Abstract
Objective. To quantify healthcare resource utilization (HRU), work loss, and annual direct and indirect healthcare costs among patients with systemic sclerosis (SSc) compared to matched controls in the United States.
Methods. Data were obtained from a large US commercial claims database. Patients were ≥ 18 years old at the index date (first SSc diagnosis) and had ≥ 1 SSc diagnosis in the inpatient (IP) or emergency room (ER) setting, or ≥ 2 SSc diagnoses on 2 different dates in the outpatient (OP) setting between January 1, 2005, and March 31, 2015; continuous enrollment was required during the followup period (12 months after the index date). Individuals with no SSc diagnoses were matched 1:1 to patients with SSc. Wilcoxon signed-rank and McNemar tests were used for comparisons and regressions with generalized estimating equations for adjusted OR (aOR) and incidence rate ratios (IRR) between 2 cohorts.
Results. There were 2192 pairs of patients with SSc and matched controls included (mean age 57.6 yrs; 84.3% female); of these, 233 were eligible for work loss/indirect cost analyses. Compared to matched controls, patients with SSc had significantly higher HRU and costs during the 1-year followup period, IP admissions (adjusted IRR = 2.4), IP hospitalization days (adjusted IRR = 3.1), ER visits (adjusted IRR = 2.0), OP visits (adjusted IRR = 2.3), and days of work loss (adjusted IRR = 2.6). The adjusted difference in annual direct and indirect costs was US$12,820 and $3103, respectively (all p < 0.0001).
Conclusion. Patients with SSc had a high direct and indirect economic burden postdiagnosis.
- SYSTEMIC SCLEROSIS
- HEALTHCARE RESOURCE UTILIZATION
- DIRECT HEALTHCARE COSTS
- WORK PRODUCTIVITY LOSS
- INDIRECT COSTS
Scleroderma or systemic sclerosis (SSc) is a chronic autoimmune disease of the connective tissue characterized by fibroproliferative vasculopathy and progressive fibrosis of the skin and internal organs1. The prevalence of SSc is estimated to range from 50 to 300 per million across the world, with one of the highest prevalence rates observed in the United States at 240 per million2,3. Cumulative survival rates of 5 and 10 years for patients with SSc were estimated at 74.9% and 62.5%, respectively4; lung involvement was the most common cause of death among patients with SSc-related deaths4.
Because of its clinical heterogeneity, SSc is a challenging disease to manage5,6. Currently, there are no disease-modifying therapies approved by the US Food and Drug Administration and treatment mostly depends on the type and severity of organ involvement5,6,7. Despite limited evidence supporting their effectiveness, nonselective immunosuppressive treatments are commonly used6,8. Hematopoietic stem cell transplantation has been shown to be effective in select patients with more severe skin thickening9,10,11. Biologic agents may be administered in refractory cases, though the evidence in support of such an approach is limited6,8.
Given the considerable comorbidity burden of SSc and the availability of limited, and only partly effective, treatment options, it is not surprising that substantial healthcare costs associated with SSc have been reported12. However, the economic burden of SSc has been estimated in a limited number of studies12–17, with only 2 such studies18,19 including US patients. One of them (Furst, et al18) found that patients with SSc had significantly higher annual healthcare costs and healthcare resource utilization (HRU) than matched controls but only included data up to 2008, and thus may be outdated and not reflective of the current treatment landscape. More recently, Fischer, et al19 also found high annual healthcare costs among patients with SSc in the United States but did not compare them with those of controls and did not include HRU or indirect costs. Moreover, neither of these US studies assessed work productivity loss among patients with SSc.
Therefore, to provide a more comprehensive and current estimate of the burden of illness of SSc, this study quantified annual HRU, work loss, and direct and indirect costs (i.e., costs associated with work productivity loss due to disability or medically related absenteeism) among patients with SSc compared to individuals without SSc in US clinical practice.
MATERIALS AND METHODS
Data source
Data for this study were obtained from the OptumHealth Care Solutions Inc. database covering 84 self-insured Fortune 500 companies, including medical and pharmaceutical claims for over 19 million privately insured individuals and their families. The database covers a range of industries, job classifications, and locations across the United States. Work productivity loss data (i.e., claims associated with short- and longterm disability or medically related absenteeism) are available for the employees of 44 out of the 84 companies in the database. All the data used in this study were de-identified and fully compliant with the patient confidentiality requirements of the Health Insurance Portability and Accountability Act. No ethical review by an institutional review board was required. Data from Q3 2004 to Q1 2016 were used.
