Direct Costs of COPD in the U.S.: An Analysis of Medical Expenditure Panel Survey (MEPS) Data

Abstract

Chronic obstructive pulmonary disease (COPD) is a costly cause of morbidity and mortality in the U.S. The objective of this study was to use contemporary national data—specifically, those from the 2000 Medical Expenditure Panel Survey (MEPS)—to estimate direct costs of COPD in the U.S. from an all-payer perspective. Due to constraints of MEPS data, indirect costs were excluded from our analyses, as were costs of long-term oxygen therapy and costs from nursing homes and long-term care facilities. Two methods of cost estimation were employed. First, we estimated resources used and expenditures incurred by individuals with COPD that were directly attributable to the disease (attributable cost approach). Second, we compared overall medical expenditures of patients with COPD to those of the non-COPD population; the resulting difference represented excess costs of COPD. Approximately 1.7% (n = 144) of the nearly 8,300 persons in the analysis data set aged ≥ 45 years used medical resources and incurred expenditures related to treatment of COPD. Mean attributable costs per patient were estimated at $2,507, with more than one-half of these costs ($1,365) associated with hospitalization. Mean excess costs of COPD, after adjustment for sociodemographic factors and smoking status, were substantially higher, at $4,932 per patient. Results of our study indicate that COPD-associated healthcare utilization and expenditures are considerable, and that annual per-patient costs of COPD are comparable to those of other chronic diseases of the middle-aged and elderly.

Key words: :

• Chronic obstructive pulmonary disease
• Costs and cost analysis
• Medical economics
• Databases
• Statistical data interpretation

Introduction

Chronic Obstructive Pulmonary Disease (COPD) represents an important cause of morbidity and mortality in the U.S., and is a major source of healthcare costs [1]. It has been estimated that 10–16 million persons in the U.S. have been diagnosed with some form of COPD, and that as many as 14–16 million more remain undiagnosed [2-5]. COPD currently ranks as the fourth leading cause of death in the U.S. after heart disease, cancer, and cerebrovascular disease, with about 120,000 annual deaths [4-7]. The prevalence, morbidity, and mortality of COPD have increased over time, and are expected to increase in tandem with the aging of the U.S. population [8&9].

By definition, COPD is a slowly progressive disease of the airways characterized by airflow limitation and gradual loss of lung function that is not fully reversible [5][8]. In the U.S., the term COPD includes chronic bronchitis, chronic obstructive bronchitis, emphysema, and combinations of these conditions [5]. Symptoms of COPD include wheezing, cough, sputum production, and dyspnea, with the latter being the most prominent and disabling symptom and the most common reason for patients to seek medical care [10]. Diagnosis is usually made on the basis of personal medical history, physical examination, and results from pulmonary function testing [9]. Treatment options for COPD are aimed largely at symptom control; these include inhaled and oral bronchodilators, anti-inflammatory drugs, and supplemental oxygen [9]. However, cessation of smoking is the only intervention that has successfully proven to reduce rate of decline in lung function [11-13]. Surgical treatment, such as bullectomy, lung volume reduction surgery, or lung transplant, remains an option for relatively few patients [9][14]. Pulmonary rehabilitation has emerged as a means to alleviate symptoms, promote better physical and social functioning, and improve quality of life of COPD patients, and it is now considered essential for many patients formerly deemed untreatable [15].

COPD poses a substantial economic burden to the U.S., with direct and indirect costs estimated at $37.2 billion in 2004 [5][9]. Direct costs (i.e., those related to hospital care, physician and other professional services, home care, nursing home care, and pharmacy) accounted for $20.9 billion, while indirect costs (lost earnings due to illness and lost future earnings resulting from death) comprised $16.3 billion [5][9]. The major resource utilization drivers for COPD are physician office and hospital outpatient services (8–12 million visits annually), emergency department services (1.5 million visits annually), and hospitalizations (more than 725,000 annually in which COPD is listed as the primary discharge diagnosis) [4&5]. Long-term oxygen therapy may also contribute significantly to costs [3][16].

