Alsarraf R, Jung CJ, Perkins J, Crowley C, Alsarraf NW, Gates GA. Measuring the Indirect and Direct Costs of Acute Otitis Media. Arch Otolaryngol Head Neck Surg. 1999;125(1):12–18. doi:10.1001/archotol.125.1.12
To test a method of measuring the related cost of acute otitis media (AOM) and to provide a preliminary calculation of the indirect and directs costs associated with a single, medically treated episode of AOM.
The Otitis Media Diary was used to measure indirect and direct costs associated with AOM in a prospective cohort study. Measured values included the parental time spent in otitis-specific child care and the number and type of medications used. A previously developed economic model was used to calculate the monetary costs associated with the value of caregiver time and the total opportunity cost of AOM.
The pediatric clinic of Madigan Army Medical Center, Tacoma, Wash.
A cohort of 25 children (12 with AOM and 13 controls) aged 1 to 3 years.
Main Outcome Measures
Caregiver time and medication use.
The total cost attributable to AOM in the 3-month period following diagnosis was $1330.58 (95% confidence interval, $1008.75-$1652.43), with the majority of that cost stemming from the indirect, rather than direct, costs of illness. After conservative estimates of unmeasured expenses, such as clinic visits and transportation, were accounted for, indirect costs, accrued primarily by parental time, accounted for nearly 90% (95% confidence interval, 87.1%-92.3%) of the total 3-month cost associated with AOM and its medical treatment. The cost items of the Otitis Media Diary were also highly correlated with each other and with other measures of clinical and functional health status.
Otitis Media Diary measures of parental time and medication use appear to provide a more accurate means of calculating the real social costs attributable to the AOM disease process in this cost-effectiveness analysis.
ACUTE OTITIS media (AOM) is the most common cause of pediatric office visits in the United States, one of the most common diseases of childhood, and has been estimated to cost the United States more than $3 billion annually.1- 3 Combined with its sequela, chronic otitis media with effusion (OME), this figure rises to more than $5 billion every year.2 The treatment for children with uncomplicated AOM is medical (antibiotics), with tympanostomy tube placement and/or adenoidectomy reserved for those cases that become recurrent or long-term.4- 6 The long-term effects of OME are still being debated, and no study to date has comprehensively measured the outcomes of the treatment of this illness process in terms of clinical health status, patient-related terms, such as quality of life or functionality, and economic or social costs.7,8
Cost-effectiveness analyses can be done in many forms, with outcomes measured in terms of the monetary equivalents of disease, disease-specific effectiveness, global quality or quantity of life, and various scales of utility rating.9- 12 Inherent to each of these methods is the ability to quantify the actual costs attributable to the illness in question or estimate these costs with a reasonable degree of certainty. In addition, most authors11- 13 differentiate between the direct costs attributable to an illness and the indirect costs associated with that same disease process.
Previous cost-effectiveness analyses of AOM and its treatment have relied on estimations of both the direct and indirect costs of AOM and its medical treatment.2,6 The direct costs defined in these studies include both the cost of clinic visits and the cost of medications taken secondary to an episode of AOM. The indirect costs defined in these studies include both the cost of parental time lost from work and the cost of transportation associated with these same episodes. These studies attempt to quantify the total costs attributable to AOM based on a range of estimated clinic visits from 2.5 to 4.6 for each acute episode, 1 to 1.5 courses of antibiotic therapy for each infection, and 12 day of parental work and transportation associated with each AOM visit. Gates2 estimated the indirect and direct cost of an episode of AOM to be $233, while the Agency for Health Care Policy and Research6 estimated this amount to be $406, with approximately half of this amount accrued within the first 3 months following diagnosis. Other studies14- 16 that rely on estimates from theoretical models, telephone interviews, or physician ratings produce similar estimates of the monetary cost of a typical AOM episode.
