Prescribing Patterns and Retreatment Rates in Patients With Otitis Media

T.I. Shireman and K.A. Kelsey

Clin Drug Invest. 2002;22(5) 

In This Article


The purposes of this study were to explore factors leading to the treatment of OM and antibiotic utilisation patterns, and to determine the effect of empiric antibiotic selection on short-term retreatment rates in a paediatric Medicaid population. The decision to use antibiotics in the treatment of OM was primarily influenced by the specific type of diagnosis. Suppurative OM was twice as likely to be treated with antibiotics compared with nonsuppurative OM. Children with a diagnosis of chronic OM, however, were less likely to be treated. Antibiotic use was higher in children with a concurrent bacterial or viral respiratory infection (sinusitis, pharyngitis, tonsillitis, bronchitis or pneumonia).

Amoxicillin was used as a first-line agent in 40% of all cases of OM and accounted for 52% of all antibiotic prescriptions. Our rate of amoxicillin use was almost identical to a Tennessee Medicaid analysis conducted using 1993 data (52 vs 53%).[6] Both our study and the Tennessee analysis had lower rates of amoxicillin use than a Colorado Medicaid study based upon 1991-1992 data (65.7%).[7] Our overall treatment rate, 77.5%, was also higher than the 65% reported in the Colorado study,[7] reinforcing the notion of a trend toward greater empirical use of antibiotics for OM.[15] Results from a Canadian study indicated that 80% of children under 5 years of age with a diagnosis of acute OM were treated with an antibiotic in 1995.[2]

Children with chronic OM or a concurrent respiratory infection were less likely to receive amoxicillin. In turn, children who received amoxicillin were less likely to receive a second antibiotic regardless of the type of OM diagnosis or concurrent infections. Sulfa and sulfa combinations also had a lower rate of retreatment. Combined, these results suggest that amoxicillin and sulfa drugs, typical first-line agents, were appropriately and successfully used in the treatment of children with OM. Cases complicated by other infections were less likely to receive amoxicillin but their rates of retreatment were no different when the diagnosis variable was considered. This may suggest that they needed expanded coverage for other pathogens. Retreatment was not substantially or significantly different between the other antibiotics.

There was little difference between paediatricians, general practitioners/internal medicine physicians, and emergency room physicians in the prescribing of amoxicillin. These results stand in contrast to a survey of paediatric and family practice physicians where family medicine physicians were 1.5 times as likely to use a second-or third-line agent than paediatricians.[5] However, the latter study focused on the management of persistent and recurrent OM cases. Other physician specialties in our study were far less likely to use amoxicillin than paediatricians.

Our results regarding race were consistent with other studies.[4,6,7,8,16] African-American children were 20% more likely to be empirically treated with an antibiotic and 70% more likely to receive amoxicillin than Caucasian children. In turn, they actually had better outcomes, e.g. lower retreatment rates. Hispanic children were 50% more likely to receive amoxicillin, but the wide confidence interval for the odds ratio suggested this may not be a consistent result. Their outcomes were not different from Caucasians.

There were several limitations to our study. The first concerns the accuracy of the ICD-9 diagnostic codes. We were unable to validate the accuracy of the diagnostic codes for either the OM or concurrent infection codes because we were using deidentified claims data. However, our rate of new onset cases of either type of OM (11.5%) for children under 7 years of age was consistent with other published studies. For instance, Berman and colleagues reported rates of new onset for children aged less than 30 months of 14.2% in 1991 and 16.5% in 1992.[7] Considering that the incidence of OM declines with age, our rate may have been slightly lower since we included older children. In addition, 14% of the children had a concurrent infection that may have influenced antibiotic selection.

Treatment failures can be classified as early treatment failure, failure late in therapy, or end-of-therapy failure.[15] Early treatment failure occurs when symptoms persist 48 to 72 hours after initial treatment and may require a change in antibiotic. Failure late in therapy typically indicates poor compliance or a secondary viral infection, and end-of-therapy failure occurs when the symptoms persist despite completion of the antibiotic course.[15] Our outcome variable, retreatment, would have excluded persistent infections that did not receive a second course of antibiotic and may have overestimated the treatment success for any given agent. In addition, we would have attributed changes in therapy for legitimate concerns (e.g. allergic responses or resistance to initial antibiotic), to failure of the initial treatment option.

Prescriber specialty codes were self-reported to Ohio Medicaid by the providers and may not have reflected actual board certification in a specialty. Specialty codes were also only available on 81% of the prescriptions. This was consistent with published results from the Tennessee Medicaid acute OM study (which noted specialties in 80% of their cases), but higher than the Colorado Medicaid study (which noted specialties in about 65% of their cases).[6,7]

Prescription claims capture what was dispensed by a pharmacy. It is inappropriate to conclude that these claims fully depict prescribing practices and patient behaviour. Patients may have received samples, may not have filled a written prescription, may not have adhered to the prescribed regimen, or may have been infected with a resistant pathogen not covered by the empiric therapy. Each of these events could have reduced the reliability of our results. Adverse effects and poor patient acceptance of antibiotics could also have contributed substantially to antibiotic selection and short-term outcomes.[17,18]