It is very common for hospitalized patients to receive more than one antimicrobial agent simultaneously. The rationale for using antimicrobials in combination is not always clearly defined, and there are a number of potential disadvantages to combination therapy. The basis for using combination therapy is considered in this section.

Possible Reasons for Using Combination Antimicrobial Therapy

Clinical indications for using combination antimicrobial therapy fall into one of five categories. Two of these categories (empirical therapy and polymicrobial infections) relate to maximizing the likelihood that at least one agent of the combination will be active against known or suspected pathogens. The other three reasons for using combinations (minimizing toxicity, preventing the emergence of resistance, and obtaining syner-gistic inhibition or killing) attempt to exploit unique advantages of the combinations as compared with any component drug alone.

Broad Coverage during Empirical Therapy

A common reason for using more than one antimicrobial in hospitalized patients is to provide broad coverage against potential pathogens and to maximize the likelihood of delivering an active antimicrobial agent as quickly as possible to seriously ill patients. When the pathogen is unknown, the choice of antimicrobials will often include an agent broadly active against gram-positive bacteria, including methicillin-resistant S. aureus (MRSA), such as vancomycin, as well as an agent active against aerobic or facultative gram-negative bacteria. Selection of the latter will be strongly influenced by local patterns of antimicrobial resistance specific to the institution and might include an extended-spectrum cephalosporin, an aminoglycoside, a fluoroquinolone, a β-lactam–β-lactamase inhibitor drug, or a carbapenem. The latter two choices would also provide activity against gram-negative anaerobes if desired, such as for an intra-abdominal infection. Alternatively, one could add an agent such as metronidazole to provide anaerobic activity. In some situations, for patients who are not severely ill, a β-lactam might be used empirically as monotherapy when MRSA or other β-lactam–resistant gram-positive organisms are deemed unlikely pathogens.

Because of the high frequency of antibiotic resistance in Pseudomonas aeruginosa isolates, in settings in which P. aeruginosa is encountered frequently, empirical use of two agents with antipseudomonal activity could be justified to maximize the likelihood that at least one of the agents will inhibit the organism.

Combination therapy is widely used in the initial treatment of hospitalized patients with community-acquired pneumonia. Commonly used regimens include a third-generation cephalosporin such as ceftriaxone with a macrolide or fluoroquinolone. The cephalosporin provides antimicrobial activity against S. pneumoniae, Haemophilus influenzae, Moraxella catarrhalis, and several other “typical” bacterial pathogens associated with community-acquired pneumonia, with the notable exception of MRSA, which can sometimes cause pneumonia in outpatients. The macrolide—azithromycin is commonly used—or the fluoroquinolone inhibits “atypical” bacteria that cause pneumonia, including M. pneumoniae, C. pneumoniae, and Legionella sp. Although fluoroquinolones approved for respiratory tract infections are likely to cover most or all of the organisms targeted by the cephalosporin, isolates of S. pneumoniae resistant to fluoroquinolones do exist, so many clinicians prefer the extra security of combination therapy in patients with severe pneumonia requiring hospitalization.

Treatment of Documented Polymicrobial Infections

For many infections from which two or more pathogens are recovered, it is possible to provide adequate coverage with a single, broadly active antimicrobial agent. Switching to a single agent reduces exposure of the patient to potential antibiotic toxicities, is usually more convenient for nursing, and may be less expensive. For some patients, susceptibility profiles or allergy to broad spectrum agents justifies the use of antibiotic combinations for treatment of polymicrobial infections.

Attempts to Reduce Toxicity

It is theoretically possible to use two or more drugs of different classes, with additive antimicrobial activities and independent toxicities, each at relatively low doses to achieve sufficient potency while avoiding toxicity. However, there are no instances in current bacterial therapy in which the approach of using submaximal doses of multiple agents is predictably effective in accomplishing this goal. This does not exclude the possibility that in isolated instances a successful response may be attained when drugs with marginal activity are combined. In antifungal therapy, combinations of antifungal agents can potentially improve the clinical response when poor results are noted with the usual doses of individual drugs and toxicities preclude further dose escalation of a single agent.

Preventing the Emergence of Drug Resistance

Antimicrobial use in the treatment of tuberculosis provides the paradigm for using combination drugs in an attempt to prevent the emergence of resistance to any one agent. The basis for this approach is that if resistance to two different agents occurs by independent mechanisms, the probability of resistance developing to both drugs will be the product of the probability of resistance to each drug, which is likely to be very low, so resistance should not emerge. For example, if the frequency of resistance to drug A is 10-6 and that of resistance to drug B is 10-7, resistance to both drugs should be encountered no more frequently than once in 1013 colonies.

