Treatment patterns and medical cost of systemic lupus erythematosus patients in Japan: a retrospective claims database study

Abstract

Aim: To describe the characteristics and medication treatment patterns, healthcare resource utilization (HRU), and associated costs in Japanese patients with systemic lupus erythematosus (SLE).

Methods: Claims data from the Japan Medical Data Center (JMDC) database were used to identify patients with SLE-related claims from 2010 to 2017. Algorithms were developed to retrospectively categorize patients by disease severity, treatment experience, and SLE-related manifestations such as lupus nephritis and central nervous system lupus. Descriptive and multivariate analyses were used to describe treatment pattern and estimate HRU and associated costs for the SLE cohort overall and by disease severity and complications.

Results: Among 4,733 eligible patients, 2,072 (43.8%) were treatment naïve, 2,214 (46.8%) were previously treated for SLE, and 447 (9.4%) did not receive any treatment. Mean (SD) age of the total SLE cohort was 45.2 (13.1) years and mean (SD) follow-up duration was 1,137.3 (758.0) d. Based on disease severity, 1,383 (29.2%) patients had mild, 2,619 (55.3%) patients had moderate, and 731 (15.4%) patients had severe SLE. Patients on glucocorticoids (both oral and parenteral) received add-on medications the most (35.5%, p < .001). Mean annual cost per SLE patient in the post-index period, inclusive of hospitalizations, outpatient visits, and pharmacy was ¥436,836; ¥1,010,772; and ¥2,136,780 for patients with mild, moderate, and severe SLE, respectively.

Limitations: The database only captured information on patients up to 75 years of age. Due to the nature of the database, biases regarding conditions that attribute to the spectrum of SLE severity, flare incidences, or individual physical status cannot be ruled out.

Conclusions: This study describes the treatment patterns and economic burden experienced by Japanese patients with SLE. The results highlight an unmet need to derive better treatment strategies for patients with SLE to effectively address the disease burden in Japan.

Keywords:

• Autoimmune disease
• direct cost
• treatment patterns
• healthcare resource utilization
• systemic lupus erythematosus

JEL CLASSIFICATION CODES:

• I11
• Z39

Introduction

Systemic lupus erythematosus (SLE) is a chronic systemic autoimmune disorder characterized by fluctuating clinical manifestations such as recurrent flares (exacerbations) and subsequent remissions, and could progressively inflict organ damage in the body¹. Although research has substantially added to existing knowledge regarding the pathology of SLE, there is currently no cure for SLE, resulting in life-threatening scenarios when major organs are affected².

The estimated global prevalence of SLE varies from 21 per 100,000 in Canada to 25–91 per 100,000 in European countries³. In Japan, the prevalence of SLE ranges from 3.7 to 37.7 per 100,000⁴. Globally, SLE has been reported to occur in both males and females; however, females are predominantly affected⁵. Moreover, there is evidence of poor health-related quality of life in patients with SLE compared with other rheumatic diseases⁶ and this is persistent over long-term follow-up^7,8. Although this multifactorial disease is associated with significant morbidity and mortality^9,10, the survival rate of patients with SLE has increased with improvements in diagnosis and therapy¹¹.

The British Society for Rheumatology¹² and The European League Against Rheumatism (EULAR)¹³ recommend initiating SLE management with hydroxychloroquine and glucocorticoids before moving on to immunosuppressants, and eventually biologics such as rituximab or belimumab for refractory patients. Similar guidelines for the management of SLE do not exist in Japan. However, Tanaka et al.¹⁴ after retrospectively analyzing claims data of Japanese patients who were diagnosed with SLE between April 2010 and March 2012 and had follow-up data for 3 years, reported corticosteroids to be the mainstay of SLE therapy in Japan, with other medications – except hydroxychloroquine – being utilized more or less in line with the guidelines of other countries.

The economic impact of SLE is two-fold, including both cost of treatment and productivity losses due to complications of SLE, as 25–60% of patients develop renal disease over time and may require expensive treatments and procedures such as dialysis and a transplant^15–17. Tanaka et al.¹⁴ reported that all patients with SLE in Japan had at least one outpatient visit, while 39.3% had been hospitalized at least once. The mean total all-cause direct medical cost per patient over 3 years was ¥2,913,509 with costs rising proportionally to disease severity. However, it is difficult to understand long-term outcomes and costs due to the episodic and recurring nature of SLE¹⁸.

Limited evidence exists in the literature with respect to comorbid health and physical conditions, treatment patterns persistency, and rates of switching to or augmentation with alternative therapies^14,19,20. The latest study¹⁴ evaluating the above outcomes assessed claims data for a maximum of 3 years from the index diagnosis period (2010–2012). Moreover, the number of patients who met the inclusion criteria was small (n = 295). Therefore, to provide an update and address existing knowledge gaps, this retrospective cohort study analyzed insurance claims data to describe the treatment patterns, healthcare resource utilization (HRU), and associated costs in Japanese patients with SLE.

Methods

Study design and data source

This retrospective observational cohort study analyzed claims data retrieved from the Japan Medical Data Center (JMDC) database between 1 January 2010 and 31 March 2017 (Figure 1). The database has been capturing patient information and diagnosis from large samples of primary and secondary care physicians in Japan comprising approximately 4% of the total population of Japan (5.6 million) since January 2005²¹. It contains inpatient, outpatient, and pharmacy administrative claims and has been widely used in health services research^22–25. Patient medical consultation information can be continuously tracked in a chronological order, even if the patient attends multiple medical institutions or transfers to another hospital, as long as the individual belongs to the Health Insurance Association.

Figure 1. Study design. A summary of extracting SLE cohort by applying an algorithm when working with data from claims database. Abbreviation. SLE, systemic lupus erythematosus.

The information retrieved from JMDC database was anonymized to protect patient privacy under the “Act on the Use of Numbers to Identify a Specific Individual in the Administrative Procedure”. This study was approved by the sponsor’s internal approval committee in accordance with Japanese ethical and legal guidelines for Japan. The informed consent of patients was not required as per the Ethical Guidelines for Epidemiological Research issued by the Japanese Ministry of Health, Labor, and Welfare (MHLW). Reporting of claims data was based on International Society for Pharmacoeconomics and Outcomes Research guidelines²⁶.

Patient selection

The index date was identified as the earliest date observed of the medical claim with a diagnosis of SLE. To categorize SLE patients naive to treatment from the treatment experienced group, a “wash out” period was defined for treatment naive SLE patients by retracing back at least 6 months from the index date to confirm that no treatment was received prior to the index date.

Patients aged ≥18 years at index date with a confirmed SLE diagnosis i.e. at least two International Classification of Diseases, 10th revision diagnosis codes M32.1²⁷, at least 6 months of continuous insurance coverage/enrollment before the index date, and at least 6 months of continuous insurance coverage/enrollment after the index date to ensure adequate follow-up on study measures were included. All eligible patients with or without comorbidities (e.g. mononeuropathy/polyneuropathy, other renal impairment, seizure, pleurisy/pleural effusion, arterial/venous thrombosis, aortitis, stroke/transient ischemic attack, and nephrotic syndrome) during study period were included. Patients who did not have a confirmed diagnosis of SLE, had other complications not related to SLE, were enrolled in the database for less than 6 months prior to or after the index date, or did not have any SLE treatment within the database during follow-up were excluded from the study (Figure 2).

