INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Considerable progress has been made in treating reversible airway disease, yet the prevalence and burden of asthma has increased in recent years. Current estimates of asthma prevalence indicate that 14 to 15 million persons in the United States are afflicted with asthma (1). In 1993, 198,000 hospitalizations and 342 deaths were attributable to asthma in persons younger than 25 yr of age (2). The hospitalization rate for those with asthma younger than 25 yr of age increased by 28% from 1980 to 1993 (2). In the general population, the overall death rate from asthma increased from 13.4 per million population (3,154 deaths) in 1982 to 18.8 per million population (5,106 deaths) in 1991, an overall increase of 40% (3). Between 1982 and 1992, the age-adjusted self-reported prevalence of asthma increased by 42%, from 34.7 to 49.9 cases per 1,000 (3).

The appropriate use of maintenance medications and medical therapy allows many asthmatics to control their asthma, but the cost of treatment can be high (4). Treating acute attacks also consumes considerable medical resources (5). In addition, asthma symptoms often result in work and school absenteeism and lead to a decreased quality of life (6).

Work by Weiss and coworkers (7) suggests that the economic burden of asthma in the United States was approximately $6.2 billion in 1990. This point estimate was derived using multiple data sources and charges rather than costs, and by simplifying assumptions regarding medication use. The purpose of this study was to provide a more accurate estimate of the cost of asthma by using expenditure-based data from a single data source, the 1987 National Medical Expenditure Survey (NMES) and by calculating point estimates for costs and their associated confidence intervals.

Data Source

The 1987 NMES was a national probability sample of approximately 35,000 subjects representative of the noninstitutionalized, civilian U.S. population. The NMES collected person-level information on demographics and income as well as data on all health care expenditures and encounters. Reported expenditures for medications and ambulatory visits reflect total payments (rather than charges), including consumer out-of-pocket and third-party payer expenditures. The NMES also captures information on health-related periods of disability. As part of conducting the 1987 NMES, interview data on health care use and expenditures were supplemented by a separate medical provider survey to reduce potential bias from relying solely on self-reported data (8). The medical provider survey focused on medical events where self-reporting was expected to be least accurate, and where inaccurate or missing data would have the greatest impact on the precision of expenditure estimates. In particular, high-cost episodes of care involving multiple providers (e.g., hospital stays), and persons with comprehensive coverage (e.g., Medicaid) were targeted (9).

Population weighing factors allowed prevalence and cost estimates to be made for the entire U.S. population. The weights also allowed the calculation of confidence intervals for prevalence and cost estimates for prescribed medicines, physician visits, hospital visits, and missed work or school days using SUDAAN software (10).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

We identified persons with asthma by searching the NMES data tapes for any occurrence of code 493 of the International Classification of Diseases-Ninth Revision (ICD-9) (11) associated with any medical event. Each event (i.e., hospitalization, office/clinic visit, ER visit, prescription, and disability days) was associated with as many as four ICD-9 codes. Population weighting factors were used to estimate the number of persons in the United States who sought medical care, or missed work or school, or were otherwise limited by asthma. Estimates were derived for both children (persons younger than 18 yr of age) and adults. Because of the NMES survey design, our estimates included only those who incurred direct or indirect expenditures associated with asthma in 1987.

Expenditures were inflated using factors based on 1987 to 1994 changes in per capita expenditures for prescription drugs, professional medical services, and hospital and related services (12). Estimates for work or school loss and restricted activity days were inflated to 1994 dollars using the annual percent change in earnings and compensation from 1987 to 1994 (12). All expenditures are reported in 1994 dollars unless otherwise stated.

Direct expenditures included payments for ambulatory care visits, hospital outpatient services, hospital inpatient stays, emergency room visits, physician and facility services, and prescribed medicines. The NMES survey did not contain questions relating to self-treatment with over-the-counter (OTC) medications, so these costs were not included. Indirect costs of asthma morbidity included the economic value of lost productivity because of missed work, restriction from normal work activities, and missed school.

