We attempted to determine whether antimicrobial prescription patterns were associated with resistance patterns among S. pneumoniae isolates from young children with acute otitis media in 2 populations treated by the same medical system, but living in separate communities with different lifestyles. We used prescription rates of ≈20% of the children <5 years of age in each population and a large collection of middle ear fluid S. pneumoniae isolates obtained from these 2 populations during the same time period. This unique opportunity enabled us to not only relate resistance dynamics to prescriptions dynamics for each population, but also to observe whether differences in antimicrobial prescription rates between populations could explain some of the differences in antimicrobial resistance patterns.
The association of azithromycin prescriptions with antimicrobial resistance patterns among S. pneumoniae isolated from middle ear fluid is noteworthy for several reasons. First, the prescription rate for azithromycin was the only one that was higher among Jewish children than among Bedouin children. Second, the azithromycin prescription rate pattern closely paralleled both macrolide and multidrug resistance in each population. Third, higher penicillin MIC values were associated with macrolide resistance, which explained, at least in part, the higher rates of S. pneumoniae isolates from middle ear fluid with a penicillin MIC ≥1.0 μg/mL among Jewish children who received relatively fewer amoxicillin (with or without clavulanate) prescriptions than Bedouin children who received more amoxicillin prescriptions (with or without clavulanate).
The distinct pattern of reducing total antimicrobial drug prescriptions that resulted from reducing prescribed penicillins, although azithromycin use increased, was reported in other regions, including the United States (21
) and western Europe (23
). The pattern of reduced antimicrobial drug use could be the result of campaigns such as those conducted in the United States following the initiative by the Centers for Disease Control and Prevention for the judicious use of antimicrobial drugs, which recommended the first-line use of amoxicillin to treat acute otitis media (21
). However, the increase in azithromycin prescriptions, along with a reduction in penicillin prescriptions, could be partly the result of commercial promotion campaigns for the use of azithromycin, which were launched in parallel with campaigns to reduce the overall use of antimicrobial agents.
The increase in the azithromycin prescription rate in our study, as in Europe and North America, is partly attributable to the properties that make this drug an attractive agent for children. The long half-life of azithromycin (≤72 h) (24
) makes a convenient dose regimen of once a day for ≤5 days. However, it is eliminated very slowly and remains at subinhibitory concentrations in tissues of persons with pneumococcal infections. Subinhibitory concentrations of antimicrobial agents favor the selection of resistant mutants. This has been shown in vitro for S. pneumoniae
exposed to subinhibitory macrolide concentrations (25
). In a clinical trial, children treated with azithromycin harbored significantly more resistant strains in their oral flora than those who randomly received other macrolides. After 6 weeks, 85% still had macrolide-resistant organisms (26
). In European countries, the increase in prescriptions of long-acting macrolides resulted in selection for macrolide resistance in S. pneumoniae
Our finding that azithromycin prescriptions were associated with S. pneumoniae
multidrug resistance was noteworthy. The ability of certain antimicrobial agents to promote resistance to other drug classes has been previously reported. Several studies showed that use of long-acting macrolides was an important factor in increasing penicillin resistance in a given community (29
). This could explain the higher rates of resistant S. pneumoniae
with MIC≥1.0 μg/mL in Jewish than in Bedouin children, despite significantly lower prescription rates for penicillins, but significantly higher prescription rates for azithromycin. This pattern of increasing penicillin and macrolide resistance in association with increased prescribing of azithromycin was also observed in the United States, where it was predicted that in the absence of pneumococcal conjugate vaccine, by 2004 ≈40% of all S. pneumoniae
isolates would be resistant to both penicillin and macrolides and that the increased rate would be exponential (31
The differential effect of azithromycin versus amoxicillin on nasopharyngeal carriage of antimicrobial-resistant S. pneumoniae
in patients was demonstrated in a study conducted in southern Israel. In this study, carriage of both macrolide- and multidrug-resistant S. pneumoniae
markedly decreased in children with acute otitis media receiving amoxicillin-clavulanate, but increased markedly in those receiving azithromycin (32
). This differential effect lasted more than 1 month.
Our study has 3 limitations. First, the factors contributing to differences in antimicrobial drug use between the 2 populations could not be controlled. Antimicrobial drug prescriptions could not be matched with individual use. In addition, potential confounders such as family structure and daycare exposure could not be assessed. The higher prescription rate for antimicrobial agents in Bedouin children could be explained by differences in accessibility to healthcare facilities. However, since there is no financial burden for healthcare in Israel and all clinics belong to the same health plan, acute otitis media is unlikely to be treated differently in either population. The similar reduction in antimicrobial drug prescriptions in both populations suggests that no difference in prescribing policies existed between the 2 populations. Lower socioeconomic status and overcrowding in the Bedouin population, which led to a higher rate of respiratory illness in this group (16
), may explain the difference in rates of antimicrobial drug prescriptions.
Second, S. pneumoniae were obtained only from children with acute otitis media. However, S. pneumoniae is part of the normal nasopharyngeal flora and is exposed to antimicrobial agents regardless of the diagnosis for which the agent was prescribed. Therefore, we believe that the effect of prescribing antimicrobial drugs in the community on resistance patterns of S. pneumoniae isolated from middle ear fluid represents the effect on the entire spectrum of S. pneumoniae disease.
Third, this was not an intervention study; therefore, we could not demonstrate unequivocally the causative effect of azithromycin use on macrolide and multidrug resistance in S. pneumoniae. However, the association demonstrated in this study, together with published data, strongly suggest such a causative effect.
The introduction of the 7-valent pneumococcal conjugate vaccine to infant and toddler immunization programs in the United States was associated with a reduction in invasive diseases (33
) and acute otitis media (35
) caused by antibiotic-resistant S. pneumoniae
. However, persistence of antimicrobial resistance within vaccine and nonvaccine serotypes (37
) suggests that vaccine alone may not reduce antimicrobial resistance, and that if the use of antimicrobial drugs is not controlled, the ability of the pneumococcal conjugate vaccine to reduce antimicrobial-resistant S. pneumoniae
may be only transient.
During the last 2 years of this study, prescription rates were reduced in both populations. This reduction could be partly explained by efforts of pediatric infectious diseases specialists to educate pediatricians and family physicians in the study area to reduce use of antimicrobial drugs, especially oral use of cephalosporins and macrolides-azalides. However, this effect may not be the main reason for this decrease. A decrease in macrolide and multidrug resistance of S. pneumoniae observed in the last year of this study may indicate that the effect of azithromycin use on antimicrobial resistance is reversible. Continuous monitoring of antimicrobial prescriptions and resistance in respiratory pathogens should help determine if a further decrease in azithromycin prescriptions would be followed by a further decrease in antimicrobial resistance of S. pneumoniae.
In conclusion, azithromycin prescriptions were associated with macrolide, penicillin, and multidrug resistance among S. pneumoniae isolated from the middle ear fluid of children with acute otitis media. Such an association was not found with amoxicillin (with or without clavulanate) prescriptions. When promoting judicious use of antimicrobial drugs, selective reduction in prescribing specific antimicrobial drugs such as azithromycin should be emphasized, in addition to total reduction in antimicrobial use.