Figure 2. Patient selection flowchart. Abbreviations: CNS- Central Nervous System, ICD-10 – International Classification of Disease 10th revision, JMDC, Japan medical data center; SLE, systemic lupus erythematosus.

Study measures

All demographic characteristics in the available patient’s claims data set were documented including age and sex. The SLE cohort extracted from the claims database was divided into three groups: sub-cohort of SLE with organ and system involvement, sub-cohort of SLE prescribed regimen, and sub-cohort of SLE severity.

Sub-cohort of SLE with organ and system involvement included patients with a confirmed i.e. at least two diagnoses of lupus nephritis (but no central nervous system [CNS] lupus), patients with a confirmed i.e. at least two diagnoses of CNS lupus (but no lupus nephritis), patients with confirmed i.e. at least two diagnoses each of lupus nephritis and CNS lupus, and patients without a confirmed diagnosis of lupus nephritis or CNS lupus.

Sub-cohort of SLE prescribed regimen included patients naive to treatment regimen and experienced patients with treatment regimen of interest. Prescriptions of drugs were retrieved using the JMDC Drug table where all drug prescriptions were recorded and identified according to the Anatomical Therapeutic Chemical (ATC) codes. Treatment naive patients were defined as patients with SLE who did not have any record of treatment claims within the pre-index period. In a given line of treatment (LoT), analyses of treatment-related events, and HRU was performed by type of SLE regimen prescribed by the drug classes. These included topical steroids, nonsteroidal anti-inflammatory drugs (NSAIDs), immunosuppressants/immunomodulators (including methotrexate, azathioprine, and mycophenolate mofetil), glucocorticoids (oral or parenteral), and hydroxychloroquine.

Patients in the sub-cohort of SLE severity were categorized based on disease severity as patients with severe, moderate, or mild disease. Disease severity was determined using a proxy algorithm defined based on steroid dosage, immunosuppressant use, or appearance of SLE-related symptoms. The SLE severity algorithm was determined by using a study by Garris et al.²⁸ and in reference to EULAR recommendations for management of SLE^13,28,29 along with inputs from Japanese clinicians. Severe SLE patients were those who had at least one prescription for cyclophosphamide or rituximab or oral glucocorticoid with a prednisone equivalent dose higher than 60 mg/d, or at least one claim with a diagnosis listed as ‘severe’; and had severe conditions such as acute confusional state/psychosis, aortitis, arterial/venous thrombosis, aseptic meningitis, cardiac tamponade, cranial neuropathy, intestinal pseudo-obstruction, end-stage renal disease, optic neuritis, pulmonary hemorrhage, and stroke or transient ischemic attack. Moderate SLE patients were those with absence of “severe” disease; at least one filled prescription for an oral glucocorticoid with a prednisone-equivalent dose between 7.5 and 60 mg/d or for an immunosuppressive agent (other than cyclophosphamide), or at least 1 claim with a diagnosis of a condition listed as “moderate”; and had moderate conditions such as acute pancreatitis, chorioretinitis, demyelinating syndrome/acute inflammatory demyelinating polyradiculoneuropathy, episcleritis/scleritis, hemolytic anemia, hepatitis (non-viral), ischemic necrosis of bone, nephritis, renal impairment other than nephritis or end-stage renal disease, lupus enteritis/colitis, mononeuropathy/polyneuropathy, myelopathy, myocarditis, pericarditis, pleurisy/pleural effusion, pseudotumor cerebri, seizure, and uveitis vasculitis (excluding aortitis). Mild SLE patients were those who did not have moderate or severe disease.

Identification of treatment patterns

An algorithm was developed to detect start dates and end dates of LoTs of SLE therapy for each patient based on several assumptions. This algorithm converted all prescriptions of treatment for SLE into a set of treatment sequences using a succession of operations which included selection of prescriptions of interest, defining theoretical end of prescriptions, and identifying treatment sequences and combination therapies.

Prescriptions for SLE were retrieved using the JMDC Drug table where all drug prescriptions were recorded and identified according to ATC codes. The following treatment prescriptions were considered: topical steroids, NSAIDs, glucocorticoids (oral or parenteral), immunosuppressants/immunomodulators (including methotrexate, azathioprine and mycophenolate mofetil, and cyclophosphamide), hydroxychloroquine, and biologic therapies. See Appendix A for a list of drugs identified.

For topical therapy, one prescription was assumed to last for 14 d. If the prescription date was missing, approximation was performed using the date of the related claim. Certain assumptions were made for treatment overlaps. It was assumed that a treatment overlapping between topical therapy and other SLE LoT would not be considered as a period of concomitant therapy. Thus, it was assumed that: (a) patients would have stopped topical therapy when oral therapy would have started and (b) an overlap between topical glucocorticoids and other types of SLE treatment could occur because the discontinuation of glucocorticoid treatment required progressive dose tapering.

Daily defined dosages were used as the reference to detect start and end dates of a LoT i.e. duration of therapy, for each patient. Moreover, for overlap between topical therapy and other SLE therapy, it was assumed that if an overlap period of longer than 14 d was observed between two therapies that would be considered as combination therapy. A LoT was considered as “discontinued” if there was no prescription renewal for a period longer than the maximum allowed gap duration, also termed the “grace period”. A grace period of 60 d was used.

Treatment-related events of SLE therapies

Once the start and end dates of a LoT were determined, patients with LoT duration not exceeding the follow-up period were categorized as (temporarily) discontinued treatment, switched to other treatments, or having an add-on or a reduction in SLE therapy.

Treatment (temporarily) stop/discontinuation – Patients were defined to have (temporarily) stopped treatment at the end of a LoT if no SLE drug category was prescribed within the specified “grace period” after the end of the LoT.

Add-on therapy – The start of a new line was defined as an add-on therapy if SLE drug categories prescribed in the previous line were still prescribed in the new LoT and one or several new SLE drug categories, not prescribed in the previous line, were initiated. In addition, the period between the end of the previous line and the start of the new line was shorter than the specified “grace period”.

Switch to another therapy – The start of a new LoT was defined as switching if one or several new SLE drug categories prescribed in the previous treatment-line were no longer prescribed in the new LoT and one or several new SLE drug categories, not prescribed in the previous line, were initiated in the new line. The period between the end of the previous LoT and the start of the new LoT was shorter than the specified “grace period”.

Treatment reduction – The start of a new LoT was defined as a reduction if one or several new SLE drug categories prescribed in the previous treatment-line were no longer prescribed in the new LoT and no new SLE drug categories were prescribed in the new line. The period between the end of the previous line and the start of the new line was shorter than the specified “grace period”.

Continuous use of the LoT drugs (censored) – Continuous use of LoT SLE drug categories was defined as no switch, stop/discontinuation, add-on or reduction in treatment until end of follow-up.

Healthcare resource utilization

As HRU is influenced by factors such as the local healthcare system, number of specialists, and ease of healthcare access, understanding the existing healthcare system before analyzing HRU is essential. Japan’s healthcare system has been well explained by Zhang et al.³⁰ and Sakamoto et al.³¹

In addition to the treatment pattern measures, specific HRU measures were assessed for each patient. These measures included outpatient/physician office visit, hospitalization, and pharmacy claims. The proportion of patients with at least one occurrence for each LoT was calculated and reported by type of resource utilized. Each LoT identified had at least one occurrence and the number of occurrences per patient-time, for each of the healthcare resource categories. The numbers of occurrences per patient-time were calculated by dividing the number of occurrences by the duration of the LoT. For the number of resources used per person per month, the average number of occurrences per person-time were calculated during the different LoTs and reported by type of resource utilized and treatment regimen.