Similar reasoning has justified the use of combination regimens when rifampin is required for the treatment of nonmycobacterial infections. Rifampin is not used alone (with rare exceptions, such as brief courses for eradication of meningococcal carriage) because resistance to this agent emerges quickly. As mentioned earlier, rifampin is particularly useful in the treatment of infections related to foreign devices because of its activity against biofilm-associated bacteria. In such use, it is combined with another active antimicrobial, such as vancomycin for coagulase-negative staphylococcal prosthetic valve endocarditis (usually with a brief course of gentamicin as well to reduce the bacterial inoculum further at the beginning of therapy) or a fluoroquinolone for orthopedic device–related infections.

It has been more difficult to show unequivocally that combination therapy provides protection against the emergence of resistance to antimicrobial drugs in other situations, including infections caused by P. aeruginosa or Enterobacter sp. There are two plausible explanations why combinations may not prevent resistance predictably. First, there may be differential penetration of the two antimicrobials to an infected site or differences in activity at the site of infection. Thus, a more readily penetrating agent may be left relatively unprotected in a privileged site of infection. Second, for many commonly encountered bacteria, resistance mechanisms against unrelated antimicrobial classes may not be truly independent. Some bacterial efflux pumps recognize chemically unrelated substrates, so upregulation of pump activity may confer resistance to several classes of antimicrobials. In other instances there may be coordinated upregulation of efflux mechanisms and downregulation of outer membrane protein channels (porins), again potentially conferring resistance simultaneously to two or more antimicrobial classes.

Use of Combinations to Attain Synergism

Decades ago, the surprising benefits of using penicillin and streptomycin together for the treatment of enterococcal endocarditis were discovered empirically. Penicillin alone usually inhibits but does not kill enterococci, and failure rates were high when penicillin G was used alone to treat enterococcal endocarditis. Streptomycin has no significant activity against enterococci at clinically relevant concentrations. However, the combination results in bactericidal synergism in vitro and high cure rates in patients with enterococcal endocarditis. Detailed studies of this phenomenon demonstrated that in the presence of a cell wall–active antibiotic, uptake of the aminoglycoside into the bacterial cell increases substantially. Unfortunately, increasing rates of high-level resistance to streptomycin (MIC >2000 μg/mL) or to gentamicin (MIC >500 μg/mL), or both, has nullified the benefit of such combinations against a substantial number of enterococcal isolates today. An example of bactericidal synergism between vancomycin and gentamicin against an Enterococcus isolate is illustrated in Figure 301-1A .

Bactericidal synergism and antagonism

FIGURE 301-1  Bactericidal synergism and antagonism. A, Bactericidal synergism between vancomycin (diamonds) and gentamicin (triangles) against an isolate of Enterococcus sp. Killing by the combination (squares) is substantially greater than by each agent alone. B, Antagonism of the bactericidal activity of oxacillin (triangles) against an isolate of Staphylococcus aureus by the more slowly bactericidal agent vancomycin (diamonds). Killing by the combination (squares) is less than that by oxacillin alone. Growth in the absence of antibiotics is shown by the solid line without markers. CFU = colony-forming units.

Since the early demonstrations of synergistic killing of enterococci, combinations of cell wall–active agents plus aminoglycosides have been shown to achieve synergistic killing against a broad range of gram-positive and gram-negative bacteria when tested in vitro. Modest clinical benefits were shown when short courses of gentamicin were added to nafcillin for the treatment of S. aureus endocarditis. Against strains of viridans streptococci that are relatively insensitive to penicillin, the addition of an aminoglycoside for the first 2 weeks of a 4-week course of penicillin G is believed to result in a higher likelihood of cure.

Once considered important in the treatment of gram-negative bacterial infections, especially in immunocompromised (e.g., neutropenic) patients, the clinical value of a synergistic combination of a cell wall–active agent and an aminoglycoside has been difficult to prove in recent experience. To a large extent, the introduction of agents with very potent activity against gram-negative bacteria has diminished interest in and the perceived value of synergistic combinations. Nevertheless, there is some evidence that administering two or more active drugs for empirical therapy may achieve a better outcome than possible with a single active agent for P. aeruginosa infections, especially in neutropenic patients. Soon after the introduction of ciprofloxacin, high rates of inhibitory synergism against P. aeruginosa were demonstrated in vitro when antipseudomonal β-lactams were combined with this fluoroquinolone. In recent years, however, an increasing proportion of isolates have become resistant to fluoroquinolones, as well as to antipseudomonal β-lactams. Although resistance does not preclude the possibility of synergistic interactions, it is unlikely that such positive interactions would occur as frequently today as was documented 2 decades ago.

