Background. Treatment of otitis media is the most frequent reason for administering antibiotics to children in the United States. However, only limited data are available on medical effectiveness of antibiotic prescribing patterns for otitis media and their associated expenditures or the factors that influence antibiotic prescribing.
Methods. The study population consisted of 131 169 children during 1991 and 157 065 children during 1992 who were ≤13 years of age and enrolled in Colorado's fee-for-service Medicaid program. Among these children, 5127 (1991) and 7254 (1992) were enrolled in the cohort treated for a “new” episode of acute otitis media. An analysis using this cohort was performed to document the antibiotics used to treat a new episode of acute otitis media, factors influencing antibiotic selection, and the short-term outcomes of therapy. An analysis using the entire Medicaid population was performed to document the annual use of antibiotics for otitis, the associated antibiotic expenditures, and factors influencing antibiotic selection.
Results. In the cohort analysis, office-based physicians prescribed second- and third-generation cephalosporins more often than did physicians in other settings (17% vs 9.7% and 11.8%), whereas hospital clinics prescribed trimethoprim plus sulfamethoxazole more frequently than did office-based physicians (19.2% vs 7.1% and 10.9%). Family physicians prescribed second- and third-generation cephalosporins more often than did pediatricians (16.6% vs 12.3%) but trimethoprim plus sulfamethoxazole and erythromycin plus sulfisoxazole less often than did pediatricians (10.5% vs 17%).
The average rate of prescribing a second course of antibiotics within 24 days after initial antibiotic treatment of a new acute otitis media episode was 11.6% when less expensive antibiotics (amoxicillin, trimethoprim plus sulfamethoxazole, or erythromycin plus sulfisoxazole) were prescribed, and 13.2% when more expensive antibiotics (cefaclor, amoxicillin plus clavulanate, or cefixime) were prescribed. The average adverse drug reaction rate was 5.9% when less expensive antibiotics were prescribed, compared with 6.1% when more expensive antibiotics were prescribed.
In each of the two study years, amoxicillin accounted for almost half of the total antibiotic fills but only 9% to 10% of the expenditures. Low-cost antibiotics (amoxicillin, trimethoprim plus sulfamethoxazole, and erythromycin plus sulfisoxazole) were prescribed for 66% to 67% of the total fills and accounted for 21% of the total projected expenditures. More expensive antibiotics (cefaclor, cefixime, amoxicillin plus clavulanate) prescribed for 30% of the fills generated 76% to 77% of expenditures. Cefaclor, prescribed for 17% to 18% of the total fills, generated 43% to 45% of total antibiotic expenses.
Conclusions. The findings of this study document a preference for amoxicillin as the initial antibiotic for a new episode of acute otitis media. Although there was a wide variation in the selection of antibiotics to treat otitis, the more expensive antibiotics were not associated with better outcomes. This wide variation has important financial implications because of differences in antibiotic costs. Changes in prescribing patterns among initially uncomplicated children that reduce the use of high-cost antibiotics could reduce expenditures substantially without compromising short-term outcomes.
Treatment of otitis media is the most frequent reason for administering antibiotics to children in the United States. Understanding the factors that influence which antibiotics are prescribed for otitis media has important financial and public health considerations. Prescribing patterns impact medical expenditures because of the large variability in the costs of different antibiotics used to treat otitis media. Clinical trials of antibiotic treatment for acute otitis media fail to document the superiority of any antibiotic.1 Despite these financial considerations, no published studies document the patterns of otitis media-related antibiotic use and their relative short-term effectiveness for large pediatric populations. The major barrier to obtaining this information has been the limited availability to researchers of large data bases that link antibiotic use with diagnoses, outpatient and emergency department visits, and types of providers. We overcame this barrier by using the Colorado Medicaid data base, which links this data for individual patients and allows longitudinal analysis. This study describes antibiotic prescribing by physicians, factors associated with prescribing, the medical effectiveness of antibiotics used to treat a new episode of acute otitis media in uncomplicated patients, and annual Medicaid antibiotic expenditures for otitis media.
The study population consisted of 131 169 children during 1991 and 157 065 children during 1992 who were ≤13 years of age and enrolled in Colorado's fee-for-service Medicaid program during 1991 and 1992. Colorado Medicaid eligibility during this period conformed to mandatory federal requirements. During the period of the study, >95% of the state's Medicaid recipients obtained services in a fee-for-service environment.