These proportions of patients with SLE were estimated using a Kaplan–Meier survival method in terms of measuring the length of time patients continued treatment or medical resource used before discontinuing or without event occurrence at last follow-up (e.g. censored data). First occurrence for the healthcare resource category of interest was the target event and censoring if the patient reached the end of follow-up period without any claim related to this type of resource.

For inpatient stays, the length of stay was described for patients with at least one inpatient stay including number of inpatient days shown by mean and its standard deviation (SD).

Medical cost evaluation

Cost data were extracted from the JMDC database and were reported as Japanese yen (¥). An overall cost per person-time was described by type of healthcare resource costs, treatment regimen costs, and costs related to SLE severity levels. The average cost per person-month was calculated and reported by type of resource utilized for the follow-up period. Patients contributed to these rates only when they were alive and in the health plan. Costs were not adjusted for inflation as Japan has a low inflation rate and prices have been flat for the last decade³². The low inflation rate is also one of the drivers for the expansive monetary policy of the Bank of Japan in recent years. Additionally, long-term interest rates (which could be used for discounting) are close to 0³³.

Statistical analysis

Data for all study measures were summarized descriptively as mean values, SD, median, and range for continuous variables and frequency distribution for categorical variables. Proportions were compared statistically using one-tailed two-sample proportion Z-test while continuous variables were compared statistically using two-sample T-test. Kaplan–Meier analyses were used to examine time to treatment-related events of interest such as time to switch to different treatment class or time to treatment discontinuation. All statistical analyses were conducted using SAS version 9.3 software (SAS Institute Inc., Cary, NC).

Results

Patient demographics

A total of 62,497 patients had at least one claim for SLE diagnosis in the study period and 4,733 patients among them fulfilled the inclusion criteria. The demographic characteristics of these patients are summarized in Table 1.

Table 1. Baseline demographics and comorbidities of patients with SLE (by Severity Level) during selection period.

Characteristics	All patients				Naïve patients
	Mild (N = 1,383)	Moderate (N = 2,619)	Severe (N = 731)	Total (N = 4,733)	Mild (N = 573)	Moderate (N = 1,114)	Severe (N = 385)	Total (N = 2,072)
Follow-up duration (days), mean (SD)	939.9 (687.0)	1,199.9 (769.0)	1,286.5 (774.3)	1,137.3 (758.0)	1,190.3 (679.9)	1,392.3 (707.8)	1,436.9 (683.0)	1,344.7 (702.0)
Age at index date, mean (SD)	45.4 (13.5)	44.8 (13.1)	46.4 (12.5)	45.2 (13.1)	46.5 (12.9)	45.2 (12.5)	47.1 (12.1)	45.9 (12.4)
Age at index date by categories, n (%)
18–34 years	302 (21.8)	607 (23.2)	125 (17.1)	1,034 (21.8)	104 (18.2)	235 (21.1)	62 (16.1)	401 (19.4)
35–44 years	365 (26.4)	703 (26.8)	208 (28.5)	1,276 (27.0)	155 (27.1)	325 (29.2)	98 (25.5)	578 (27.9)
45–54 years	331 (23.9)	637 (24.3)	192 (26.3)	1,160 (24.5)	146 (25.5)	279 (25.0)	117 (30.4)	542 (26.2)
55–64 years	270 (19.5)	509 (19.4)	157 (21.5)	936 (19.8)	120 (20.9)	210 (18.9)	82 (21.3)	412 (19.9)
65–74 years	106 (7.7)	155 (5.9)	47 (6.4)	308 (6.5)	44 (7.7)	61 (5.5)	26 (6.8)	131 (6.3)
75 + years	9 (0.7)	8 (0.3)	2 (0.3)	19 (0.4)	4 (0.7)	4 (0.4)	0	8 (0.4)
Gender: Male, n (%)	357 (25.8)	531 (20.3)	129 (17.6)	1017 (21.5)	145 (25.3)	205 (18.4)	68 (17.7)	418 (20.2)
Insurance status
Employees	613 (44.3)	1,077 (41.1)	266 (36.4)	1,956 (41.3)	325 (56.7)	697 (62.6)	251 (65.2)	1,273 (61.4)
Dependents	770 (55.7)	1,542 (58.9)	465 (63.6)	2,777 (58.7)	248 (43.3)	417 (37.4)	134 (34.8)	799 (38.6)
Comorbidity at baseline, n (%)
Mononeuropathy/polyneuropathy	26 (1.9)	200 (7.6)	62 (8.5)	288 (6.1)	3 (0.5)	19 (1.7)	5 (1.3)	27 (1.3)
Other renal impairment	28 (2.0)	63 (2.4)	23 (3.1)	114 (2.4)	4 (0.7)	6 (0.5)	3 (0.8)	13 (0.6)
Seizure	0	75 (2.9)	25 (3.4)	100 (2.1)	0	5 (0.4)	5 (1.3)	10 (0.5)
Pleurisy/pleural effusion	1 (0.1)	13 (0.5)	8 (1.1)	22 (0.5)	0	3 (0.3)	0	3 (0.1)
Arterial/venous thrombosis	1 (0.1)	5 (0.2)	125 (17.1)	131 (2.8)	0	0	10 (2.6)	10 (0.5)
Aortitis	0	0	0	0	0	0	0	0
Stroke/transient ischemic attack	2 (0.1)	2 (0.1)	25 (3.4)	29 (0.6)	1 (0.2)	1 (0.1)	2 (0.5)	4 (0.2)
Nephrotic syndrome	0	233 (8.9)	53 (7.3)	286 (6.0)	0	8 (0.7)	1 (0.3)	9 (0.4)
At least one comorbidity of interest at baseline, n (%)	56 (4.0)	527 (20.1)	219 (30.0)	802 (16.9)	6 (1.0)	41 (3.7)	21 (5.5)	68 (3.3)
CCI, mean (SD)	0.7 (1.3)	1.1 (1.6)	1.3 (1.8)	1.0 (1.6)	0.4 (1.1)	0.3 (0.9)	0.3 (1.0)	1.0 (1.0)
CCI, median	0	0	0	0	0	0	0	0

Abbreviations. CCI, charlson comorbidity index; SD, standard deviation; SLE, systemic lupus erythematosus.

Among the 4,733 included patients, 2,072 (43.8%) were treatment naïve with no treatment within the pre-index period, 2,214 (46.8%) were previously treated for SLE, and 447 (9.4%) were those who did not receive any treatment. Mean (SD) age of total SLE cohort was 45.2 (13.1) years and mean (SD) follow-up duration was 1,137.3 (758.0) d. Majority of the patients with SLE (78.5%) were females, with the incidence ranging from 21.8% to 24.5% in child bearing age of 18–45 years. Based on insurance status, 58.7% were dependents. In terms of disease severity level, 1,383 (29.2%) of the patients had mild, 2,619 (55.3%) had moderate, and 731 (15.5%) had severe SLE. Among treatment naive patients, 573 (27.7%) had mild, 114 (53.8%) had moderate, and 385 (18.6%) had severe SLE. Based on organ and system involvement, the sub-cohort included 941 (19.8%) patients with lupus nephritis, 40 (0.8%) with CNS lupus, 47 (1.0%) with both lupus nephritis and CNS lupus while 3,705 (78.3%) had no confirmed diagnosis of CNS lupus or lupus nephritis (Appendix B).