Most COPD cost analyses performed to date have important limitations, with many focusing on a single payer (e.g., Medicare or a commercial insurer) or using outdated national data [3][7][17-23]. In the time since data for these studies were collected, dramatic changes have occurred in the organization, financing, and delivery of health care in the U.S. Moreover, changes over time in sociodemographic and clinical factors (i.e., treatment of COPD) associated with COPD incidence and prevalence suggest that previous estimates of these costs may misstate the true economic burden of COPD [24]. The objective of the present study was to use contemporary national data—specifically, those from the 2000 Medical Expenditure Panel Survey (MEPS)—to estimate direct costs of COPD in the U.S. from an all-payer perspective.

Methods

Data Source

Data used in this study were obtained from electronic files from the 2000 Full Year Consolidated File of MEPS, an effort co-sponsored by the Agency for Healthcare Research and Quality (AHRQ) and the National Center for Health Statistics. MEPS is the third in a series of AHRQ national probability surveys on the financing and use of medical care in the U.S. The National Medical Care Expenditure Survey (NMCES, also known as NMES-1) was conducted in 1977 and the National Medical Expenditure Survey (NMES-2) in 1987 [25]. Since 1996, MEPS has continued this series with design enhancements and efficiencies that provide a more current data resource to capture the changing dynamics of the healthcare delivery and insurance systems [25].

MEPS provides nationally representative estimates of healthcare use and expenditure, sources of payment, and insurance coverage in the civilian non-institutionalized U.S. population through the use of three surveys [25]. The first and most comprehensive of these is the Household Component (HC), which collects detailed data on demographic characteristics, health conditions, health status, use of and expenditures for medical care services and prescribed medications, access to care, satisfaction with care, health insurance coverage, income, and employment. The Medical Provider Component (MPC) is a follow-back survey that collects data through telephone interviews and record abstraction pertaining to clinical and financial characteristics of medical and pharmacy events. These events are identified in a sample of healthcare providers and pharmacies used by survey participants in 2000. Finally, the Insurance Component (IC) collects data on health insurance plans obtained through private and public sector employers. Together, the three surveys yield comprehensive data that provide national estimates of the level and distribution of healthcare use and expenditures, support health services research, and can be used to assess the implications of healthcare policy [25].

Study data consisted of surveys completed in Rounds 3–5 of Panel 4 and Rounds 1–3 of Panel 5 (i.e., all rounds for MEPS panels conducted during calendar year 2000 comprising the 2000 Full Year Consolidated Data File); all final person-level variables were consolidated into a single file [25]. In total, MEPS contains survey data for 25,096 persons [25].

Study Population

Information on demographics, health status, and healthcare utilization and expenditures was collected from survey participants. Our analyses included survey participants with self-reported COPD, and, for analytical comparison, those who did not report having COPD (i.e., the remaining survey sample). We identified persons with self-reported COPD in the MEPS database through three-digit codes from the International Classification of Disease, Ninth Revision (Clinical Modification) (ICD-9-CM) system that had been applied to medical conditions reported by survey participants. Included were ICD-9-CM codes 491 (chronic bronchitis), 492 (emphysema), and 496 (chronic obstructive airway disease, not elsewhere classified). Only persons aged 45 years and older were included in the analyses; attribution of medical resource utilization to COPD is uncommon in persons younger than 45 years old [3].

Analyses

Sociodemographic and Clinical Characteristics

Prior to conducting data analyses, we assessed the sociodemographic and clinical characteristics of persons with and without COPD. Sociodemographic characteristics included age (45–54, 55–64, 65–74, and 75+ years), sex, race/ethnicity, marital status, level of education (in years), employment status, annual personal income, and type of medical insurance coverage. Clinical characteristics included health status, smoking status, presence of particular comorbid conditions, and MEPS eligibility status (i.e., whether full data were available, or whether data were missing due to death or other causes and required imputation).