Patients with SSc
Patients with SSc were eligible for inclusion if they had at least 1 medical claim with a diagnosis of SSc (International Classification of Diseases, 9th Revision, Clinical Modification code 710.1) in the inpatient (IP) or emergency room (ER) setting, or at least 2 medical claims with diagnoses of SSc on 2 different dates in the outpatient (OP) or other settings between January 1, 2005, and March 31, 2015. Two medical claims were required to limit potential misclassification, which tends to be higher in the OP setting compared to the IP and ER settings, as commonly done in the literature18,19. The date of the first observed SSc diagnosis was defined as the index date. Patients were further required to be at least 18 years of age at the index date, have continuous enrollment in their healthcare plan for 6 months prior to the index date (baseline period) and 12 months following the index date (followup period), and have no claims with a diagnosis of SSc during the baseline period.
To be eligible for work loss and indirect cost analyses, patients were additionally required to be primary policyholders (this is because the database contains salary information only for primary policyholders), actively employed on the index date, and continuously employed during the 12-month followup period with available work loss data.
Matched controls
Controls were required to have no diagnosis of SSc in their entire claims history. For the controls, the index date was selected as the same calendar date as that of the patients with SSc to whom they were matched. Controls were matched exactly with a 1:1 ratio to patients with SSc on age at the index date (measured in years), sex, region of residence (i.e., Northwest, Midwest, South, West, unknown), and insurance type (i.e., preferred provider organization, point of service, indemnity, unknown/other). Specifically, to be matched to a patient, a control was required to have the exact same value for each and every characteristic of the patient. If there were multiple potential controls who matched a certain patient exactly, the control was randomly selected from the pool of potential matches; those who were not matched to that particular patient were returned to the pool of potential matches for the next round of matching. The controls were further required to have continuous enrollment in their healthcare plan during the baseline and followup periods.
Baseline characteristics
Baseline characteristics included demographics (e.g., age, sex) at the index date, the index year (defined as the year of the index date), a modified Charlson comorbidity index (CCI)20 that excluded SSc diagnoses, individual CCI components, and other SSc-related comorbidities. All-cause HRU and healthcare costs (i.e., medical and pharmacy costs) were also measured during the baseline period.
HRU and clinical outcomes in the followup period
During the followup period, HRU measures included IP admissions and total IP hospitalization days, ER visits, OP visits, specialist visits, and other visits (including laboratory tests and medical services at home). In addition, selected procedures, medication use, and SSc-related comorbidities, as well as the modified CCI, were also measured.
Direct healthcare costs in the followup period
Annual direct healthcare costs measured during the followup period included medical costs and pharmacy costs and were calculated for both patients with SSc and matched controls. Total medical costs included IP costs, ER costs, OP costs, and costs for other visits (including laboratory tests and medical services at home, etc.). All costs were inflated to 2016 US dollars (US$).
Work loss in the followup period
Work loss analyses were conducted among individuals with available work loss data. Work loss during the followup period included total days of work loss due to disability and days of medically related absenteeism. These were identified based on the use of medical services (e.g., an office or hospital inpatient visit) during business days (Monday through Friday) as well as the gap days between the injury date (i.e., the date of the injury that resulted in the disability claim; obtained from medical claims) and the start date of disability (obtained from disability claims). It was assumed that each hospitalization day or emergency department visit accounted for a full day of work loss, while each outpatient or other visit accounted for half a day of work loss.
Indirect costs in the followup period
Annual indirect costs were calculated among the subset of patients with available work loss data. Indirect costs during the followup period included medically related absenteeism costs and disability costs. Medically related absenteeism costs were calculated based on individual employees’ income information obtained from eligibility files and days of medically related absenteeism, defined above. Annual disability costs were calculated based on individual employees’ income information obtained from eligibility files and days of disability, which were identified based on the dates of disability provided in disability claims. All indirect costs were inflated to 2016 US$.
