Treatment patterns and health care resource utilization among Japanese patients with ankylosing spondylitis: A hospital claims database analysis

Abstract

Objectives

To understand the current state of treatment patterns and health care resource utilization among patients in Japan with ankylosing spondylitis (AS) managed in the real-world setting.

Methods

Patient records from the Medical Data Vision database were analyzed to identify patients with ICD-10 AS from April 2009 through July 2017. Measures evaluated included demographic, clinical, and other characteristics at diagnosis; treatment patterns; health care resource utilization; and costs.

Results

Four hundred and seventeen patients met the study’s inclusion criteria. Treatments observed during the first year after the initial AS diagnosis included nonsteroidal anti-inflammatory drugs (79.6%), corticosteroids (39.3%), methotrexate (22.3%), sulfasalazine (16.8%), adalimumab (14.2%), and infliximab (12.2%). At any time during the mean 33 months of study follow-up, biologic disease-modifying antirheumatic drugs (bDMARDs) were initiated by 115 patients. During the study follow-up, patients who initiated bDMARDs had higher median total per-patient annual health care costs ($26,937 vs $15,323), lower median per-patient hospitalization costs ($29,817 vs. $39,509), and fewer median hospital days per admission (7.0 vs. 11.0 days) compared with the overall group of patients diagnosed with AS.

Conclusion

This database study provides knowledge of patient characteristics, treatment patterns, HCRU, and costs for patients with AS in Japan. The study outcomes demonstrate a need for increased awareness of proper AS management.

Keywords:

• Ankylosing spondylitis
• biologic disease-modifying antirheumatic drugs
• claims database
• Japan
• spondyloarthritis

Introduction

Spondyloarthritis encompasses a group of inflammatory autoimmune conditions that can lead to severely impaired quality of life among those affected. In Japan, spondyloarthritis is relatively uncommon, with an estimated prevalence of 9.5 per 100,000 person-years [1]. However, the current diagnosis rate for spondyloarthritis has been increasing and suggests spondyloarthritis has historically been both underdiagnosed and misdiagnosed in Japan [2]. The difficulty in accurately diagnosing this condition could be due to a lack of information on clinical characteristics of spondyloarthritis in Japanese patients [2]. Ankylosing spondylitis (AS) is the most common type of spondyloarthritis, comprising 68.3% of spondyloarthritis cases in Japan [1]. Although the presence of human leukocyte antigen B27 (HLA-B27) is common among patients with AS, the prevalence and incidence of AS varies across populations because of ethnic differences in the presence of HLA-B27 [3]. In Japan, HLA-B27 is uncommon, which could plausibly explain a low prevalence and incidence of AS [1,3].

As there is no cure for AS, treatment options focus on symptom control, preventing joint damage, and maintaining or achieving disease remission. Typical treatment options include nonsteroidal anti-inflammatory drugs (NSAIDs), biologic disease-modifying antirheumatic drugs (bDMARDs), and conventional synthetic disease-modifying antirheumatic drugs (csDMARDs). Nonsteroidal anti-inflammatory drugs are often prescribed as first-line treatment for control of pain and stiffness [4]. If NSAIDs are unsuccessful, bDMARDs are recommended based on substantial evidence of rapid and sustained response, particularly among young patients with a short disease duration, high level of inflammatory markers, and good functional status [4]. Biologics, including infliximab, adalimumab, secukinumab, and ixekizumab have been approved for treatment of AS in Japan for patients who failed to respond to NSAIDs. In Japan, AS may be diagnosed and managed by a variety of specialists (e.g. rheumatologists, orthopedists), each of whom may approach therapeutic treatment differently. To avoid inconsistencies with regard to treatment approaches for managing patients with AS, the Japanese Ministry of Health, Labour and Welfare (MHLW) is preparing the first treatment guidelines for AS in Japan; however, there is still a lack of evidence for understanding the current as well as historical treatment patterns, health care resource utilization (HCRU), and economic burden.

We conducted a retrospective study using a Japanese hospital database to address the significant gap in the literature on the understanding of patient characteristics, treatment patterns, HCRU, and costs among patients managed in the real-world setting. The evaluation of treatment patterns included assessment of medication discontinuation, persistence, estimated adherence, augmentation, AS-related concomitant medication use, and switching among patients who initiated bDMARDs.