When comorbid conditions were determined using Charlson comorbidity index (CCI) as a reference, the mean (SD) CCI was 1.0 (1.6) and 802 (16.9%) patients had at least one comorbidity of interest at the baseline. About 6.1% of the patients were recorded with mononeuropathy or polyneuropathy, 6.0% had nephrotic syndrome, and 2.8% had arterial thrombosis.

Treatment patterns

Japanese patients with SLE were prescribed various types of treatment during the post-index period, as outlined in Table 2. The treatment patterns mainly comprised topical steroids, NSAIDs, glucocorticoids, immunosuppressants, and hydroxychloroquine. In treatment naïve patients with SLE, NSAIDs (31.9%) and glucocorticoids (both oral and parenteral; 31.0%) were largely used as the first LoT; with a similar pattern continuing up to the final LoT. Significantly higher NSAID (31.9% vs. 28.5%, p < .05) and glucocorticoid (31.0% vs. 19.0%, p < .001) utilization was observed in the first LoT than the second. However, the next LoTs did not show any significant differences. Parenteral glucocorticoids were administered to patients who had been prescribed a high dose of steroids in the first LoT. More than half of the treatment naive patients had moderate SLE in all the LoTs (Appendix C).

Table 2. Treatment regimen per treatment line in all treatment naïve patients with SLE.

Treatment regimen (n, %)	Line 1 (n = 2,072)	Line 2 (n = 1,508)	Line 3 (n = 1,096)	Line 4 (n = 758)	Line 5 (n = 539)	Line 6 (n = 392)
Monotherapy
Topical steroids	244 (11.8)**	218 (14.5)	156 (14.2)	106 (14.0)	75 (13.9)	53 (13.5)
NSAIDs	662 (31.9)*	430 (28.5)	322 (29.4)	229 (30.2)	142 (26.3)	98 (25.0)
Immunosuppresants	77 (3.7)	53 (3.5)	40 (3.6)**	49 (6.5)	33 (6.1)	25 (6.4)
Glucocorticoids	643 (31.0)***	287 (19.0)	219 (20.0)	146 (19.3)	111 (20.6)	76 (19.4)
Hydroxychloroquine	2 (0.1)	1 (0.1)	3 (0.3)	1 (0.1)	2 (0.4)	1 (0.3)
Combination therapy
NSAIDs + glucocorticoids	76 (3.7)***	219 (14.5)***	102 (9.3)	72 (9.5)	45 (8.3)	44 (11.2)
NSAIDs + hydroxychloroquine	0	0	0	0	0	1 (0.3)
Glucocorticoids + hydroxychloroquine	0**	4 (0.3)	3 (0.3)	2 (0.3)*	6 (1.1)	5 (1.3)
NSAIDs + glucocorticoids + hydroxychloroquine	0	0	0	1 (0.1)	2 (0.4)	0
NSAIDs + immunosuppressive agent	8 (0.4)***	21 (1.4)	20 (1.8)	7 (0.9)*	11 (2.0)	11 (2.8)
Immunosuppressive agent + glucocorticoids	293 (14.1)***	136 (9.0)**	140 (12.8)*	75 (9.9)	60 (11.1)	36 (9.2)
Immunosuppressive agent + hydroxychloroquine	0	0	0	0	0	0
NSAIDs + immunosuppressive agent + glucocorticoids	67 (3.2)***	132 (8.8)	84 (7.7)	64 (8.4)	51 (9.5)	39 (9.9)
NSAIDs + immunosuppressive agent + hydroxychloroquine	0	0	0	1 (0.1)	0	0
Immunosuppressive agent + glucocorticoids + hydroxychloroquine	0***	7 (0.5)	6 (0.5)	5 (0.7)	1 (0.2)	2 (0.5)
NSAIDs + immunosuppressive agent + glucocorticoids + hydroxychloroquine	0	0	1 (0.1)	0	0	1 (0.3)

Each line of treatment was compared against the subsequent line only using a simple two-sample proportion Z-test. E.g. line 1 vs. line 2, line 2 vs. line 3, etc. *p < .05, **p < .01, and ***p < .001. As the study does not test a preconceived hypothesis, patients who had repeated the medication were ruled out.

Abbreviations. NSAID, non-steroidal anti-inflammatory drug; SLE, systemic lupus erythematosus.

Treatment-related events during therapy

Discontinuation was the most frequent treatment-related event for most of the therapies administered in every LoT. NSAIDs (75.9%) and topical steroids (73.4%) were more significantly discontinued compared with other treatments (p < .001 for all). NSAIDs were used for the shortest duration (median 7 d) while glucocorticoids were used for the longest (median 100.5 d). Switching was the second highest treatment-related event after discontinuation. Switching was significantly higher in topical steroids (16.4%) than in other treatments except hydrochloroquine (p < .001 for all) and in NSAIDs compared with glucocorticoids (11.8% vs. 7.4%, p < .001). Glucocorticoids (35.5%) and immunosuppressants (24.2%) were more significantly used as add-on treatments than others (p < .001 for all, Table 3).

Table 3. Treatment persistence/patterns and treatment-related events according to the treatment regimen dispensed.

Treatment line	Topical steroids	NSAIDs	Immunosuppressants	Glucocorticoids	Hydroxychloroquine
First line
Number of patients	244	662	77	643	2
Duration of the line, mean (SD)	71.4 (125.6)	66.1 (257.2)	422.7 (481.6)***^###	675.2 (727.7)***^###	202.5 (207.2)
Duration of the line, median	28	7	240	350	202.5
Treatment-related events
Discontinuation	198 (81.2)	493* (74.5)	27 (35.1)***^###	176 (27.4)***^###	1 (50.0)
Switch	30 (12.3)	98 (14.8)	6 (7.8)^#	36 (5.6)***^###	0
Add-on	0	34 (5.1)***	28 (36.4)***^###	260 (40.4)***^###	0
Reduction	0	0	0	0	0
No change	16 (6.6)	37 (5.6)	16 (20.8)***^###	171 (26.6)***^###	1 (50.0)**^##
Second line
Number of patients	218	430	53	287	1
Duration of the line, mean (SD)	67.4 (154.6)	35.4 (124.4)**	231.2 (419.2)***^###	279.8 (455.7)***^###	14.0 (–)
Duration of the line, median	28	7	63	70	14
Treatment-related events
Discontinuation	173 (79.4)	351 (81.6)	20 (37.7)***^###	103 (35.9)***^###	0*^#
Switch	31 (14.2)	32 (7.4)**	2 (3.8)*	22 (7.7)**	0
Add-on	0	8 (1.9)*	16 (30.2)***^###	101 (35.2)***^###	1 (100.0)***^###
Reduction	0	0	0	0	0
No change	14 (6.4)	39 (9.1)	15 (28.3)***^###	61 (21.3)***^###	0
Third line
Number of patients	156	322	40	219	3
Duration of the line, mean (SD)	51.9 (74.3)	34.4* (78.4)	207.7 (315.2)***^###	296.2 (481.0)***^###	135.7 (159.6)^#
Duration of the line, median	28	7	75	78	45
Treatment-related events
Discontinuation	114 (73.1)	257 (79.8)*	15 (37.5)***^###	77 (35.2)***^###	0^**###
Switch	21 (13.5)	23 (7.1)*	4 (10.0)	14 (6.4)*	1 (33.3)^#
Add-on	0	6 (1.9)*	8 (20.0)***^###	74 (33.8)***^###	0
Reduction	0	0	0	0	0
No change	21 (13.5)	36 (11.2)	13 (32.5)**^##	54 (24.7)^**###	2 (66.7)**^##
All lines
Number of patients	977	2,187	388	1,692	10
Duration of the line, mean (SD)	66.9 (122.1)	42.7 (160.1)***	210.7 (341.9)***^###	396.5 (583.7)***^###	91.3 (129.8)
Duration of the line, median	28	7	57	100.5	43.5
Treatment-related events
Discontinuation	717 (73.4)	1,659 (75.9)	161 (41.5)***^###	526 (31.1)***^###	2 (20.0)***^###
Switch	160 (16.4)	257 (11.8)***	38 (9.8)***	125 (7.4)***^###	2 (20.0)
Add-on	0	77 (3.5)***	94 (24.2)***^###	600 (35.5)***^###	2 (20.0)^***##
Reduction	0	0	0	0	0
No change	100 (10.2)	194 (8.9)***	95 (24.5)***^###	441 (26.1)***^###	4 (40.0)^**###