Statistical analysis
For baseline characteristics and clinical outcomes during the followup period, mean and SD were reported for continuous variables; frequency and percentage were reported for categorical variables. Descriptive comparisons were made between patients with SSc and matched controls using Wilcoxon signed-rank tests for continuous variables and McNemar tests for categorical variables.
For HRU and work loss during the followup period, the proportions of patients with any event and mean counts of events were described and compared between patients with SSc. In addition, aOR and incidence rate ratios (IRR) between patients with SSc and matched controls were calculated using multivariable regressions with generalized estimating equations and with binomial distributions for OR and Poisson distributions accounting for overdispersion for IRR. To improve the precision of the coefficient estimates, the regressions controlled for baseline covariates, including age, sex, index year (i.e., 2005–2010 vs 2011–2015), region of residence (i.e., South vs other), insurance type (i.e., preferred provider organization vs other), industry category (i.e., manufacturing, transportation, government vs other), and selected baseline comorbidities that were not related to SSc and with prevalence > 5% in any cohort (i.e., diabetes without chronic complication, any malignancy including leukemia and lymphoma).
For annual direct healthcare and indirect costs, descriptive costs for the followup period were compared between patients with SSc and matched controls. In multivariable analyses, differences in mean adjusted costs between patients with SSc and matched controls were estimated using multivariable regression with generalized estimating equations and with the Tweedie distribution, controlling for the same covariates as those used in the HRU and work loss analyses.
RESULTS
Baseline period
A total of 2192 patients with SSc and 2192 matched controls were included in the study; of these, 233 pairs were eligible for inclusion in the work loss and indirect cost analyses (Figure 1).
Sample selection. 1Index date was defined as the date of the first SSc diagnosis between January 1, 2005, and March 31, 2015. 2The baseline period was defined as the 6-month period before the index date. 3The followup period was defined as the 12-month period following the index date. 4Patients were eligible for work loss analyses if they were primary policyholders, actively employed at the index date, and continuously employed during the followup period. SSc: systemic sclerosis.
Baseline characteristics are presented in Table 1. For both patients with SSc and matched controls, mean age at the index date was 57.6 years and the proportion of females was 84.3%. The modified CCI score, which excluded SSc, was significantly higher for patients with SSc than for matched controls (0.7 vs. 0.3, p < 0.0001; Table 1). Rheumatic disease (excluding SSc) and chronic pulmonary disease were the 2 most common comorbidities for patients with SSc and were significantly more prevalent among patients with SSc than matched controls (rheumatic disease: 21.0% vs 1.6%, p < 0.0001; chronic pulmonary disease: 14.5% vs 5.8%, p < 0.0001; Table 1).
Baseline characteristics.
All-cause HRU measures at baseline were significantly higher for patients with SSc than matched controls (all p < 0.0001; Table 1). Similarly, all-cause costs at baseline were also significantly higher for patients with SSc than for matched controls (all p < 0.0001; Table 1). Specifically, baseline total costs were $9382 for patients with SSc and $3058 for matched controls (p < 0.0001), with medical costs accounting for $7363 for patients with SSc and $2121 for matched controls.
HRU and clinical outcomes in the followup period
Unadjusted measures of clinical outcomes and rates of medication use were generally higher among patients with SSc than matched controls (Table 2). Unadjusted and adjusted HRU measures were all significantly higher among patients with SSc than among matched controls (Table 2 and Table 3). In particular, patients with SSc versus matched controls had more IP admissions (0.7 vs 0.3 events; adjusted IRR 2.4, 95% CI 2.2–2.6), days of IP hospitalization (3.7 vs 1.1 days; adjusted IRR 3.1, 95% CI 2.9–3.4), ER visits (0.7 vs 0.3 events; adjusted IRR 2.0, 95% CI 1.9–2.2), OP visits (23.8 vs 10.3 events; adjusted IRR 2.3, 95% CI 2.2–2.4), and other visits (3.0 vs 1.0 events; adjusted IRR 2.9, 95% CI 2.7–3.1; all p < 0.0001). Further, the odds of any IP admission (aOR 3.8, 95% CI 3.2–4.5), ER visit (aOR 1.9, 95% CI 1.7–2.1), and any other visit (aOR 2.2, 95% CI 2.0–2.6) were higher among patients with SSc than among matched controls (all p < 0.0001; Table 3).