Resource Utilization

We assessed use of inpatient, outpatient, and pharmacy services in 2000 among patients included in the study. Inpatient resource utilization data (which were contained in the Hospital Inpatient Stays File) included direct hospital care, including room and board, diagnostic and laboratory work, x-rays, and similar charges, as well as any physician services included in hospital charges. Outpatient resource utilization included data retrieved from event files pertaining to healthcare visits in office or clinic settings, emergency rooms, and in the home. These data were obtained from the Office-Based Medical Provider, Outpatient Visits, Home Health, and Emergency Room Files. Visit-level records in each file contain the visit date, medical condition codes (i.e., ICD-9-CM diagnosis codes identifying the purpose of the visit), expenditures, and sources of payment associated with the visit.

Pharmacy resource utilization was recorded in the Prescribed Medication File and collected directly from pharmacy records. Only payments for prescription medication were included in our analyses, as payments for over-the-counter medications and alternative care services are not included in MEPS total expenditures. The Prescribed Medications File includes a mix of nonproprietary and trade names, National Drug Codes for each medication, and associated expenditure data, including the amount paid for each prescription and the basis of source of payment.

Expenditures Assessment

Cost measurement focused on annual direct expenditures (i.e., inpatient, outpatient, and pharmacy) for COPD from an all-payer perspective. Expenditures in MEPS reflect payments for healthcare services, defined as the sum of third-party payments from all sources and out-of-pocket payments. Indirect payments not related to specific medical events, such as Medicaid Disproportionate Share and Medicare Direct Medical Education subsidies, were not part of the MEPS database and, therefore, were excluded from our analyses. Also, expenditures for long-term oxygen therapy were not included in the analyses because of insufficient data in MEPS. Oxygen therapy supplies, along with various other types of equipment unrelated to COPD, were categorized as “medical equipment” in the Other Medical Expenses File. Therefore, we could not disaggregate costs in this category. Furthermore, indirect costs associated with output losses and family costs as a consequence of illness (e.g., loss of work time and productivity) were not included in this study. Due to constraints of the design of MEPS, the costs of procedures and laboratory tests were not listed separately; only the total amount paid for each visit (including payment for these services) was identified. Also, because there was not a clear distinction between clinic visits and office-based provider visits, costs for both types of outpatient visits were combined.

Our methodology for estimating direct costs of COPD in this analysis employed two distinct approaches. First, we used an attributable cost approach to estimate resources used and expenditures incurred by individuals with COPD. In this approach, services were attributed to treatment of COPD if corresponding records in the MEPS database listed the ICD-9-CM codes used to define COPD. We then used an excess cost approach to estimate and compare overall medical expenditures among persons with and without a diagnosis of COPD; the resulting difference represented excess costs associated with COPD. All cost estimates are stated as 2000 U.S. dollars.

Statistical Considerations

MEPS employs a complex, probabilistic survey design that entails stratification, clustering, and multi-stage sampling criteria (including over-sampling within strata). Analyses and between-group comparisons of patient characteristics were performed on unweighted measures. MEPS sampling weights were used to generate point estimates for all measures of resource utilization and costs. Measures of variability were based on estimates of standard error derived through use of Taylor series linearization methods to account for the complex survey design. Standard error estimates were computed using the SURVEYMEANS and SURVEYREG procedures found in the Statistical Analysis System (SAS) software package [26], which was used for all analyses. In our excess-cost approach for cost estimation, we used multivariate linear least-squares regression to adjust the raw estimates for potentially confounding factors, including age, sex, race, current smoking status, marital status, and number of years of education. We also performed additional analyses using the excess-cost approach that involved stratifying patients by age (i.e., < 65 years vs. ≥ 65 years) and then controlling for comorbid conditions (i.e., cardiovascular and other circulatory disorders, arthritis and other musculoskeletal disorders, diabetes and other endocrine disorders, depression and other mental disorders, neoplasms, and neurological diseases).

