Cost-of-illness studies in systemic lupus erythematosus: A systematic review

Authors

Errata

This article is corrected by:

  1. Errata: Erratum: Cost-of-illness studies in systemic lupus erythematosus: A systematic review Volume 63, Issue 7, 1074, Article first published online: 29 June 2011

Abstract

Objective

To summarize the state of knowledge regarding the economic burden of systemic lupus erythematosus (SLE) and to evaluate the quality of cost-of-illness (COI) studies conducted to date.

Methods

Relevant literature was retrieved from the PubMed database in April 2010. The evaluation of identified articles was based on 7 key elements of COI studies derived from previous literature on health economics. Costs derived from each study were converted into 2008 US dollars using the Consumer Price Index and the purchasing power parity conversion rate.

Results

A total of 11 articles were included in the evaluation. The average direct costs per patient-year ranged from $3,735–$14,410. Costs of inpatient care were found to be the largest component of direct costs in most of the studies. The employment rate varied from 35.8–55%. The average duration of annual short-term sick leave ranged from 7.0–64.8 days. The mean annual indirect costs per patient ranged from $1,093–$14,614, depending on the valuation method.

Conclusion

There is a substantial economic burden, in terms of health care resource utilization, associated with SLE, as well as losses of productivity due to work capacity impairment. The cost estimates in this review should be used with caution due to significant discrepancies in methodologies across studies. Future studies should address several methodologic considerations in order to measure the true costs related to SLE. Incidence-based COI studies are needed to evaluate the lifetime costs of SLE.

INTRODUCTION

Cost-of-illness (COI) studies measure the monetary burden that a disease imposes on society caused by morbidity and premature mortality in terms of the consumption of health care resources and losses of productivity. In 1967, Rice first outlined a methodologic framework for calculating costs of illness/disability/death (1). In 1982, Hodgson and Meiners provided guidelines in detail for those intending to perform COI studies (2). Results from COI studies are valuable in providing informative data to emphasize the scope of the problem of a disease and highlight the profile of patients with that disease. They also have the potential to serve as the basis of a principal component for economic evaluations, such as cost-effectiveness analysis.

A comprehensive COI study includes direct, indirect, and intangible costs associated with the disease. The specific focus of a study may make one or the other unnecessary. Direct costs represent the opportunity costs of all kinds of resources used for treating a disease (2). They usually include direct medical costs and direct nonmedical costs. Direct medical costs refer to the medically related inputs used directly to provide the treatment, including costs associated with the diagnosis, treatment, continuing care, emergency care, and rehabilitation, whereas direct nonmedical costs refer to costs to patients and their families that are directly associated with a disease but are not medical in nature, including transportation costs, costs for household expenditures, and informal care. Indirect costs represent productivity losses related to morbidity and mortality, which can be related to labor or nonlabor activities. Indirect costs usually account for a large proportion of total costs in most COI studies. Intangible costs refer to patients' psychological pain, discomfort, anxiety, and depression, and distress related to an illness or the related treatment (3). They are very difficult to quantify in monetary terms; therefore, they are usually omitted in COI studies or presented as quality of life.

Rheumatic diseases attract enormous attention in COI studies, considering their chronic disease course, morbidity, and long-term disability (4). However, most of this attention is given to rheumatoid arthritis, which is a chronic crippling disorder with a higher prevalence among other rheumatic diseases (5, 6). It appears that there are few and sparse data available for other inflammatory rheumatic diseases, such as systemic lupus erythematosus (SLE).

SLE is a prototypical autoimmune disorder with a broad range of clinical and laboratory presentations involving almost all organ systems. The prevalence is estimated to be 52 per 100,000 in the US, 21 per 100,000 in Canada, and 25–91 per 100,000 in European countries (7). There is a peak age of onset in young women after their late teens and before their early 40s. Women are affected 9 times more frequently than men. SLE is a chronic complex disease characterized by recurrent flares (exacerbations) and subsequent remissions. There is currently no cure for SLE and this condition can be life-threatening when major organs are affected.