Unadjusted HRU and clinical outcomes in the followup period1.
Unadjusted and adjusted all-cause HRU in the followup period1.
Direct healthcare costs in the followup period
Unadjusted total annual direct healthcare costs during the followup period were significantly higher for patients with SSc ($22,016, of which $16,269 were medical costs and $5746 pharmacy costs) than matched controls ($5977, of which $4076 were medical costs and $1901 pharmacy costs; all p < 0.0001). After adjustment, the adjusted difference between patients with SSc and matched controls was $12,820 (SSc: $18,069; matched controls: $5250) for total annual direct healthcare costs; $9756 (SSc: $13,024; matched controls: $3268) for total annual medical costs, and $2942 (SSc: $4625; matched controls: $1684) for annual pharmacy costs (all p < 0.0001; Table 4).
Unadjusted and adjusted direct healthcare and indirect costs (2016 US$) in the followup period1.
Work loss in the followup period
Unadjusted and adjusted work loss was also significantly higher among patients with SSc than matched controls (Table 5). In particular, patients with SSc had more days of work loss (21.3 vs 6.8; adjusted IRR 2.6, 95% CI 2.1–3.2), composed of days of medically related absenteeism (11.0 vs 4.8; adjusted IRR 2.3, 95% CI 2.0–2.6) and days of disability (10.2 vs 2.0; adjusted IRR 2.1, 95% CI 1.7–2.7; all p < 0.0001). Additionally, the odds of any medically related absenteeism were significantly higher (aOR 6.3, 95% CI 2.4–16.6; p = 0.0002) and odds of any day of disability were numerically higher (aOR 2.2, 95% CI 0.9–5.1; p = 0.0760) for patients with SSc versus matched controls (Table 5).
Unadjusted and adjusted work loss in the followup period1.
Indirect costs in the followup period
Indirect costs (i.e., costs associated with work productivity loss due to disability or medically related absenteeism) were significantly higher for patients with SSc than for matched controls in both the unadjusted and adjusted analyses. Unadjusted indirect costs were $5170 (disability: $1554; medically related absenteeism: $3616) for patients with SSc versus $1663 (disability: $137; medically related absenteeism: $1526) for matched controls (all p < 0.0001). The adjusted difference was $3103 (SSc: $4554; matched controls: $1451) for total annual indirect costs, with $1361 (SSc: $1513; matched controls: $151) for disability costs and $2029 (SSc: $3362; matched controls: $1333) for medically related absenteeism costs (all p < 0.0001; Table 4).
DISCUSSION
Our present study sought to provide an updated estimate of the direct economic burden of SSc in the United States as well as to provide an estimate of the indirect economic burden of SSc, which has not been characterized in the literature. Using claims data, we found that patients with SSc had substantially greater HRU, work loss, and annual direct healthcare and indirect costs than their matched controls without SSc. More specifically, compared to their matched controls, patients with SSc had IP admissions about 3 times more frequently and incurred all-cause annual direct healthcare costs that were nearly 3 times higher during the first year after diagnosis. The adjusted difference in total annual direct healthcare costs between patients with SSc and matched controls was $12,820, mostly driven by medical costs (difference of $9756). In addition, patients with SSc had about 14 more days of work loss due to disability or medically related absenteeism in the first year after diagnosis. In line with these results, total annual indirect costs were significantly higher for patients with SSc than their matched controls, with an adjusted cost difference of $3103.
The finding that direct healthcare costs are significantly higher for patients with SSc than controls is consistent with the literature. A claims data study by Furst, et al18 reported unadjusted annual healthcare (medical and pharmacy) costs of $21,625 for patients with SSc and $6859 for matched controls, resulting in a cost difference of $14,766. These costs are comparable to the unadjusted annual healthcare cost ($22,016) for patients with SSc and cost difference ($16,039) found in our current study. Furst, et al also reported adjusted annual healthcare costs of $22,909 for patients with SSc and $6620 for matched controls, resulting in a cost difference of $16,288. The lower adjusted cost difference found in the current study ($12,820) may be explained by the different definition of matched controls used in Furst, et al18. The difference may also reflect that the Furst, et al study was conducted among patients in a managed care organization, where patients included in the current study were enrolled in fee-for-service health plans. Nevertheless, our study and that by Furst, et al found largely comparable annual direct healthcare costs for patients with SSc and high incremental burden between patients with SSc and matched controls. This suggests that the economic burden of SSc has remained high over the past decade despite the application of different treatment options or approaches.