The indirect and direct costs of AOM have not previously been directly measured by an otitis-specific instrument. We developed 3 new measures of otitis-specific clinical and functional health status for the pediatric population aged 1 to 3 years; the reliability and validity of these instruments have been previously established.17 One of these instruments, the Otitis Media Diary (OMD), provides a measure of pediatric functional health status plus 2 important components of the indirect and direct costs of the illness as well. The OMD records, on a daily basis, the amount of parental time spent with the child secondary to an episode of AOM, in addition to the number and type of medications taken (antibiotics, cold medicine, and pain or fever medicine).
The purpose of this study was to examine the indirect and direct costs attributable to episodes of AOM and its medical treatment, as measured by the OMD instrument. The temporal relation of cost accrual was also evaluated to determine the overall cost of illness in the 3-month period following AOM diagnosis. In addition, these cost items were correlated among themselves and with other measures of clinical and functional health status to evaluate if parental time and medication use are accurate indicators of AOM severity or functional impact.
A cohort of 25 children (12 with AOM and 13 healthy controls), aged 1 to 3 years, was serially enrolled and followed up during a 3-month period as a part of a larger study at the pediatric clinic of Madigan Army Medical Center (MAMC), Tacoma, Wash. The number of enrolled children was limited by the total number of subjects of this previous pilot study (N=51).17 Inclusion criteria for the patients with AOM included a diagnosis of AOM from the results of a standardized physician examination, with no history of ear surgery, sensorineural hearing loss, or any form of craniofacial disorder or named syndrome. Criteria for the inclusion of healthy children included the same restrictions, with results negative for AOM in a standardized physician examination. Acute otitis media was diagnosed by the presenting symptoms of otalgia, irritability, fever, and sometimes otorrhea, confirmed by pneumatic otoscopy by the standardized physician (C.J.J.). Two follow-up visits were scheduled for each child at 6 and 12 weeks from the initial enrollment visit, with 100% follow-up rate. A total of 50 visits were accrued, with 1 of the 13 original healthy children subsequently developing AOM and 6 of the 12 original children with AOM developing a recurrent episode. Two temporally related case groups of subjects were thus identified: (1) children with an episode of AOM who were within the first 6 weeks following diagnosis, and (2) children who were 6 to 12 weeks past an episode of AOM. Of the total 50 visits, there were 25 healthy children, 19 children with AOM episodes with evaluation in the 0- to 6-week period following diagnosis, and 6 children recovering from AOM with evaluation in the 6- to 12-week period. On enrollment into this study and at the 2 follow-up visits, the clinical health status of each child was evaluated by the standardized physician at the pediatric clinic (C.J.J.), and each parent completed an Otitis Media Functional Status Questionnaire to evaluate the functional health status of the child. The OMD was sent home with every parent to be completed during the two 6-week periods in between visits and served as the outcome measure of indirect and direct costs.
The OMD was developed as a take-home, daily measure of otitis media health status. The reliability and validity of this measure, in addition to the Otitis Media Functional Status Questionnaire, have been established elsewhere.17 The psychometric properties of this survey include an internal consistency of .96 (Cronbach α) excellent reproducibility, and a high construct validity (P<.01). This diary measures the daily presence and severity of AOM symptoms in each child, the amount of time spent by the primary caregiver, and the medications taken secondary to such episodes. Parental time is measured on a daily basis as the amount of extra time that the parent was required to spend with the child due to AOM signs or symptoms for that calendar day. This item is scored on a 0 to 4 scale for the entire 24-hour day (no time, 14 day,
12 day, 34 day, and all day) to yield an OMD time score similar to the symptom-based OMD health status score. In addition, the total number of days of parental time spent due to AOM is measured by summing the individual daily scores for the period of interest. Medication use is recorded on a daily basis for antibiotics, cold medicines (such as antihistamines or decongestants), and pain or fever medicine (such as acetaminophen). The daily use of each medication is then summed, for the total days of medication use for each subject.