Data obtained from an extract of the Medicaid Medical Events Database (MMED) included patient demographics; provider information; and claims paid by Medicaid for prescriptions, outpatient office and clinic visits, emergency department visits, and inpatient hospitalizations. A detailed description of MMED and the artificial intelligence software developed by the research team has been described in previous publications.2,3 The analyses followed individual patients over time, using the unique Medicaid recipient identification numbers that are maintained across interrupted periods of enrollment, name changes, and moves to different counties.
International Classification of Diseases, 9th revision, diagnostic codes were used to identify visits for otitis media. No attempt was made to use these codes to distinguish chronic from acute otitis designations, because documentation of physicians' coding practices suggests sufficient variability that renders such distinctions invalid.4 An antibiotic was considered to be prescribed to treat otitis media if dispensed up to 24 hours before or within 48 hours after a diagnosis of otitis media.5,6 Identification of the sites for the visits for otitis media (a hospital emergency department, a hospital outpatient department, a community clinic, an office, a federally qualified health center) was determined using “provider-type” and current procedural terminology code for all but emergency department visits, which were determined using only current procedural terminology codes. A site was listed as “indeterminate” if a recipient had more than one otitis-related visit on the same day with two different providers.
The patient demographic variables included age, sex, urban/rural residence, ethnicity, and enrollment category (Aid to Families with Dependent Children, etc). Residence referred to the first county of residence during the study year, with “urban” defined by the Census Bureau's Metropolitan Statistical Areas. The classification for ethnicity included Caucasian, Hispanic, African-American, Native American, and Other. We calculated days of enrollment for all children and assigned each child to one of two enrollment categories: continuously enrolled children who were either eligible for all of the study year or born during the study year and enrolled from birth to the end of the study year; all other children were considered to be discontinuously enrolled. When appropriate, analyses include only continuously enrolled children rather than the entire population. When study results for both study years are similar, they are combined in the same table.
Cohort Analysis of Treatment of a New Acute Otitis Media Episode
An analysis, using the cohort of 36 170 children during 1991 and 43 973 children during 1992 who were ≤30 months with a “new” episode of otitis, was performed. The index visit for this “new” episode was defined as an ambulatory visit with a diagnosis of otitis media and an antibiotic fill when in the previous 3 months otitis media was not diagnosed and an antibiotic was not prescribed unless there was a non-otitis diagnostic explanation. This means a child evaluated for inclusion in the cohort could have had an antibiotic in the previous 3 months associated with another diagnosis, such as impetigo, but would have been excluded by our criteria if this antibiotic prescription had been filled without an associated diagnosis. This criteria excluded most children with recurrent acute otitis or persistent middle ear effusion from the cohort who were receiving antibiotic prophylaxis but not having visits. Children with an appropriate index visit had to meet the following additional criteria for inclusion: age at the time of the index visit ≤30 months; subsequent 6 months of continuous Medicaid enrollment after the index visit; no hospitalization within 24 days of the index visit with diagnoses of sepsis, bronchiolitis, croup, meningitis, sinusitis, pneumonia, or mastoiditis; and no index visit with a secondary diagnosis of mastoiditis, hearing loss, cholesteatoma, or perforation. This cohort of children without associated infections or complications was used to document the antibiotics prescribed for a “new” episode of acute otitis media and short-term clinical outcomes.
New antibiotic fills within the first 24 days after the initial antibiotic treatment for the otitis at the index visit were documented. Reasons for prescribing a new antibiotic were classified as unresponsive acute otitis media (UAOM) or adverse drug reaction (ADR)/side effect. UAOM was defined as a change in antibiotic within 24 days of the index visit, accompanied by an ambulatory visit that had no diagnosis consistent with a drug reaction/side effect or diagnosis other than otitis. An ADR or side effect was defined as a change in antibiotic within 24 days either 1) accompanied by another visit and a diagnosis of an allergic reaction, urticaria, diarrhea, or emesis, or 2) without a visit. When a new antibiotic was prescribed by telephone without a visit, it was conservatively considered as an ADR/side effect, even though it was not possible to determine whether the new antibiotic was prescribed because of an adverse reaction or clinical unresponsiveness. A follow-up visit for otitis was defined as a visit occurring within 90 days of the index visit; the interval between the index visit and follow-up visits was categorized as <10 days, 11 to 24 days, 25 to 45 days, and 46 to 90 days. It was not possible to determine when an antibiotic administered during a follow-up visit after 24 days represented a new or continued infection.