This review summarized the state of knowledge regarding the economic burden of SLE and the weaknesses of the studies conducted to date in order to inform future COI studies. Due to the difficulties of accurately quantifying the intangible costs, they are not within the scope of this review.

MATERIALS AND METHODS

Literature search.

Abstracts were retrieved from PubMed in April 2010 using the term “systemic lupus erythematosus” in combination with “cost-of-illness,” “cost analysis,” “costs,” “economic,” “disease burden,” or “burden-of-illness” in the title, abstract, or key words. Titles and abstracts of the retrieved articles were then used to assess the relevance to this systematic review. Reference lists of each potentially relevant study were then checked for any articles not identified by the database search. The inclusion criteria used for this review included articles in English and studies that measure the costs of SLE from the perspective of either the society, the individual, the payer, or the health care system. Exclusion criteria included non-English studies; studies that measure the costs of rheumatic diseases other than SLE; no costs quoted in the results section; no primary cost data reported; no breakdown of direct costs; economic evaluations of treatment, therapy, or drugs of SLE; review articles; and conference papers.

Evaluation of identified articles.

There are currently no formal guidelines for COI studies. Therefore, the evaluation of identified studies was based on 7 key elements derived from previous health economics literature (5, 8, 9) (Table 1). Each element is briefly discussed below.

Table 1. Key elements of cost-of-illness studies
  1. Perspective of the study clearly stated

  2. Study population clearly stated

  3. All relevant direct medical and/or nonmedical costs included, and their data sources clearly stated

  4. All relevant productivity losses due to morbidity and/or mortality included, and their data sources clearly stated

  5. Incremental/attributable costs calculated

  6. Where appropriate, discounting of costs should be applied

  7. Sensitivity analysis performed to address parameter uncertainties

The first key element of COI studies is the perspective of the analysis. Perspective of a COI study describes which costs are relevant based on the purpose of the study (2). Societal perspective is the most comprehensive perspective when performing a COI study. Using societal perspective, all types of costs are relevant, irrespective of who pays for them eventually. Other perspectives, such as the perspective of the health care system, the government, and the payer, provide information about costs to the particular group. The study population of a COI study should be clearly defined, including the source of patient recruitment and the sociodemographic and disease characteristics, so as to facilitate the comparison among studies. A COI study should include all of the important and relevant costs defined according to the perspective of the study, and also indicate the sources of data in detail. Where possible, only the costs generated by the disease of interest or incremental costs (using a disease-free population as a comparison) should be calculated. Discounting is an economic method to capture the time value associated with money (2). It is relevant for both direct and indirect costs that accrue past the first year. The discount rate converts a stream of future money values into its present value. Since the proper value for the discount rate is uncertain and the effect of alternative values of the discount rate is not uniform for all diseases, multiple discount rates are recommended to determine their effect on the cost estimates (2). Finally, it is recommended to conduct a sensitivity analysis applying different values for factors where uncertainty exists to determine the possible range of costs.

The study design of the COI studies will be assessed as follows. The incidence-based COI studies estimate the discounted, lifelong costs based on all cases with onset of disease within the period of study, usually a year. The prevalence-based studies focus on the costs of an illness in one period, usually a year, and on a cohort of typical patients, irrespective of the onset of the disease (2). The top-down approach is an epidemiologic approach that divides the total national costs of illness (e.g., the national health care expenditures) between different diseases (10). The bottom-up approach first estimates the quantity of health resources used and then estimates the unit costs of the resources used. Using this approach, patient and disease characteristics of a certain group of patients can be thoroughly investigated and related to resource use and costs (8).

Costs of SLE.

Sociodemographic and disease characteristics of the study population, categories of costs, and average costs were obtained from each study. Main cost domains included direct medical costs, direct nonmedical costs, and indirect costs. Indirect costs are commonly measured using two methods: the human capital approach (HCA) and the friction cost method (FCM) (11). The HCA uses wages as a proxy measure of the output of work time to evaluate the losses of productivity during the time absent from work. The HCA can also include the value of nonlabor activities. The FCM assumes that the short-term work loss caused by one employee can be made up by another one or the employee themselves, and employees absent from work for a long period can be replaced from the internal labor market or by an unemployed individual (concept of worker replacement) (12). Therefore, productivity losses due to short-term absence from work will not be considered, and for long-term absence, the FCM limits costs to a friction period (the amount of time before the losses of productivity are restored). Furthermore, all nonlabor activities will not be pertinent using the FCM. Due to the significant differences in indirect costs calculated by these 2 methods, the results were presented separately.

