Principles of Antibiotic Formulary Selection for P&T Committees
Burke A. Cunha, MD
Part I: Antimicrobial Activity
A
ntimicrobial agents that are being considered for inclusion
on a formulary should have a high degree of activity
against the pathogens they are intended to treat. If an anti-
microbial is going to be used primarily for urinary tract infections
(UTIs) or urosepsis, its spectrum of activity should be greatest
against aerobic gram-negative bacilli and enterococci because
these two pathogen groups are the primary causes of UTIs in
hospitalized patients. Antibiotics that are used to treat UTIs do
not require activity against respiratory pathogens (e.g., pneu-
mococci) or against organisms that affect the skin (e.g., Staphy-
lococcus aureus). Furthermore, because anaerobes are not
uropathogens, anti-anaerobic activity is not needed in drugs
whose intended primary use is for UTIs or urosepsis.
Frequently, pharmaceutical companies present data to P&T
committees showing a wide range of activity, often against
organisms that are either clinically irrelevant or so rare as to be
clinically unimportant. These data are often displayed against
organisms that are almost always colonizers and that do not
cause infection and, therefore, do not warrant treatment.
Ordinarily, organisms that colonize respiratory secretions,
wounds, urine, feces, or skin are not treated. Therefore, anti-
biotics that have a high degree of activity against these organ-
isms might be suitable for treating infection but are inappro-
priate and not cost-effective for use against colonizing
organisms. Most organisms that have been recovered from the
sites mentioned are indeed colonizers and cause infection only
infrequently (e.g., Citrobacter freundii, Enterobacter agglomerans,
Stenotrophomonas maltophilia, Burkholderia cepacia, and so
on). Other organisms, such as methicillin-resistant S. aureus
(MRSA) or vancomycin-resistant enterococci (VRE), do cause
infection, but more than 90% of the time they colonize the skin,
respiratory secretions, urine, or feces.
MINIMAL INHIBITORY CONCENTRATION
A company’s pharmaceutical literature might also empha-
size differences in the minimal inhibitory concentration (MIC)
of its products compared with that of the competition. Remem-
ber, the MIC is an in vitro determination and is an important—
but not the only—measure of an antibiotic’s activity. The MIC
is isolated and meaningless if it is taken out of the clinical con-
text.
It is a popular misconception that drugs with a lower MIC are
more effective, or that they kill more efficiently or quickly, than
drugs with a higher MIC if an organism is sensitive to both types
of antibiotics. If an organism is sensitive, all antibiotics with an
MIC of 2, 0.2, or 0.02 mcg/ml kill at precisely the same rate with
the same efficacy and effectiveness. There is no advantage to
paying a premium for drugs with the lowest MIC.
Occasionally, a pharmaceutical firm’s promotional literature
might present differences in MICs as if they were important
when, in fact, the differences are so small as to be unimportant.
If an antibiotic has an MIC of 0.3 mcg/ml, it should not be
assumed to be inferior to an agent with an MIC of 0.03 mcg/ml;
these concentrations are very low and should be easily achiev-
able in most body tissues in comparisons of sensitive antibiotics.1
The MIC should also be considered in relation to achievable
blood levels. The ratio of peak serum concentration to MIC is
taking in vitro testing one step further toward predicting in vivo
efficacy.
For example, if the MIC of a first-generation cephalosporin
against methicillin-sensitive Staphylococcus aureus (MSSA), with
an MIC of 1 mcg/ml, were compared with a third-generation
cephalosporin having an MIC of 3 mcg/ml, it appears that the
first-generation cephalosporin, with an MIC of 1 mcg/ml, would
be three times as active as the one with an MIC of 3 mcg/ml
against this strain of S. aureus. If the achievable blood levels with
normal dosing were 100 mcg/ml for the drug with the MIC of
1 mcg/ml, the serum to MIC ratio would be 100:1. If the anti-
biotic with an MIC of 3 mcg/ml had achievable blood levels of
600 mcg/ml, the serum/MIC ratio of the second drug would be
200:1. Viewed from this perspective, the antibiotic with a higher
serum concentration/MIC ratio would be preferred, all other fac-
tors being equal.2
SUSCEPTIBILITY TESTING
P&T committees should also become more familiar with the
details of antimicrobial susceptibility testing. Committee mem-
bers should realize that in vitro susceptibility testing is just
that—an approximation of activity in vitro that does not always
correlate with clinical efficacy in vivo. Susceptibility testing is
highly dependent on the methodology and the microorganism;
susceptibility is also dependent on the concentration, an impor-
tant point to remember for appropriate interpretation of data and
antibiotic usage.
When cefoxitin (Mefoxin®, Merck) is given as a 2-g IV dose,
for example, approximately 85% of Bacteroides fragilis activity is
inhibited. When a 1-g IV dose of cefoxitin is administered every
six hours as a cost-saving benefit, approximately only 20% of
B. fragilis activity is inhibited. Therefore, the in vitro suscep-
tibility determination is based on achievable serum levels using
the recommended doses of antibiotics.
Doses may not be changed without consideration of the
microbiologic implications, as determined by pharmacokinetic
parameters. Clearly, infectious disease representatives on P&T
committees or an infectious disease clinician should be con-
sulted if there are any questions about interpreting in vitro sus-
ceptibility data.1–3
There are important differences between in vitro susceptibility
and in vivo efficacy. Certain antimicrobials appear to be sensi-
tive in vitro but are ineffective in vivo. Knowing which antibiotics
are in this category is important, not only for formulary consid-
eration but also in terms of antimicrobial prescribing for clini-
cians.
Some common examples of in vivo versus in vitro discrep-
ancies include the relative inactivity of beta-lactams and
trimethoprim–sulfamethoxazole (TMP–SMX) (Bactrim®,
Roche; Septra®, Monarch) against Bartonella. These antibiotics
Vol. 28 No. 6 • June 2003 • P&T® 397
 Antibiotic Formulary Selection

