University of Pennsylvania Medical Center Guidelines for Antibiotic Use



Routine blood cultures should be performed on any patient in whom there is a suspicion of bacteremia or candidemia.

Isolator blood cultures have exceptionally poor yield for anything other than detection of disseminated M. avium-intracellulare infection in patients with AIDS. However, they might increase yield over Bactec blood cultures on patients suspected of having one of the following conditions:

1. Endocarditis with multiple negative BACTEC system cultures. It may be appropriate to use the BACTEC system for the initial workup of endocarditis of any suspected etiology. Isolators should be used after multiple BACTEC cultures are obtained and fail to reveal an etiologic agent (including HACEK). No endocarditis patient at HUP has ever had a positive Isolator blood culture and negative Bactec blood culture.

2. Suspected deep fungal infection, such as histoplasmosis and blastomycosis. Ordinarily, cultures of other sites, such as tissue biopsy, and in some cases serological tests such as antigen and antibody tests, are more helpful than blood culture. Only one HUP patient has ever had a positive Isolator culture for histoplasma in over 25 years.

3. Suspected mycobacteremia caused by Mycobacterium avium-intracellulare, particularly in HIV patients with CD4 counts <50. This is the preferred test for diagnosing disseminated MAI infection.

4. Suspected disseminated gonococcal infection. No HUP patient with disseminated gonococcal infection has ever had a positive Isolator and negative Bactec culture.

5. Suspected bartonellosis. No HUP patient has ever had a positive blood culture for Bartonella. Serologic testing is preferred for diagnosis. PCR can be done on excised infected non-fixed valve tissue.

6. Suspected candidemia or disseminated cryptococcosis in patients for whom routine cultures have not detected Candida species or Cryptococcus neoformans, respectively. This yielded positive results in Isolator cultures fewer than five times over the last 25 years.

7. Suspected Malassezia furfur infection, an agent of catheter-associated infection in patients receiving intravenous lipid.

The indication for use of isolators should be given to the microbiology resident (pager #215 9606941), who will release the tubes for pickup if the use fits the above guidelines. Clinicians of patients fitting indication #3 above can simply call 6623406 for tubes.The tubes must be hand picked up and delivered as transit via the specimen transport "tube" system damages them.

The following is not an indication for blood cultures of either type: Surveillance for infection before the clinical suspicion of infection exists


Blood cultures should be drawn prior to the institution of antibiotics whenever possible. If empiric treatment is an emergency, blood cultures should still be drawn as soon as possible after institution of antibiotics. There are no data to suggest that the timing of culture in relation to the appearance of fever or chills will maximize the yield.


There is a direct relationship between the volume of blood obtained and the yield of a blood culture set. Forty to 60 ml of blood should be obtained per episode (in other words, 2-3 sets with 20 ml per set, and 10 ml per bottle).


Single sets should not be used to evaluate any patient with suspected bacteremia or candidemia. The optimal yield is obtained with three or four sets of blood cultures. No more than four blood cultures should be obtained for any given 24 hour period.


Blood should be obtained from peripheral venous or arterial sites. Obtaining blood cultures from central venous catheters, arterial lines and inguinal vessels increases the likelihood of obtaining a false positive blood culture. The practice of drawing blood for culture from catheters or the groin should never be performed when a peripheral (i.e., non-catheterized) site is available.


Labeling the site of each set of blood cultures, particularly regarding whether a set was drawn from a catheter, the groin, or not, is of utmost importance in helping to distinguish pathogens from contaminants in those cases in which no peripheral access can be found.


1. After the vessel site is selected, a 5 cm area of skin should be disinfected by swabbing concentrically with 70% alcohol, from the venipuncture site outward.

2. The site should be cleansed once again, this time with 2% chlorhexidine gluconate again in a circular motion.

3. Allow the chlorhexidine to dry completely before performing venipuncture. This should take 1 - 2 minutes.

4. While waiting for the site to dry, the plastic cap covering each blood culture bottle should be removed, and the rubber stopper should be decontaminated with 70% alcohol.

5. 20 ml of blood should be withdrawn from the puncture site.

6. Do not change needles between venipuncture and inoculation of the bottles, or between bottles. The risk of needlestick is increased, while the chance of contamination is not significantly lessened.


updated 3/11/13 P Edelstein

Back to Table of Contents Back to Table of Contents