Topical steroids and NSAIDs groups were compared against other groups. Continuous variables were compared using the two-sample T-test and proportions were compared using two-sample proportion Z-test. *p<.05, **p<.01, and ***p<.001 vs. topical steroids. ^#p<.05, ^##p<.01, and ^###p<.001 vs. NSAIDs.

Abbreviations. NSAID, non-steroidal anti-inflammatory drug; SD, standard deviation.

Distribution of glucocorticoids

Among the 4,733 patients included in the study, 3,115 (65.8%) received oral or parenteral glucocorticoids during the course of their treatment, while 2,666 (56.3%) were prescribed only oral and 1,510 (31.9%) were prescribed only parenteral glucocorticoids. High-dose glucocorticoids (>60 mg) were more likely to be administered parenterally compared with orally (36.0% vs. 1.3%, p < .001). However, oral high-dose glucocorticoids were prescribed for the longest duration (mean [SD] 13.8 [14.9] d, p < .001 compared with other routes). Low glucocorticoid dosage (<7.5 mg) was highly prescribed to patients with oral glucocorticoids (84.0%, p < .001 compared with other doses) as well as used for the highest cumulative duration per-patient (mean [SD] 1,009 [981.7] d, p < .001 compared with other doses) (Table 4). Figure 3 shows the distribution of the duration of oral glucocorticoid doses by calendar year. No significant changes were observed from 2010 to 2016.

Figure 3. Distribution of oral glucocorticoid doses per calendar year. Days are represented as mean.

Table 4. Distribution of glucocorticoid doses after index date in patients with SLE.

Distribution of glucocorticoids	Oral	Parenteral	Oral or parenteral
Patients (n)	2,666	1,510	3,115	–
Patients with high-dose glucocorticoids (>60 mg; n [%])	34 (1.3)	543 (36.0)***	561 (18.0)***	–
Cumulative duration of high-dose glucocorticoid prescription (days)
Mean (SD)	13.8 (14.9)	5.2 (8.4)***	5.8 (9.6)***	–
Median (IQR)	9 (2−25)	3 (1−6)	3 (1−6)	–
Patients with oral glucocorticoids (n = 2,666)	<7.5 mg	7.5−40 mg	40−60 mg	≥60 mg
Patients (n, %)	2,239 (84.0)	1,747 (65.5)***	326 (12.2)***	107 (4.0)***
Cumulative duration per-patient (days) (mean, SD)	1,009 (981.7)	500.4 (597.2)***	26.1 (22.2)***	16.4 (16.9)***

Proportion of patients with high-dose glucocorticoids was compared between the oral and other groups using two-sample proportion Z-test. Cumulative duration of high-dose glucocorticoid prescription was compared between oral and other groups using two-sample T-test. Proportion of patients on oral glucocorticoids was compared between <7.5 mg and other groups using two-sample proportion Z-test. Cumulative duration per-patients was compared between <7.5 mg group and other groups using two-sample T-test. ***p<.001.

Abbreviations. IQR, interquartile range; SD, standard deviation; SLE, systemic lupus erythematosus.

Healthcare resource utilization and associated costs

At least one hospitalization was reported by 32.4% of patients overall; patients with severe SLE reported higher proportion of hospitalizations (58.1% vs. 33.0% and 17.7%, p < .001 compared with both moderate and mild groups). Irrespective of SLE severity, number of hospitalizations per-patient per-month was similar, with less than two occurrences per year. Almost every patient reported an outpatient visit (99.4%); patients with severe SLE reporting higher number of visits per-patient per-month (mean [SD] 3.7 [5.3]) compared with moderate (3.1 [4.0]) and mild groups (2.6 [2.9]) (p < .001 for both). Magnetic resonance imaging (MRI) was used more than X-rays (34.4% vs. 8.9%; Table 5).

Table 5. Healthcare resource utilization during follow-up period.

Parameter	Mild (n = 1,383)	Moderate (n = 2,619)	Severe (n = 731)	Total (n = 4,733)
Hospitalization
At least one admission (n, %)	245 (17.7)	864 (33.0)***	425 (58.1)***^###	1,534 (32.4)
No. admissions per patient/ month^¥ (mean, SD)	0.09 (0.17)	0.07 (0.12)***	0.08 (0.11)^#	0.07 (0.13)
Outpatient visits
At least one visit (n, %)	1,374 (99.4)	2,603 (99.4)	727 (99.5)	4,704 (99.4)
No. visits per patient/ month^¥ (mean, SD)	2.6 (2.9)	3.1 (4.0)***	3.7 (5.3)***^###	3.1 (4.0)
X-Ray
At least one occurrence (n, %)	95 (6.9)	239 (9.1)**	77 (10.5)**	411 (8.9)
No. occurrences per patient/ month^¥ (mean, SD)	0.11 (0.12)	0.09 (0.09)***	0.08 (0.06)***^##	0.09 (0.09)
MRI
At least one occurrence (n, %)	293 (21.2)	962 (36.7)***	374 (51.2)***^###	1,629 (34.4)
No. occurrences per patient/ month^¥ (mean, SD)	0.13 (0.35)	0.09 (0.15)***	0.10 (0.14)*	0.10 (0.20)

*Among patients with at least one occurrence.

All the groups were compared against each other. Proportions were compared using two-sample proportion Z-test and continuous variables were compared using the two-sample T-test. ^¥p<.05, **p<.01, and ***p<.001 compared with mild group. ^#p<.05, ^##p<.01, and ^###p<.001 compared with moderate group.

Abbreviations. MRI, magnetic resonance imaging; SD, standard deviation.

