Long-term medical costs and resource ... - Wiley Online Library

Arthritis & Rheumatism (Arthritis Care & Research) Vol. 61, No. 6, June 15, 2009, pp 755–763 DOI 10.1002/art.24545 © 2009, American College of Rheumatology

ORIGINAL ARTICLE

Long-Term Medical Costs and Resource Utilization in Systemic Lupus Erythematosus and Lupus Nephritis: A Five-Year Analysis of a Large Medicaid Population TRACY LI,1 GINGER SMITH CARLS,2 PANTELIS PANOPALIS,3 SARA WANG,2 TERESA B. GIBSON,2 4 AND RON Z. GOETZEL

Objective. To estimate the long-term direct medical costs and health care utilization for patients with systemic lupus erythematosus (SLE) and a subset of SLE patients with nephritis. Methods. Patients with newly active SLE were found in the MarketScan Medicaid Database (1999 –2005), which includes all inpatient, outpatient, emergency department, and pharmaceutical claims for more than 10 million Medicaid beneficiaries. The date a patient became newly active was defined as the earliest observed SLE diagnosis code, with a 6-month clean period prior to the diagnosis. This method identified 2,298 patients with a consecutive followup of 5 years. A reference group of patients without SLE was constructed using propensity score matching. Nephritis was assessed based on diagnosis and procedure codes involving the kidney. Results. Mean annual medical costs for SLE patients totaled $16,089 at year 1, which is significantly greater (by $6,831) than that for reference patients. Costs decreased slightly at year 2 but then increased yearly at an average rate of 16% through year 5, to $23,860. SLE patients without nephritis (n ⴝ 1,809) had costs $967–3,756 higher than the reference patients. SLE patients with nephritis (n ⴝ 489) had costs $13,228 –34,907 greater than the reference group. Inpatient visits for the nephritis subgroup were 0.6 –1.0 per capita, which are approximately twice the rate for all SLE patients and 3 to 4 times higher than the reference group. Conclusion. SLE is a costly condition to treat. Medical expenses incurred by SLE patients increase steadily over time, particularly for patients with nephritis.

INTRODUCTION Systemic lupus erythematosus (SLE) is a chronic systemic autoimmune disorder affecting multiple organ systems in the body. Many SLE patients (25– 60%) develop renal disease over time and may require expensive treatments and procedures such as dialysis and a transplant (1–3). How-

Supported by a research grant from Bristol-Myers Squibb. 1 Tracy Li, PhD: Bristol-Myers Squibb, Princeton, New Jersey; 2Ginger Smith Carls, MA, Sara Wang, PhD, Teresa B. Gibson, PhD: Thomson Reuters, New York, New York; 3Pantelis Panopalis, MD: University of California, San Francisco; 4 Ron Z. Goetzel, PhD: Thomson Reuters, New York, New York, and Rollins School of Public Health, Emory University, Atlanta, Georgia. Dr. Li owns stock in Bristol-Myers Squibb. Address correspondence to Tracy Li, PhD, PO Box 4000, Mail Stop: J23-02, Bristol-Myers Squibb, Princeton, NJ 08543. E-mail: [email protected]. Submitted for publication August 8, 2008; accepted in revised form February 23, 2009.

ever, the long-term outcomes and costs of SLE are not well understood, in part because large numbers of patients with SLE can be difficult to follow due to the episodic and recurring nature of this relatively rare condition. Although previous studies have found that patients with SLE have high medical care costs, few studies have assessed costs over a long period of time (beyond 1 year), particularly in the US (4 –14). Therefore, it is not clear if these high costs are sustained over time. Recent evidence suggests that few SLE patients experience prolonged remission (15). Instead, most patients experience frequent cycles of disease flare and remission or experience near constant disease activity. Previous studies have not considered lupus nephritis as a co-occurring condition, which may be important to SLE treatment costs. The annual cost to treat a patient with chronic kidney disease has been estimated at $14,000 –22,000 (3). The objective of this study is to examine long-term medical costs and utilization patterns of SLE patients and a subset of SLE patients with nephritis. 755

756

Li et al

Table 1. Diagnosis and procedure codes used to find systemic lupus erythematosus patients with nephritis* Condition Acute glomerulonephritis Nephrotic syndrome Chronic glomerulonephritis Nephritis, not otherwise specified Acute renal failure Chronic kidney disease Renal failure, unspecified Kidney biopsy

