Use and overuse of antibiotics is associated with the development and spread of resistant bacteria 1; a problem continuing to gain attention from national organizations as a significant threat to the public health 2. It is important to avoid unnecessary pediatric antibiotic use because children represent a large reservoir for resistant organisms 3. Acute otitis media (AOM) accounts for up to 40% to 50% of antibiotics prescribed for United States children less than 2 years of age 4. Encouragingly, overall pediatric antibiotic prescribing has declined recently 56, and visits for AOM have decreased, apparently in conjunction with implementation of the pneumococcal conjugate vaccine (PCV)47, although PCV only has a marginal effect on decreasing AOM episodes for an individual child 8. Visit-rate prescribing of antibiotics for AOM, however, has not decreased 6.

Antibiotic treatment is recommended for most children with AOM in the United States 910 with therapy targeted primarily at Streptococcus pneumoniae, the most frequent, clinical pathogen 1112. Amoxicillin remains the recommended first-line therapy for uncomplicated cases 911, but as resistant strains of Streptococcus pneumoniae have increased considerably in the past 20 years 131415 there is concern of increased prescribing of broad-spectrum agents. A counteracting factor to this concern is the effectiveness of PCV in reducing resistant strains of Streptococcus pneumoniae 1617. By providing reassurance that common pathogens are still sensitive to amoxicillin, PCV may mitigate antibiotic choices for broader coverage. It is important for the Centers for Disease Control and Prevention (CDC) and other national organizations involved in curtailing antibiotic resistance to gain a better understanding on how physicians have responded to these influences in their prescribing choices for AOM.

To measure changes in the rate of broad-spectrum and other types of antibiotics prescribed to children seen with a diagnosis of AOM, data was analyzed from the National Ambulatory Medical Care Survey (NAMCS) and National Hospital Ambulatory Medical Care Survey (NHAMCS) from 1998 to 2004. The NAMCS and NHAMCS are the only surveys of outpatient settings in the United States that collect prescribing information and produce unbiased national estimates. These data allow for a comprehensive assessment of the antibiotic prescribing for children with AOM on a national level over time.

Although amoxicillin remained the recommended antibiotic for AOM and PCV had become widely implemented during the study period, we found that among children receiving antibiotics, broad-spectrum prescribing steadily increased from 34% of visits in 1998 to 45% of visits in 2004, a 32% increase over the 7 year interval. An 88% increase in amoxicillin/clavulanate prescribing and a 67% increase in macrolide prescribing were largely responsible for the progressive trend in broad-spectrum usage. So although fewer children were being diagnosed with AOM, those that presented for care and received an antibiotic prescription had a high likelihood of receiving a broad-spectrum agent by 2004.

Other studies have looked at the types of antibiotics prescribed for AOM. Broad-spectrum antimicrobials were used to treat 28% of uncomplicated new AOM infections in a sample of visits to community pediatricians in 1999–2000 22. More recently, it was found that for children with AOM less than 2 years of age, 66% received penicillins, 21% cephalosporins, and 11% macrolides including erythromycin 4. Amoxicillin/clavulanate prescriptions were not reported separately. Curiously, in that study, antibiotics were prescribed in only 58% of visits related to AOM. Nonetheless, it is difficult to make comparisons between these results and our analysis which is derived from a nationally representative sample of visits over a 7 year period. The increased prescribing of macrolides is consistent with another study 23 and is noteworthy because of increased streptococcus pneumoniae resistance to macrolides 24.

We had hypothesized that implementation of PCV after 2000 might have provided physicians with reassurance in the continued effectiveness of amoxicillin for AOM and thus lessened their tendency to prescribe broad-spectrum agents. PCV contains polysaccharides for the five serotypes of Streptococcus pneumoniae that account for most of the antibiotic resistance among pneumococci including high-level resistance to penicillin, macrolide resistance, and multidrug resistance 25. Since its introduction, implementation of PCV has become widespread; coverage (≥ 3 doses) has increased from 41% of children aged 19 – 35 months in 2002 to 73% in 2004 26. Postlicensure studies have shown a decline in the prevalence of serotypes with resistance to penicillin 27282930, a reduction in treatment failures, and a reduction in frequent episodes and tube procedures 31. Other studies have suggested that standard dose, rather than high dose, amoxicillin may once again be sufficient as first-line therapy in children who have received ≥ 3 doses of PCV 3233. And in France, a country with a high prevalence of antibiotic-resistant pneumococci, implementation of PCV, combined with a reduction in antibiotic use, decreased the carriage of penicillin nonsusceptible pneumococci in children with AOM 34. One randomized trial, however, found no reduction in recurrent AOM episodes in the vaccine group 35 and a Cochrane review estimated that AOM episodes would only decrease by 6% to 7% 8.