The human capital approach was taken to value lost productivity using a person's daily personal wage rate (13). A lost work or school day was defined in NMES as a day when the respondent missed at least half a day. A restricted activity day was defined in NMES as a day when the respondent was unable to engage in normal activities for at least half a day. The adult daily wage rate was calculated by dividing the person's annual income from wages and tips by the number of days worked during the year. The economic loss caused by missed work was estimated by multiplying the self-reported number of work days missed because of asthma by the daily wage rate. This figure was then multiplied by the population weighting factor and summed across all individuals to yield population estimates.

Because the degree of restriction was not specified for restricted activity days, two levels of the wage rate, 50% and 25%, were used in calculations of economic loss caused by restricted activity. Given the lack of empirical evidence on the valuation of restricted activity days, 50% was used in the primary analysis and varied in sensitivity analyses. Because restricted activity days were not differentiated into work or leisure time, each restricted activity day was multiplied by five-sevenths, under the assumption that each day of the week had an equal chance of being a restricted activity day, and that each subject worked five days per week. Restricted activity day costs were calculated only for adults.

The cost of asthma attributable to missed housekeeping work was calculated for those who indicated they were taking care of the home/ family. The 1987 U.S. Bureau of Labor's median wage for private household service ($6,760) (14) was imputed as the wage for housekeepers. The median annual wage among all wage earners with asthma was $11,000, making the $6,760 amount appear to be a reasonable and conservative estimate. Lost productivity (bed days and restricted activity days) associated with housekeeping was valued at five-sevenths, assuming a five-day work week.

Missed school days for children 5 to 17 yr of age and bed days for children younger than 5 yr of age were valued by attributing lost productivity to the parent or guardian, under the assumption that they stayed home to take care of the sick child. In cases where two persons were linked to the child, the lowest income was used to estimate lost productivity. Approximately 39% of the time, one of the parents/ guardians did not have any income. In these cases, the median wage of private household service was imputed as previously described for housekeeping. Bed days for children younger than 5 yr of age were valued at five-sevenths, assuming a five-day work week for the parent/ guardian.

Because of concerns regarding misclassification of asthma in those 0 to 4 and 5 to 11 yr of age and in those 45 yr and older, we conducted sensitivity analyses on the direct expenditures for these two groups. In the younger ages, nonasthmatic conditions such as bronchiolitis and wheezing may have been diagnosed as asthma, whereas in the older population, the primary concern was that in subjects with chronic obstructive pulmonary disease (COPD), the COPD might have been misdiagnosed as asthma. There were as many as four ICD-9 codes associated with each event. Even though all events in this study had at least one ICD-9 of asthma associated with it, not all the ICD-9 codes for each event were necessarily for asthma. In an attempt to create a "pure" asthma estimate in the younger age groups, the sensitivity analysis was done by excluding events with ICD-9 codes unrelated to asthma. For those 45 yr of age and older, the sensitivity analysis was done by excluding all costs for those who had any COPD-related events.

Because there was a concern that asthma may have been misclassified as COPD among subjects 40 yr of age or younger (15), those 40 yr of age or younger with COPD-related events (ICD-9 codes 491, 492, and 496) were also classified as having asthma for the purpose of this study. The impact of this assumption was evaluated using sensitivity analysis.

Lastly, we plotted cumulative direct expenditures against cumulative population in order to explore the data for a natural dichotomous break point regarding per-patient expenditures (i.e., high-cost and low-cost patients). These groups were analyzed with respect to direct costs, age, race, and sex.

Data Validation

In assessing potential outlier data points, the highest cost records were examined and excluded if the clinical procedure codes were not clinically related to asthma.