ICD-9-CM 580.0, 580.4, 580.8, 580.9 581.1, 581.2, 581.3, 581.8, 581.81, 581.9 582.0, 582.1, 582.2, 582.4, 582.8, 582.81, 583.9 583.0, 583.1, 583.2, 583.4, 583.6, 583.7, 583.8, 583.81, 583.9 584, 584.5, 584.6, 584.7, 584.8, 584.9 585, 585.1, 585.2, 585.3, 585.4, 585.5, 585.6, 585.9 586 55.23, 55.24, 55.39, 55.21

Hemodialysis or peritoneal dialysis

39.95, 88.7, 54.98, 39.42, 39.27, 96.6, 99.15, 99.05, 99.06, 99.19, 99.21, 99.24, 99.25, 99.28

Kidney transplant

55.51, 55.52, 55.53, 55.54, 55.61, 55.69

CPT-4

50200, 50205, 50555, 50557, 50559, 50574, 50576, 50578 90935, 90937, 93990, 90945, 90947, 36832, 36145, 36833, 36831, 90918, 90919, 90920, 90921, 90922, 90923, 90924, 90925, 90939, 90940, 90989, 90993, 90997, 90999, 99512, 99559, 99601, 99602 00862, 00868, 50300, 50320, 50323, 50325, 50327, 50328, 50329, 50340, 50360, 50365, 50370, 50380, 50547

* ICD-9-CM ⫽ International Classification of Diseases, Ninth Revision, Clinical Modification; CPT-4 ⫽ Current Procedural Terminology 4.

PATIENTS AND METHODS Data source. The patient sample for this study was drawn from the MarketScan Medicaid Claims and Encounters Research Database (Thomson Reuters, Montvale, NJ), which was constructed from unidentified health insurance claims and represents the health care experience of more than 10 million Medicaid beneficiaries in 8 geographically dispersed states from 1999 to 2005. Medicaid is a state-run health insurance program for low-income and categorically “needy” people (both requirements must be met), funded jointly by each US state and the federal government (16). Medicaid covers children, the elderly, those who are visually impaired and/or have a disability, and others who are eligible for financial assistance. This large database enables longitudinal analysis of a rare condition such as SLE. For this study, we followed patients with SLE for up to 5 years after a treatment episode. The Medicaid Database includes demographic and enrollment information; inpatient, outpatient, and outpatient pharmacy claims; and claims related to home medical care and other medical services covered by Medicaid programs. Both fee-for-service and managed-care plans are represented in the database. Patients. SLE patients were identified as having at least 1 inpatient claim with a diagnosis code for SLE (International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] code 710.0), or 2 SLE outpatient claims during an office visit and/or emergency department visit that were at least 30 days apart; the second outpatient claim was required to exclude “rule-out” diagnoses found mainly on laboratory or radiology claims and to ensure

that patients had active disease, as reflected by utilization of medical services. For patients who met these criteria, the earliest date of SLE diagnosis was defined as the index date. We found newly active patients by searching for all patients who had at least a 6-month period without SLE diagnosis at the start of a treatment episode. Because it captures the entire index episode of care, this approach facilitates the interpretation of results. It is also similar to the approach used in other claims-based studies of SLE (8,9) and other conditions (17). Patients selected in this study may include both newly diagnosed patients and existing SLE patients with a new episode of care; therefore, we refer to the patients as newly active patients. Additionally, we restricted inclusion to SLE patients who had at least 1 year of continuous enrollment after the index date. The results shown in the tables and figures included in this study are based on the cohort of patients who were continuously followed for 5 years after the index date. We also tracked SLE patients who met our criteria and received treatment related to nephritis. We classified patients as having nephritis if they had claims with any of the diagnoses or treatment codes shown in Table 1 during the 5-year followup period. Those patients with any dialysis or renal transplant during the 5-year followup period were classified as patients with end-stage renal disease (ESRD). We dropped outliers, i.e., patients with claims totaling more than $800,000 in a given year, and this excluded 12 patients from the SLE group. To assess the incremental costs associated with SLE, we constructed a matched reference group of patients without SLE but who otherwise shared similar characteristics. Patients eligible for the comparison group were drawn from a 10% random sample of the entire Medicaid population