All costs were converted into 2008 US dollars using the Consumer Price Index derived from the statistics division of each country and the 2008 purchasing power parity conversion factor (from local currency to international dollar) derived from the United Nations, Statistics Division.

RESULTS

Literature search.

The literature search identified 246 articles. Of these, only 16 fulfilled the inclusion criteria (13–28). Five studies were further excluded due to lack of breakdown of direct costs (26) or lack of primary cost data (14, 15, 23, 28). The evaluation included 11 articles, of which 4 were from a longitudinal Tri-Nation cohort including patients from Canada, the UK, and the US (16–19); 3 were from the US (22, 25, 27); and 1 each were from Canada (13), Hong Kong (24), the UK (20), and Germany (21).

Evaluation.

Table 2 shows the assessment of the 11 identified COI studies against the key elements. An explicit statement of perspective of the study was evident in 5 of the 11 studies. In the other studies, the perspective of the evaluation was not clearly stated but could be inferred by the cost data collected. Societal perspective was most commonly used except for 1 study by Pelletier et al, in which costs were defined from the perspective of national health plans (27).

Table 2. The presence/absence of the 7 elements in each of the 11 cost-of-illness studies in systemic lupus erythematosus*
Author, year (ref.)PerspectivePopulationDirect costsProductivity lossDiscountingIncremental/ attributable costsSensitivity analysis
  • *

    Notation based on Rothfuss et al (9). [+] = partly fulfilled; + = present; N/A = not applicable; 0 = absent.

Patient-reported data       
 Clarke et al, 1993 (13)[+]+++N/A[+]+
 Clarke et al, 1999 (17)[+]++0N/A0+
 Clarke et al, 2000 (19)[+]+0+N/A00
 Clarke et al, 2004 (16)[+]++000+
 Panopalis et al, 2007 (18)[+]+0+000
 Sutcliffe et al, 2001 (20)++++N/A00
 Huscher et al, 2006 (21)+[+]++N/A0+
 Panopalis et al, 2008 (22)++++N/A0+
 Zhu et al, 2009 (24)++++N/A00
Medical claims data       
 Pelletier et al, 2009 (27)+[+]+0N/A0+
 Carls et al, 2009 (25)[+][+]++N/A+0

There were 3 studies without the description of the activity or severity of SLE, 2 of which used administrative medical claims data and 1 of which used a national rheumatology database. Two studies did not measure direct costs and 3 studies did not measure indirect costs. A clear description of the cost categories included in the valuation, along with a clear indication of the source of the data (e.g., source of unit price), was provided by most of the studies. However, there were only a few studies that provided results expressed in physical units, such as number of physician visits, duration of hospitalization, and number of sick days off of work (13, 20, 22, 24).

Two studies calculated 4-year cumulative costs of the Tri-Nation cohort using constant unit values for measurement; however, discounting was not employed (16, 18). Only 1 study employed a matched control group of patients without SLE to assess the incremental costs associated with SLE (25), rather than limit costs to those incurred by SLE patients only. Sensitivity analysis was used in 6 studies in order to test the robustness of the conclusion against variation in underlying assumptions and estimations.

Study design.

Among studies using the patients' self-reported data, the clinic-based cohort was most commonly used. Only the study by Panopalis et al (2008) indicated community sources of their cohort (66%) (22). All of the identified studies adopted a prevalence-based design, which involved the cost estimation of SLE in a given period. A 1-year period was most commonly used, and only 2 studies estimated costs for longer than a 1-year period (16, 18). The bottom-up design was employed in all of the identified studies. Most of the studies relied on patients' self-reported health resources utilization collected by questionnaires, except 2 studies that used administrative medical insurance claims data (25, 27). In the study by Huscher et al, utilization of health care resources was also derived from a national database (21). In the study by Zhu et al, both patients' self-reported data and review of medical records were used in data collection (24). A retrospective design was used in most of the identified studies, except the 2 studies reporting the cumulative costs of the longitudinal Tri-Nation cohort (16, 18).