Because of the self-report nature of the data, some medications that were linked to asthma were obviously not used in the treatment of asthma (e.g., gemfibrozil). For this reason, three physicians independently categorized each medication into one of three groups based on the clinical use of the medication in each physician's experience. The criteria for the groupings were: Group 1 (most likely used for asthma), Group 2 (possibly used to treat asthma or some complication of asthma), and Group 3 (definitely not used for asthma). Differences in classification for a particular medication were resolved via discussion and majority consensus. Our primary analysis includes only Groups 1 and 2. A sensitivity analysis was undertaken with only Group 1 medications included.

Because of the NMES survey design, there are overlapping days of disability in the disability file. In order to avoid double counting, this decision rule was followed: if a school loss or work loss occurred on the same day as a bed day, the bed day was not counted. There were no instances in our data of school loss and work loss overlap. Children younger than 5 yr of age had no occurrences of school loss.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Sample Characteristics

The weighted sex and racial characteristics by age categories for persons with economic expenditures for asthma in 1987 are shown in Table 1. There were more male subjects with asthma at the younger ages, but the prevalence was greater in female subjects after age 18. Among children 0 to 4 and 5 to 17 yr of age, African-American account for 19 and 27% of the total, respectively. Among adults, however, African-Americans account for only 12% of the total.

An estimated 4.7 million people sought care or incurred indirect costs for asthma in 1987. More than 2.7 million were adults and 2 million were children.

Direct Costs

Prescribed medicines. Approximately 30.5 million (95% CI, 24.4 to 36.8 million) prescriptions were filled for asthma in 1987. The average cost per prescription was $27, with an annual average of 6.5 prescriptions per person with asthma. Theophylline accounted for 25.9% of the total prescriptions filled, followed by albuterol (22.8%), metaproterenol (10.0%), combination products (7.8%), inhaled corticosteroids (6.2%), and prednisone (5.3%). Expenditures for prescribed medicines were estimated to be $817.0 million (95% CI, $640.1 to $994.0 million) (Table 2).

Office and clinic visits. More than 13 million (95% CI, 10.2 to 16.6 million) office and clinic visits related to asthma occurred for a total cost of more than $616.2 million (95% CI, $481.7 to $750.7 million), for an annual average of $131 per person.

Hospital outpatient visits. More than 1.5 million (95% CI, 0.7 to 2.3 million) hospital outpatient visits were made for the treatment of asthma. Total estimated expenditures for hospital outpatient visits were $566 million (95% CI, $276 to $856 million); the annual average expense was $120 per person.

Emergency Room (ER) visits. Approximately 1.2 million (95% CI, 0.9 to 1.4 million) ER visits were made in 1987. The annual average cost per person was $74. The total cost of these visits was $348 million (95% CI, $245 to $451 million).

Hospital stays. More than 445,000 hospital stays (95% CI, 299,000 to 624,000) related to asthma occurred in 1987. The mean length of stay associated with an asthma hospitalization was 4.7 days. Expenditures for these hospitalizations were estimated to be $2.8 billion (95% CI, $1.7 to $3.9 billion), for an average annual expense of $595 per person.

The 1994 total direct costs were estimated to be $5.1 billion dollars (95% CI, $3.3 to $7.0 billion) (Table 2), for an annual cost of $1,096 per person with asthma. Hospitalizations accounted for the greatest proportion (54.4%) of direct costs, prescribed medicines accounted for 15.9%, office and clinic visits 12.0%, hospital outpatient visits 11.0%, and emergency room visits 6.8%.

Indirect Costs

Work and housekeeping. The total value of lost work productivity (housekeeping loss plus work loss) was estimated to be $242.7 million (95% CI, $80.7 to $404.8 million) (Table 2). Work loss accounted for approximately 2.1 million days and $222.1 million. Housekeeping wages were estimated to be $20.6 million for the 769,000 days of lost productivity.

School loss. The economic value of caregiver time associated with the estimated 3.6 million school days missed for children 5 to 17 yr of age was $194.5 million (95% CI, $123.9 to $265.1 million).