Table 1 Antibiotic Organism Combinations for Which in Vitro Susceptibility1 Testing Is Unreliable

Antibiotic “Sensitive” Organism

Penicillin Haemophilus influenzae, Yersinia pestis
Trimethoprim–sulfamethoxazole Klebsiella, enterococci, Bartonella
(TMP–SMX) (Bactrim®, Roche;
Septra®, Monarch)
Polymyxin B Proteus, Salmonella
Imipenem (Primaxin®, Merck) Stenotrophomonas maltophilia2
Gentamicin (Garamycin®, Schering) Mycobacterium tuberculosis
Vancomycin (Vancocin®, Eli Lilly) Erysipelothrix rhusiopathiae
Aminoglycosides Streptococci, Salmonella, Shigella
Clindamycin (Cleocin®, Fusobacteria, Clostridia, enterococci, Listeria
(Pharmacia & Upjohn)
Macrolides Pasteurella multocida
First-generation and second-generation Salmonella, Shigella, Bartonella
cephalosporins3
All antibiotics except vancomycin MRSA4
Minocycline (Minocin®, Wyeth-Ayerst),
quinupristin/dalfopristin (Synercid®, Rhone-Poulenc
Rorer/Aventis), and linezolid (ZyvoxTM, Pharmacia)

MRSA = methicillin-resistant Staphylococcus aureus; VRE = vancomycin-resistant enterococci.

1 In vitro susceptibility does not predict in vivo activity, and susceptibility data cannot be relied on to guide therapy.
2 Formerly called Pseudomonas and Xanthomonas.
3 Cefoperazone is the only cephalosporin with clinically useful anti-enterococcal activity (Enterococcus faecalis, not Enterococcus faecium [VRE]).
4 Despite apparent in vitro susceptibility of many antibiotics against MRSA, only vancomycin, quinupristin/dalfopristin, linezolid, and minocycline are

effective in vivo.
Adapted from Cunha BA. Antibiotic Essentials. Royal Oak, MI: Physicians’ Press, 2003.