Study population and costs of SLE.

Table 3 shows the demographic and clinical characteristics of patients of the identified studies. The percentage of women in each study cohort varied from 88.4% to 100%. The study that had only women in the cohort focused on the productivity losses among female subjects. The mean ages of the cohorts hovered at approximately 40 years old and the mean disease duration ranged from 8.6–13.7 years. As to measures of disease activity, 3 studies used the revised Systemic Lupus Activity Measure (SLAM-R) (16–18) and 1 each used the original SLAM (20), the Systemic Lupus Activity Questionnaire (22), and the Systemic Lupus Erythematosus Disease Activity Index (24). The Systemic Lupus International Collaborating Clinics/American College of Rheumatology Damage Index (SDI) was the only index used to measure disease damage of SLE among the identified studies. The SDI score ranged from 0.71 (24) to 1.8 (17).

Table 3. Characteristics of cohorts by study
Author, year (ref.)CountryNo. of subjectsWomen, %Mean age, yearsMean disease duration, yearsMean disease activity score*Mean SDI score*Employed, %Annual sick days off of work (%) or average duration (days)
  • *

    The Systemic Lupus Activity Measure (SLAM) was used in the study by Sutcliffe et al (20). The revised version of the SLAM was used in the study by Clarke et al (1999) (17), Clarke et al (2004) (16), and Panopalis et al (2007) (18). The Systemic Lupus Activity Questionnaire was used by Panopalis et al (2008) (22) and the Systemic Lupus Erythematosus Activity Index was used by Zhu et al (24). In all of these measures, higher scores indicate higher disease activity. The Systemic Lupus International Collaborating Clinics/American College of Rheumatology Damage Index (SDI) was used to measure disease damage, with higher scores indicating more disease damage.

  • The 1989 and 1990 costs were based on 164 patients and 155 patients, respectively.

  • Provided complementary data on a longitudinal cohort.

  • §

    Calculation of indirect costs due to absenteeism and shorter-term disability was based on 140 and 260 patients, respectively.

Patient-reported data         
 Clarke et al, 1993 (13)Canada16488.44513.54413 days in 1989, 16 days in 1990
 Clarke et al, 1999 (17)Canada, US, UKCanada: 229Canada: 91Canada: 43.3Canada: 10.2Canada: 7.1Canada: 1.8
 US: 268 US: 95 US: 39.0 US: 8.6 US: 4.1 US: 1.5
 UK: 211 UK: 95 UK: 40.7 UK: 10.0 UK: 6.0 UK: 1.2
 Clarke et al, 2000 (19)Canada, US, UK648100419.648
 Clarke et al, 2004 (16)Canada, US, UKCanada: 231Canada: 93.5Canada: 42.4Canada: 9.9Canada: 7.3Canada: 1.8
 US: 269 US: 95.1 US: 39.1 US: 8.6 US: 4.1 US: 1.7
 UK: 215 UK: 94.8 UK: 40.0 UK: 10.0 UK: 6.3 UK: 1.3
 Panopalis et al, 2007 (18)Canada, US, UKCanada: 231Canada: 93.5Canada: 42.4Canada: 9.9Canada: 7.3Canada: 1.8Canada: 48.7Canada: 9.1 days
 US: 269 US: 95.1 US: 39.1 US: 8.6 US: 4.1 US: 1.7 US: 45.3 US: 7.0 days
 UK: 215 UK: 94.8 UK: 40.0 UK: 10.0 UK: 6.3 UK: 1.3 UK: 52.6 UK: 8.1 days
 Sutcliffe et al, 2001 (20)UK1059439.910.56.31.154.3 at study entry, 52 at 6 months10.5 days at study entry, 8.2 days at 6 months
 Huscher et al, 2006 (21)Germany84490425530% required sick leave with an average 64.8 days
 Panopalis et al, 2008 (22)US81292.648.213.712.448.7
 Zhu et al, 2009 (24)Hong Kong30696419.62.50.714685% required sick leave with an average 14 days
Medical claims data         
 Pelletier et al, 2009 (27)US15,59088.748
 Carls et al, 2009 (25)US6,269§89.746.8135.8