Bed days. The economic value of caregiver time associated with bed days for children younger than 5 yr of age was estimated to be $18.5 million (95% CI, $5.5 to $31.4 million) for the 369,000 days.

Restricted activity. The estimated cost associated with restricted activity days (valued at 50% of a full work day) was $217.5 million (95% CI, $60.8 to $374.2 million) for the 18.3 million days of reduced activity.

Total indirect costs were $673.2 million (Table 2). Work loss accounted for 33.0% of the total indirect costs, restricted activity accounted for 32.3%, school loss days accounted for 28.9%, housekeeping accounted for 3.1%, and bed days (for children younger than 5 yr of age) accounted for 2.7%.

Total Costs

It can be seen in Table 2 that the total cost for asthma was $5.8 billion (95% CI, $3.6 to $8.1 billion). Indirect costs accounted for 12% and direct costs accounted for 88% of the total. The average annual cost per patient was $1,238 ($5.82 billion/4.7 million patients). The distribution of direct costs by age is shown in Table 3. Hospitalization is the greatest cost component in every age category, and for children 0 to 4 yr of age, hospitalization made up more than 74% of direct expenditures. Hospitalization expenditures are proportionally the lowest for children 5 to 17 yr of age, but they tend to use more ER services and prescribed medicines.

In an attempt to understand the distribution of direct costs among the sample, individual annual direct costs were cumulated and plotted against cumulative number of persons with asthma having direct costs as shown in Figure 1. This graph reveals that approximately 20% of persons with direct costs account for slightly more than 80% of all direct costs. The annual per-patient direct costs stratified by high-cost patients (the 20% of asthmatics accounting for 80% of direct costs) and low-cost patients (the 80% of persons with asthma account for 20% of direct costs) are shown in Table 4. The average annual cost was $2,584 for the high-cost patients versus $140 for the low-cost patients. Hospitalizations account for the largest portion of direct costs (65%) among high-cost patients, whereas among low-cost patients, hospitalizations account for only 1% of the annual direct costs. The average age of the high-cost patients was 36 yr compared with 29 yr for the low-cost patients. In the high-cost group, 79.1% of the patients were Caucasian and 13.0% were African-American, whereas in the low-cost group, 75.5% of the patients were Caucasian and 18.3% were African-American. High- and low-cost patients were not different with respect to sex distribution.

View larger version (27K): [in this window] [in a new window]   Figure 1.   Cumulative proportion of people with direct expenditures of asthma are plotted on the y-axis against the cumulative proportion of their direct expenditures.

Sensitivity Analysis

Consistent with accepted economic methods, sensitivity analyses were conducted to evaluate the robustness of the cost estimates when key assumptions were modified. Four sensitivity analyses were conducted to test assumptions made in the primary analysis.

When only medications most likely to be used for asthma (i.e., Group 1 medications) were used in the analysis. the estimated total direct cost decreased by about 2%, from $5.1 to $5.0 billion.

Changing the restricted activity day wage rate from 50 to 25% decreased the total cost estimate by less than 2%, from $5.8 to $5.7 billion.

When only "pure" asthma events (i.e., those events with ICD-9 codes of asthma only) were included, direct costs for those 0 to 4 yr of age decreased by 14%, from $791.4 to $679.7 million, and direct costs for those 5 to 11 yr of age decreased by 15%, from $455.7 to $385.7 million. In children 0 to 4 yr of age, the three most common ICD-9 codes in addition to asthma were otitis media (17%), bronchitis (13%), and acute nasopharyngitis (10%). For those 5 to 11 yr of age, the most common ICD-9 codes in addition to asthma were allergic rhinitis (30%), strep throat (11%), and pneumonia (10%). In those 45 yr of age and older, excluding all costs for persons with any COPD event, decreased direct costs by 12%, from $2.7 to $2.4 billion. The most common concurrent codes in this group were emphysema (62%), bronchitis (7%), and heart diseaseill defined (5%).