Analysis of Entire Medicaid Population
An analysis using the entire Medicaid population of children ≤13 years of age was carried out to document the annual use of antibiotics for otitis and the associated antibiotic expenditures. We calculated age of recipients as of the end of the study year for this analysis. We estimated the expenditures for an antibiotic using the type of drug and age of the child, because the actual cost to Medicaid for any antibiotic was not included in the MMED data base. We assigned to antibiotics a cost based on published average wholesale prices or the federal maximum allowable cost for the specific formulation available on the Medicaid formulary during the study period (Table1). The method used to estimate the cost of a 10-day course took into account the age-appropriate formulation and the amount to be dispensed. We calculated the amount by using the recommended milligrams-per-kilogram dosage and the published weight for the 95th percentile of children at a given age. When more than one formulation of a drug was available within an age group, we chose the least expensive for our calculations.
Antibiotics Prescribed for a “New” Episode of Acute Otitis Media
Among children <30 months of age enrolled in Medicaid, 5127 of 36 170 (1991) and 7254 of 43 973 (1992) were enrolled in the cohort treated for a “new” episode of acute otitis media. Males comprised 53% of the cohort. Caucasians accounted for 54%, Hispanics 33%, African-American 11%, and Native American and Other 2%. A total of 66% of the cohort were Medicaid-eligible through Aid to Families with Dependent Children, 32% through the Medicaid expansion, and 1% through programs for foster care, blind/disabled, and other. A total of 81% of the cohort lived in urban areas and 19% in rural areas. The sociodemographic characteristics of the children <30 months of age with a new episode of otitis were similar to the characteristics of the entire Medicaid population <30 months of age.
A total of 581 physicians prescribed an antibiotic at the index visit, with 179 physicians treating more than five patients in the cohort. A pediatrician diagnosed the otitis at 32% of the index visits in 1991 and 33% in 1992. A family physician diagnosed the otitis in 34% of the index visits in 1991 and 30% in 1992. Otolaryngologists made the diagnosis in <1% of children during both years. The specialty of the clinician could not be determined in 34% of the 1991 index visits and 37% of the 1992 visits.
The total number of courses of antibiotics prescribed for children during the 6-month follow-up period after a “new” episode of acute otitis according to age is shown in Table2. Younger patients were more likely to have more antibiotic courses. For example, more than twice as many infants 3 to 6 months of age received three or more courses of antibiotics (52.6%), compared with children 25 to 30 months of age (22.3%) (P < .001). This difference could be attributable to younger infants having more unresponsive infections, more episodes of persistent middle ear effusion, and/or more recurrent episodes. During 1991, an antibiotic was prescribed at 38% of the first follow-up visits and at 51% to 71% of the other follow-up visits.
Although amoxicillin was the most frequent choice for treating a new otitis media episode, a cephalosporin was prescribed to at least one child at an index visit by 320 (55%) of the 581 physicians who cared for the cohort. Among the 179 physicians who followed more than five patients in the cohort, 83 (46.4%) prescribed a cephalosporin for ≥20% of the children treated at their index visit.
The antibiotic selections at the index visit are shown according to the type of setting (private office, hospital clinic, federally qualified health center) and type of physician (pediatrician, family physician, otolaryngologist, or indeterminate) in Tables3 and 4, respectively. Office-based physicians prescribed second- and third-generation cephalosporins more often than did physicians in other settings (17% vs 9.7% and 11.8%) (P < .0001), whereas hospital clinic personnel prescribed trimethoprim plus sulfamethoxazole (TMP/SMZ) more frequently than did office-based and emergency department physicians (19.2% vs 7.1% and 10.9%) (P < .0001). Family physicians prescribed second- and third-generation cephalosporins more often than did pediatricians (16.6% vs 12.3%) (P < .0001) but trimethoprim plus sulfamethoxazole and erythromycin plus sulfisoxazole less often than did pediatricians (10.5% vs 17%) (P < .0001).