Methods

Study design

This retrospective cohort study analyzed patient records from the Medical Data Vision (MDV) database. The MDV database is a large, hospital-based database derived from hospital claims data recorded at approximately 369 hospitals across Japan [5,6]. The database is composed of inpatient and outpatient visits, procedure, and pharmacy claims data linked by an anonymized patient identifier for more than 20 million patients seeking care since January 2008. The age distribution of the database population is similar to that of the general population in Japan, and data from the MDV has been used in multiple health care database studies [7–11]. All patients with a diagnosis of AS according to International Classification of Diseases, Tenth Revision (ICD-10; codes M08.1, M45.0, M45.4, M45.9) [12] during the study selection window (April 1, 2009, through July 31, 2017) were screened for study inclusion. To increase the specificity of the sample, included patients were additionally required to have a second AS diagnosis at least 30 days after the first AS diagnosis. The MDV database indicates each diagnosis code as either ‘confirmed’ or ‘doubt’ diagnosis and only ‘confirmed’ diagnoses were used for this study. The date of the first AS diagnosis during the selection window was used to define the index date. All patients were required to be at least 10 years of age on their index date. Patients were also required to have at least 3 months of continuous data available before their index date (termed baseline period) and at least 12 months of continuous data available after their index date to allow for sufficient time to observe the study measures. To ensure patients with AS were newly diagnosed with the condition, we excluded from the analysis those who had AS diagnoses during the baseline period. The periods of continuous data availability before and after the index date were established by selecting the earliest and most recent medical encounters between 1 April 2008, and 31 July 2018 (time-period for which database was available when the study was conducted). To be eligible for the study, patients were required to have a subsequent diagnosis of AS that was at least 30 days after the index date. A subgroup of patients with AS who initiated a bDMARD at any time during the available follow-up post-index were identified by the presence of at least one agent-specific health claim code(s) associated with anatomical therapeutic chemical (ATC) codes for bDMARDs (L04B, L04C, M01C) combined with product generic and brand names (Supplemental Material, Table S-1). The first bDMARD identified after the index date was used to define the index bDMARD.

Assessments

Demographic, clinical, and other characteristics at diagnosis

Age and gender were measured at the index date. The Charlson Comorbidity Index (CCI) score was calculated to obtain a measure of patients’ overall comorbidity burden during the baseline period [13] (Supplemental Material, Table S-2). Comorbidities or complications related to AS that were not captured in the CCI were also examined during the baseline period (Supplemental Material, Table S-3). Additionally, the distributions of physician departments and imaging procedures associated with the index diagnosis were reported (Supplemental Material, Table S-4). Since the database did not provide the exact day (only month and year) when a visit was made to a particular department, a hierarchy-based algorithm was used to assign the index department to each patient. If a patient had multiple visits to different departments during the index month, including a visit to rheumatology, the patient was assigned first to rheumatology. If no visit to rheumatology was observed, the next preference was given to orthopedics, then to internal medicine, then dermatology, and then other specialties. The total follow-up for a given patient was defined as the time from the index date to the last recorded medical encounter in the MDV database for that patient.

Treatment patterns

Longitudinal treatment patterns were reported. These patterns included the number and percentage of patients with AS-related therapies received during the first 12-month follow-up period, along with the average time from the index date to the first observed treatment. To define treatment line, the first prescription fill of AS-related therapies after AS diagnosis was defined as the start of a line of treatment. Any additional treatment administered within 30 days of start of treatment defined a combination therapy. End of a specific line of treatment was defined as: (1) line continues until the end of follow-up period; (2) no drug administered within 90 days of exhaustion of the last administration; or (3) previous line of treatment was interrupted by a new agent/drug class. To assess treatment sequences, this process was repeated for three times during the available follow-up period. Second- and third-line treatments following the top five most frequently administered first-line treatments were described. Treatments were identified based on specific health claims codes that were associated with respective ATC codes and product generic and brand names (Supplemental Material, Table S-1). Procedure codes for physical therapy and surgeries are presented in Table S-4 (Supplemental Material).

