Treatment of Otitis Media With Observation and a Safety-Net Antibiotic Prescription

METHODS

RESULTS

A total of 194 patients were enrolled at 11 practice sites. A total of 100% of those eligible were approached, and only 5 patients declined to participate. Ninety percent (n = 175) completed the follow-up interview. Five (33%) of 15 nonwhite children were lost to follow-up, compared with 14 (7.9%) of 178 white children (χ² = 11.45; P < .001; race was missing for 1 child). The proportion of children followed up did not differ by gender. The age distribution of children lost to follow-up was slightly younger than those not lost, but the difference was not statistically significant. Among the 175 who completed the study, the average age of the children was 5.0 years (range: 1–12 years). Forty-four percent were girls, and 56% were boys. Ninety-four percent of parents described their children as white, 3% as black, and 3% as other. Seventy-one percent of families had private insurance, 27% had Medicaid, and 2% were self-pay. Table 3 shows the pain medications and antibiotics prescribed as a SNAP.

A total of 120 (69%) of 175 families did not fill the antibiotic prescription (95% CI: 61.7%–75.5%). Of these 120 parents, 117 (97.4%) said that they were willing to use pain medication without antibiotics in the future (95% CI: 94.4%–100%). Of the 55 families who did fill the prescription, 33 filled the prescription within 48 hours of diagnosis. Parents’ reasons for filling the prescriptions, based on the structured interview, are described in Table 4. (Note that the responses are not mutually exclusive.)

Of the 175 children who completed the study, 161 had had at least 1 previous episode of AOM. For 2 of the subjects, data on previous episodes were missing. For the 159 remaining, all of the parents of these children reported being given antibiotics in the past. Of these 159 children, 155 (97%) had used antibiotics during their last AOM episode, whereas only 52 (33%) used antibiotics during this episode (McNemar χ² = 101.00; P < .0001).

In trying to understand further which factors predicted the parents’ behavior, we examined a number of demographic variables. Child’s age, income, insurance status, parental education, and diagnostic criterion did not predict whether parents did not fill the prescription (Table 5). In addition, practitioner site did not predict parent behavior (data not shown). Previous episodes of AOM was the only variable that we analyzed that seemed to explain parents’ behavior. When the child had had 2 or more previous episodes of AOM, parents were significantly more likely to fill the prescription than parents of a child who had 1 or no previous episodes (83.9% vs 65.3%; χ² = 4.09; P = .04).

There were no complications reported as a result of the study. The course of 1 patient, however, is worth noting. A 16-month-old boy was diagnosed as having AOM and given the SNAP. The amoxicillin prescription was filled at 48 hours when his symptoms did not resolve, but he did improve after 48 hours of antibiotic therapy. He was seen 6 weeks later with AOM in the opposite ear, treated with antibiotics, and had postauricular swelling suggestive of early mastoiditis. He received a diagnosis of postauricular cellulites by our hospital otolaryngology team, responded to intravenous antibiotics, and had no additional difficulties.

DISCUSSION

Antibiotic resistance has become an increasing clinical problem over the past decade. Duchin et al¹⁰ described that more than half of Streptococcus pneumoniae cultured from nasopharyngeal swabs of children who attended child care in 1 community were penicillin-resistant. Looking specifically at AOM, Block et al¹⁹ demonstrated that the pneumococcal isolates from middle-ear fluids were 16% relatively resistant and 15% highly resistant to penicillin. With growing resistance in mind, several authorities have suggested guidelines for more judicious use of antibiotics.^1,20

Although studies have shown that there is little if any benefit in treating AOM, it is not clear whether a strategy of watchful waiting is practical in the United States, where antibiotics traditionally have been used for this infection. Watson et al,¹⁵ in a survey of 366 physicians, showed that 97% recognized that overuse of antibiotics contributed to resistance. Still, 46% of these physicians were prescribing antibiotics for the common cold. Parents also have conflicting concerns over the use of antibiotics. Palmer and Bauchner¹³ showed that the vast majority of parents (85%) believed that there were problems with antibiotic overuse, but 93% thought that antibiotics were necessary for the treatment of AOM. Adding to the pressure on practitioners to prescribe antibiotics is the concern that the child may legitimately need antibiotics if the infection does not respond to 48 hours of watchful waiting. This may lead to an additional office visit and add to the expense and inconvenience of the infection.

Any successful treatment strategy for AOM must take into account that there is a hesitancy of both physicians and parents not to have antibiotics available for this infection. The SNAP pioneered by Cates is an attractive method in AOM treatment, as it gives both practitioners and parents the security of having antibiotics available if the child’s infection does not respond to watchful waiting. In our study, the majority of parents did not fill the SNAP and reported that they would be willing to treat AOM without antibiotics in the future. The majority of parents also believed that their children had adequate pain control, and there was a significant lowering of antibiotic use compared with previous episodes as reported by parents. Also, no significant complications were reported in those who were treated by observation alone or those who went on to fill their SNAP.

The most common major complication of AOM is progression to mastoiditis.²¹ Although none of the episodes of AOM in the study progressed to mastoiditis, 1 child who was enrolled in the study developed what may have been mastoiditis 6 weeks after treatment in the opposite ear. Whether children who are treated initially for AOM are at lower risk for mastoiditis is not clear.²² Historically, approximately half of children who develop mastoiditis do so on antibiotics. Although uncommon, the incidence of mastoiditis is approximately twice as high in countries where practitioners treat AOM with observation compared with countries in which antibiotics are used initially, such as the United States.²³ The incidence of mastoiditis, however, is increasing in the United States and may be related to the increasing frequency in antibiotic resistance in common AOM pathogens.^24–26 Use of the SNAP could potentially reduce the risk of the development of mastoiditis compared with watchful waiting alone, as antibiotics are readily available if a child’s condition worsens or does not improve after an adequate observation period. A larger study with a longer follow-up is needed to determine this potential added benefit.

Potential concerns with our study are that the diagnosis of AOM was a clinical one and that AOM may have been overdiagnosed. Children who did not actually have AOM would presumably respond to watchful waiting. If more selective criteria were used, such as a bacteriologic diagnosis by tympanocentesis, then the acceptance of the SNAP and antibiotic usage may have been altered. In our study, we chose a clinical definition of AOM to reflect how clinicians make the diagnosis in real practice settings. All children who received the SNAP in our study would have received antibiotics based on the practitioners’ past performance. We believe that the acceptance of the SNAP and decreased antibiotic use in a real-practice setting demonstrate the effectiveness of the SNAP approach.

There are several other limitations to our study. Our sample size, as in most AOM studies, was small. The population, although having a broad socioeconomic spread, was lacking in minority enrollment. It certainly is not clear whether these results can be generalized to other settings. It is also unclear how this select group of patients compares with all children with AOM and whether the results can be extended to children with more severe disease. Another concern is that the criteria that we used for the diagnosis of AOM differ from other proposed criteria. Finally, the follow-up period was relatively brief, and it is uncertain how many children ultimately will go on to receive antibiotics shortly after the acute episode. We are planning a larger study with a broader range of disease and a longer follow-up period to settle some of these issues.

CONCLUSIONS

Our results suggest that the use of a SNAP can safely reduce the use of antibiotics in children with uncomplicated AOM and that a population of parents in the United States find this strategy acceptable. Antibiotic prescribing strategies, such as the SNAP, may help alter the trend of antibiotic overprescribing and development of resistant organisms occurring in the United States.