Indirect and direct costs attributable to episodes of AOM were evaluated based on the OMD measurements of parental time spent and medications used. The value of parental time spent as a caregiver due to this illness is the best measure of the indirect cost of AOM in this age range, since this time would alternatively have been spent participating in other normal, daily activities.10,13 In addition, because these children are not yet in school, do not work, experience primarily short-term morbidity, and are not, in general, threatened with mortality due to AOM, there are few other measures of indirect cost in this setting. Other indirect costs associated with AOM are the travel and parking expenditures required to transport the child to and from clinic visits, the hospital, or the pharmacy. The value of the antibiotic, cold medicine, and pain or fever medicine taken was used to measure one aspect of the direct costs associated with episodes of AOM. Direct costs not measured in this study include clinic visits, other treatments (such as surgical procedures), laboratory tests, diagnostic evaluations, and the management of any complications of either the illness or the treatment modalities.11
An economic model was used to convert OMD measurements of parental time into monetary costs. This model sums a weighted average of the value of parental activities that were lost secondary to the time attributable to an AOM episode, based on the employment characteristics of the population and a range of monetary inputs. The economic model is represented by the equation
where OC is the total opportunity cost of caregiver time; X, Y, Z, and W are variables representing the value of normal parental activities, namely, employment (X), housekeeping (Y), leisure (Z), and sleep (W); a/A, b/B, c/C, and d/D are the proportion of time spent by the caregiver in each daily activity for the children with AOM and healthy children, respectively; and t/T represent the children with AOM and healthy children caregiver times, as measured by the OMD. For the most conservative, lower limit estimate of opportunity cost, the nonlabor variables of leisure and sleep were accorded no monetary value. A range of costs may be produced in addition to this conservative, lower limit value; the upper limit value is that in which all activities (including leisure and sleep) are valued by the appropriately weighted wage rate. These upper limit values are provided for comparison purposes only. Although traditional cost-effectiveness analyses use similar methods to estimate the monetary value of patient or caregiver time, no prior study has directly measured this time and converted it to the monetary value of the opportunity cost of normal daily activities.10,13,18 Thus, an important goal of this study was to test the feasibility of using the OMD and this economic model to measure the monetary costs associated with AOM.
Cost calculations were based on published national monetary values. Medication use was calculated as a mean number of days per patient in each group, and this figure was multiplied by the average daily cost for each medication type, based on the usual age-specific dosage and derived from these published figures.19,20 Since in many cases parents did not give a full course of antibiotics and other over-the-counter medication use was sporadic as well, these daily measurements provide more complete information than assumptions based on pharmacy data of prescription. As the most common, first-line treatment of AOM, amoxicillin elixir was used as the standard for cost comparisons. For cold and pain or fever medications, respectively, phenylpropanolamine hydrochloride or brompheniramine maleate elixir and acetaminophen elixir were used for cost calculation. Given these assumptions, cost calculations for these 3 medications were based on an average daily cost of $0.79 for antibiotic, $0.54 for cold medicine, and $0.33 for pain medicine. Parental time, in days per patient, was used as the indicator of time lost from normal daily activities. The primary caregiver of the child was the mother for all these calculations, and it was assumed that, on average, mothers of children in this age group ranged in age from 20 to 39 years. Average daily income was derived from US Bureau of Labor statistics.21 Income was calculated to account for both actual earnings and the imputed value of housekeeping work for both employed and unemployed mothers. Each of these earnings was derived from 1985 figures and inflated to 1996 dollars using a conservative 3% average annual increase of individual income.13 For instance, cost calculations were based on an average annual salary of $26,903 for women in this group, with the value of average annual housekeeping services to be $10,865 for employed women and $17,065 for unemployed women.21 The total number of days lost due to AOM was converted into equivalent workday earnings, based on these given figures using the above economic model.