Estimation of Medical Costs in SLE and Lupus Nephritis

757

represented in the database and were randomly assigned an index date. We restricted inclusion in the reference group to beneficiaries with at least 6 months of enrollment before the index date and 1 year of continuous enrollment after the index date. Each SLE patient was matched to a member of the potential reference group using propensity score matching, a technique that summarizes patient characteristics into a single propensity score (18,19). Propensity score matching has been applied in similar studies of SLE (9) and other conditions such as rheumatoid arthritis (17). Patient characteristics used in matching included the following (measured at the index date or during the 6-month clean period): age categories, sex, race/ethnicity, state Medicaid program, whether the patient was enrolled in a capitated insurance plan, year of the index date, Medicaid eligibility category, dual eligibility for both Medicaid and Medicare, the Charlson Comorbidity Index (20), and the number of psychiatric diagnostic groups (21). Study outcomes and analytical methods. The outcomes for this study were annual medical costs and resource utilization, examined on an annual basis for 5 years. Medical costs were measured by total payments made to providers, including any copayment that may have been required from the patient and any payments made by other payers (e.g., Medicare). Medical costs included all services that generated a claim to Medicaid, such as costs associated with all inpatient and outpatient care, including diagnostic and laboratory tests; inpatient stays; emergency department visits; prescription drugs; and other medical services paid by Medicaid, such as home health care. In addition to the payments made by the patient and Medicaid, we also captured payments made by Medicare, if any. All costs were adjusted to 2006 US dollars using the Consumer Price Index (22). Costs to SLE patients and the matched reference group were compared using t-tests. Approximately 41% of the patients in our sample were enrolled in capitated plans. Although service-level payments were not available for these patients, detailed use and type of services (e.g., outpatient visits, length of hospital stays, etc.) were captured in the database. Therefore, payments were imputed for patients in capitated plans using standard methods (23). Specifically, encounters for patients in capitated plans were matched to claims for similar patients who were enrolled in fee-for-service indemnity plans. This matching was based on procedure code, the year of service, and the state. We then assigned the payment on the fee-for-service claim to the matching capitated claim. Measures of annual health care utilization included inpatient admissions, length of stay, emergency department visits, outpatient doctor visits, dialysis treatments, outpatient prescription drug fills, home health visit, renal transplant, and kidney biopsy. These measures were defined using service, diagnosis, and procedure codes available in the claims data. Inpatient and outpatient services were differentiated based on the presence of room and board claims that indicated an overnight stay.

Figure 1. Cohort diagram showing the selection of 2,298 systemic lupus erythematosus (SLE) patients with 5 years of followup for the study.

RESULTS We found 2,298 SLE patients who met our inclusion criteria and had 5 years of followup (Figure 1). Twenty-one percent (n ⫽ 489) of the SLE patients with 5 years of followup had evidence of nephritis during the followup, although in any given year during the followup, a smaller number (n ⫽ 240 –286) of SLE patients had evidence of treatment for nephritis. Table 2 shows characteristics of the SLE patients with 5 years of followup and the subgroup with evidence of nephritis during the followup period. SLE patients were predominantly women (93.3% versus 89.8% of the nephritis subgroup). Approximately one-third of the SLE sample was classified as white, 28.4% were African American, 27.5% were classified as other, and 10.2% were Hispanic. The nephritis subgroup was on average younger (age 40.6 years), had less white patients (19.6%), and had more African American patients (37.8%). The reference group (not shown) was well matched to the total SLE group, with no statistically significant differences between the SLE patients and the reference group for the characteristics shown in Table 2 (P ⱕ 0.05 for all). Mean ⫾ SD medical expenditures were significantly higher for SLE patients ($16,089 ⫾ $26,041) compared with the matched reference group ($9,258 ⫾ $17,100) at

758

Li et al

Table 2. Characteristics of systemic lupus erythematosus (SLE) patients and the subset of patients with nephritis with 5 years of followup

Number of patients Age, mean years Female, % Race, % White African American Hispanic Other or missing/unknown Medicaid eligibility category (measured at index date), % Elderly, blind, or disabled Temporary aid to “needy” families Other Eligibility changed during study period Medicare eligibility, % Dual eligibility for the entire study period Dual eligibility for part, but not all, of the study period

year 1 (P ⬍ 0.001) (Figure 2). Costs for SLE patients decreased in year 2 but then increased yearly at an average rate of 16% until year 5, when annual costs totaled a mean ⫾ SD of $23,860 ⫾ $47,700. The mean annual costs for SLE patients were ⬃$3,800 – 8,200 greater than for the matched reference group over the 5-year period (P ⬍ 0.001). Costs increased for the reference group by 14% annually. The annual mean ⫾ SD cost for the SLE nephritis subgroup in the first year was $27,463 ⫾ $39,610 followed by a 10% decrease in costs in the second year, and then a

