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 study design was a cohort review of Ohio Medicaid medical and prescription claims for children with a new diagnosis of OM. The study population included all children who were 7 years of age or younger at the beginning of the period and continuously enrolled in the Medicaid fee-for-service programme for 15 months. We identified the first episode of OM per child occurring during a 12-month study period (1 August 1997 to 31 July 1998). Acute OM cases were identified through diagnosis codes (International Classification of Diseases Clinical Manual, 9th edition or ICD-9CM codes) recorded in medical claims. Acute OM diagnoses included were: 381.0, 381.00-381.03, 382.00, 382.01.[14] The 381.xx codes reflected nonsuppurative cases (including OM with effusion), and the 382.xx codes indicated suppurative cases (necrotising and purulent cases). Since cases of OM may present with concurrent infections, we screened medical claims for the presence of other upper respiratory infections within a 3-day window on either side of the date of the OM diagnosis: sinusitis, pharyngitis, tonsillitis, laryngitis, bronchitis and pneumonia (461-466.xx, 481.xx, 485.xx, 486.xx, 490.xx). In addition, we identified which children had a concurrent diagnosis of chronic OM (381.1x-381.9x; 382.1x-382.9x).

Children who had diagnoses of acute OM in the 3 months preceding the study period (1 May 1997 to 31 July 31 1997) were excluded to eliminate children with ongoing treatment. Claims data for cases were followed for 30 days after the diagnosis to identify antibiotic use and retreatment, as described in the next section. Since the cohort included children continuously enrolled for the 3 months prior to the study period, the minimum age of the final cohort was 3 months. Age, gender and race were determined from recipient eligibility files. Race was grouped as Caucasian, African-American, Hispanic or other.

Primary outcome measures included the initial antibiotic treatment rate, the likelihood of receiving amoxicillin as initial treatment, and treatment failures (retreatment within 30 days). Treatment was defined as an antibiotic prescription dispensed within the 24 hours before or 48 hours after diagnosis of OM by matching medical and prescription claims.[7] Antibiotics were sorted into six categories:[1] amoxicillin (amoxicillin, penicillin, ampicillin),[2] amoxicillin plus clavulanic acid,[3] new macrolides (azithromycin, clarithromycin),[4] sulfa and sulfa combinations [cotrimoxazole (sulfamethoxazole with trimethoprim), sulfafurazole (sulfisoxazole), erythromycin ethylsuccinate with sulfisoxazole (EES)],[5] second and third generation cephalosporins (cefaclor, cefpodoxime, cefuroxime, loracarbef, cefprozil, cefixime, ceftibuten), and[6] erythromycin (erythromycin estolate, erythromycin ethylsuccinate) plus the first generation cephalosporins (cefadroxil, cefalexin). Although the amoxicillin group included ampicillin and penicillin, these two agents accounted for less than 0.5% of the prescriptions in the group. Therefore, the amoxicillin group primarily reflected the use of amoxicillin. In the latter group, erythromycin and first generation cephalosporins were grouped together since neither is indicated for the treatment of OM, and there was extremely low use of these agents. We operationalised retreatment as the presence of a prescription claim for a second antibiotic prescription dispensed between day 3 and day 30 following diagnosis of OM. Prescriptions dispensed between the initial antibiotic and day 3 were not examined.

Analyses included descriptive statistics and multivariate logistic regressions. We compared the study population with the eligible Medicaid population on the basis of race, gender and age. We used logistic functions to model the probability that a case was treated empirically with an antibiotic, and for cases that were treated, the probability that amoxicillin was prescribed. The final model examined the probability that the OM cases were retreated. Predictors examined included age, race, gender, suppurative diagnosis, chronic OM and concurrent upper respiratory infections (URI). Provider specialty was added to the model predicting the use of amoxicillin. Prescriber specialties included paediatrics, general practice and internal medicine (including family medicine), emergency medicine, and other (specialties with <2% frequency). Prescriber specialty was self-reported as part of the Medicaid provider profile.

The study protocol was ruled exempt from review by the University of Cincinnati Institutional Review Board since the project involved analysis of secondary claims data that had identifying information removed.