Healthy children and those with AOM episodes were not significantly different in any characteristic other than age (Table 1). Children with AOM were older than healthy children in this study (25.8 months vs 15.4 months) because healthy children were enrolled at visits that are scheduled at 12, 18, and 24 months while children with AOM were enrolled at any walk-in visit, provided they were between 1 and 3 years of age. Nine (75%) of the 12 mothers of children with AOM vs 7 (54%) of the 13 mothers of healthy children were employed outside the home. Although the differences are not statistically significant, children with AOM were more likely to be in day care (8 [67%] of 12 vs 5 [39%] of 13) than healthy children, consistent with the known causes of AOM. In addition, healthy children were more likely to have experienced another medical problem during the preceding 6-month period (2 [15%] of 13 vs 1 [8%] of 12). None of these factors were thought to bias the results of this study, since medication use and parental time measured were restricted to otitis-specific recording alone. Furthermore, factors, such as age or day care use, although related to the development of otitis are not considered to be associated with the severity or functional significance of these episodes, as evidenced in a previous study of this same population.17
The amount of parental time spent with subjects secondary to AOM and the type and amount of medication taken for these episodes were both appreciably different between the 3 groups of children categorized in this study (Table 2). The OMD time scores showed a significant linear trend between healthy children, recovering children, and children with AOM episodes (r=0.51), with the actual number of parental days devoted to caregiving varying from 0.62 to 3.83 to 6.12 per patient in each of these 3 categories. In terms of medication use, similarly significant linear trends were found for antibiotic, cold medicine, and pain or fever medicine use alike (r=0.29-0.57). Mean antibiotic use for the 0- to 6-week AOM group was 10.74 days per patient, and the attributable amount was 9.5 days once the baseline healthy-children use was subtracted. Recovering and healthy children showed similar amounts of antibiotic use, illustrating a similar lack of infection in these 2 groups. Cold and pain or fever medicine use was less specifically associated with the 0- to 6-week AOM group.
Conversion of these attributable days of parental time and medication use to monetary values was accomplished as described in the "Methods" section. Table 3 shows the results for the economic model of the opportunity cost derived from the MAMC study population's actual employment characteristics, in which parental time has been valued only based on lost employment and housekeeping activities. This is not truly a point estimate but represents the most conservative, lower limit of the range of indirect cost calculations based on the series of measured and unmeasured inputs described herein. In addition, more traditional analyses of these data (results not shown) produced cost calculations that did not differ significantly from these results. For instance, applying the OMD measures of parental time to cost inputs derived entirely from US statistics (ie, rather than the employment characteristics of this MAMC population) yielded a value of $1128.73 for the monetary value of AOM-related parental time. The total expenditures for each category of subject was $811.21 per patient for the 0- to 6-week AOM group, $504.42 per patient for the recovering 6- to 12-week AOM group, and $72.40 per patient for the healthy control group. Subtracting this control figure yields total attributable costs of parental time and medication use of $738.82 per patient for the AOM period and $432.01 per patient for the recovering AOM period. The combined use of medications for this 3-month period totaled an attributable cost of $10.97 per patient for the AOM group and only $2.72 per patient for the recovering AOM group. Totaling the average attributable costs for this 3-month period results in a total 3-month cost of $1170.83 per patient for parental time and medication use alone. This figure far surpasses the estimations of the cost of AOM, which can be found in the literature.1,2,6 Of this total, 63% of the costs were accrued in the first 6-week period following diagnosis, while 37% of the costs were contributed from the second 6- to 12-week period.
These derived costs were used for further calculation of the total costs attributable to an episode of AOM. If one accepts the conservative, previously published estimates for those aspects of indirect and direct cost not directly measured in this study, a more complete picture of the costs of AOM results (Table 4). Based on the findings of Gates2 and the Agency for Health Care Policy and Research guidelines,6 direct costs for AOM can be approximated with the cost of each clinic visit and the cost of medication, while indirect costs can be estimated with the cost of parental time and the cost of transportation. From these 2 sources, the best estimate for the cost of clinic visits associated with AOM is $35 per visit, with an average of 2.5 to 4.6 visits per episode. Similarly, the best estimate for the cost of transportation is $10 per visit multiplied by these same 2.5 to 4.6 visits. If one sums these estimated values along with the measured values of parental time and medication use found in this study, the total 3-month cost of an episode of AOM is $1330.58 (95% confidence interval, $1008.75-$1652.43) for indirect and direct costs of illness combined. Of this sum, 89.7% (95% confidence interval, 87.1%-92.3%) is attributable to the indirect costs of illness, with the bulk of this cost due to the parental time lost from actual work or work imputed to housekeeping.