Results

Sociodemographic Characteristics

A total of 144 persons aged 45 years or older in the MEPS 2000 Full Year Consolidated File reported using medical resources or incurring expenditures due to COPD (Table 1). Clinical characteristics of the 8,150 non-COPD persons in the analysis dataset are reported in Table 1 as well, as these persons were used in the analyses featuring our excess-cost approach. The majority of persons with COPD were older than age 65 years, white, married, and had completed 12 or fewer years of education. Patients without COPD tended to be younger and had completed more years of education. Males were slightly overrepresented among COPD patients, and underrepresented among persons without COPD. Only about one-fifth of patients with COPD were employed, with the remainder either retired or unemployed due to disability; most had personal incomes of less than $20,000. Non-COPD persons were more likely to be employed and had higher incomes. Private insurance was the predominant insurance type among all persons.

Table 1.  Sociodemographic characteristics of COPD and non-COPD patients

Variables	COPD patients (n = 144)		Non-COPD patients (n = 8,150)
	N	%	N	%
Age (years):
45–54	18	12.5%	3,260	40.0%
55–64	36	25.0%	2,032	24.9%
65–74	49	34.0%	1,523	18.7%
75+	41	28.5%	1,335	16.4%
Mean age (years)	67.7		60.7
Sex:
Men	76	52.8%	3,718	45.6%
Women	68	47.2%	4,432	54.4%
Race/Ethnicity:
Non-Hispanic White	116	80.6%	5,596	68.7%
Non-Hispanic African-American	17	11.8%	1,094	13.4%
Hispanic	9	6.3%	1,239	15.2%
Asian	1	0.7%	184	2.3%
Other	1	0.7%	37	0.5%
Marital status:
Never married	8	5.6%	475	5.8%
Married	71	49.3%	5,265	64.6%
Separated or divorced	26	18.1%	1,185	14.5%
Widowed	39	27.1%	1,222	15.0%
Education (highest level achieved):
< 12 years	67	46.5%	2,150	26.4%
12 years	36	25.0%	2,596	31.9%
>12 years	39	27.1%	3,287	40.3%
Unknown	2	1.4%	117	1.4%
Employment status:
Employed	32	22.2%	4,213	51.7%
Retired	55	38.2%	2,046	25.1%
Unemployed due to disability	33	22.9%	564	6.9%
Unemployed, not due to disability	2	1.4%	891	10.9%
Unknown	22	15.3%	436	5.3%
Personal annual income (2000):
Less than $20,000	97	67.4%	4,098	50.3%
$20,000–39,999	29	20.1%	2,166	26.6%
$40,000–59,999	14	9.7%	1,098	13.5%
$60,000 or more	4	2.8%	778	9.5%
Insurance coverage:
Public only	58	40.3%	1,762	21.6%
Any private	82	56.9%	5,580	68.5%
Uninsured	4	2.8%	808	9.9%

Percentages calculated from sample size, and due to rounding, may not total 100% within each category.

^†If data from the third interview round in 2000 were not available, values were estimated from the second and first rounds.

*COPD patients were identified by the three-digit ICD-9-CM codes 491, 492, and 496.

^‡Data collected in third round of MEPS 2000 survey.

^§When first entered MEPS.

Self-Reported Health Status and CoMorbidities

On average, more than one-half of patients with COPD reported their health status as “fair” or “poor”—by contrast, only 20% of non-COPD persons reported “fair” or “poor” health (Table 2). About 28% of COPD patients were smokers, versus only 17% of persons without COPD. The majority of COPD patients had cardiovascular and other circulatory disorders and nearly one-half had arthritis or other musculoskeletal disorders. Overall, non-COPD persons had fewer comorbidities. Fifteen of the identified patients with COPD died during 2000.