Among treatment naïve patients, those who received glucocorticoids were hospitalized the most (21%, p < .001 compared with others), while those who received immunosuppressants reported the highest number of admissions per-patient per-month (mean [SD] 5.1 [12.2], p < .05 compared with others). More than 90% of patients receiving topical steroids, immunosuppressants, or glucocorticoids reported at least one outpatient visit. However, patients receiving NSAIDs reported the highest number of outpatient visits per-patient per-month (mean [SD] 15.2 [20.4], p < .001 compared with others). Similar to the overall cohort, MRI was used more frequently than X-rays by treatment naive patients irrespective of the administered treatment, with patients on NSAIDs reporting the highest number of X-ray visits per-patient per-month (mean [SD] 1.3 [2.2]) and MRI visits per-patient per-month (mean [SD] 1.9 [2.9]) (p < .0.01 for both, Table 6).

Table 6. Healthcare resource utilization in treatment naïve patients by first treatment line and by treatment classes.

Parameter	Topical steroids (n = 244)	NSAIDs (n = 662)	Immunosuppressants (n = 77)	Glucocorticoids (n = 643)	Hydroxychloroquine (n = 2)
Hospitalization
At least one admission (n, %)	5 (2.1)	52 (7.9)***	6 (7.8)**	135 (21.0)***^###	0
No. admissions per patient/ month^¥ (mean, SD)	0.70 (0.47)	3.3 (5.1)***	5.1 (12.2)***^#	0.73 (2.5)^###	–
Outpatient visits
At least one visit (n, %)	229 (93.9)	515 (77.8)***	73 (94.8)^###	596 (92.7)^###	2 (100.0)
No. visits per patient/ month^¥ (mean, SD)	4.5 (4.0)	15.2 (20.4)***	3.2 (4.4)*^###	5.01 (13.6)^###	3.5 (2.8)
X-Ray
At least one occurrence (n, %)	0	9 (1.4)*	1 (1.3)*	32 (5.0)***^###	0
No. occurrences per patient/ month^¥ (mean, SD)	–	1.3 (2.2)	0.05 (–)^###	0.14 (0.18)^###	–
MRI
At least one occurrence (n, %)	6 (2.5)	45 (6.8)**	7 (9.1)**	125 (19.4)***^###	0
No. occurrences per patient/ month^¥ (mean, SD)	0.61 (0.57)	1.9 (2.9)***	0.09 (0.04)***^###	0.85 (4.0)^###	–

^¥Among patients with at least one occurrence, – Indicates not calculable due to low or zero number of patients.

Topical steroids and NSAIDs groups were compared against other groups. Proportions were compared using two-sample proportion Z-test and continuous variables were compared using two-sample T-test. *p < .05, **p < .01, and ***p < .001 compared with topical steroids. ^#p < .05, ^##p < .01, and ^###p < .001 compared with NSAIDs.

Abbreviations. MRI, magnetic resonance imaging; NSAID, non-steroidal anti-inflammatory drug; SD, standard deviation.

A greater proportion of patients with SLE and lupus nephritis were hospitalized compared to naive or patients without CNS lupus and lupus nephritis (45.8% vs. 36.1% vs. 28.2%, p < .001 for both comparisons). However, the number of hospitalizations per-patient per-month was highest in patients without CNS lupus and lupus nephritis (mean [SD] 0.08 [0.13], p < .05 compared with other groups). Almost every patient with SLE reported at least one outpatient visit; patients without CNS lupus and lupus nephritis reported the highest number of visits per-patient per-month (mean [SD] 3.15 [4.05], p < .05 compared with other groups). MRI was utilized more than X-rays in every sub-cohort, with <1 occurrence per-patient per-month for every sub-cohort (Appendix D).

Mean (SD) annual inpatient, outpatient, and pharmacy costs per-patient for the overall cohort were ¥379,380 (¥491,587), ¥390,036 (¥388,814), and ¥247,596 (¥212,336), respectively, resulting in a total annual per-patient cost of ¥1,017,012 (¥707,221). Patients with severe SLE reported the highest expenses regardless of type of cost (p < .001 for all comparisons) while patients with moderate SLE reported higher expenses than patients with mild SLE for all types of costs (p < .001 for all). Outpatient costs were higher than inpatient costs for patients with mild and moderate SLE. However, the reverse was true for patients with severe SLE (Figure 4). For informational purposes, the costs are also reported in US$ using purchasing power parity for GDP data from 2017³⁴. The reported values are not for comparison as the healthcare markets and financing systems vary across countries and regions. Mean (SD) annual inpatient, outpatient, and pharmacy costs per-patient for the overall cohort were $3,600 ($4,665), $3,701 ($3,690), and $2,350 ($2,015), respectively, resulting in a total annual per-patient cost of $9,651 ($6,711).

Figure 4. Direct healthcare cost of Japanese SLE patients. Total healthcare costs of SLE in Japanese patients, estimated during entire follow up period of the study. Data are plotted as mean of the total patient number in the cohort/sub-cohort. Costs of patients with severe, moderate, and mild SLE were compared against each other using two-sample T-test. ***p<.001 compared with mild SLE group and ^###p<.001 compared with moderate SLE group.

Discussion

SLE is a complicated and fickle disease that inflicts a considerable humanistic and economic burden on patients, caregivers, and the health services. This retrospective cohort study allowed us to observe longitudinal treatment patterns and HRU in patients with SLE in Japan. Our study identified differences among patients in terms of sex, age, disease duration, and comorbidity at baseline, which were consistent with a previous epidemiology report³⁵. Both males and females were susceptible during their productive years, with SLE being highly prevalent among females. Based on insurance category, more than half of the overall patients with SLE were identified as dependent household members, suggesting that they were either unemployed or unable to work due to SLE. On the other hand, the majority of claims in the treatment naïve group were employment-related. This observation might imply that these patients were on a treatment break or undergoing a washout period for 6 months wherein they did not experience any symptoms. However, the fact that such patients restarted treatment suggests that these patients experienced SLE-related symptoms which could, hypothetically, negatively impact their ability to work. It should be noted that the existing data are not comprehensive enough to fully support our speculation. Moreover, the difference in type of insurance category claims could also have resulted from a skewed patient proportion due to the treatment naïve selection algorithm.

In our study, the proportion of patients with moderate SLE was greater compared to patients with mild or severe SLE. Due to the complexity of SLE symptom presentation, treatment administration could significantly vary depending on factors such as comorbidities, signs and symptoms, extent of organ involvement, and access to different specialists for care³⁶. Patients with moderate SLE could be at risk of progressing to severe SLE in case of delayed or incorrect management. Moreover, glucocorticoid-induced side effects would add to the clinical burden.

We identified SLE-related claims for five major drug classes (topical steroids, NSAIDs, immunosuppressants, glucocorticoids, and hydroxychloroquine and their combinations). The meagre number of drug classes observed could be explained by the stringent treatment algorithm that defined the LoT in this study as well as by the presence of a standard treatment algorithm for the clinical management of SLE^12,13. Our treatment pattern assessment was comparable to a previous Japanese study¹⁴ as well with reports from other countries^37,38. Utilization of antimalarials (e.g. hydroxychloroquine) was low in our and Tanaka et al.’s¹⁴ study, since hydroxychloroquine for SLE was approved in 2015 and adequate capture of this treatment in the database would require a longer observation period.