Treatment patterns with bDMARDs

Medication discontinuation, persistence, estimated adherence, augmentation, AS-related concomitant medication use, and switching were evaluated for the first bDMARD therapy. Treatment discontinuation was defined as ending treatment with the specific bDMARD without evidence of another refill within 90 days of exhausting the drug supply for the prior prescription. Days’ supply for each bDMARD administration was imputed based on the prescribing guidelines of each agent as these are injection-based therapies. The date of exhaustion of the days’ supply from the last prescription before the initiation of an observed treatment gap (if applicable) was considered the date of treatment discontinuation. Treatment persistence was defined as continuation of bDMARD treatment at the end of 12 months and 24 months of study follow-up. Estimated treatment adherence was defined using the medication possession ratio (MPR), which was defined as the proportion of patients’ time on drug with medication supply on hand (i.e. sum of days’ supply of all fills for the drug divided by total patient time on drug; time on drug was defined as the days between the initiation of a bDMARD and the end of the days’ supply of the last administration). The MPR was computed for the overall study follow-up as well as at the end of 12-months and 24-months of study follow-up. Switching was defined as receipt of a new prescription for a different bDMARD and/or csDMARDs following discontinuation of the index bDMARD. Combination regimens were determined by including treatments that were initiated within 30 days of the first agent that the patient switched to after discontinuation of the index bDMARD. Augmentation was defined as the uptake of additional AS therapies after 90 days of treatment with an index bDMARD along with continued use of the index bDMARD [12]. All other AS-related treatments that the patient received during first 90 days after initiating index bDMARD were considered as concomitant medications to the index bDMARD. There was no limit to the time window for switching.

HCRU and costs

Health care resource utilization and costs were reported overall and for the subset of patients who initiated a bDMARD anytime during the study follow-up. All-cause resource utilization and costs were stratified by the service sector (i.e. inpatient visits, hospital outpatient encounters, hospital outpatient pharmacy) in which they occurred based on the inpatient/outpatient designation and claim categorization codes associated with each claim. The subset of HCRU and costs specific to AS-related hospitalizations and AS-related outpatient pharmacy dispenses, identified as claims associated with an AS diagnosis or related-treatment, were reported separately. AS-related hospitalizations were defined as hospital confinements in which AS was recorded as a confirmed diagnosis and linked with diagnosis and medical treatment data. All cost data were updated to 2018 Japanese yen using the Consumer Price Index for Japan [14], then converted to United States (US) dollar using the 2018 mid exchange rate of 1 US dollar = 110.8 Japanese yen [15].

Statistical analysis

All analyses were conducted using SAS version 9.4 (SAS Institute Inc., Cary, North Carolina). Descriptive analyses displayed mean values, medians, ranges, and standard deviations (SDs) of continuous variables and frequency distributions for categorical variables. Treatment discontinuation was assessed using the Kaplan-Meier method. The censored group included those who had treatment ongoing at the end of the study follow-up, and the event group were those with a discontinuation. The association between patient demographics, baseline comorbidities of interest, baseline medications received, and the subsequent initiation of a bDMARD in the 12 months after diagnosis of AS was assessed using a multivariable logistic regression model. The selection of covariates was based on clinical importance and a literature review [16,17].

Results

Patient population

A total of 417 patients met inclusion criteria and were included in the overall analysis for patients with an AS diagnosis (Figure 1). Of these patients, 115 (27.6%) had initiated bDMARDs during the entire study follow-up period and were included in a subanalysis population. Demographics and patient characteristics can be found in Table 1. Patients had a mean (SD) age of 60.6 (16.4) years and tended to be male (60.2%). All patients had a minimum of 12 months’ follow-up, with follow-up duration ranging from 12.1 to 110.7 months and a mean (SD) of 33.0 (18.7) months. Most patients (82.3%) received at least one imaging procedure during the baseline period. Patients had a mean (SD) CCI score of 1.3 (1.7). Gastroesophageal reflux disease (GERD) (33.8%), low back pain (28.3%), rheumatoid arthritis (28.1%), and hypertensive diseases (24.5%) were the most prevalent comorbidities among all patients diagnosed with AS (Table 2). The majority of patients diagnosed with AS had an index diagnosis attributable to a physician department of orthopedics (54.7%), followed by internal medicine (20.6%), rheumatology (13.2%), other specialties (9.8%), and dermatology (1.7%). The subset of 115 patients who initiated bDMARDs was younger, with a mean (SD) age of 51.6 (14.3) years, and was more evenly split between sexes (52.2% male). The categorization of physician departments associated with the diagnosis had less variance, with 30.4% of diagnoses attributable to orthopedics, followed by internal medicine (29.6%), rheumatology (27.0%), other specialties (8.7%), and dermatology (4.4%).

Figure 1. Sample selection flow chart. AS: ankylosing spondylitis; DMARD: disease-modifying antirheumatic drug.

Table 1. Patient demographics and clinical characteristics.

