NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Marcy M, Takata G, Shekelle P, et al. Management of Acute Otitis Media. Rockville (MD): Agency for Healthcare Research and Quality (US); 2001 May. (Evidence Reports/Technology Assessments, No. 15.)

  • This publication is provided for historical reference only and the information may be out of date.

This publication is provided for historical reference only and the information may be out of date.

Cover of Management of Acute Otitis Media

Management of Acute Otitis Media.

Show details

Appendix A. Economic Burden of Illness

The Societal Cost Burden of Acute Otitis Media

Previous Estimates of the National Cost of Otitis Media

The medical literature was reviewed to identify estimates of the annual national cost burden of AOM in the U.S. Three estimates of the cost of otitis media (OM) were found (Gates, 1996b; Stool and Field, 1989; Stool, Berg, Berman, et al., 1994 ). A fourth study reported the cost per episode of "persistent middle ear infection" and an estimate of the total annual number of episodes of the illness (Berman, Roark, and Luckey, 1994). These figures yielded a fourth estimate. The four estimates are summarized in Table 51. To provide for comparability, each was inflated to 1995 price levels using the medical care component of the Consumers Price Index and the Employment Cost Index.

Table 51. Estimates of the U.S. National Cost of Otitis Media.


Table 51. Estimates of the U.S. National Cost of Otitis Media.

As the table indicates, the four estimates of the national cost burden differed with respect to definition of illness and age group for which costs were derived. They also differed with respect to specification of the scope of costs and the pricing of cost components. The estimates are described in detail in the following attachment to this appendix.

The National Cost of AOM in 1995

At the conclusion of the literature review, the annual national cost of AOM in children younger than18 years of age was estimated from existing data summaries and published reports of costs and utilization per episode of illness. Because of the dearth of usable epidemiological and cost-related data, it was necessary to employ a number of assumptions to derive the cost estimate. The principal features and underlying assumptions of the estimate are the following: Otitis media with effusion (OME) and chronic (or persistent or recurrent) middle ear infections were considered possible complications of AOM or the results of unsuccessful treatment of AOM. Because the costs of complications of an illness and of treatment failures are conventionally and correctly regarded as costs of the illness, the cost of episodes of OME/chronic middle ear infection that follow episodes of AOM was considered a cost of AOM. National illness costs were derived for the year 1995. Certain data used in deriving and validating the estimates were not yet available for 1996 and later years when the report was begun. The cost estimate was constructed to be as consistent as possible with published data on national medical care utilization for middle ear infections.

The major results and implications of the cost estimation procedure are these:

  • The baseline estimated total national cost estimate of AOM was $2.98 billion in 1995. The figure is shown in Table 51.
  • The estimated direct cost of AOM was $1.96 billion (65.8 percent of the total) in 1995, and the estimated indirect cost was $1.02 billion (34.2 percent of the total). The large percentage of indirect costs in the total suggests that AOM imposes a considerable economic burden on patients' families.
  • Medical care prices rose 12.5 percent from 1995 to the first quarter of 1999 and workers'earnings per hour rose 14.7 percent. Applying these inflators to the estimates of direct and indirect costs suggests that the annual national cost of OM increased to $3.38 billion in early 1999 because of price increases alone.
  • Seventy percent of the direct cost of AOM and 63 percent of the total cost was attributable to episodes of OME/chronic middle ear infection progressing from AOM. These figures imply that effective treatments for AOM and prevention of subsequent infection would achieve significant national cost savings.
  • The analytic procedure produced an estimate of 5.18 million treated episodes of AOM in 1995, 2.22 million treated episodes of OME/chronic middle ear infection, and 1.04 million episodes of treated OME/chronic middle ear infection assumed to be progressions or complications of AOM. Although these values were constrained to yield national ambulatory care visit totals reported by NAMCS, the estimated total number of episodes of OM, 7.40 million, is smaller than the number conjectured in other cost estimates.
  • Insofar as it is comparable to it, the cost estimate itself is smaller than the estimated cost of AOM assumed by Gates (1996b). The lower value is due partly to the generally conservative assumptions used here and partly to what seems to be an exaggerated estimate of the annual prevalence of AOM used by Gates.