The antibiotic fills prescribed at the index and subsequent visits are shown in Table 5. The use of amoxicillin decreased progressively during the subsequent otitis visits. During the first, second, and third follow-up visits, use of cefaclor and trimethoprim plus sulfamethoxazole showed the greatest increase. Use of amoxicillin plus clavulanate and cefixime increased most after the second follow-up visit.
Outcomes: UAOM and ADRs/Side Effects
The rates of unresponsive episodes and ADRs are shown in Table6 according to age and in Table7 according to the type of antibiotic prescribed at the initial index otitis media visit. The frequency of unresponsive episodes within 24 days of diagnosis had a range of 10.% for children 25 to 30 months of age to 13.6% for infants 7 to 12 months of age. This data suggest that different rates of unresponsive otitis do not account for the large difference in the number of antibiotic courses prescribed to younger infants. Therefore, younger patients probably had more persistent middle-ear effusions and/or more recurrent acute episodes.
The average rate of prescribing a second course of antibiotics within 24 days after initial antibiotic treatment of a new acute otitis media episode was 11.9% overall. The rate was 11.6% when less expensive antibiotics (amoxicillin, trimethoprim plus sulfamethoxazole, or erythromycin plus sulfamethoxazole) were prescribed and 13.2% when more expensive antibiotics (cefaclor, amoxicillin plus clavulanate, or cefixime) were prescribed (P = .029). The average ADR rate within 24 days was 5.9% when less expensive antibiotics were prescribed, compared with 6.1% when more expensive antibiotics were prescribed (P value not significant). Therefore, a second antibiotic was filled within 24 days of the initial treatment in 17.5% of cases for either unresponsive otitis or ADR when less expensive antibiotics were prescribed versus 19.3% of cases when more expensive antibiotics were prescribed.
Table 8 displays the antibiotic selection for cases of unresponsive otitis according to the initial antibiotic prescribed at the index visit. Physicians who initially treated the patient with amoxicillin most often used trimethoprim plus sulfamethoxazole or cefaclor to treat unresponsive otitis. Physicians who initially selected trimethoprim plus sulfamethoxazole for acute otitis treated UAOM with a second- or third-generation cephalosporin in 42% of cases. Physicians selecting cefaclor to treat the acute episode had no clear preference for unresponsive cases. Physicians selecting a low-cost antibiotic to treat the acute episode selected another low-cost antibiotic for unresponsive otitis in 54% of cases. Physicians selecting a high-cost antibiotic to treat the acute episode selected another high-cost antibiotic for unresponsive otitis in 31% of cases.
Annual Use of Antibiotics for Otitis Media
The age-specific rates for antibiotic fills associated with a diagnosis of otitis media per child-year are shown in Table9 for children with otitis enrolled continuously in Medicaid during 1991 and 1992. The per child-year rates differ slightly from the per child rates, because many of the children were enrolled for a portion of the year. Children <30 months of age have at least two antibiotic fills per child-year. The highest child-year rate for antibiotic fills, about 2 fills occurred for children 13 to 24 months of age. It is not clear from this data whether these children had higher antibiotic fill rates because they experienced a higher frequency of acute episodes, had a higher proportion of unresponsive episodes requiring multiple antibiotic treatments, or received more antibiotics for otitis with persistent effusions.
Antibiotic Expenditures for Treatment of Otitis Media
Table 10 displays the projected expenditures for the antibiotics prescribed to treat otitis media diagnosed in all Medicaid children ≤13 years during 1991 and 1992. In each of the two study years, amoxicillin accounted for almost half of the total antibiotic fills, but for only 9% to 10% of the expenditures. Low-cost antibiotics (amoxicillin, trimethoprim plus sulfamethoxazole, and erythromycin plus sulfisoxazole) were prescribed for 66% to 67% of the total fills and accounted for 21% of the total projected expenditures. Cefaclor, prescribed for 17% to 18% of the total fills, generated 43% to 45% of total antibiotic expenses. More expensive antibiotics (cefaclor, cefixime, amoxicillin plus clavulanate), prescribed for 30% of the fills, generated 76% to 77% of expenditures.