This finding is also supported by a recent study by Fischer, et al19, which estimated first-year healthcare (medical and pharmacy) costs for patients with SSc at $22,245 (2016 US$) considering claims data from 2003 to 2014 but not including matched controls. The amount is similar to the first-year costs found in this study of $22,016. Besides first-year costs, the current study also estimated all-cause annual direct healthcare costs during the second and third years after a diagnosis of SSc among patients with continuous health plan enrollment as an exploratory analysis (data not reported here). In the second and third years, average annual total costs for patients with SSc were estimated at $19,119 and $18,742, respectively ($5921 and $6108, respectively, for matched controls), which are comparable to $20,050 in the second year and $19,707 in the third year (all costs inflated to 2016 US$), as reported in the Fischer, et al study19. Taken together, these results point to a substantial direct economic burden among patients with SSc. Moreover, it should be noted that neither Furst, et al18 nor Fischer, et al19 reported the indirect costs associated with SSc, which, as found in our current study, are significantly high. Thus, when both annual direct healthcare and indirect costs are taken into consideration, the overall burden of illness related to SSc becomes even more pronounced, further underscoring the existence of a high unmet need for more effective management of SSc.
Some limitations should be considered when interpreting the results of our study. First, our study may have underestimated the burden of illness of SSc given that it was assessed using claims submitted during the first year after diagnosis. Because its clinical heterogeneity makes SSc difficult to diagnose in a timely manner, disease onset may have occurred before a diagnosis of SSc was made and thus patients may have incurred SSc-related costs and HRU before an SSc-related claim was submitted. Therefore, the true burden of SSc may be higher than the burden during the first year after a diagnosis of SSc. In addition, the later stages of the disease tend to have more chronic organ involvement, which may be associated with higher direct and indirect costs. Further, the study may have included patients who were initially diagnosed with SSc prior to the data period; because disease burden is also relatively high at diagnosis, the burden may have been underestimated by not identifying these patients’ true diagnosis date. Further studies are needed to quantify such burden from disease onset over a longer period (> 3 yrs), given the progressive features of the disease. Additionally, patients who died or lost their job and/or insurance coverage because of SSc within 12 months post-index may have contributed higher costs but were not included, possibly leading to an underestimate of the overall cost burden of SSc. Second, some patients over 65 years of age could be partially covered by Medicare. The data record the portion of costs covered by the private insurers but not the portion of costs covered by Medicare and thus their claims data may have been incomplete, leading to underestimation of the burden within this group of patients. Third, our study assumed that each hospitalization day and emergency department visit accounted for a full day of work loss, while each outpatient or other visit accounted for half a day of work loss, which may not reflect true work loss incurred by patients. Other limitations inherent to claims data should also be taken into consideration. Disease cohorts were defined based on diagnosis codes; because claims data are subject to error or omissions, patients may have been misclassified and incorrectly included in or excluded from one of the 2 cohorts. Further, because claims data do not provide information on disease severity, our study could not assess the effect of SSc severity on the burden of illness. Also, because the data used in our study were limited to patients with employer-sponsored health plans, the results may not be generalizable to other patient populations. Finally, as with any retrospective observational study, associations but no causal relationships should be inferred.
This claims data study found that patients with SSc had a substantially higher burden of illness than their matched controls, as evidenced by greater HRU, work loss, and annual direct healthcare and indirect costs. This represents an updated estimate of the direct economic burden and the first estimate of the indirect economic burden associated with SSc in the United States. The substantial burden of illness of SSc found in this study underscores the need for more effective disease management that can potentially improve clinical outcomes and thus reduce HRU, costs, and work loss among patients with SSc.
Acknowledgment
Medical writing assistance was provided by Cinzia Metallo, PhD, an employee of Analysis Group Inc.
Footnotes
Full Release Article. For details see Reprints and Permissions at jrheum.org
This study was funded by Boehringer Ingelheim Pharmaceuticals Inc.
- Accepted for publication October 16, 2018.
Free online via JRheum Full Release option