Among the 9 studies that measured direct costs, costs for health professional visits, medication, and inpatient care were measured. Patients' out-of-pocket expenses and nonmedical costs were omitted in most studies. Table 4 shows the mean direct costs per patient-year of the 9 studies. Costs calculated by different studies varied dramatically in all of the cost categories. Among the studies using patients' self-reported data, the average direct costs per patient-year ranged from $3,735–$14,410. Among the countries with a primarily publicly funded health care system, including the UK, Canada, and Hong Kong, average direct costs per patient-year were estimated between $4,852 and $8,878 (excluding cumulative costs by Clarke et al [2004] [16]). The highest annual direct costs were observed in the study in the US by Panopalis et al (2008) (22), of which the study population had community patient sources, the oldest mean age (48.2 years), and the longest disease duration (13.7 years). The results from the 2 studies from the US using the medical claims data were quite comparable, with average annual direct costs per patient both estimated to be approximately $13,000. For all of the studies except 2 (16, 27), costs for inpatient care were found to be the largest component of annual direct costs. The hospitalization rate among studies ranged from 15% (13, 27) to 27% (20, 24).

Table 4. Mean annual direct costs of SLE per patient by study*
Author, year (ref.)Health professional visitEmergency room visitMedicationDiagnostic procedureOutpatient surgeryInpatient careRehabilitation facilityOther medical costsOut-of-pocket expensesNonmedical costsTotal direct costs
  • *

    Costs are shown as 2008 US dollars adjusted by the Consumer Price Index derived from the statistics division of each country and the 2008 purchasing power parity conversion factor derived from the United Nations, Statistics Division. SLE = systemic lupus erythematosus.

  • Including costs for 1) medical devices and nontraditional therapies (13); 2) emergency room visits, outpatient surgery, nonacute hospital stays, dialysis, and assistive devices (16); 3) aids (20); 4) physical treatment (21); 5) dialysis (22); and 6) injectables (other than corticosteroids), intravenous infusion administration, and other outpatient services (27).

  • Including costs for 1) babysitter and transportation (13) and 2) transportation, private household help, and adaptation to houses (24).

  • §

    The largest component of total direct costs of the cohort.

  • Costs are shown as the average annual costs of the cumulative 4-year costs.

  • #

    Including costs for dialysis.

  • **

    The largest component of total direct costs of the cohort. Including costs for outpatient surgery.

  • ††

    The largest component of total direct costs of the cohort. Including costs for rehabilitation facility.

  • ‡‡

    Costs are shown as the aggregate costs from the 2 groups of patients (with and without nephritis).

  • §§

    Costs are shown as the incremental costs of SLE relative to a control group without diagnosis of SLE.

Patient-reported data           
 Clarke et al, 1993 (13) (1989 costs)9191141,1231,5371002,618§621541587,244
 Clarke et al, 1993 (13) (1990 costs)1,063959431,2692254,982§10129 1728,878
 Clarke et al, 1999 (17)Canada: 893Canada: 134Canada: 1,058Canada: 605Canada: 126Canada: 2,404§Canada: 5,062
 US: 815 US: 119 US: 1,275§ US: 759 US: 149 US: 960 US: 5,512
 UK: 689 UK: 53 UK: 1,206 UK: 533 UK: 106 UK: 1,346§ UK: 4,965
 Clarke et al, 2004 (16)Canada: 547Canada: 1,156§Canada: 357Canada: 884Canada: 568Canada: 3,735
 US: 595 US: 1,399§ US: 570 US: 1,335 US: 785 US: 4,772
 UK: 501 UK: 1,380 UK: 344 UK: 1,416§ UK: 371 UK: 4,160
 Sutcliffe et al, 2001 (20)1,318122833#6931,852**374,852
 Huscher et al, 2006 (21)2801,09341421,684§2651535454,103
 Panopalis et al, 2008 (22)2,1114213,6974657,013§20250114,410
 Zhu et al, 2009 (24)699392721,0743,557††9442056,788
Medical claims data           
 Pelletier et al, 2009 (27)‡‡9553743,3891,4563,0924,038§13,305
 Carls et al, 2009 (25)§§5,1022061,3536,831§13,491