appear to be sensitive in vitro but are ineffective in vivo for the
treatment of Bartonella infections. Similarly, TMP–SMX appears
to be sensitive to Klebsiella strains in vitro but is relatively
ineffective in vivo. Penicillin is commonly reported as being
effective against Haemophilus influenzae but is ineffective in
vivo. Streptococci are usually reported as being sensitive to
aminoglycosides but, in fact, have no anti-streptococcal activity
unless they are combined with another agent. For this reason,
penicillin plus gentamicin is effective in treating streptococcal
endocarditis. Gentamicin (e.g., Garamycin®, Schering) alone,
although sensitive by in vitro testing, is ineffective clinically and
works only in combination because of the synergy between
penicillin and gentamicin (Table 1).4–10
BACTERIOSTATIC AND
BACTERICIDAL ANTIBIOTICS
Pharmaceutical companies sometimes underscore the pur-
ported advantages of bactericidal antibiotics (destructive to
organisms) over bacteriostatic antibiotics (inhibitory to the
growth of organisms); “bactericidal” and “bacteriostatic,” how-
ever, are in vitro determinations. Both categories of drugs kill
with equal speed and equal efficacy if the organism is sensitive
to both types.
Bactericidal drugs have no advantage over bacteriostatic anti-
biotics except in three situations: febrile neutropenia, meningitis,
and endocarditis. Even in these areas, there are important excep-
tions. For example, chloramphenicol (e.g., Chloromycetin®,
Parke-Davis/Pfizer) has been successfully used for decades in
the treatment of meningitis, even though it is a bacteriostatic
antibiotic by in vitro determinations. As stated earlier, suscep-
tibility is concentration-dependent. According this perspective,
then, it can be shown that penicillin—in a low concentration—
acts as a bacteriostatic agent and that chloramphenicol—in a
high concentration—acts as a bactericidal agent, further blur-
ring the artificial distinction between “cidal” and “static” anti-
microbials. For patients with endocarditis, in whom bacterio-
static drugs are preferred, they are by no means the only way
to effectively treat endocarditis. For example, Q fever and Legion-
ella endocarditis have been treated successfully with static
agents.
Some antibiotics work well both bacteriostatically and bac-
tericidally, depending on the in vitro methodology used in test-
ing. Linezolid (ZyvoxTM, Pharmacia), for instance, has bacteri-
cidal as well as bacteriostatic activity. Because it doesn’t matter
whether an antibiotic is bactericidal or bacteriostatic, this is a

398 P&T® • June 2003 • Vol. 28 No. 6

 Antibiotic Formulary Selection
moot point. Linezolid has been used to treat MRSA endocarditis
with good results. The clinical outcome is the most important
determination; it is immaterial whether endocarditis is cured
bacteriostatically or bactericidally.
Because microbiological aspects of susceptibility testing are
complex with many clinical nuances, P&T committees should
rely on input from infectious disease experts by including an
infectious disease clinician in the decision-making process; it is
imperative that the clinician be a specialist in antimicrobial ther-
apy. If possible, such individuals should participate as P&T
committee members or should be available as consultants. How-
ever, clinicians from various disciplines, including infectious dis-
ease, who have been involved in antimicrobial drug trials should
not automatically be viewed as “experts.” An authority in anti-
biotics will have undertaken specific training in microbiology
and infectious disease and will have demonstrated expertise in
the field of antimicrobial therapy; he or she will not have expe-
rience only in the treatment of infectious disease.2,3,7
REFERENCES
1. National Committee for Clinical Laboratory Standards (NCCLS).
Performance standards for antimicrobial susceptibility testing:
Fourth Information Supplement, M100-S4. Villanova, PA: NCCLS;
1992.
2. Johnson CC. In vitro testing: Correlations of bacterial suscepti-
bility, body fluid levels, and effectiveness of antibacterial therapy.
In: Lorian V, ed. Antibiotics in Laboratory Medicine, 4th ed. Balti-
more: Williams & Wilkins; 1996:813–834.
3. Cunha BA. Antibiotic Essentials. Royal Oak, MI: Physicians’ Press;
2003.
4. York MK. Reporting susceptibility test results: Are we really com-
municating with physicians? Clin Microbiol Newslett 1996;18:
177–182.
5. Biedenbach DJ, Jones RN. Interpretive errors using an auto-
mated system for the susceptibility testing of imipenem and aztre-
onam. Diagn Microbiol Infect Dis 1995;21:57–60.
6. Marshall SA, and the Aztreonam Study Group. Serious inter-
pretive error among three commercial systems for susceptibility
testing of aztreonam. Diagn Microbiol Infect Dis 1995;22:249–251.
7. Cunha BA, Ortega AM. Antibiotic failure (Review). Med Clin
North Am 1995;79:663–672.
8. Lorian V, Sabath LD. Effect of pH on the activity of erythromycin
against 500 isolates of gram-negative bacilli. Appl Microbiol 1970;
20:754–756.
9. Nicolle LE. Measurement and significance of antibiotic activity in
the urine. In: Lorian V, ed. Antibiotics in Laboratory Medicine, 4th
ed. Baltimore: Williams & Wilkins; 1996:793–812.
10. Cunha BA. Problems arising in antimicrobial therapy due to false
susceptibility testing. J Chemother 1997;9:25–35.
COMING NEXT MONTH:
Principles of Antibiotic Formulary Selection,
Part 2: Pharmacokinetics and Pharmacodynamics
Vol. 28 No. 6 • June 2003 • P&T® 399