The combination of sulfamethoxazole with trimethoprim, agents that block sequential steps in folic acid synthesis, can also achieve bacteri-cidal (or bacteriostatic) synergism against a number of important gram-positive and gram-negative pathogens. This agent has been available for more than 30 years and still enjoys wide use. Quinupristin and dalfopristin are streptogramin antibiotics that display inhibitory activity against gram-positive organisms. Combining these two agents results in bactericidal synergism against organisms susceptible to both. Quinupristin-dalfopristin is provided as a 30:70 mixture of the two components and is approved for the treatment of staphylococcal or streptococcal complicated skin and skin structure infections and for infections caused by vancomycin-resistant enterococci that are associated with bacteremia.

β-Lactam–β-lactamase inhibitor antimicrobials represent another example of synergistic combinations. Four drugs of this category are currently marketed in the United States: amoxicillin-clavulanate, ampicillin-sulbactam, ticarcillin-clavulanate, and piperacillin-tazobactam. The β-lactamase inhibitors themselves, clavulanic acid, sulbactam, and tazobactam, are devoid of significant antimicrobial activity, with rare exceptions. However, by inhibiting the common β-lactamases that are sensitive to these agents, the inhibitors restore the activity of the hydrolyzable penicillins against many target pathogens elaborating these enzymes.


As mentioned earlier, using antimicrobial combinations does have possible disadvantages, including exposure of the patient to the potential toxicities of the multiple individual components and the added cost to purchase and administer multiple antibiotics. However, antibiotic combinations can sometimes also result in microbiologic antagonism, with the result that the combination may have reduced activity when compared with the most active single agent of the treatment regimen. In vitro, antagonism can be demonstrated with assays for killing of bacteria in broth culture. For example, tested alone in broth culture, oxacillin exerts a bactericidal effect against (methicillin-susceptible) S. aureus, usually defined as ≥3 log10 CFU/mL killing over a 24-hour period of incubation. In such assays, vancomycin alone tends to kill organisms more slowly. When the drugs are combined, the result may more closely resemble the slower killing seen with vancomycin. Time-kill curves illustrating this antagonism are shown in Figure 301-1B . In this example, the more slowly bactericidal drug, vancomycin, antagonizes the killing effect of the intrinsically more bactericidal drug, oxacillin. Antagonistic interactions against S. aureus between less bactericidal (linezolid) and more bactericidal (vancomycin) antimicrobials have also been demonstrated in vivo in experimental endocarditis.

The D-zone test for detection of inducible resistance to clindamycin, described earlier, illustrates in vitro antagonism as measured by a reduction of the inhibitory activity of clindamycin in the presence of inducing concentrations of erythromycin. In vitro antagonism can also be demonstrated when certain β-lactams are tested in combination against gram-negative bacteria with inducible β-lactamases. Here, exposure to one β-lactam can de-repress the synthesis of inducible β-lactamases, which then degrade the second antibiotic.

The most striking example of clinically important antagonism emerged from studies on the treatment of bacterial meningitis. In 1951, Lepper and Dowling reported pneumococcal meningitis mortality rates of 21% in patients treated with penicillin alone. For patients who also received chlortetracycline, the mortality rate was greatly increased at 79%. In this case, inhibition of bacterial growth by the bacteriostatic agent (chlortetracycline) is believed to have compromised the bactericidal activity of penicillin, which is greatest against actively growing bacteria.

Nevertheless, it is uncommon to encounter clinically apparent antagonism between antibiotics in the patient care setting, in part because offending combinations such as the ones described in the preceding paragraphs are not very likely to be used in routine clinical care today. However, if in desperation unusual antimicrobial combinations are used increasingly against isolates exhibiting multiple drug resistance, it is possible that clinically relevant antagonism will be encountered more often in the future. Antagonism of bactericidal activities may also be difficult to detect in clinical practice because most common infections (with the exception of endocarditis and meningitis) do not unequivocally benefit from bactericidal therapy. As long as one agent maintains inhibitory activity, it is unlikely that failure resulting from antagonism will be observed.