Indirect costs exceed direct costs in most studies. However, there were 2 studies that reported direct costs being the largest component of total costs (59% [22] and 62% [24]). The highest ratio of indirect costs:direct costs was 4.51 (21). Table 5 summarizes the mean indirect costs in the 8 studies. Four studies restricted indirect costs to subjects within working age. The employment rate varied from 35.8% (25) to 55% (21). The average duration of annual short-term sick leave away from work ranged from 7.0 (18) to 64.8 days (21). The framework of valuing indirect costs varied among studies, with productivity losses due to short-term sick leave from work and due to SLE-related unemployment being the most commonly measured categories. However, only 2 studies provided details of the breakdown of indirect costs. HCA was the most common method used to calculate indirect costs, with only 3 studies indicating using FCM as well. The study by Clarke et al (2000) (19) used the FCM and 4 different approaches using the HCA (approaches varied in whether to include nonlabor market productivity and the wage level) to calculate indirect costs among female patients (19). Also, the study by Panopalis et al (2008) calculated indirect costs by measuring changes in hours of work productivity since diagnosis of SLE (22). When excluding these 2 studies, among the studies using patients' self-reported data, the mean annual indirect costs per patient varied from $4,188–$14,614 using the HCA and from $1,093–$4,277 using the FCM.

Table 5. Mean annual indirect costs of SLE per patient by study*
Author, year (ref.)Restricted to working ageIndirect costs frameworkCalculation methodIndirect costsDue to short-term sick leaveDue to permanent work disabilityDue to daily activities limitation
  • *

    Costs are shown as 2008 US dollars adjusted by the Consumer Price Index derived from the statistics division of each country and the 2008 purchasing power parity conversion factor derived from the United Nations, Statistics Division. SLE = systemic lupus erythematosus; HCA = human capital approach; FCM = friction cost method.

  • Restricted to age <65 years or according to the study design.

  • Depending on calculation methods and wage rates.

  • §

    Costs are shown as the average annual costs of the cumulative 4-year costs.

  • Costs are calculated as incremental costs related to a matched control. Negative costs due to absenteeism indicated that patients with SLE had less absenteeism costs compared with the control group.

Clarke et al, 1993 (13)NoSLE-related unemployment, lost work hour, household chores limitationsHCA8,504 (1989 costs) 8,962 (1990 costs)
Clarke et al, 2000 (19)Not specifiedLimitations of labor market activity (lost work hour, extra work hour if SLE free, unpaid labor, domestic and volunteer work) and nonlabor market activityHCA and FCM1,485–23,577
Panopalis et al, 2007 (18)§Not specifiedSame as Clarke et al (2000)HCA and FCMHCA Canada: 10,395 US: 14,614 UK: 11,971
    FCM Canada: 1,539 US: 1,093 UK: 1,177   
Sutcliffe et al, 2001 (20)NoLost work hour, unemployment, household chores limitationsHCA9,784
Huscher et al, 2006 (21)YesSick leave and early retirementHCA and FCMHCA: 14,427 FCM: 4,2772,360HCA: 12,068 FCM: 1,917
Panopalis et al, 2008 (22)YesChanges in productivity at the time of the study and at the time of diagnosisChanges in annual income since diagnosis of SLE9,869
Zhu et al, 2009 (24)YesShort-term sick leave, early retirement, daily activity limitationHCA4,1881,01415,3533,559
Carls et al, 2009 (25)YesAbsenteeism and short-term disabilityHCAAbsenteeism: −605 Short-term disability: 1,305

Incremental costs of SLE were calculated in 1 study (25) using medical claims data. Direct medical costs, indirect costs due to absenteeism, and indirect costs due to short-term disability in patients with SLE were estimated at $21,499, $3,824, and $2,474, respectively, compared with $8,008, $4,430, and $1,169, respectively, in persons without SLE. Patients with SLE had less indirect costs due to absenteeism than persons without SLE.