The OMD instrument and the Otitis Media Functional Status Questionnaire have been shown to be valid and reliable measures of otitis-specific functional health status, and this study also evaluated the correlation of parental time and medication use with these health status instruments (Table 5). The OMD time score, actual parental time spent, antibiotic use, and cold medicine use all correlated significantly with both the OMD rating and the Otitis Media Functional Status Questionnaire score (P<.001-.004). In addition, the OMD time score, actual parental time spent, and antibiotic and cold medicine use were all highly correlated with each other (P<.001-.002). These indirect and direct cost measures thus appear to be relatively accurate indicators of otitis-specific health status, with a high degree of consistency within their individual measurement ratings.
To analyze the cost-effectiveness of treatments for OME, one must first be able to measure both the direct and indirect costs associated with each treatment option. For the medical treatment of AOM in the young pediatric population, direct costs can be defined as the cost of each visit plus the cost of medications taken secondary to the episode of AOM, while indirect costs can be defined as the cost of parental time plus the cost of transportation incurred because of the illness.2 This study does not measure the number or cost of clinic visits, because each visit was scheduled for the purposes of study follow-up, and so a range of visit estimates is used for direct cost calculations. Similarly, this study does not measure the cost of transportation (such as gas and parking), which is thought to depend on the number of visits, so this same range is applied to this aspect of indirect cost. In all cases, however, conservative estimates are used to limit any bias toward inflated figures of AOM cost expenditures.
The measurement of parental time and medication use accomplished by the OMD appears to provide a real approximation of the indirect and direct costs of the medical treatment of AOM in children aged 1 to 3 years. In addition, the economic model used in this study provides a simple means of converting measurements of parental time into the monetary value of lost opportunity. It was found that indirect costs, accrued primarily by parental time lost from actual work or work imputed to housekeeping, account for approximately 87% to 92% of the total costs attributable to AOM during the 3 months following diagnosis. The conservative estimate of the total expenditures for this period was $1330.58 (95% confidence interval, $1008.75-$1652.43), once the other, unmeasured aspects of direct and indirect costs were added. This figure far exceeds the estimates of the cost of medical treatment of AOM in the literature, which ranged from $233 per episode to $406 per episode, depending on the number of follow-up visits assumed with each infection. In addition, this figure is composed primarily of the lower limit of a spectrum of the value of parental time ($1157). The upper limit of this spectrum is $3471, a figure that would overwhelmingly make up the majority of AOM-related costs determined in this study (95%). Since it is likely that the most accurate value for the real costs of parental time associated with AOM falls somewhere in this range, we have presented the most conservative, lower limit of this spectrum to avoid an overestimation of these total costs. Although the precision of our calculations is limited by the small number of visits accrued in this study, the large differences found between indirect and direct cost suggest that, in general, these estimates provide a realistic sense of the proportion of total AOM-related indirect and direct costs.
We found that approximately 63% of the costs associated with AOM in the 3 months following diagnosis were incurred in the first 6 weeks, while only 37% of these costs were incurred in the second 6 weeks of the illness. The Agency for Health Care Policy and Research guidelines6 for the treatment of OME contend that only 48.2% of the costs attributable to AOM are accrued in the initial 3-month period. If this is the case, then the costs measured in this study would only represent approximately half the total costs attributable to AOM. These total costs would then equal approximately $2660, if this assumption is correct.