SLE patients

SLE patients with nephritis

2,298 44.0 93.3

489 40.6 89.8

33.9 28.4 10.2 27.5

19.6 37.8 12.3 30.3

82.1 16.1 1.9 4.2

80.2 17.0 2.8 4.7

36.6 8.8

41.1 10.2

yearly increase at an average rate of 31% that reached $50,578 at year 5, which is more than double the costs of the overall SLE group. Patients who ever had ESRD during the study (n ⫽ 156) had the highest annual mean ⫾ SD costs, ranging from $43,614 ⫾ $44,044 to $106,982 ⫾ $123,870. Patients who had nephritis but not ESRD (n ⫽ 333) had similar costs as the overall SLE group in the first 3 years (2% lower to 12% higher), but much greater costs in years 4 and 5 (46 – 61% higher than the overall SLE group). SLE patients who never had active nephritis during the study period (n ⫽ 1,809) had lower medical costs

Figure 2. Mean annual medical costs per patient over time for the total systemic lupus erythematosus (SLE) sample and subgroups who ever had end-stage renal disease (ESRD), ever had nephritis, never had nephritis, and matched reference patients.


759

Figure 3. Median annual medical care costs per patient over time for the total systemic lupus erythematosus (SLE) sample and subgroups who ever had end-stage renal disease (ESRD), ever had nephritis, never had nephritis, and matched reference patients.

than any of the other SLE subgroups, but were still $1,000 –3,800 more than the matched reference group. Costs for the reference group increased yearly at an average rate of 14%. Median costs were lower than mean costs and mirrored mean costs in terms of cost growth and comparisons between the groups (Figure 3). The only exception was nephritis patients without ESRD, where median costs did not

increase dramatically in years 4 and 5, suggesting that the cost increase occurred within the top 50% of patients with nephritis without ESRD. Median costs for the overall SLE group were $7,952 in year 1, which decreased to $7,459 in year 2 and increased to $9,777 in year 5. Median annual costs for the nephritis subgroup from year 1 to year 5 ranged from $11,144 –21,500, and for the nephritis group without ESRD ranged from $8,706 –12,441.

Figure 4. Components of medical costs for the systemic lupus erythematosus (SLE) patients and the nephritis subgroup.

760 Outpatient care, including dialysis, office visits, and other services, accounted for the largest share of costs in each year for the SLE populations, comprising 49 –56% of annual costs (Figure 4). Inpatient care accounted for the next largest share of costs, at 16 –27% of annual costs, followed by prescription drug fills at 19 –23% of costs. Results were similar for the subset of patients with nephritis, as shown in Figure 4. Outpatient costs, including office visits, dialysis, and other care, accounted for 51– 64% of annual costs. Inpatient services accounted for 18 –33% of costs, and prescription drug fills accounted for 10 –15% of costs for patients with nephritis. Utilization rates of medical care services by SLE patients, the reference group, and the subset of SLE patients with nephritis are shown in Table 3. Approximately onequarter of SLE patients had an inpatient admission in the first year followed by ⬃18% in each year afterward, compared with ⬃10 –11% of the reference group (P ⬍ 0.001). A higher percentage of SLE patients with nephritis (29.4 – 38.4%) had an inpatient admission. For the SLE group, the mean length of stay per admission was highest in the first year (6.0 days), decreased in the second year to 4.4 days, and then fluctuated between 5.2 and 4.7 days for the remaining years. The average length of stay in the reference group was 4.2–5.8 days. The subgroup of SLE patients with nephritis had the longest average length of stay, at 7.2 days in the first year of followup, declining to 4.5 days, and then increasing to 6.3 days at year 5. Forty-four percent of the SLE patients had an emergency department visit in year 5, compared with 33% of the reference group (P ⬍ 0.001). Of the SLE patients with nephritis, 53% had an emergency department visit in the first year of followup, with the percentage dropping in the next 2 years before increasing to 52% of the patients. Annual outpatient office visits ranged from 6 to 7 for the SLE group, from 3 to 4 for the reference group, and were ⬃7 for the subgroup of SLE patients with nephritis. Outpatient pharmaceutical drug fills increased over time in the SLE group, from 55 fills to 63 fills annually per patient compared with the reference group that had 37– 47 fills annually (P ⬍ 0.001). Home health visits increased from 3.3 visits to 5.6 visits on average per SLE patient. Approximately 4 –5% of the SLE patients received dialysis each year compared with ⬃1% of the matched reference group. Of the SLE patients with nephritis, 36 – 46% were receiving dialysis. We observed an average of 21–26 dialysis treatments per patient each year.