Cost predictors of SLE.

Cost predictors refer to independent determinants of high direct or indirect costs. Impaired physical or mental health, young age, high disease activity, and more disease damage were constantly found to be associated with high direct costs (Table 6). Poor physical or mental health, low social help, high education level and unemployment, and high disease activity were associated with high indirect costs. There were also some disparities among studies. Long disease duration was found to be associated with high direct costs by Panopalis et al (22), in contrary to the findings by Huscher et al (21) and Zhu et al (24), where the short disease duration was associated with higher direct costs. Young age predicted high indirect costs in the study by Zhu et al (24), opposite of the findings in the studies by Huscher et al (21) and Panopalis et al (22).

Table 6. Cost predictors associated with high direct and indirect costs*
Author, year (ref.)Predictors for high direct costsPredictors for high indirect costs
  • *

    NPSLE = neuropsychiatric systemic lupus erythematosus.

Clarke et al, 1993 (13)Impaired physical function, high serum creatinineLow level of global well-being, impaired physical function, low level of social support, high education level and unemployment
Sutcliffe et al, 2001 (20)Young age, high education level, high disease activity, more disease damage, impaired physical functionHigh education level, high disease activity, impaired physical function
Huscher et al, 2006 (21)High disease activity, poor function, short disease durationPoor function, long disease duration, old age
Panopalis et al, 2008 (22)Young age, high disease activity, long disease duration, poor physical health, poor mental healthOld age, high disease activity, poor physical health, poor mental health
Zhu et al, 2009 (24)Shorter disease duration, higher disease activity, more disease damage, more episodes of NPSLEYounger age, more episodes of NPSLE

Organ involvement, disease flare, and costs of SLE.

Several studies have attempted to elucidate costs among different subgroups of patients with different organ involvement. There were 2 studies that estimated direct medical costs in patients with or without nephritis using medical claims data (25, 27). One of them also estimated indirect costs due to absenteeism and short-term disability; however, the number of patients with available data for this calculation was small (only 10–20 patients with nephritis) (25). Both studies found that SLE patients with nephritis consumed more health care resources and incurred significantly higher direct medical costs. The majority of added costs were attributable to inpatient care and outpatient services and less were attributable to medication. A subgroup analysis by Clarke et al, using the longitudinal Tri-Nation cohort, found that SLE patients with renal damage incur higher direct costs, but not indirect costs (28). An additional analysis was performed in a Chinese cohort from Hong Kong to investigate the relationship between neuropsychiatric SLE and disease costs (24). Neuropsychiatric SLE was found to be associated with increased higher direct and indirect costs and also an independent predictor for both direct and indirect costs.

The impact of lupus flare (exacerbation of disease) on costs of SLE was evaluated by Zhu et al (23). Results suggested that patients with flares incur significantly higher direct and indirect costs compared with those without flares. Multiorgan flares or renal/neuropsychiatric flares were more costly than single-organ flares or other minor organ flares.

DISCUSSION

This review found relative paucity of data on the costs of SLE. Results showed that SLE is associated with substantial economic burden, in terms of health care resources utilization, as well as losses of productivity due to work capacity impairment. Average costs may be more relevant to decision makers considering estimating total costs for a cohort of patients, although they have a risk of overestimation due to the positively skewed distribution of cost data. Among the 5 studies that reported median costs as well (13, 19–21, 24), agreement in direct costs was much improved with a narrower range between $2,056 (21) and $3,049 (24), but for indirect costs, there was still a wide range of median costs between $0 (21) and $4,038 (20). Generally, costs for inpatient care represented the largest proportion of direct costs, suggesting that at least 17% and up to 56% of annual direct costs were incurred by less than 27% of the population. Furthermore, the estimates of indirect costs are most likely to be underrated because these studies rarely take into account the productivity losses incurred by unpaid caregivers.