Characteristic	All patients with AS diagnosis (n = 417)	Patients with AS diagnosis who initiated biologic DMARDs^b (n = 115)
All patients, n (%)	417 (100.0)	115 (27.6)
Age at index date, mean (SD)	60.6 (16.4)	51.6 (14.3)
Male sex, n (%)	251 (60.2)	60 (52.2)
Follow-up time
Mean (SD), months	33.0 (18.7)	29.9 (15.2)
At least 18 months, n (%)	331 (79.4)	91 (79.1)
At least 24 months, n (%)	269 (64.5)	67 (58.3)
CCI score, mean (SD)	1.3 (1.7)	1.3 (1.2)
Imaging procedures during baseline period and the month of AS diagnosis
Patients with at least one imaging procedure, n (%)	343 (82.3)	82 (71.3)
Number of imaging procedures, mean (SD)	3.2 (3.0)	3.1 (2.2)
Physician department for index AS diagnosis, n (%)^a
Dermatology	7 (1.7)	5 (4.4)
Internal medicine (general physician)	86 (20.6)	34 (29.6)
Rheumatology	55 (13.2)	31 (27.0)
Orthopedic	228 (54.7)	35 (30.4)
Other specialties	41 (9.8)	10 (8.7)

AS: ankylosing spondylitis; CCI: Charlson Comorbidity Index; DMARD: disease-modifying antirheumatic drug; SD: standard deviation.

^a Characterized based on visits during the month of initial AS diagnosis.

^b Initiated biologic DMARDs during the entire follow-up period.

Table 2. Comorbidity burden using the charlson comorbidity index and other comorbidities related to AS.

	All Patients With AS Diagnosis (n = 417)		Patients With AS Diagnosis Who Initiated Biologic DMARDs (n = 115)^a
	n	%	n	%
Comorbidities per Charlson Comorbidity Index
Rheumatologic disease	119	28.5	62	53.9
Peptic ulcer disease	55	13.2	15	13.0
Liver disease – mild	48	11.5	14	12.2
Chronic pulmonary disease	43	10.3	11	9.6
Congestive heart failure	40	9.6	6	5.2
Any malignancy	35	8.4	7	6.1
Cerebrovascular disease	31	7.4	5	4.4
Diabetes – without chronic complications	19	4.6	6	5.2
Peripheral artery disease	15	3.6	3	2.6
Diabetes – with chronic complications	15	3.6	3	2.6
Renal disease	14	3.4	2	1.7
Myocardial infarction	9	2.2	2	1.7
Dementia	9	2.2	0	0.0
Metastatic carcinoma	6	1.4	0	0.0
Paraplegia and hemiplegia	3	0.7	0	0.0
Other comorbidities related to ankylosing spondylitis
GERD	141	33.8	57	49.6
Low back pain	118	28.3	28	24.4
Rheumatoid arthritis	117	28.1	62	53.9
Hypertensive diseases	102	24.5	17	14.8
Constipation	76	18.2	12	10.4
Osteoporosis	72	17.3	29	25.2
Hyperlipidemia	69	16.6	18	15.7
Chronic gastritis	58	13.9	19	16.5
Osteoarthritis	44	10.6	12	10.4
Ischemic heart diseases	38	9.1	4	3.5
Neck pain	23	5.5	5	4.4
Depression	18	4.3	7	6.1
Ulcerative colitis	18	4.3	9	7.8
Anxiety	14	3.4	5	4.4
Chronic renal disease	12	2.9	2	1.7
Psoriatic arthritis	12	2.9	6	5.2
Psoriasis (other than psoriatic arthritis)	11	2.6	8	7.0
Uveitis	11	2.6	5	4.4
Enthesitis	9	2.2	6	5.2
Dactylitis	6	1.4	4	3.5
Crohn's disease	4	1.0	4	3.5

AS: ankylosing spondylitis; DMARD: disease-modifying antirheumatic drug; GERD: gastroesophageal reflux disease.

^a Initiated biologic DMARDs during the entire follow-up period.

Treatment patterns

Most patients (88.7%) received treatment for AS during the 12-month follow-up period, with the initiation of the first treatment occurring at a mean (SD) of 29.7 (55.1) days from the index date (Table 3). First treatments administered to patients during the 12-month follow-up period included oral NSAIDs (51.6%), topical NSAIDs (30.0%), corticosteroids (24.7%), methotrexate (14.2%), sulfasalazine (10.3%), adalimumab (5.8%), and infliximab (4.6%). Multiple combinations of first treatments administered were observed (first-line combinations), including NSAIDs only (36.2%), corticosteroid + NSAIDs (9.8%), NSAIDs + csDMARD (7.7%), corticosteroid + NSAIDs + csDMARD (7.4%), and corticosteroid only (6.5%) (Figure 2). Over the course of the 12-month follow-up period, patients received NSAIDs (79.6%), corticosteroids (39.3%), csDMARDs (33.8%), bDMARDs (25.2%), physical therapy (13.7%), and AS-related surgery (5.3%) (Table 3). Methotrexate and sulfasalazine comprised the majority of csDMARD treatments, which were used by 22.3% and 16.8% of patients, respectively.