DISCUSSION In this study, we found SLE to be a costly condition to treat over the long term, particularly for patients with nephritis. Costs incurred by SLE patients were ⬃$3,800 – 8,200 higher than the matched reference group. The nephritis subgroup appeared to drive increasing costs for the SLE group, with annual costs more than doubled in the overall SLE group. Even when ESRD patients were excluded, costs for nephritis patients with ESRD were 46 – 61% higher in years 4 and 5. In contrast, SLE patients without nephritis had costs 17–30% lower than the overall SLE group, but

Li et al still higher than the reference group. Similarly, the use of medical care services by the SLE patients was substantially higher than the reference group. We found that costs for SLE patients decreased between the first and second year. A possible explanation is that we selected newly active patients who needed intensive medical care at year 1, and then the costs decreased at year 2 when the disease was managed. The cost increases for years 2 through 5 may be due to disease flares or disease progression after the index episode. This is consistent with previous work that found that few SLE patients had long periods of remission, and instead experienced either recurrent disease flares or had continuously active disease (15). We also found large SDs indicating a wide variation in the costs of individual patients, so we also examined median costs. Patterns were generally similar, albeit at lower cost levels for median values, compared with the means. This observed wide variation in costs among individual patients is consistent with the disease characteristic where some patients may experience low disease activity and low costs, and others may experience high disease activity and higher costs. Compared with previous studies examining direct medical costs for SLE patients (4 –14), our cost estimates at the first year are very similar to those of Brunner et al ($15,949 in 2006 US dollars) (7), Gironimi et al ($15,586 in 2006 US dollars) (13), Nichol et al (⬃$13,000 in 2006 US dollars, based on inferences from graphs) (8), and Carls et al ($20,131 in 2006 US dollars) (9). All of these studies included US patients and were based on medical claims data, except for the study by Gironimi et al, who used a standard questionnaire to collect utilization data and surveyed local providers to obtain cost data. In a cross-sectional estimate, Gironimi et al found that US direct medical costs for SLE patients were more than twice as high as those for Canadian patients. Higher US prices of health services and a more severe disease case-mix of patients in the US explained this cost difference, but did not explain higher utilization of medical care services. Our first-year direct medical cost estimates were ⬃1.6 – 4.5 times the cost estimates that used Canadian fee schedules, after converting estimates to 2006 US dollars (4 – 6,12,14). Our direct medical care costs were ⬃4.8 times higher than those found in Germany (10) and 3.1 times higher than those in the UK after making the same currency and inflation adjustments (11). In terms of health care utilization, emergency department visit rates in our study were 1.2 (7,24) to 2.4 (4) times higher than for previous studies, with the more similar estimates coming from US studies. The percentage of patients in our study with an inpatient stay was almost identical to that reported in a study by Yelin et al (24), who compared utilization rates for patients in fee-for-service and health maintenance organization plans. Our percentage of patients with an inpatient stay was 1.6 –2.2 times higher than for patients in 2 Canadian studies (4,12), but much lower than for patients in a study by Brunner et al (7). These comparisons with previous studies suggest that a major cause of differences in cost estimates across studies is the difference in the fees charged for care in various countries. Questionnaire-based studies tend to have lower


761

Table 3. Average medical care utilization over 5 years for the entire SLE sample (n ⴝ 2,298), the reference group (n ⴝ 2,298), and the subgroup of SLE patients with nephritis (n ⴝ 489)*

Inpatient admissions Any admissions, % Total SLE Reference group Nephritis subgroup Rate per patient Total SLE Reference group Nephritis subgroup Length of stay per admission, mean days Total SLE Reference group Nephritis subgroup Emergency department visits Any visits, % Total SLE Reference group Nephritis subgroup Rate per patient Total SLE Reference group Nephritis subgroup Rate per emergency department user Total SLE Reference group Nephritis subgroup Outpatient care Outpatient/doctor office visits per patient Total SLE Reference group Nephritis subgroup Prescription drug fills per patient Total SLE Reference group Nephritis subgroup Home health visits per patient Total SLE Reference group Nephritis subgroup ESRD-related services, % of patients Dialysis Total SLE Reference group Nephritis subgroup Kidney biopsy Total SLE Reference group Nephritis subgroup Kidney transplant Total SLE Reference group Nephritis subgroup