In our study, glucocorticoids and NSAIDs were the most used first LoT in treatment naïve Japanese patients with SLE. Oral glucocorticoids (7.5–40 mg) were administered for a cumulative duration of 17 months for each patient. This shows that a large proportion of patients with SLE were unlikely to successfully complete glucocorticoids tapering within the average 17 months period. In the EULAR recommendation on corticosteroid management, long-term use of dosages above 7.5 mg/d can substantially increase the risk of harmful effects (e.g. irreversible organ damage)¹³. Based on our observation of long-term usage of glucocorticoid doses >7.5 mg, it is suspected that patients who received these doses were likely to suffer from SLE-related flares and to have high risk of organ-threatening conditions. This assumption was made in reference to SLE flare measurement by treatment change as SLE flare has not yet reached a globally accepted definition¹³. This could indicate that current option of SLE treatment may not be optimal for patients with SLE to prevent flares or reduce detrimental effects.

HRU in this study was similar to the results reported by Tanaka et al.¹⁴ in a retrospective, observational cohort study conducted using claims data of Japanese patients who were diagnosed with SLE between April 2010 and March 2012. These patients were followed-up for 3 years and data on HRU and costs were analyzed. All the patients in both this study and Tanaka et al.’s study had at least one outpatient visit while 30−40% had an inpatient visit. These visits tied into the direct costs as well, as inpatient and outpatient costs were higher than pharmacy costs. The high occurrence of outpatient visits or hospitalizations in patients who received NSAIDs or immunosuppressants could have resulted in the high annual cost observed in this study. At the same time, long duration of treatment (e.g. glucocorticoids) with increasing add-on medications to control frequent flare activity could have increased the costs too^39,18. Although patients with SLE in Japan are eligible for funding support from the intractable disease program⁴⁰, the costs were quite high in the severe group compared to mild and moderate group. Mean overall direct costs in the current study were lower than Tanaka et al.’s¹⁴ results (¥1,017,012 vs. ¥2,975,066). The large difference in mean direct medical costs could have occurred due to the different inclusion criteria. While this study included patients who were enrolled in the JMDC database for 6 months prior to and after the index date, Tanaka et al. only included patients who were enrolled for 6 months prior to and 3 years after the index date. Patients present in the database for such a long time would have been undergoing continuous treatment and hence experienced greater direct costs. Nevertheless, both the studies reflect the massive economic burden inflicted by SLE, especially in patients with severe disease.

Steroid exposure has been associated with unwanted organ damage⁴¹ including an increased risk of cataract and osteoporosis in patients with SLE. However, steroids have remained the backbone of SLE therapy in Japan. This scenario could have changed after late 2017 when MHLW approved the use of belimumab as adjunctive therapy in autoantibody-positive patients with SLE who do not respond adequately to existing therapies⁴². As novel treatment therapies for SLE – such as biologics – emerge, there would be greater potential for clinicians and patients to expand therapeutic strategies in order to improve prognosis for SLE, reduce treatment burden, and enable maintenance of long-term remission to prevent flares.

Limitations

The findings of this study should be considered within the context of several limitations inherent to the JMDC data source which could not rule out biases toward SLE patients’ conditions that attribute to the spectrum of SLE severity, flare incidences, or individual physical status (e.g. female hormonal or reproductive condition). The elderly are underrepresented in the JMDC database as enrollment into the database is linked to employment status. Patients who were not represented in the database might be receiving healthcare benefits under other insurance types. Loss of productivity and other indirect costs could not be evaluated in the claims database. Our study was descriptive and not comparative in nature. The algorithm of SLE and disease severity using claims databases has not been previously validated. The patient’s medical history might have influenced a physician’s decision to modify or switch treatments. The database did not have an account of reasons underlying specific treatment-related events such as discontinuation, switching, and augmentation, which could not correlate to the outcomes of the patients. In terms of drug claims capturing, some recently approved drugs for SLE may not reflected in this analysis; for example, hydroxychloroquine for SLE, mycophenolate mofetil for lupus nephritis, and belimumab for SLE were approved in 2015 or later in Japan. In this case, the number of these drug claims may not fully account in the treatment pattern in our follow-up time frame. Hence, interpretation of the treatment patterns and treatment-related events should be made with caution.

Conclusion

This is one of the few Japanese studies that analyzed current treatment patterns, treatment-related events, and economic burden of SLE from an administrative database. Glucocorticoids were the backbone of SLE therapy to which other medications were added on. However, the myriad risks associated with steroids should shift healthcare providers and policy makers toward more effective treatment for SLE, resulting in lower SLE-associated burden.

Transparency

Declaration of funding

Data acquisition, research, and preparation of the manuscript were funded by Janssen Pharmaceutical K.K. The funding body did not have any additional role in the study design, data collection, and analysis, decision to publish, or preparation of the manuscript.

Declaration of financial/other interests

CM, RS, JM, and WJ were employees of Janssen Pharmaceutical K.K. at the time the study was conducted.

The peer reviewers on this manuscript have received an honorarium from JME for their review work. In addition, a reviewer on this manuscript has disclosed that they are a shareholder in a company which has received income for scientific services and expenses from Celgene, Janssen and Novartis, amongst others. Per company policy, employees/shareholders cannot contract individually with sponsor organizations and cannot hold equity in sponsor organizations. The reviewers have no other relevant financial relationships or otherwise to disclose.

Author contributions

CM conceived the idea and drafted the manuscript. RS performed data acquisition. CM and JM analyzed and interpreted the data. WJ also assisted in data interpretation. WJ, RS, and JM reviewed and revised the manuscript. All authors provided final approval for the manuscript.

Acknowledgements

The authors thank Leo J. Philip Tharappel from Siro Clinpharm Pvt. Ltd, Mumbai and Dr. Kiran Saqib for providing medical writing assistance and reviewing the manuscript for final submission.

Data availability

The datasets analyzed during the current study are not publicly available due to copyright reasons but are available from the corresponding author on reasonable request.

Appendix A. List of drugs identified during all follow-up.

Table

Drug class	Drug name
Topical steroids	Alclometasone Dipropionate
	Beclometasone Dipropionate
	Betamethasone Butyrate Propionate
	Betamethasone Dipropionate
	Betamethasone Valerate
	Clobetasol Propionate
	Clobetasone Butyrate
	Deprodone Propionate
	Dexamethasone
	Dexamethasone Propionate
	Dexamethasone Valerate
	Diflorasone Diacetate
	Diflucortolone Valerate
	Difluprednate
	Fludroxycortide
	Fluocinolone Acetonide
	Fluocinonide
	Hydrocortisone Butyrate
	Hydrocortisone Butyrate Propionate
	Mometasone Furoate
	Prednisolone Farnesylate
	Prednisolone Valerate Acetate
NSAIDs	Acemetacin
	Amfenac Sodium Hydrate
	Ampiroxicam
	Bucolome
	Celecoxib
	Diclofenac Sodium
	Etodolac
	Flufenamate Aluminum
	Flurbiprofen
	Flurbiprofen Axetil
	Ibuprofen
	Indometacin
	Indometacin Farnesil
	Ketoprofen
	Lornoxicam
	Loxoprofen Sodium Hydrate
	Mefenamic Acid
	Meloxicam
	Naproxen
	Neurotropin
	Pranoprofen
	Sulindac
	Tiaprofenic Acid
	Tiaramide Hydrochloride
	Zaltoprofen
Immunosuppressants/ immunomodulators	Abatacept (Genetical Recombination)
	Adalimumab (Genetical Recombination)
	Anti-human Thymocyte Immunoglobulin, Rabbit
	Azathioprine
	Basiliximab (Genetical Recombination)
	Certolizumab Pegol (Genetical Recombination)
	Ciclosporin
	Cyclophosphamide Hydrate
	Etanercept (Genetical Recombination)
	Golimumab (Genetical Recombination)
	Infliximab (Genetical Recombination)
	Leflunomide
	Lenalidomide Hydrate
	Methotrexate
	Mizoribine
	Mycophenolate Mofetil
	Salazosulfapyridine
	Tacrolimus Hydrate
	Tocilizumab (Genetical Recombination)
	Tofacitinib Citrate
	Ustekinumab (Genetical Recombination)
Glucocorticoids	Betamethasone
	Betamethasone Acetate/Betamethasone Sodium Phosphate
	Betamethasone Sodium Phosphate
	Cortisone Acetate
	Dexamethasone
	Dexamethasone Metasulfobenzoate Sodium
	Dexamethasone Palmitate
	Dexamethasone Sodium Phosphate
	Fludrocortisone Acetate
	Hydrocortisone
	Hydrocortisone Sodium Phosphate
	Hydrocortisone Sodium Succinate
	Methylprednisolone
	Methylprednisolone Acetate
	Methylprednisolone Sodium Succinate
	Prednisolone
	Prednisolone Sodium Succinate
	Triamcinolone
	Triamcinolone Acetonide
Hydroxychloroquine	Hydroxychloroquine Sulfate