The cost estimates have important limitations, primarily due to conflicting epidemiological evidence regarding the prevalence of AOM and OME. In particular:

  • The annual national numbers of ambulatory care visits for OM implied by epidemiological studies of OM are 50 percent or higher than national visit totals reported from NAMCS and the National Hospital Ambulatory Medical Care Surveys (NHAMCS). Because the epidemiological studies were conducted in small geographic areas and nearly all of them were performed before 1990, it was considered that NAMCS and NHAMCS data yielded more reliable estimates of the national prevalence of OM than the epidemiological literature. However, it has been claimed -- although without hard supporting evidence -- that NAMCS reports understate the true national visit totals (Glass, Lew, Gangarosa, et al., 1991). If the claim is true, the national cost estimate is biased downward.
  • To deduce the annual numbers of episodes of AOM and OME/chronic middle ear infection following AOM in the under-18-year-old age group from NAMCS and NHAMCS data, it was necessary to employ two major assumptions. The first was that OME/chronic middle ear infection represents 30 percent of all episodes of OM. The second was that 20 percent of the episodes of AOM progress to OME/chronic middle ear infection. Little empirical evidence exists with which to confirm or reject either assumption.

  • Sensitivity analyses of the national cost estimate showed that variations of 10 percent in most of the price and utilization parameters changed the estimate by 4 percent or less. However, the cost estimate was sensitive to the assumed national ambulatory care visit total for AOM, the percentage of episodes of OM classifiable as OME/chronic middle ear infection, and the percentage of episodes of AOM that progress to OME/chronic middle ear infection. Sensitivity analyses indicated that plausible values of these three parameters raise the estimated national cost of AOM to as much as $6 billion in 1995 prices. Thus, it is reasonable to believe that the true annual national cost of AOM was $3 billion to $6 billion in 1995.

Cost-Effectiveness Studies of Therapies for AOM

Literature Search

The literature search for economic evaluations of therapies for treating OM identified nine studies published since 1965.1 All nine were cost-effectiveness, cost-utility, cost-minimization, or similar analyses. One of the nine was undertaken in Nigeria (Amadasun, 1997), and it was not used in the cost-analysis because of potential differences in medical care costs and practice between that country and the U.S. Three of the remaining four studies defined OM either nonspecifically or as a recurrent illness (Banz, Schwicker, and Thomas, 1998; Berman, Roark, and Luckey, 1994; Bisonni, Lawler, and Pierce, 1991; Callahan, 1988). Only four studies explicitly addressed treatments for AOM (Branthaver, Greiner, and Eichelberger, 1997; Landholt and Kotschwar, 1994; Oh, Maerov, Pritchard, et al., 1996; Weiss and Melman, 1988), and they are discussed here. The salient characteristics of the four studies are summarized in Table 52.

Table 52. Summary of Cost-Effectiveness Studies of Therapies for Acute Otitis Media.


Table 52. Summary of Cost-Effectiveness Studies of Therapies for Acute Otitis Media.

Overview of Methods and Materials in the Literature

Except in the broadest terms, the four economic evaluations of antibiotic treatments for AOM have little in common. All four studies were performed using standard decision-analysis or decision-tree models. In three studies, amoxicillin-clavulanate combination therapy was compared with one or more other antibiotic therapies. In the fourth (Weiss and Melman, 1988), amoxicillin monotherapy was compared with cefaclor monotherapy. Three of the four were performed, or seemed to have been performed, from the payer's perspective. The fourth was conducted from the perspective of the child's family under the assumption that the family was uninsured (Weiss and Melman, 1988).