Because the antibiotics for all children enrolled in Medicaid during the study period were provided without any copayments in a noncapitated, fee-for-service environment, there were no financial incentives or disincentives related to antibiotic selection for the patient or physician. Yet, antibiotic selection differed by practice setting and type of physician. There are several possible reasons why physicians in private practice more often prescribed expensive antibiotics, compared with physicians working in clinics. Private practice physicians may have been more influenced by pharmaceutic detail representatives. It is also possible that prescribing patterns for commercially insured families influenced physicians in private practice to prescribe higher-cost antibiotics for otitis media more frequently to their Medicaid patients. Alternatively, physicians practicing in hospital clinics and community health centers may have adhered to a restricted formulary. These physicians may also have been more likely to develop a common approach to managing otitis that stressed cost containment or cost effectiveness. The finding that family physicians were also more likely than were pediatricians to prescribe certain higher cost antibiotics for otitis may represent differences in residency training, graduate medical education, and/or the influence of pharmaceutic detailing.
This study assessed the frequency with which clinicians prescribed a second antibiotic within 24 days of diagnosing and treating initially uncomplicated cases of acute otitis media because of unresponsive otitis or an ADR. The 11.9% rate of unresponsive otitis found in this medical effectiveness study of actual clinical practice is higher than the 7% reported in a metaanalysis of randomized antibiotic clinical trials for acute otitis media.1 Physicians in practice may have less skill in performing pneumatic otoscopy compared with clinicians who participate in otitis media clinical trials, and may be less likely to confirm their findings with tympanometry. Therefore, they may be more likely to treat children with another course of antibiotics. Other factors such as differences in ages, ethnicity, and socioeconomic status may have contributed to the difference in the rates of unresponsive episodes.
The frequency of prescribing a second course of antibiotics for unresponsive acute otitis was slightly less when a low-cost drug such as amoxicillin, trimethoprim plus sulfamethoxazole, or erythromycin plus sulfisoxazole was initially prescribed, compared with a high-cost drug such as cefaclor, cefixime, or amoxicillin plus clavulanate. It is possible that clinicians used more expensive antibiotics when they thought children had findings of more severe disease or were more likely to be at higher risk of unresponsive otitis because of differences in age or history of recurrent otitis media. However, although clinicians selected a more expensive antibiotic most often when the patient was 19 to 24 months of age, the rates of unresponsive otitis were slightly lower at this age compared with rates for younger children.
There are limitations in this study to determining antibiotic expenditures accurately. Estimating the cost of a 10-day course of a particular drug for a particular age for acute otitis media has many difficulties such as (1) manufacturer rebates and agreements between health systems and pharmacies that affect acquisition price but are not in the public domain; (2) the fact that the average wholesale price of a drug is formulation-specific (strength-form packaging) and that there may be several formulations that are appropriate for children of a given age; and (3) the absence of published standards for crushing pills in food or about the age at which children can be expected to chew or swallow a medication. In addition, the methodology for cost calculation in this study is unlikely to reflect actual costs in non-Medicaid systems, because Pharmacy Benefit Management firms now use strategies to obtain manufacturer rebates based on market share, discounts by selectively contracting with pharmacies, and formulary restrictions of either entire drugs or certain formulations. However, the relative cost index (ie, that some drugs cost 10 times more than others) is likely to hold across all systems.
Although there was a wide variation in the selection of antibiotics to treat a new episode of acute otitis media, the outcomes associated with more expensive antibiotics were not superior to less expensive antibiotics. Changes in prescribing patterns for acute otitis media among initially uncomplicated children that reduce the use of high-cost antibiotics could reduce expenditures substantially without compromising outcomes. For example, in 1992, if half of the cefaclor, cefixime, and amoxicillin plus clavulanate prescriptions (8359) were written for amoxicillin instead, the Medicaid program would have realized savings of $399 412.
This study was supported by a grant from the Agency for Health Care Policy and Research (RO1 HS07816-03).
- Received December 30, 1996.
- Accepted March 3, 1997.
Reprint requests to (S.B.) Children's Hospital, 1056 East 19th Ave, B032, Denver, CO 80218.
- MMED =
- Medicaid Management Information System •
- UAOM =
- unresponsive acute otitis media •
- ADR =
- adverse drug reaction
- Copyright © 1997 American Academy of Pediatrics