From our results it does not appear that PCV has affected prescribing choices. Children less than 2 years of age, for instance, the group most likely to have been administered the complete schedule of PCV doses during the study period, were just as likely as older children to have received broad-spectrum antibiotics. What may be particularly concerning about these prescribing trends, in fact, is their concurrence with PCV implementation. One author has posed that resistant isolates of Streptococcus pneumoniae may reflect an adaptive response of nonvaccine serotypes to the widespread use of broad-spectrum antibiotics 36. This may represent the type of adaptive response that has contributed to the emergence of multiresistant serotypes 36. So, rather than providing reassurance to clinicians to continue treating AOM with amoxicillin, PCV, in combination with broad-spectrum prescribing, may actually be a contributor to antibiotic resistance. Consistent with a Swedish study which showed that more children with AOM are probably being treated at home without antibiotics 37, we also found that visits for children with AOM have decreased in recent years. In the US this trend has been attributed to the national PCV program 47.

We also noted that as broad-spectrum antibiotic prescribing increased, amoxicillin prescribing remain stable, despite the operant guideline for this time period recommending high dose amoxicillin as first-line therapy for most children 11. Of note, the study period ended before the most recent guidelines on AOM 9, recommending amoxicillin/clavulanate as the initial choice for children with severe infection, would have had a significant impact on amoxicillin prescribing. Encouragingly, we found a significant decrease in the proportion of patients receiving narrow-spectrum antibiotics other than amoxicillin from 1998 to 2004. These antimicrobials have been shown to be less effective options for AOM and are no longer recommended therapy in the United States 3839.

We found that broad-spectrum prescribing was less likely in those children with ear pain and was not associated with age or fever; factors characteristic of more severe infections that might have justified broader coverage 4041. Instead, significant associations were for the nonclinical factors of race, insurance status, geographical location, and ambulatory setting. Over-prescribing of antibiotics to pediatric white patients has been reported in another study 42. High antibiotic prescribing rates have been found in the South in other studies 4344. And lower rates of nonrecommended (broad-spectrum) antibiotics for children with sore throats seen in emergency departments have been reported as well 42. In contrast, general practitioners in the Netherlands were more likely to overprescribe antibiotics, according to Dutch guidelines, to younger children who were more severely ill 45. Unfortunately, due to limitations in the data, we were unable to explore associations with other risk factors linked to the presence of bacterial species likely to be resistant to amoxicillin. These include attendance at child care and recent prescription (less than 30 days) of an antibiotic 4647. However, there is no reason to believe that these factors would have changed significantly over the 7 year period.

There were some aspects of the data that may limit the conclusions that can be drawn from our results. First, the NAMCS and NHAMCS survey format does not allow for direct linkage between diagnosis and medication. Even though we attempted to use a rigorous method for linking the antibiotic prescribed during a visit to an AOM diagnosis, it is possible that the antibiotics were actually prescribed for a different condition than was addressed in the visit. Second, as mentioned previously, multiple visits by the same individual are not accounted for because individual identifiers are not coded in NAMCS and NHAMCS. Lastly, we were unable to determine if high dose amoxicillin (80 to 90 mg/kg/day) was prescribed to children who received an amoxicillin prescription. High dose amoxicillin was recommended as first line treatment for AOM by a CDC consensus panel in 1999 due to concerns of drug-resistant Streptococcus pneumoniae 11. It is possible that the rate of prescribing of high dose amoxicillin, in keeping with these guidelines, increased despite the overall stable trend in amoxicillin prescribing.

The CDC has championed national efforts to address antibiotic overuse through education of healthcare providers and the public 48. And although overall pediatric antibiotic prescribing rates 5 and visits for AOM have decreased 47, it is concerning that prescribing for AOM, now known to be coupled with a high likelihood of a broad-spectrum antibiotic prescription, has not 6. Perhaps the best means of avoiding unnecessary broad-spectrum antibiotics for AOM is to limit those children receiving any antibiotic prescription. In this regard, we support the national AOM practice guidelines that allow for the option of observation as the first-line treatment of uncomplicated AOM in children with mild disease 9. These guidelines have been further supported by recent, randomized, controlled trials in multiple settings as well as several older metaanalyses 4950515253. Adherence to the guidelines would not only directly decrease antibiotic prescriptions for broad-spectrum and other agents, but perhaps, also indirectly, as more parents came to learn that a doctor's visit is not always necessary when their child's symptoms are mild.

In conclusion, the prescribing of broad-spectrum antibiotics for AOM has increased dramatically from 1998 to 2004 and the trend occurred in the midst of the implementation of a national immunization program effective in decreasing the severity and incidence of this common pediatric infection. Physicians have increased their prescribing of broad-spectrum antibiotics for AOM without correlation to clinical factors such as age or severity of illness. If primary care physicians decide antibiotics are needed for AOM, prescribing amoxicillin as initial therapy is recommended for most children, especially with the reassurance of its effectiveness provided by PCV.

The authors declare that they have no competing interests.

AC conceived and designed the study and participated in the analysis and writing. MH participated in the design and analysis. AG coordinated and participated in the analysis and writing of the study. All authors read and approved the final manuscript.