Figure 2. Treatment sequencing following the top five first-line combinations for all patients with ankylosing spondylitis diagnosis. AS: ankylosing spondylitis; cs: conventional synthetic; DMARD: disease-modifying antirheumatic drug; NSAID: nonsteroidal anti-inflammatory drug. Notes. Forty-seven patients (11.3%) had no observed treatment during available follow-up, and 21.1% had other treatment sequences not shown in the top 5 treatments. ‘Other treatments’ comprises second-line treatments received by fewer than five patients.

Table 3. Overall treatment categories during the 12-Month follow-up period^a.

Characteristic	All patients with AS Diagnosis	Patients with AS diagnosis who initiated biologic DMARDs at any time post-index
Any treatment received, n (%)	370 (88.7)	115 (100.0)
Days from index date to initiation of first observed treatment, mean (SD)	29.7 (55.1)	20.0 (34.3)
NSAIDS, n (%)	332 (79.6)	97 (84.4)
Days from index date to first observed treatment, mean (SD)	40.0 (67.5)	39.9 (66.2)
csDMARDs, n (%)	141 (33.8)	76 (66.1)
Days from index date to first observed treatment, mean (SD)	44.7 (73.5)	44.0 (68.6)
Methotrexate^a, n (%)	93 (22.3)	58 (50.4)
Sulfasalazine^a, n (%)	70 (16.8)	32 (27.8)
Biologic DMARDs, n (%)	105 (25.2)	105 (91.3)
Days from index date to first observed treatment, mean (SD)	59.3 (81.4)	59.3 (81.4)
Adalimumab^a, n (%)	59 (14.2)	59 (51.3)
Infliximab^a, n (%)	51 (12.2)	51 (44.4)
Corticosteroids, n (%)	164 (39.3)	60 (52.2)
Days from index date to first observed treatment, mean (SD)	53.9 (84.5)	60.2 (92.0)
Physical therapy, n (%)	57 (13.7)	10 (8.7)
Days from index date to first observed physical therapy, mean (SD)	57.5 (84.3)	109.0 (120.1)
AS-related surgery, n (%)	22 (5.3)	4 (3.5)
Days from index date to first observed surgery, mean (SD)	83.2 (94.7)	244.0 (90.5)
Opioids, n (%)	16 (3.8)	6 (5.2)
Antidepressants, n (%)	30 (7.2)	13 (11.3)
Anxiolytics, n (%)	136 (32.6)	35 (30.4)
Folic acid, n (%)	77 (18.5)	45 (39.1)

AS: ankylosing spondylitis; cs: conventional synthetic; DMARD: disease-modifying antirheumatic drug; NSAID: nonsteroidal anti-inflammatory drug; SD: standard deviation.

^a Only showing agents received by more than 5% of patients.

Adalimumab and/or infliximab were the two most commonly used bDMARDs and were used by 14.2% and 12.2% of all AS patients, respectively (Table 3). The majority of patients receiving bDMARDs remained on treatment, with persistence rates of 71.1% and 61.0% observed at 12 and 24 months’ follow-up, respectively (Table 4). Persistence rates were observed to be similar for adalimumab, infliximab, and overall bDMARDs at multiple time points through 60 months (Figure 3). The proportion of patients adherent (MPR ≥ 0.8) to the index bDMARD declined from 83.2% at 12 months to 64.9% at 24 months. A relatively greater proportion of patients were adherent to adalimumab as compared with infliximab (Table 4). Among the 41 patients (35.7%) who discontinued bDMARD treatment, 31 (75.6%) switched to a different treatment or treatment combination. Of the patients treated with bDMARDs, 102 (88.7%) exhibited use of AS-related concomitant medication, which included oral NSAID use in 64 patients (62.8%), methotrexate use in 50 patients (49.0%), corticosteroid use in 48 patients (47.1%), topical NSAIDs use in 29 patients (28.4%), and sulfasalazine use in 27 patients (26.5%) (Supplemental Material, Table S-5). The logistic regression analysis (Figure 4) demonstrated that age at index of 45–59 years (odds ratio [OR] = 0.335), age at index of 60+ years (OR = 0.286), a GERD baseline comorbidity (OR = 2.187), and csDMARD baseline treatment (OR = 3.071) were significant factors associated with receipt of bDMARDs in the 12 months following diagnosis.