Costs of SLE are primarily determined by factors related to disease status, such as disease duration, disease activity, and damage, as well as patients' physical and mental health status, which are influenced, at least in part, by the disease progression and cumulative damage. Major organ involvements, such as lupus nephritis and neuropsychiatric SLE, along with exacerbation in disease status (lupus flare), are also shown to be associated with both increased direct and indirect costs, independent of demographics and other disease status. It is anticipated that interventions that can effectively control disease activity, prevent disease flare, and delay and slow disease progression may potentially save large amounts of costs attributable to disease damage and major organ involvement.

We found great discrepancies in both direct and indirect costs across studies, which cannot be simply explained by the demographic and clinical differences across the population. Such discrepancies can be attributed to the absence of well-defined guidelines for COI studies. It should be noted that there are significant differences in the cost framework, the methods of costs calculation, the health care system, and the pattern of practice across studies. Furthermore, the cost estimates cited in this review are derived from several different types of studies from many different countries over a period of at least 16 years. Changes in the state of knowledge of this condition, medical technology, and patterns of practice may have substantial influences on the cost estimates.

There are several methodologic considerations that arise from the evaluation of studies. First, the source of the study population should be clearly addressed. A clinic-based cohort is easy to assemble and more likely to be homogeneous in terms of diagnosis and disease characteristics, but there may be a risk of overrepresentation of patients with more severe disease status. On the other hand, a community-based cohort ensures a more representative patient population, but the diagnosis of this cohort may rely on administrative data and/or medical notes, which are certainly less precise (29). Second, the source of the data, unit prices, and calculation methods should also be clearly stated, so that the cost estimates can be reworked. There are mainly 2 methods to calculate indirect costs and there are great discrepancies between these 2 methods. When possible, both methods should be employed. Third, the categories of costs should be elaborated in detail. Direct and indirect costs should be reported separately. Nonmedical costs and patients' out-of-pocket expenses should be calculated in order to capture the comprehensive burden of SLE. Furthermore, thus far, analysis of incremental costs of SLE was only performed in one study. In order to avoid the assumption that without SLE a person would incur no disease costs, case–control studies calculating incremental costs of having SLE are suggested. Finally, currently, there is no incidence-based COI study measuring lifetime costs of SLE. Considering SLE as a chronic fluctuating disease with significant long-term morbidities and the advances in new treatments, the incidence approach allows analysis regarding how costs are distributed while the disease progresses.

Numerous COI studies have been performed in different countries and different disease groups in the last several decades. However, the reliability, consistency, and usefulness as a decision-making tool have been criticized since its inception. The main debate includes questioning the practicality of realizing the potential “cost savings” if a disease is eradicated and the ability of COI studies to guide efficient and equitable health care resource allocation (30). Other attacks relate to the inability of COI studies to accurately quantify intangible losses on patients' well-beings and the variations in methodologies employed in different studies that limit the comparability of findings (31, 32). Therefore, results from COI studies should be interpreted and used with caution. As descriptive studies, COI studies have the potential to educate and inform policymakers and to provide information to support the complex decision-making process. To achieve these goals, it is important that the design of COI studies is innovative enough to measure the true costs of the disease and is capable of identifying subjects who bear the costs and explaining the variability of costs (8).

Most of the COI studies in SLE conducted to date provided cost estimates specific to only several countries or several types of health care systems. In the existing literature, several methodologic problems were identified. It is important, therefore, for more studies to be conducted to highlight the magnitude of the problem of SLE to the society and individual in other countries, and these studies should also have an innovative design capable of addressing these methodologic considerations.

At present, new therapies specifically targeting the immune system have been developed and claimed to be able to control disease activity and prevent major organ flare of SLE (33). Costs of these therapies are likely to be much higher than conventional therapies. Given the substantial costs associated with SLE, it is promising that the potential benefits of these therapies will be commensurate to their costs. Economic evaluations will give us more information regarding the economic properties of these novel therapies. The estimates from this review provide a baseline and framework for such evaluations.

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. Zhu 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. Zhu, Tam, Li.

Acquisition of data. Zhu.

Analysis and interpretation of data. Zhu.

Ancillary