The inclusion of those patients deemed to be misclassified as having COPD (i.e., those younger than 40 yr of age and reporting COPD-related events) added $31 million or 0.6% to the direct cost estimate.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The total estimated cost of asthma in the United States in 1994 was $5.8 billion (95% CI, $3.6 to $8.1 billion). Direct costs represent the majority (88%) of the total, and indirect costs comprise only 12%. Hospitalizations represent the single greatest cost category, accounting for 48% of the total cost overall. For children 0 to 4 yr of age, hospitalizations make up a substantially larger proportion of costs (74%).

Among those children 5 to 17 yr of age, hospitalizations are still the largest cost category, but ER visits are 16.5% of direct costs, in contrast to less than 7% for both the younger and the older ages. Possible reasons for this difference include access to care issues, disease severity differences, or poor management of disease.

The sample used in this study consisted only of those patients with asthma who had direct medical or indirect expenditures. Our findings indicate that there are differences in the proportion of racial minorities seeking care by age group. Specifically, there is a greater proportion of African-Americans in the younger versus the older age groups. This may be a reflection of limited health care access for African-Americans older than 18 yr of age. An alternative explanation is that the prevalence of asthma among this group decreases as age increases. This alternative is not a likely explanation because several studies have shown that African-Americans, at any age, have a higher prevalence of asthma than do Caucasians (16, 17). Given this, it is expected that the proportional change in African-Americans seeking care for asthma with increasing age would be at least as great as the proportional change in Caucasians. Our findings, however, indicate a slightly lower number of African-Americans 18 yr of age and older sought care for asthma than did those younger than 18 yr of age whereas the number of Caucasians who sought care almost doubled for those older than 18. This issue should be examined more closely in future studies to assess the issue of changes in access as it relates to the management of asthma.

A prevalence estimate of the cost of disease can provide insight regarding the categories of cost that can be targeted for economic savings (18). This prevalence estimate of the cost of asthma indicates that hospitalizations account for the largest proportion of costs, and thus represent the area with the largest potential for savings. This is consistent with findings from other studies (7, 19, 20).

The findings of this study indicate that in 1987, theophylline was the most commonly used medication (26%), followed by albuterol (23%), with inhaled corticosteroids making up only 6.2% of the total. More recent data suggest that inhaled beta agonists are the most frequently used medication in a health maintenance organization (21). In light of asthma treatment recommendations from the National Asthma Education Program (22, 23), it is expected that these therapy patterns have shifted toward greater use of inhaled corticosteroids. Information from these studies can be useful to help track changes in the pharmacologic treatment of asthma over time and the impact that drug therapies may have on healthcare use.

The comparison of the direct costs of high-cost patients to those of low-cost patients is important because it identifies areas of potential intervention. As with the nonstratified analysis, hospitalizations accounted for the greatest proportion of direct costs. The estimated per-patient annual hospitalization expense of almost $1,700 for the high-cost patients (versus less than $2 for the low-cost patients) may indicate patients with severe disease and/or poorly controlled disease. Among the high-cost patients, there is a clear potential for savings by preventing hospitalization or perhaps by shortening length of stay. The high-cost patients were found to have a somewhat higher mean age than the low-cost patients (36 versus 29 yr, respectively), and were somewhat more likely to be Caucasian, but no differences were evident by sex. The area of high- versus low-cost patients is one that deserves more attention to help elucidate differences in health status, severity of disease, insurance coverage, and demographic characteristics that may be contributing to the observed differences in direct costs.

In order to gain a broader perspective on cost of illness estimates, it is helpful to compare them with the overall cost of care and to other diseases. In 1993 dollars, an estimate of the overall direct cost of medical care for all diseases was $572 billion, and the direct cost of all respiratory diseases was $38.3 billion (20). That analysis was done using the 1987 NMES data, and costs were inflated to 1993 dollars. Our estimate of the direct cost of asthma is a small fraction (0.9%) of the overall direct costs of disease, but it is a substantial percentage (13%) of the cost of all respiratory diseases.