Figure 3. Persistence rate estimated through Kaplan-Meier analysis. SE: standard error.

Figure 4. Logistic regression analysis to assess factors associated with receipt of biologic DMARDs in the 12 months After Diagnosis of AS. AS: ankylosing spondylitis; CCI: Charlson Comorbidity Index; CI: confidence interval; DMARD: disease-modifying antirheumatic drug; NSAID: nonsteroidal anti-inflammatory drug; ref.: reference.

Table 4. Treatment patterns during all available follow-up among patients receiving biologic DMARDs.

Characteristic	Patients With AS Diagnosis and Biologic DMARD^b	Patients With AS Diagnosis and Adalimumab^b	Patients With AS Diagnosis and Infliximab^b
All patients receiving treatment, n (%)	115 (100.0)	59 (51.3)	50 (43.5)
Patients with treatment discontinuation, n (%)	41 (35.7)	19 (32.2)	18 (36.0)
Permanent discontinuation	30 (73.2)	14 (73.7)	13 (72.2)
Temporary discontinuation (i.e. subsequent reuptake)	11 (26.8)	5 (26.3)	5 (27.8)
Persistence rate, 12 months, % (SE)	71.1 (4.3)	71.4 (6.1)	73.8 (6.3)
Persistence rate, 24 months, % (SE)	61.0 (5.0)	63.2 (7.1)	64.0 (7.1)
MPR, n (%)	112 (97.4)	56 (94.9)	50 (100.0)
Mean, SD	0.9 (0.1)	1.0 (0.1)	0.8 (0.1)
MPR ≥ 0.8, n (%)	89 (79.5)	53 (94.6)	30 (60.0)
MPR at 12 months, n (%)	101 (90.2)	50 (89.3)	46 (92.0)
Mean, SD	0.9 (0.1)	1.0 (0.1)	0.8 (0.1)
MPR ≥ 0.8, n (%)	84 (83.2)	48 (96.0)	31 (67.4)
MPR at 24 months, n (%)	57 (50.9)	28 (50.0)	28 (56.0)
Mean, SD	0.9 (0.1)	0.9 (0.1)	0.8 (0.1)
MPR ≥ 0.8, n (%)	37 (64.9)	25 (89.3)	11 (39.3)
Patients with treatment switching (among patients with discontinuation), n (%)	31 (75.6)	12 (63.2)	15 (83.3)
Methotrexate	7 (22.6)	3 (25.0)	3 (20.0)
Methotrexate + Sulfasalazine	5 (16.1)	3 (25.0)	2 (13.3)
Infliximab + Methotrexate	3 (9.7)	3 (25.0)	0 (0.0)
Adalimumab + Methotrexate	2 (6.5)	0 (0.0)	2 (13.3)
Sulfasalazine	2 (6.5)	1 (8.3)	1 (6.7)
Others where n = 1	12 (38.8)	2 (16.7)	7 (46.7)
Other AS-related treatment augmentation (among all patients who received treatment)^a, n (%)	45 (39.1)	20 (33.9)	22 (44.0)
NSAIDS – oral	21 (46.7)	8 (40.0)	12 (54.6)
NSAIDS – topical	16 (35.6)	7 (35.0)	7 (31.8)
Corticosteroids	9 (20.0)	4 (20.0)	4 (18.2)
Methotrexate	9 (20.0)	4 (20.0)	5 (22.7)
Sulfasalazine	3 (6.7)	1 (5.0)	1 (4.6)

AS: ankylosing spondylitis; DMARD: disease-modifying antirheumatic drug; MPR: medication possession ratio; PDC: proportion of days covered; NSAID: nonsteroidal anti-inflammatory drug; SD: standard deviation; SE: standard error.

^a Only showing agents received by more than 5% of patients.

^b Initiated biologic DMARDs during the entire follow-up period.