Year 1

Year 2

Year 3

Year 4

Year 5

24.4 10.4 38.4

17.5 10.8 29.4

18.1 11.5 30.4

18.3 12.0 31.5

18.1 11.7 30.0

0.5 0.2 1.0

0.3 0.2 0.7

0.3 0.2 0.6

0.3 0.2 0.7

0.4 0.2 0.7

6.0 4.2 7.2

4.4 4.9 4.5

5.2 4.5 5.0

4.7 4.6 5.7

5.2 5.8 6.3

48.8 30.5 53.0

47.6 32.8 52.4

46.6 33.4 48.2

46.1 33.4 51.3

44.2 32.6 51.5

1.6 0.8 1.8

1.5 0.8 1.9

1.5 0.9 1.8

1.4 0.9 1.6

1.3 0.9 1.5

3.3 2.7 3.4

3.2 2.6 3.6

3.3 2.7 3.6

3.0 2.7 3.1

3.0 2.7 2.9

6.9 3.7 7.4

6.4 3.8 7.1

6.0 3.7 6.7

5.9 3.6 6.7

5.6 3.4 6.6

54.5 37.4 57.5

55.8 40.2 58.7

58.3 43.3 62.0

61.2 45.1 65.6

62.9 47.2 68.7

3.3 2.6 3.2

3.9 3.0 3.9

4.5 3.5 4.7

5.0 3.8 4.9

5.6 4.3 5.6

3.6 1.0 36.2

3.9 1.2 38.9

4.2 1.2 41.9

4.6 1.3 46.2

4.6 1.6 46.2

0.8 0.0 4.2

0.4 0.3 2.0

0.4 0.1 1.6

0.4 0.0 2.3

0.4 0.1 2.0

0.2 0.0 0.9

0.2 0.1 0.9

0.3 0.0 1.3

0.2 0.1 0.9

0.4 0.0 1.8

* SLE ⫽ systemic lupus erythematosus; ESRD ⫽ end-stage renal disease.

costs (except Gironimi et al, who found approximately the same costs as in our study) (13), which could be the result of patient recall bias or because our claims database captures a broader set of medical services than clinic-based studies. The higher emergency department and inpatient visit rates in our population suggest that we may have a

more complex group of patients, although we did not have a measure of SLE disease activity to test this variable. Medicaid patients may also tend to use the emergency department more often if they are accustomed to doing so during prior periods when they lacked insurance. Our study is among the first to demonstrate a long-term

762 trend in medical costs of SLE. A comprehensive assessment of the costs of SLE has been difficult to perform due to the rarity of the disease and its complex nature. Since SLE affects multiple organ systems and a number of health care professionals with different specialties may be simultaneously involved in patient care, capturing the totality of the services used by the patients is challenging for clinicbased data sources. Furthermore, the disease activity of SLE fluctuates over time, and consequently, shorter-term estimates do not provide an overall view of the cost burden of the disease. In the present study, we used a large Medicaid database of more than 10 million enrollees. There are several advantages of using this medical claims database. First, the size of the database and the availability of a long-term followup enable us to capture a large population of SLE patients, including a relatively large sample of patients with lupus nephritis. Second, the Medicaid database includes detailed and comprehensive information on service utilization and payments to providers of all types. Our cost information is likely to be more complete than data from selected centers, and is more representative of actual spending than data sets that rely on charges or other approximations of costs. Third, utilization information is more reliable in administrative data compared with estimates derived from surveys that require patients to recall their medical visits. We acknowledge that there are potential limitations related to the use of an administrative claims database. There was no information on reasons why patients left our sample, such as losing eligibility for Medicaid or death (except for the rare cases when death occurs in the hospital). Therefore, our study is essentially a cross-sectional study of patients with 5 years of Medicaid eligibility and could be subject to survivorship bias. To investigate this issue, we examined patients with shorter periods of followup because mean annual medical costs were significantly higher the shorter amount of time a patient was enrolled in Medicaid. SLE patients with 1 year of followup had annual average costs of ⬃$25,000, suggesting our 5-year sample may have less severe disease than those who were in the Medicaid database for less than 5 years. Time trends for patients with 2– 4 years of enrollment were similar to the group with 5 years of enrollment, despite higher cost levels. We identified SLE patients based on the ICD-9-CM diagnosis codes on their insurance records, which can be less precise than physicians’ clinical assessments captured in medical charts (25). To address this concern, patients included in our SLE population using an outpatient diagnosis were required to have at least 2 outpatient claims a minimum of 30 days apart to avoid the inclusion of patients with “rule-out” diagnoses of SLE. Since we do not have patient history, we view our sample as SLE patients at the start of an episode of care, including both newly diagnosed patients and patients who were asymptomatic 6 months prior to the index date. Another limitation of administrative data in the US is that care for patients with ESRD is almost completely covered by Medicare, and therefore our data may not capture all the costs of care for patients with ESRD. Medicare