Table 2.  Clinical characteristics of COPD and non-COPD patients

Variables	COPD subjects (n = 144)		Non-COPD subjects (n = 8,150)
	N	%	N	%
Health status
Excellent	3	2.1%	1,504	18.5%
Very good	23	16.0%	2,402	29.5%
Good	35	24.3%	2,559	31.4%
Fair	35	24.3%	1,117	13.7%
Poor	42	29.2%	506	6.2%
Unknown	6	4.2%	62	0.8%
Current smoking status
Nonsmoker	86	59.7%	5,874	72.1%
Smoker	40	27.8%	1,404	17.2%
Unknown	18	12.5%	872	10.7%
Selected comorbidities
Cardiovascular and other circulatory disorders	80	55.6%	2,282	28.0%
Arthritis and other musculoskeletal disorders	68	47.2%	2,347	28.8%
Diabetes and other endocrine disorders	34	23.6%	1,256	15.4%
Depression and other mental disorders	27	18.8%	912	11.2%
Neoplasms	24	16.7%	618	7.6%
Neurological diseases	19	13.2%	580	7.1%
MEPS population eligibility status (unweighted)
Eligible subjects with complete data	128	88.9%	7,904	97.0%
Subjects died during 2000 (missing data imputed)	15	10.4%	139	1.7%
Subjects disaggregated from MEPS for other reasons (missing data imputed)	1	0.7%	107	1.3%

Percentages calculated from sample size, and due to rounding, may not total 100% within each category.

^†If data from the third interview round in 2000 were not available, values were estimated from the second and first rounds.

*COPD patients were identified by the three-digit ICD-9-CM codes 491, 492, and 496.

Resource Utilization and Expenditures

Attributable Cost Approach

Direct medical costs attributable to COPD were estimated at $2,507 (standard error [SE] ± $65) per patient (year 2000 U.S. dollars) (Table 3). More than one-half ($1,365; 54%) of this estimate was generated by inpatient hospitalizations for COPD, which were experienced by approximately 7.2% of COPD patients, with an average of 1.4 admissions per patient and a mean length of stay of 5 days.

Table 3.  Mean ( ± SE) per-person direct medical expenditures attributable to COPD in 2000

Variables	COPD-related
Inpatient	$1,365 ($37)
Outpatient
Office/clinic visits	$312 ($6)
Emergency room visits	$66 ($2)
Home healthcare visits	$389 ($11)
Subtotal	$767 ($18)
Pharmacy	$375 ($10)
Total	$2,507 ($65)

SE = standard error. Standard errors were estimated by the Taylor-series linearization method.

*COPD-related costs account for services associated with self-reported conditions ascribed to one or more ICD-9-CM diagnosis codes 491, 492, or 496.

Home health visits and pharmacy were the next largest categories of expenditure, contributing $389 (16%) and $375 (15%), respectively, to attributable costs. Approximately 20% of patients with COPD had home health care visits in 2000 for COPD and other causes; they required, on average, 48 visits during the year, or nearly one visit per week. Almost all patients with COPD were prescribed medications, with nearly 87% of patients prescribed COPD-related medications. For COPD patients receiving medication, the mean number of medications dispensed per patient was 9.3, with 3.1 (33%) of these specifically for treatment of COPD. Most patients (93.1%) had office/clinic visits for COPD and other causes, averaging about one visit per month, with the COPD-attributable visits generating $312 (12%) in total COPD expenditure. Although 28% of COPD patients had emergency room visits for COPD and other causes, with a mean of 1.8 visits per patient, the expenditures for COPD-attributable visits averaged only $66 per patient, or approximately 3% of total expenditure.