HCRU

All-cause HCRU and costs were explored for all patients with an AS diagnosis and a smaller subset of patients with AS who initiated bDMARDs (Table 5). The overall AS patient group had lower median annual per-patient costs compared with patients who initiated bDMARDs $15,323 vs. $26,937). Both groups had similar percentages of patients who experienced at least one hospital admission and who had at least one hospital outpatient encounter. The majority of the patients (95.4%) in the overall AS group had at least one prescription filled at an outpatient pharmacy. Notably, the average patient who initiated bDMARDs spent fewer days in the hospital per stay (median, 7.0 vs. 11.0 days), and median hospitalization costs per patient were lower by approximately $10,000 ($29,817 vs. $39,509). Similar trends were observed for specific HCRU and costs related to AS (Table 5). Patients who were hospitalized and who initiated bDMARDs, spent fewer days receiving AS-related care as admitted hospital patients than did the overall group of patients diagnosed with AS (median, 8 days vs. 18 days, respectively) and had lower annual AS-related median hospitalization costs of $28,926 per patient compared with $43,222.

Table 5. All-Cause and Other specific health care resource use and costs during anytime in the Follow-Up period.

Characteristic	All patients with AS Diagnosis (n = 417)	Patients with AS diagnosis who initiated biologic DMARDs^b (n = 115)
All-cause health care resource use and costs^a
Total per-patient health care use and costs
Number of encounters per year, median	10.8	13.8
Annual per-patient costs, median (USD)	$15,323	$26,937
Hospitalizations
≥ 1 hospital admission, n (%)	200 (48.0)	55 (47.8)
Admissions per year, median	0.7	0.7
Days of hospital care per stay, median	11.0	7.0
Annual per-patient costs, median (USD)	$39,509	$29,817
Hospital outpatient encounters
≥ 1 outpatient encounters, n (%)	416 (99.8)	115 (100.0)
Outpatient encounters per year, median	10.4	13.8
Annual per-patient costs, median (USD)	$1,078	$9,011
Hospital outpatient pharmacy
≥ 1 prescriptions filled, n (%)	398 (95.4)	115 (100.0)
Number of prescriptions per year, median	31.5	46.0
Annual per-patient costs, median (USD)	$1,196	$4,142
Specific health care resource use and costs^a
AS-related hospitalizations
≥ 1 hospital admission, n (%)	86 (20.6)	35 (30.4)
Hospital admissions per year, median	0.5	0.5
Days of hospital care per stay, median	18.0	8.0
Annual per-patient costs, median (USD)	$43,222	$28,926
AS-related outpatient pharmacy
≥ 1 prescriptions filled, n (%)	358 (85.9)	113 (98.3)
Number of prescriptions per year, median	8.7	16.9
Annual per-patient costs, median (USD)	$254	$1,459

AS: ankylosing spondylitis; DMARDs: disease-modifying antirheumatic drugs; SD: standard deviation; USD: US dollars.

^a All costs are presented in 2018 USD. JPY were converted to USD using the 2018 mid exchange rate of 1 USD = 110.76 JPY.

^b Initiated biologic DMARDs during the entire follow-up period.

Discussion

The results of this database study describe the current, real-world state of diagnosis and treatment patterns for AS in Japan. Patients had a mean age of 61 years and were mostly male (60.2%). Gastroesophageal reflux disease, low back pain, and rheumatoid arthritis were the most prevalent comorbidities among all patients diagnosed with AS. The majority of patients diagnosed with AS had an index diagnosis attributable to a physician department of orthopedics. Nonsteroidal anti-inflammatory drugs were a frequently prescribed first-line treatment, either alone or in combination with other treatments, which is consistent with the literature [12]. The prevalent NSAID use, one of the causes of peptic ulcers [18], may explain the presence of GERD as a common baseline comorbidity observed in one-third of the population (33.8%). The GERD code could enable physicians to prescribe proton pump–inhibiting medication for a longer period of time to prevent peptic ulcer formation in patients frequently taking NSAIDs. Notably, a GERD comorbidity at baseline was associated with a significantly increased likelihood of receiving bDMARDs in the 12 months following diagnosis of AS (OR = 2.187). Over the course of the 12-month follow-up period, 25% of all patients received bDMARDs. Many of these patients used bDMARDs as the first-line treatment (14.2% for adalimumab and 12.2% for infliximab). Patients who initiated treatment with bDMARDs had higher median total per-patient health care costs, lower median per-patient hospitalization costs, and fewer median hospital days compared with the overall group of patients diagnosed with AS.

