-culture tubes and that 5% of specimens drawn in 10-mL adult blood-culture tubes contained less than 5 mL of blood. Concerned that low volumes of cultured blood might be compromising the sensitivity of blood cultures for detection of bacteremia in our adult patients, we studied the yield of low-volume blood cultures at our institution. We also surveyed directors of clinical microbiology laboratories across the United States to determine the prevalence of low-volume blood culturing in adult patients.

Methods

Top Methods Results Discussion Author & Article Info References

Study of Blood-Culture Volume and Yield

All blood-culture specimens drawn from patients older than 18 years between 1 January 1988 and 31 December 1990 at the University of Wisconsin Hospital were reviewed. In our study, a standard-volume blood-culture specimen was defined as 7 to 10 mL of blood and a low-volume specimen was defined as 3.5 mL of blood or less. Standard-and low-volume blood-culture specimens drawn within 45 minutes of each other from the same patient were considered a matched pair.

If either or both cultures from a matched pair were positive, the patient's clinical record was reviewed to make a determination regarding the presence of true bacteremia or candidemia: At least one of the positive blood-culture specimens had been drawn from a peripheral vein by percutaneous venipuncture (rather than from an intravascular catheter), and a microbiologically documented local infection (for example, urinary tract infection) giving rise to the bloodstream infection had been identified, or clinical evidence of infection manifesting as a temperature of 38.0 °C or higher or leukocytosis (leukocyte count, 10.0 x 10⁹/L) had been observed. When specimens grew common blood-culture contaminants—Corynebacterium species, Bacillus species, coagulase-negative staphylococci, and Propionibacterium acnes—we required that at least two separate blood cultures be positive for the same organism within a 24-hour period for the case to be regarded as true bacteremia.

During the period reviewed, blood-culture specimens were collected in sodium-polyanethol-sulfonate-containing Vacutainer tubes (Becton-Dickinson, Diagnostic Systems, Sparks, Maryland): The BDVS 4960 was used for adult (10-mL) specimens and the BDVS 6406 for pediatric (3.5-mL) specimens. In the clinical laboratory, specimens were inoculated into vials used in the nonradiometric NR-730 BACTEC blood culture system (Becton-Dickinson). Before February 1990, BACTEC 6A vials were used for aerobic cultures; when requested, blood was first placed in a resin-containing vial (ARD, Marison Laboratories, Kansas City, Missouri) to bind antimicrobial agents present in the specimen. After February 1990, BACTEC 16A vials containing resin beads were used for aerobic cultures, and BACTEC 7A vials without resin were used during the entire study for anaerobic cultures. Isolates from positive cultures were identified using standard microbiologic methods [26]. The time to detection of growth was recorded for each positive culture.

Chi-square tests were used for testing the statistical significance of dichotomous data, and the Wilcoxon matched-pairs sign-rank test was used for comparing differences in detection time between standard- and low-volume blood cultures.

National Survey of Blood Culture Practices

Directors of U.S. clinical microbiology laboratories who were members of the American Society of Clinical Pathologists were surveyed with the assistance of the Society; voluntary responses to six closed-ended questions were requested. The goal of the survey was to determine how many clinical laboratories routinely record the volume of blood submitted for culture; how frequently low-volume blood-culture specimens are submitted from adult patients; and whether laboratories have policies regarding the acceptance of low-volume blood-culture specimens from adult patients. If the institution did not routinely record the volume of blood submitted in blood cultures, the director was requested to prospectively obtain and provide data on the prevalence of low-volume blood cultures submitted from adults in their institution.

Results

Top Methods Results Discussion Author & Article Info References

Blood Culture Volume and Yield

Eight-hundred twenty-nine matched pairs (1658 separate blood cultures or 1.4% of all blood cultures obtained in the hospital during the study period) yielded 68 true bloodstream infections, 65 single-organism bacteremias or candidemias, and 3 polymicrobial bacteremias. For episodes of true bacteremia or candidemia, standard-and low-volume blood cultures contained an average of 8.7 mL and 2.7 mL of blood, respectively.

As shown in Table 1, 92% of the 69 true bloodstream infections were detected by standard-volume cultures, whereas 69% were detected by low-volume cultures, a difference in yield of 23% (95% CI, 9% to 37%; P <0.001). The respective sensitivities of standard- and low-volume blood cultures in the detection of bacteremia were 93% and 74% for gram-positive bacteria (difference, 19%; 95% CI, 3% to 36%; P =0.01), 89% and 60% for gram-negative organisms (P =0.10), 100% (2 of 2) and 0% (0 of 2) for anaerobic bacteria, and 88% (7 of 8) and 63% (5 of 8) for yeasts. For the 19 cases of bloodstream infection in which both standard- and low-volume samples were drawn by percutaneous venipuncture and both were culture positive, the difference in mean detection time (37.8 compared with 36.3 hours) was not statistically significant.

National Blood-Culture Practices

One-hundred fifty-eight U.S. clinical microbiology laboratory directors were surveyed. Responses to most or all of the questions were received from 71 laboratories (45%) in 32 states (mean hospital size, 439 beds). Eighty-eight percent of responding laboratories reported accepting blood-culture specimens from adults that contained 5 mL of blood or less, and only 20% of laboratories reported routinely recording the volume of blood submitted. Twenty-six directors provided data on the frequency of low-volume blood-culture specimens submitted from adults in their institution (either partially filled adult tubes or collected in pediatric tubes): 0% to 85% (mean, 12%) of all adult blood cultures in these institutions contained less than 5 mL of blood; in one third of these hospitals, more than 10% of adult blood cultures contained less than 5 mL of blood. Forty-one institutions provided information on the practice of drawing blood-culture specimens from adults in pediatric tubes: 0% to 53% (mean, 6%) of all adult blood cultures in these institutions were reported as being submitted in pediatric tubes or bottles.

Discussion

Top Methods Results Discussion Author & Article Info References

The yield of blood cultures from adults is clearly volume-dependent [5-16] and increases approximately 3% (range, 0.6% to 4.7%) for each additional milliliter of blood obtained (Table 2). We found an increased yield of approximately 3.2% per milliliter in our study. Because nearly 20% of adult blood cultures in our institution contained inappropriately low volumes of blood, we believe that a substantial number of bloodstream infections have gone undiagnosed. Our survey results indicate that our institutional experience does not appear to be an isolated one; 80% of the responding U.S. laboratories reported not routinely recording the volume of blood submitted for culture, and nearly all routinely accept low-volume blood-culture specimens from adults, either in partially filled adult tubes or in pediatric tubes. Even if all nonresponding laboratories routinely recorded blood volume and rejected low-volume blood-culture specimens, nearly one half of all the U.S. laboratories surveyed would still not be performing these vital quality-assurance functions.

We believe that this situation has arisen because few clinicians or nurses are aware of the powerful influence that the volume of blood has on the sensitivity of blood cultures for detecting bacteremia and candidemia in adult patients. We strongly reaffirm the recommendation made by the American Society for Microbiology [17] and others [18] that blood cultures for adults contain at least 10 mL per culture, preferably 20 mL [5, 13, 19, 20]. We further believe that microbiology laboratory staff should routinely record the volume of blood submitted for culture from adults as part of their quality-assurance program. This is particularly important because a recent report suggests that it may be difficult to change the behavior of health care workers who draw suboptimal volumes of blood for culture [27].

Presented in part at the Annual Meeting of the American Society for Microbiology, 26-30 May 1992, New Orleans, Louisiana.