Li et al pays the costs of dialysis after 3 months and much of the costs for renal transplant (26). We found an average of 20 dialysis treatments per patient per year, which is less than what is typically observed in the clinic. An examination of specific claims suggested that we observed costs for claims that were paid by both Medicaid and Medicare, but we lacked claims paid solely by Medicare. Dialysis typically costs $8,500 a year (⬃3 treatments per week), and renal transplants cost more than $100,000 (27). Patients in this study appear to use medical care services more often than in some of the previous studies; however, since there is no detailed clinical information in the database, we cannot determine if this is due to the care-seeking patterns of Medicaid patients or patient severity. Medicaid patients in our study had low income and met the requirements for a “needy” category. Patients with low income often have difficulty obtaining health insurance and may have unmet medical needs that would lead to higher costs. Since SLE can limit one’s ability to work, it is possible that some patients became eligible for Medicaid coverage due to having a more severe case of SLE. Our analysis emphasizes relative changes in costs, e.g., the time trend and differences between the reference group and SLE patients, which may be more generalizable than the absolute levels of costs. The difference in costs between the reference group and the SLE patients should be interpreted as the amount that is attributable to SLE. Our reference group incurred costs within the range of national estimates for Medicaid beneficiaries ($13,524 per disabled beneficiary and $2,102 per nondisabled adult, fiscal year 2005) (28). We also compared enrollee characteristics of the 8 state Medicaid programs included in our sample with nationwide data and found that these state Medicaid programs were generally representative. Our state Medicaid programs included more white beneficiaries (50% versus 41% nationwide), fewer beneficiaries eligible for “aged, blind, and disabled” benefits (16% versus 24% nationwide), fewer dually eligible beneficiaries (13% versus 17% nationwide), and more children age ⬍18 years (59% versus 48% nationwide) (29). This study provides new information about medical costs and long-term health care utilization for a large sample of patients with SLE. We found that costs incurred by the SLE patient population increased steadily over time, particularly for patients with nephritis. Interventions that help physicians and patients with SLE manage this complex and costly condition may reduce the likelihood of renal impairment, improve overall health, and reduce medical care utilization and expenditures.

AUTHOR CONTRIBUTIONS All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Li had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study conception and design. Li, Carls, Panopalis, Wang, Gibson, Goetzel. Acquisition of data. Wang, Gibson, Goetzel. Analysis and interpretation of data. Li, Carls, Panopalis, Wang, Gibson, Goetzel.


763

ROLE OF THE STUDY SPONSOR This study was funded by Bristol-Myers Squibb, and used existing databases for disease-specific research. Dr. Li, as an employee of Bristol-Myers Squibb, contributed to the study design, data analysis, writing of the manuscript, and approval of the content as an author, together with the research team.

14. 15. 16.