There were several inconsistencies between the patient characteristics and treatment patterns observed in this study and the literature. The mean age of patients in this study was substantially greater than previously reported for populations in Japan, other Asian countries, and non-Asian countries [2]. The greater than expected mean age could be caused by misdiagnosis in patients with diseases other than AS, such as diffuse idiopathic skeletal hyperostosis (DISH). Both DISH and AS cause bone proliferation along the spine that can result in confusion between the two conditions [19]. In further support of the greater age observed in this study, DISH patients tend to be older; Hiyama et al. [20] reported a mean age of 72 years in a study that explored DISH prevalence in 1,479 Japanese patients aged at least 20 years. There was an unusually high percentage (28.1%) of patients who exhibited rheumatoid arthritis as a comorbidity, which does not reflect the literature [21]. The high proportion of records for diagnosis of both rheumatoid arthritis and AS in patients could plausibly be due to the process for establishing a confirmed AS diagnosis in routine clinical practice and the resulting documentation in the MDV database. This process could include ordering tests to rule out rheumatoid arthritis as a possible diagnosis as well as documenting subsequent findings to utilize the subsidy offered by the MHLW in Japan for rare and intractable conditions like AS in Japan [22,23]. Finally, we observed that about 40% of patients received corticosteroids, which is inconsistent with the recommendation of the Assessment of SpondyloArthritis international Society and European League Against Rheumatism 2016 guidelines discouraging long-term glucocorticoid use [24]. The frequent use of corticosteroids in this population may be from the data set reflecting a combination of past [25] and current practice in Japan due to the early start date and a wide range for the study window (April 2008 through July 2018). Taken together, these unexpected observations for patient demographics, comorbidities, and treatment modalities highlight the challenges of diagnosing AS and providing appropriate treatment in Japan. Currently, guidance on initial diagnosis of and treatment for spondyloarthritis, including AS, in Japan is being prepared by the MHLW spondyloarthritis research group to address these issues [26].

There are several limitations associated with this study. The MDV database includes information from only participating facilities, so findings from this study may not be generalizable to all patients with AS in Japan. The study required at least 3 months of AS-free status before the first AS diagnosis; however, patients may have received a diagnosis of or been treated for AS at a nonparticipating facility and therefore could be misclassified as having newly received an AS diagnosis. While longer durations of minimum baseline and follow-up periods would help assess the study objectives adequately, it would limit the sample sizes for this study considerably, and therefore the primary analysis was conducted using a 3-month baseline period and a 12-month follow-up period. Additionally, participating hospitals did not link individual patient data, so patients who received care at multiple hospitals would appear as multiple unique individuals. These database limitations, in combination with the tendency of some patients with AS to seek a diagnosis at several clinics and hospitals, may partially explain the discrepancy in demographics between populations found in the literature and our population, including the older mean age of patients in our study compared with AS populations in other studies. An additional limitation was that infliximab and adalimumab were approved for AS indication in Japan in 2010, which occurred after the start of the study selection window of April 2009. While actual adherence can be difficult to analyze using claims data, we were able to estimate adherence using MPR [12]. The definition of AS-related hospitalization was intended for use in an exploratory analysis and had not been validated. Lastly, information on laboratory test results, radiographical information, HLA-B27 status, and severity of AS was not available for this study; therefore, assessments of background characteristics such as CCI comorbidities relied solely on ICD-10 coding, which could have resulted in an underestimation of the prevalence of these conditions. Similarly, the definition of AS in the claims data in Japan has not been validated, so the reliance on ICD-10 coding to define AS was another limitation.

Conclusion

To our knowledge, this is the first database study to provide knowledge of patient characteristics, treatment patterns, HCRU, and costs for patients with AS in Japan. Patients tended to be older than AS populations in other studies and had a high rate of GERD and rheumatoid arthritis comorbidities. While most patients received NSAID and bDMARD treatments recommended by guidelines, a considerable proportion of patients received corticosteroids. The overall AS patient group had lower annual per-patient costs compared with patients who initiated bDMARDs; however, the average patient who initiated bDMARDs spent fewer days in the hospital per stay and had lower hospitalization costs. The study outcomes demonstrate a need for increased awareness of AS and improved care of AS consistent with established guidelines.

Acknowledgments

The authors thank Brian Samsell of RTI Health Solutions for medical writing assistance.

Conflict of interest

MS, KH, and KN are Eli Lilly Japan K. K. employees and hold stock in Eli Lilly Japan K. K. TT is a paid consultant for Eli Lilly Japan K. K. but did not receive compensation for participation in this study. EE and RP are employees of RTI Health Solutions, an independent nonprofit research organization that does work for government agencies and pharmaceutical companies.

Additional information

Funding

The present research was financially supported by Eli Lilly Japan K. K.