A comparison of the estimates by Weiss and coworkers (7) with the estimates presented here is shown in Table 5. To facilitate this comparison, the estimates of Weiss and coworkers have been inflated to 1994 dollars using the same methods as were used in the NMES-based study. Although the two estimates of total costs are similar, there are differences in both the proportion of costs accounted for by individual cost categories and the assumptions made in the estimates. The NMES-based estimate of direct costs is $5.1 billion, which is very close to the $4.8 billion estimate from the work by Weiss and coworkers. The two estimates of indirect costs are quite different, however. The NMES total indirect cost estimate is $674 million versus more than $2.9 billion in the work done by Weiss and coworkers. The estimate of Weiss and coworkers did include a mortality cost component that accounted for more than 12% ($943 million) of the total estimate, and the NMES analysis did not. A mortality estimate was not included in the NMES study because cause-specific mortality data are not available from the NMES, and inclusion of data from other sources would have limited our ability to construct confidence intervals. This deficiency may cause our estimate to be an underestimation of the true indirect cost of asthma. Even after accounting for the mortality cost difference, the estimate of Weiss and coworkers of indirect cost is still greater than the NMES estimate, by more than $1.3 billion. About $841 million of this difference is attributable to missed school days. There are two main reasons for the difference, the estimated number of missed school days and the valuation of missed school day. Weiss and coworkers, using National Health Interview Survey (NHIS) data, estimated more than 10 million lost school days. In the NMES data, asthmatics 5 to 17 yr of age missed more than 7 million days of school for any reason, and had an additional 3 million bed days for any reason, for a total of approximately 10 million days. However, asthma-specific school loss was estimated at 3.6 million days. Our valuation of caregiver time, described above, yields a per-day estimate of $59 ($195 million/3.3 million days). This is in contrast to the $104/day ($1,036 million/10 million days) estimate used by Weiss and coworkers (Table 5). The lower number of estimated school days combined with the lower valuation per day make our estimate smaller. This comparison suggests that the indirect costs of asthma may be substantially lower than previously thought.

There are several limitations that must be acknowledged in this study. First, the NMES data are subject to certain limitations (24), primarily the self-report nature of the data. Second, estimates do not reflect demographic changes and population growth that have occurred since 1987. Third, changes in treatment patterns that may have occurred since 1987 are not reflected in the estimates. Costs may be substantially lower if more preventive therapy is being utilized, especially if preventive therapy decreases hospitalizations. Fourth, self treatment with OTC medications was not included; consequently, this estimate probably understates actual expenses. Fifth, there is evidence that the prevalence of asthma has increased in recent years (2, 3). Our analysis did not adjust for any increase in prevalence of disease, so our estimate is likely somewhat of an underestimate of the true current cost of the disease.

Other limitations in this study are more theoretical in nature. When a lost work day or school day occurred, we valued time as the cost associated with lost productivity as reflected by the worker's wage. This may overstate or understate the cost of lost productivity.

The sensitivity analyses show that our results are stable even with changes in several key assumptions used in this study, namely, misclassification of asthma as COPD, medications used to treat asthma, and the wage rate associated with restricted activity. The changes in direct cost range from 12 to 16% in the sensitivity analysis that considered misclassification of asthma in the young (0 to 11 yr of age) and older (45 yr of age and older) age groups. This decrease is well within the 95% CI estimated for total direct expenditures.

Asthma is a disease that consumes a large share of economic resources. The findings of this study suggest that considerable cost savings could be achieved by minimizing hospitalizations. Because only a small proportion of those with asthma are consuming the majority of resources, interventions aimed at these high-cost patients could be an effective strategy in reducing the morbidity and cost of asthma. An important next step is to more closely define the demographic and severity characteristics of the high-cost patients so that interventions can be tailored to their particular situations.