Cost-effectiveness analyses traditionally emphasize direct costs, and it is only recently that greater attention has been given to the indirect costs associated with an illness or its treatment. This emphasis is due, at least in part, to the fact that indirect costs tend to be more difficult to measure than, for instance, the number of clinic visits or the cost of medication. With regard to AOM, the emphasis has been placed on the direct costs of illness, particularly those costs associated with clinic visits themselves. The indirect cost measured in this study only represents one possible component of indirect costs attributable to AOM; however, other parental or childhood costs, such as emotional burden or pain, are difficult factors to directly measure. The economic model used in this study attempts to provide the broadest characterization of the cost related to parental time. To remain conservative in our estimates, only the lower limit of a range of attributable indirect cost was used for ultimate cost calculation. The fact that more traditional methods of analysis of these data result in the same or very similar cost calculations also provides support for the use of a more complete model in the calculation of this AOM-related indirect cost.
It has been postulated that approximately 45% of the costs of illness due to AOM are direct in nature, while 55% stem from the indirect costs associated with this illness process.2,6 The findings of this study suggest that this is not the case. Since this study is limited by a small number of subjects, this preliminary finding should be studied further. Indirect costs accounted for the overwhelming majority of costs associated with AOM. Even when a range of clinic visits was used to estimate the unmeasured aspects of cost in this study, parental time remained the largest contributor to the costs of AOM, averaging 84% to 90% of the total costs of illness. The Agency for Health Care Policy and Research guidelines6 also suggest that the ratio of the indirect costs of illness to the costs of medication is 31.3%:6.8%, or approximately 5:1. This study found that, in the 3-month period following diagnosis, this ratio was actually 89.7%:1.0%, or, nearly 90:1. Further work needs to be done to confirm this finding, but it does appear that the costs of illness attributable to AOM and its medical treatment are overwhelmingly indirect in nature, in contrast to what was previously believed.
The assumptions made in the methods of this study are important limitations of the measurements provided herein. It is true that not all primary caregivers are mothers, and the age range, percentage of employment, and earnings figures will certainly vary considerably depending on the specific population of interest. Although most mothers in this study were employed outside the home, certain aspects, such as prescription medication and parking, are free at MAMC. To provide the most generalizable figures, however, this study used national statistics on earnings and costs, because the MAMC population is not representative of the country as a whole.13 It may seem that this population does not compare with a similar US population of mothers; however, the demographic makeup and, most importantly, employment characteristics of this group, were similar to national averages. The fact that 75% (vs 74% of the US population of mothers aged 20-39 years) of the mothers of children with AOM were employed outside the home supports the generalizability of the OMD measurements of time that these mothers produced. One must simply assume that these mothers value their own time in a way that is similar to other mothers. Of course, one is always dependent on the accurate recording of such a survey. This accuracy is supported by the fact that, on average, children were given a 10-day course of antibiotics (similar to the prescription habits at MAMC) and the high correlation of the OMD ratings with other measures of AOM-related quality of life. Additionally, the assumption that all caregivers are mothers serves as a conservative means of estimating these costs, as men tend to have higher average incomes than women.
The value of parental or caregiver time is an important issue in cost-effectiveness analyses and should not be overlooked in studies of the costs of OME. This study shows the feasibility of measuring and quantifying the monetary value of such time. For instance, the imputed value of housekeeping, whether or not the caregiver is employed outside the home, contributes greatly to the overall indirect costs associated with illness due to AOM and presumably with other forms of otitis as well. The OMD provides a measurement tool that directly assesses the daily time spent by the parent taking care of the child because of an episode of AOM. These daily recordings allow one to account for the time of the parent that may not necessarily be found if one relies on questions regarding the days lost from work alone. Time represents the underlying value that must be summed if one is to capture the true value that an illness exacts in monetary terms. In this manner, the value of parental time translates into an important aspect of the opportunity cost that can be attributed to AOM in this young pediatric population.
Accepted for publication August 11, 1998.
Presented at the 13th Annual Meeting of the American Society of Pediatric Otolarynogology, Palm Beach, Fla, May 12, 1998.
Reprints: Ramsey Alsarraf, MD, MPH, Box 356515, Department of Otolaryngology–Head and Neck Surgery, University of Washington School of Medicine, Seattle, WA 98195 (e-mail: email@example.com).