REFERENCES 17. 1. Agrawal N, Chiang LK, Rifkin IR. Lupus nephritis. Semin Nephrol 2006;26:95–104. 2. Mok CC. Prognostic factors in lupus nephritis. Lupus 2005; 14:39 – 44. 3. Thorp ML, Eastman L, Smith DH, Johnson ES. Managing the burden of chronic kidney disease. Dis Manag 2006;9:115–21. 4. Clarke AE, Esdaile JM, Bloch DA, Lacaille D, Danoff DS, Fries JF. A Canadian study of the total medical costs for patients with systemic lupus erythematosus and the predictors of costs. Arthritis Rheum 1993;36:1548 –59. 5. Clarke AE, Petri MA, Manzi S, Isenberg DA, Gordon C, Senecal JL, et al. An international perspective on the well being and health care costs for patients with systemic lupus erythematosus. J Rheumatol 1999;26:1500 –11. 6. Clarke AE, Petri M, Manzi S, Isenberg DA, Gordon C, Senecal JL, et al. The systemic lupus erythematosus Tri-Nation study: absence of a link between health resource use and health outcome. Rheumatology (Oxford) 2004;43:1016 –24. 7. Brunner HI, Sherrard TM, Klein-Gitelman MS. Cost of treatment of childhood-onset systemic lupus erythematosus. Arthritis Rheum 2006;55:184 – 8. 8. Nichol M, Shi S, Knight T, Wallace D, Weisman M. Eligibility, utilization, and costs in a California Medicaid lupus population. Arthritis Rheum 2004;51:996 –1003. 9. Carls G, Li T, Panopalis P, Wang S, Mell AG, Gibson TB, et al. Direct and indirect costs to employers of patients with systemic lupus erythematosus with and without nephritis. J Occup Environ Med 2009;51:66 –79. 10. Huscher D, Merkesdal S, Thiele K, Zeidler H, Schneider M, Zink A. Cost of illness in rheumatoid arthritis, ankylosing spondylitis, psoriatic arthritis and systemic lupus erythematosus in Germany. Ann Rheum Dis 2006;65:1175– 83. 11. Sutcliffe N, Clarke AE, Taylor R, Frost C, Isenberg DA. Total costs and predictors of costs in patients with systemic lupus erythematosus. Rheumatology (Oxford) 2001;40:37– 47. 12. Lacaille D, Clarke AE, Bloch DA, Danoff D, Esdaile JM. The impact of disease activity, treatment and disease severity on short term costs of systemic lupus erythematosus. J Rheumatol 1994;21:448 –53. 13. Gironimi G, Clarke AE, Hamilton VH, Danoff DS, Bloch DA, Fries JF, et al. Why health care costs more in the US: comparing health care expenditures between systemic lupus ery-

18. 19.

20. 21.

22. 23. 24.

25. 26. 27.

28. 29.

thematosus patients in Stanford and Montreal. Arthritis Rheum 1996;39:979 – 87. Clarke AE, Panopalis P, Petri M, Manzi S, Isenberg DA, Gordon C, et al. SLE patients with renal damage incur higher health care costs. Rheumatology (Oxford) 2008;47:329 –33. Urowitz MB, Feletar M, Bruce IN, Nez DI, Gladman DD. Prolonged remission in systemic lupus erythematosus. J Rheumatol 2005;32:1467–72. US Department of Health and Human Services. Medicaid at a glance 2005: a Medicaid information source. Baltimore: Centers for Medicare and Medicaid Services; 2005. Ozminkowski R, Burton W, Goetzel R, Maclean R, Wang S. The impact of rheumatoid arthritis on medical expenditures, absenteeism, and short-term disability benefits. J Occup Environ Med 2006;48:135– 48. Imbens, G. The role of propensity score in estimating doseresponse functions. Biometrika 2000;87:706 –10. Leuven E, Sianesi B. PSMATCH2: Stata module to perform full Mahalanobis and propensity score matching, common support graphing, and covariate imbalance testing, version 3.0.0. 2003. URL: http://ideas.repec.org/c/boc/bocode/s432001. html. Charlson M, Pompei P, Ales K, MacKenzie C. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987;40:373– 83. Ashcraft ML, Fries BE, Nerenz DR, Falcon SP, Srivastava SV, Lee CZ, et al. A psychiatric patient classification system: an alternative to diagnosis-related groups. Med Care 1989;27: 543–57. US Department of Labor, Bureau of Labor Statistics. Consumer price index news releases, January to December 2006. URL: www.bls.gov/schedule/archives/cpi_nr.htm#2006. Little RJ, Rubin DB. Statistical analysis with missing data. New York: Wiley; 1987. Yelin E, Trupin L, Katz P, Criswell LA, Yazdany J, Gillis J, et al. Impact of health maintenance organizations and fee-forservice on health care utilization among people with systemic lupus erythematosus. Arthritis Rheum 2007;57:508 –15. Baron JA, Weiderpass E. An introduction to epidemiological research with medical databases. Ann Epidemiol 2000;10: 200 – 4. US Department of Health and Human Services. Medicare coverage of kidney dialysis and kidney transplant services. Baltimore: Centers for Medicare and Medicaid Services; 2007. National Institutes of Health, US Renal Data System. USRDS 2007 annual data report: atlas of chronic kidney disease and end-stage renal disease in the United States. Bethesda (MD): National Institute of Diabetes and Digestive and Kidney Diseases; 2007. URL: www.usrds.org/adr.htm. The Henry J. Kaiser Family Foundation. State health facts. 2008. URL: www.statehealthfacts.org. US Department of Health and Human Services. Medicaid Statistical Information System data. URL: www.cms.hhs.gov/ MedicaidDataSourcesGenInfo/02_MSISData.asp.