The studies differed considerably in virtually all other respects. One was a cost-effectiveness analysis, one a cost-utility analysis, and two were cost-minimization analyses (Table 52). Only two of the three studies specified the dosage and schedule of antibiotic treatments. The cost-of-therapy estimates were derived from: a very large sample of patient records, very small samples of patient records, a small sample of pediatricians, and estimates of the authors. In two of the four studies, estimates of the efficacy of treatment were deduced from patient records. In the other two, they were taken from the literature on clinical trials of antibiotics for OM. The health-related physical outcomes of treatment were defined as the patient's quality-adjusted life-days in one study, as treatment cure and failure rates in another, as an OM severity score and clinical cure rate (but not used in the analysis) in the third, and left unspecified in the fourth (a cost-minimization analysis). The time frame over which costs and health outcomes were measured appeared to range from 2 to 8 weeks but was not expressly stated in any of the four studies.

Because payers are usually concerned only with direct costs, the three studies conducted from the health plan payer's perspective excluded indirect care-giving costs, although one of the three presented estimates of hours of lost parental work time caused by care-giving (Landholt and Kotschwar, 1994). Only one study estimated the direct costs of treatment side effects. None of the studies examined or estimated the costs of long-term sequelae of failed treatments such as recurrences of illness, surgery, or patient hearing loss and related learning disabilities. Only one of the three studies performed a sensitivity analysis of the results derived from its model (Oh, Maerov, Pritchard, et al., 1996).


Findings as to the most cost-effective antibiotic therapy for AOM differ. In two studies it was concluded that cefaclor monotherapy was more cost-effective than amoxicillin monotherapy, amoxicillin-clavulanate, and erythromycin-sulfisoxazole (Branthaver, Greiner, and Eichelberger, 1997; Oh, Maerov, Pritchard, et al., 1996). By contrast, amoxicillin monotherapy was judged more cost-effective than cefaclor monotherapy in one of the other two studies (Weiss and Melman, 1988). In the fourth analysis, amoxicillin-clavulanate and cefpodoxime-proxetil were deemed equally cost-effective (Landholt and Kotschwar, 1994).

The specific results of the studies differ as well. For example, Oh, Maerov, Pritchard, et al. (1996) found that cefaclor monotherapy and amoxicillin-clavulanate produced nearly identical quantities of health benefits per patient measured as quality-adjusted patient-days, but that direct treatment costs with cefaclor monotherapy were 10 percent lower per patient than the direct costs of amoxicillin-clavulanate. Conversely, Weiss and Melman (1988) concluded that cefaclor monotherapy yielded a 5 percent higher treatment cost per patient than amoxicillin monotherapy. And finally, Branthaver, Greiner, and Eichelberger (1997) reported that there seemed to be no differences in the direct costs per patient of five antibiotic therapies, including amoxicillin monotherapy, amoxicillin-clavulanate, and cefaclor monotherapy, but that cefaclor monotherapy and trimethoprim-sulfamethoxazole produced a higher cure rate than the other antibiotics. These conflicting findings make it unclear whether the treatment cost of cefaclor monotherapy is higher than, lower than, or the same as the treatment cost of amoxicillin and other antibiotics.


The evaluative economic literature on antibiotic treatments for AOM is too small, its methodologies too varied, and its findings too inconsistent to provide a clear-cut guide for choosing the most cost-effective antibiotic therapy. One possible reason for the inconsistencies in results is that the total treatment costs of antibiotics like amoxicillin and cefaclor may not be greatly different. Two of the four studies discussed here did claim that there appeared to be nonsignificant differences in the total treatment costs of alternative antibiotics (Oh, Maerov, Pritchard, et al., 1996; Weiss and Melman, 1988). Even the two studies that favored one antibiotic over another did not report large differences in efficacy and treatment costs, and it is conceivable that true efficacy rates and direct treatment costs do not differ materially among several of the antibiotic alternatives (Oh, Maerov, Pritchard, et al., 1996; Weiss and Melman, 1988). The issue should be explicitly addressed in future research.