Excess Cost Approach

We also generated estimates of excess direct costs of COPD (Table 4). Estimates of direct medical costs using this methodology were substantially larger than our attributable cost estimates—$5,645 (SE ± $61) per patient before multivariate adjustment and $4,932 (SE ± $90) per patient after multivariate adjustment for age, sex, race, current smoking status, marital status, and years of education. Patients aged ≥ 65 years had higher adjusted excess cost than patients aged < 65 years ($5,679 vs. $3,942, respectively). Because patients with COPD tend to have a high prevalence of comorbid conditions with potential for impacting healthcare utilization and expenditure, we performed an additional analysis in which comorbidity was added to the multivariate adjustment of our cost estimates. Our estimate of the adjusted excess cost of COPD when controlling for presence of the comorbidities listed in Table 2 (in addition to the other confounding factors mentioned above) was $3,755 (SE ± $158). Looking at the major drivers of COPD expenditures, we found results from the excess-cost approach to parallel estimates generated using the attributable-cost approach. Inpatient hospitalizations were the major driver of COPD expenditures, accounting for 58.4% of our unadjusted estimate. Outpatient expenditures (i.e., office/clinic visits, emergency room visits, and home healthcare visits) accounted for about 27.5% of our unadjusted estimate, whereas pharmacy expenditures were 14.1% of total COPD expenditure.

Table 4.  Mean ( ± SE) per-person total direct medical expenditures in 2000: COPD versus non-COPD patients

Variables	COPD subjects	Non-COPD subjects	Excess cost of COPD
Inpatient	$4,913 ($132)	$1,614 ($67)	$3,298 ($64)
Outpatient
Office/clinic visits	$1,916 ($33)	$1,108 ($3)	$808 ($3)
Emergency room visits	$199 ($5)	$90 ($1)	$109 ($1)
Home healthcare visits	$683 ($18)	$155 ($4)	$528 ($4)
Other medical supplies and equipment	$186 ($2)	$81 ($0)	$105 ($0)
Subtotal	$2,984 ($59)	$1,433 ($7)	$1,551 ($8)
Pharmacy	$1,543 ($98)	$747 ($4)	$796 ($5)
Total (unadjusted)	$9,439 ($93)	$3,795 ($65)	$5,645 ($61)
Total (adjusted)			$4,932 ($90)

SE = standard error. Standard errors were estimated by the Taylor-series linearization method.

*Adjusted for age, sex, race, current smoking status, marital status, and number of years of education.

Discussion

In this study, we report estimates of direct medical resources use and associated costs of COPD among non-institutionalized adult civilians who sought treatment for the condition in 2000. Using data from the most recent consolidated full-year data file of MEPS allowed us to estimate resource use and costs from a multi-payer perspective. To our knowledge, no other published studies to date have used these data to estimate direct medical costs of COPD.

Commonly cited COPD prevalence estimates, as well as prevalence-based estimates of direct medical cost in the U.S. adult population (i.e., $20.9 billion among 10–16 million diagnosed patients) would suggest per-patient annual costs in the range of $1,300–$2,100 [2][4][9]. Although comparison of national cost estimates is complicated by differences in the cost components included in each study, estimates in this range are lower than estimates reported by Strassels and colleagues ($1,637 in 1987 U.S. dollars, or $3,283 in 2000 U.S. dollars) [21], and somewhat higher than the $896 in 1996 U.S. dollars, or $1,025 in 2000 U.S. dollars, reported by Wilson and colleagues [7]. By comparison, our estimates of attributable and excess costs of COPD range from $2,507 to $4,932 (2000 U.S. dollars), respectively. As noted previously, our estimates do not include costs associated with use of long-term oxygen therapy and, perhaps more significantly, those incurred by institutionalized adult civilians. Had our estimates reflected such costs, they would be even higher—and would compare more closely to those reported by Mapel and colleagues ($6,039 in 1997 U.S. dollars, or $6,711 in 2000 U.S. dollars) [23], Grasso and colleagues ($4,971 in 1992 U.S. dollars, or $6,837 in 2000 U.S. dollars) [18], or the recently-published estimate of $4,120 (2000 U.S. dollars) that Halpern and colleagues [27] derived from the U.S. sample of the international COPD survey, Confronting COPD in North America and Europe. On the whole, results of our study suggest that annual per-patient costs of COPD are of similar magnitude to those of other chronic diseases of the middle-aged and elderly, including diabetes [28], arthritis [29-32], and cardiovascular disease (CVD) [33&34].