Appendix B. Baseline characteristic of SLE by organ and system involvement.

Table

Characteristics	Lupus nephritis	CNS Lupus	CNS + Lupus nephritis	No CNS lupus and Lupus nephritis	All
	n = 941	n = 40	n = 47	n = 3705	n = 4733
Follow-up duration (days), mean (SD)	1,342.2 (777.2)	1,243.5 (711.5)	1,582.3 (738.0)	1,078.5 (742.8)	1,137.3 (758.0)
Age at index date, mean (SD)	41.3 (12.6)	44.2 (10.9)	42.9 (13.3)	46.3 (13.1)	45.2 (13.1)
Age at index date by categories, n (%)
18–34 years	292 (31.0%)	8 (20.0%)	14 (29.8%)	720 (19.4%)	1034 (21.8%)
35–44 years	280 (29.8%)	12 (30.0%)	13 (27.7%)	971 (26.2%)	1276 (27.0%)
45–54 years	213 (22.6%)	13 (32.5%)	10 (21.3%)	924 (24.9%)	1160 (24.5%)
55–64 years	129 (13.7%)	5 (12.5%)	9 (19.1%)	793 (21.4%)	936 (19.8%)
65–74 years	26 (2.8%)	2 (5.0%)	0 (0.0%)	280 (7.6%)	308 (6.5%)
75 + years	1 (0.1%)	0 (0.0%)	1 (2.1%)	17 (0.5%)	19 (0.4%)
Gender: Male, n (%)	159 (16.9%)	4 (10.0%)	3 (6.4%)	851 (23.0%)	1017 (21.5%)
Insurance status
Employees	385 (40.9%)	11 (27.5%)	13 (27.7%)	1547 (41.8%)	1956 (41.3%)
Dependents	556 (59.1%)	29 (72.5%)	34 (72.3%)	2158 (58.2%)	2777 (58.7%)
CCI, Mean (SD)	1.0 (1.5)	0.9 (1.4)	1.2 (1.7)	1.0 (1.6)	1.0 (1.6)
CCI, Median	0	0	0	0	0
Comorbidity at baseline, n (%)
Mononeuropathy/polyneuropathy	39 (4.1%)	2 (5.0%)	2 (4.3%)	245 (6.6%)	288 (6.1%)
Other renal impairment	22 (2.3%)	0 (0.0%)	1 (2.1%)	91 (2.5%)	114 (2.4%)
Seizure	16 (1.7%)	6 (15.0%)	3 (6.4%)	75 (2.0%)	100 (2.1%)
Pleurisy/pleural effusion	5 (0.5%)	0 (0.0%)	0	17 (0.5%)	22 (0.5%)
Arterial/venous thrombosis	43 (4.6%)	3 (7.5%)	0 (0.0%)	83 (2.2%)	131 (2.8%)
Aortitis	0 (0.0%)	0 (0.0%)	0 (0.0%)	0 (0.0%)	0 (0.0%)
Stroke/transient ischemic attack	4 (0.4%)	0 (0.0%)	2 (4.3%)	23 (0.6%)	29 (0.6%)
Nephrotic syndrome	252 (26.8%)	0 (0.0%)	11 (23.4%)	23 (0.6%)	286 (6.0%)
At least one comorbidity at baseline, n(%)	296 (31.5%)	9 (22.5%)	15 (31.9%)	482 (13.0%)	802 (16.9%)

Abbreviations. CCI, charlson comorbidity index; CNS, central nervous system; SD, standard deviation; SLE, systemic lupus erythematosus.

Appendix C. Disease severity as per treatment line.

Table

Severity (n, %)	Line 1	Line 2	Line 3	Line 4	Line 5	Line 6
Mild	573 (27.7)	369 (24.5)	239 (21.8)	151 (19.9)	106 (19.7)	67 (17.1)
Moderate	1,114 (53.8)	833 (55.2)	613 (55.9)	434 (57.3)	306 (56.8)	226 (57.7)
Severe	385 (18.6)	306 (20.3)	244 (22.3)	173 (22.8)	127 (23.6)	99 (25.3)

Appendix D. Healthcare resource used during all follow-up in treatment naïve patients, SLE with lupus nephritis and patients without CNS lupus and lupus nephritis.

Table

All follow-up	Naïve patients	SLE with Lupus nephritis	Patients without CNS lupus and lupus nephritis
Hospitalization	n = 2,072	n = 941	n = 3,705
At least one admission, n (%)	748 (36.1%)	431 (45.8%)***	1,043 (28.2%)***^###
No. admissions per patient/ month,^¥ mean (SD)	0.06 (0.11)	0.07 (0.13)*	0.08 (0.13)***^#
Outpatient visits	n = 2,072	n = 941	n = 3,705
At least one visit, n (%)	2,065 (99.7%)	940 (99.9%)	3,677 (99.2%)*^#
No. visits per patient/ month,^¥ mean (SD)	2.9 (4.1)	2.73 (3.7)	3.15 (4.1)*^##
X-Ray	n = 2,072	n = 941	n = 3,705
At least one occurrence, n (%)	219 (10.6%)	90 (9.6%)	312 (8.4%)**
No. occurrences per patient/ month,^¥ mean (SD)	0.09 (0.09)	0.08 (0.07)**	0.10 (0.10)***^###
MRI	n = 2,072	n = 941	n = 3,705
At least one occurrence, n (%)	787 (38.0%)	374 (39.7%)	1,197 (32.3%)***^###
No. occurrences per patient/ month,^¥ mean (SD)	0.08 (0.20)	0.07 (0.08)	0.10 (0.22)***^###

^¥Among patients with at least one occurrence.

All groups were compared against each other. Proportions were compared using two-sample proportion Z-test and continuous variables were compared using two-sample T-test. *p < 0.05, **p<.01, and ***p<.001 compared with naïve patients. ^#p<.05, ^##p<.01, and ^###p<.001 compared with SLE patients with lupus nephritis.

Abbreviations. CNS, central nervous system; SD, standard deviation; SLE, systemic lupus erythematosus.