Other research problems and issues also should be addressed in new cost-effectiveness analyses of antibiotic therapies for AOM. Some of these are the following:

  • Large, national representative databases should be used for determining medical care utilization and prices. The studies discussed here all employed small local databases, and their estimates of illness and treatment costs might be unrepresentative of costs in the United States as a whole.
  • A societal perspective should be adopted, and indirect illness and treatment costs should be incorporated in evaluative studies. The national cost estimate of AOM illness suggests that indirect costs constitute nearly half of the total treatment costs of AOM. A national cost burden that large should not be ignored, and the inclusion of indirect costs may have substantial effects on the results of cost-effectiveness analyses. For example, Landholt and Kotschwar (1994) noted that amoxicillin-clavulanate caused an average of 32 hours of lost parental work time per patient in their sample (chiefly due to adverse side-effects), and that cefpodoxime-proxetil caused an average of only 16 hours of lost work time per patient. Had this time been priced and included in their analysis, it would have altered their conclusion that amoxicillin-clavulanate and cefpodoxime-proxetil are equally cost-effective. Finally, if direct treatment costs are approximately the same for all antibiotic therapies, payers have no disincentives to sanction or provide therapies with low indirect costs. Hence, the selection of low indirect-cost antibiotic treatments for AOM would benefit patients' parents without penalizing payers.
  • A standard generic measure of the health benefits of treatment, such as quality-adjusted or healthy life-years (or days), should be adopted and used in cost-effectiveness analyses. There are three reasons for this. First, a standard generic measure of treatment benefits would provide for comparability of study results. Second, the conventional two-treatment cutoff-point methodology in cost-effectiveness analysis requires the definition of a cutoff point, typically the maximum amount that society pays for a quality-adjusted or healthy year of life. As a practical matter, it is not feasible to define a meaningful cutoff point when the measure of treatment benefits for AOM is the rate of cures, a symptom severity score, or any sort of illness-specific health outcome. Third, an illness-specific or other idiosyncratic measure of treatment benefits makes it impossible for decisionmakers to allocate spending among treatments for AOM and other illnesses in a rational manner. The decisionmaker must have estimates of how its population's overall health benefits are affected when it provides or sanctions treatments for many illnesses. Unless health benefits are (or can be) defined identically for all illnesses, there are no ways to determine how a shift in spending -- for example, to more expensive or less expensive treatments for AOM -- affects the population's aggregate level of health and well-being. Quality of life in AOM recently has become a research issue (Alsarraf, Jung, Perkins, et al., 1998; Rosenfeld, Goldsmith, Tetlus et al., 1997;), but only one of the four studies attempted to estimate and incorporate patients' quality-adjusted survival time into its cost-effectiveness analysis (Oh, Maerov, Pritchard, et al., 1996).
  • The costs and health benefits of treatment failures and complications should be recognized and incorporated into evaluations of treatments for AOM. The economic impact of hearing loss on children's learning abilities and subsequent lifetime earnings capabilities should be rigorously evaluated. Unless the probability or risk of progression of AOM to more serious illness is deemed independent of the mode of treatment, the effects of disease progression on costs and patients' health outcomes should be accounted for in cost-effectiveness analyses. According to the cost-of-illness estimates obtained here, the cost per case of OME is up to 10 times as large as the treatment cost of AOM. Even small differences in treatment failure rates may therefore have large effects on the relative cost-effectiveness of different antibiotics for AOM. Although two of the studies discussed here found that cefaclor was a cost-effective antibiotic for AOM (Branthaver, Greiner, and Eichelberger, 1997; Oh, Maerov, Pritchard, et al., 1996), it has been reported that cefaclor was associated with an especially high recurrence rate of illness (Kaplan, Wandstrat, and Cunningham, 1997). If that is the case, the high cost and adverse health consequences of recurrent illness argue prima facie against the cost-effectiveness of cefaclor therapy.


  • PubReader
  • Print View
  • Cite this Page

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...