Study Limitations

Some limitations of our study bear mention. We used patient-reported data from a national survey to ascertain whether or not a participant had COPD. The possibility of faulty recall or other ascertainment bias among MEPS participants cannot be ruled out. Furthermore, the self-reports may not perfectly conform to diagnoses made by physicians. We do note, however, that medication costs in MEPS were validated with pharmacy records for prescription drugs.

Approximately 1.7% (n = 144) of the nearly 8,300 persons in the analysis data set aged ≥ 45 years used medical resources and incurred expenditures related to treatment of COPD. We acknowledge that this prevalence of COPD is lower than what has been reported from formal COPD epidemiological studies. However, it is important to note that MEPS only accounts for persons with COPD who were treated for their condition, and that people not using direct medical resources or incurring direct medical expenditures during the survey period were excluded. Because most people with COPD tend to be mildly or moderately affected, and because healthier persons are less likely to use medical resources and to incur costs as a result of their disease, COPD prevalence in MEPS is bound to be less than estimates from an all-inclusive approach [21]. Consequently, we argue that because COPD was reported by relatively few respondents, it is not possible to project results of this study to estimate a national economic impact. Yelin and colleagues reached a similar conclusion about COPD using 1996 MEPS data in their national study of medical care expenditures for respiratory conditions [22]. More to the point, MEPS was designed to collect information about healthcare utilization and costs for making public policy, not necessarily for making robust epidemiological estimates.

Due to constraints of MEPS data, costs associated with long-term oxygen therapy were not included in our direct cost estimates. Nonetheless, supplemental home oxygen is reported to be the most costly component of outpatient therapy [1]. Ward and colleagues estimated the yearly cost of long-term oxygen therapy for COPD to be $3,708 in 1994 U.S. dollars, or $4,591 in 2000 dollars—35% of their total annual direct medical costs of COPD [3]. Grasso and colleagues estimated that oxygen-dependent COPD Medicare beneficiaries had per capita total expenditures 1.6 times that of the average beneficiary with COPD ($13,786 versus $8,842 in 1992 U.S. dollars, or $18,961 versus $12,161 in 2000 U.S. dollars) [18]. Excluding long-term oxygen therapy also was an important but insurmountable limitation in the study conducted by Strassels and colleagues [21][35].

Because MEPS collects information only from the non-institutionalized civilian population, our analyses do not include utilization and costs attributable to COPD patients in nursing homes and long-term care facilities. These additional costs could be substantial. Indication of the magnitude of the additional costs can be found in the study conducted by Ward and colleagues who estimated the average yearly attributable cost for COPD-related nursing home care to be $17,868 in 1994 U.S. dollars, or $22,122 in 2000 U.S. dollars [3].

Finally, there may be a strong correlation between disease severity and healthcare utilization and costs of treating COPD. A retrospective pharmacoeconomic analysis conducted by Hilleman and colleagues showed direct costs of COPD treatment to vary from $1,681 in 1999 U.S. dollars, or $1,750 in 2000 U.S. dollars, per patient per year for patients with mild COPD, to $5,037 in 1999 dollars, or $5,245 in 2000 U.S. dollars, for patients with moderate COPD, to $10,812 in 1999 dollars, or $11,257 in 2000 U.S. dollars, for patients with severe COPD [36]. Because our data were self-reported without regard to disease severity, we were not able to stratify results of our analyses accordingly.

Conclusions

COPD remains a leading cause of morbidity and mortality in the United States. Changes that have occurred in recent years within the U.S. healthcare system, as well as changes in the epidemiology of COPD, warranted a new assessment of the direct costs of the disease. MEPS provided the best current data for this estimate. Results of our study indicate that COPD-associated healthcare utilization and expenditures are considerable when examined from both attributable and excess-cost perspectives. Our results also suggest that per-person annual costs of COPD are comparable to those of several other chronic diseases of the middle-aged and elderly, including diabetes, arthritis, and cardiovascular disease.