Rheumatol Ther https://doi.org/10.1007/s40744-018-0114-6 ORIGINAL RESEARCH Treatment Patterns of Newly Diagnosed Rheumatoid Arthritis Patients from a Commercially Insured Population . . . David M. Kern Lawrence Chang Kalyani Sonawane . . . Cynthia J. Larmore Natalie N. Boytsov Ralph A. Quimbo . . . Joseph Singer John T. Hinton Sze-jung Wu Andre B. Araujo Received: February 27, 2018 The Author(s) 2018 DMARD treatment, including conventional ABSTRACT synthetic DMARDs (csDMARD), tumor necrosis factor inhibitors (TNFi), non-TNFi, and Janus Introduction: To describe treatment patterns in kinase inhibitors (JAKi), were captured during newly diagnosed rheumatoid arthritis (RA) follow-up. patients in a large, nationally representative Results: Of the 63,101 RA patients identiﬁed, managed-care database. 73% were female; mean age was 57 years. Dur- Methods: Newly diagnosed RA patients were ing an average of 3.5 ± 2.1 years of follow-up, identiﬁed from 07/01/2006–08/31/2014. 45% of patients never received a DMARD, 52% Patients had C 1 RA diagnosis by a rheumatol- received a csDMARD (94 ± 298 mean ± SD days ogist, or C 2 non-rheumatologist RA diagnoses from index), 16% a TNFi (315 ± 448 days), 4% a C 30 days apart, or RA diagnosis followed by a non-TNFi (757 ± 660 days), and \ 1% a JAKi. disease-modifying antirheumatic drug Among DMARD recipients, the most common (DMARD) prescription ﬁll within 1 year. treatment patterns were: receiving csDMARDs Patients were C 18 years old at index (earliest only (68%), adding a TNFi as second-line ther- date fulﬁlling diagnostic criteria) and had C 6 apy after initiation of a csDMARD (12%), and and 12 months of pre- and post-index health receiving only a TNFi (6%) during follow-up. plan enrollment, respectively. Patterns of Among those not on DMARDs, the all-cause usage of an opioid was 56% and 19% had chronic opioid use (C 180 days supplied). Conclusions: Despite American College of Enhanced digital features To view enhanced digital Rheumatology recommendations for DMARD features for this article go to https://doi.org/10.6084/ treatment of RA, nearly half of newly diagnosed m9.ﬁgshare.6205478. RA patients received no DMARD therapy during follow-up. These data identify a treatment gap D. M. Kern K. Sonawane R. A. Quimbo J. Singer in RA management. S. Wu Funding: Eli Lilly & Company. HealthCore, Inc, Wilmington, DE, USA L. Chang C. J. Larmore N. N. Boytsov Keywords: Arthritis; Antirheumatic agents; A. B. Araujo (&) Eli Lilly and Company, Indianapolis, IN, USA Guideline adherence; Medication adherence; e-mail: email@example.com Practice patterns; Physicians’ rheumatoid J. T. Hinton Anthem, Inc, Indianapolis, IN, USA Rheumatol Ther Attainment of remission or low disease activity INTRODUCTION remains challenging for many patients . The management of RA has implications on the Rheumatoid arthritis (RA) affects an estimated overall cost and quality of care for patients, 1.3 million adults in the United States (US), is providers, and payers [6, 10, 17–19]. twice as prevalent in women as in men, and is The availability of recent population-based associated with a range of signs and symptoms estimates of RA treatment patterns is limited. that include pain, fatigue, and joint stiffness Such information would be of interest to parties [1–3]. Because of its debilitating symptoms, RA involved in the management of this disease. strains the resources of healthcare systems, This study aimed to address this knowledge gap exacts severe economic burdens on patients and by examining treatment patterns within a their families, and diminishes quality of life population of newly diagnosed RA patients along with patients’ ability to work and function using a large, nationally representative man- normally [4, 5]. Inﬂammation caused by the aged-care database. disease can lead to cartilage damage and destruction of the joints . RA patients may have multiple comorbidities and psychosocial METHODS impairments including cardiovascular disease, osteoporosis, interstitial lung disease, infections, Study Design and Data Source malignancies, fatigue, depression, and cognitive dysfunction, among others . Thus, it is vital This observational cohort study queried medical that RA disease activity is controlled to lessen and prescription claims from the HealthCore the disease-related burden on patients’ personal SM Integrated Research Database (HIRD) ,a lives and the healthcare system as a whole. repository of longitudinal claims data for The American College of Rheumatology approximately 40 million health plan enrollees (ACR) RA treatment recommendations specify across the US. The study population of interest the use of conventional, synthetic disease- was newly diagnosed RA patients. The study modifying antirheumatic drugs (csDMARDs) as cohort consisted of patients identiﬁed during ﬁrst-line therapy for newly diagnosed RA the period from July 1, 2006 through August 31, patients to minimize or prevent progressive 2014 based on a combination of claims-based damage to joints and improve long-term out- algorithm [20, 21]. The index date was deﬁned comes [8, 9]. Currently, methotrexate, hydrox- as the earliest date on which a patient fulﬁlled ychloroquine, sulfasalazine, or other csDMARDs one of the three algorithm criteria. Study data are well established as ﬁrst-line DMARD treat- were handled in compliance with all applicable ments for RA; however, multiple therapeutic Health Insurance Portability and Accountability options are now available beyond csDMARDs Act (HIPAA) rules. Patient anonymity and con- for patients who fail to achieve their treatment ﬁdentiality were preserved with data de-identi- goal on these agents [6, 8, 10], including bio- ﬁcation throughout. Investigational Review logic therapies and a Janus kinase inhibitor Board (IRB) informed consent requirements (JAKi) [11–13]. were waived for this non-experimental study, Despite the increasing availability of thera- which was conducted under the research peutic options for RA management, not all exception provisions of the Privacy Rule, 45 patients receive recommended therapy. Physi- CFR 164.514(e). cian-perceived barriers to therapy include cost to patient and insurance/formulary coverage Inclusion/Exclusion Criteria . Patient factors that may inﬂuence therapy choice include the fear of side effects and a belief in adequate disease control despite many Study patients were C 18 years on their index patients exhibiting discordant moderate or date and had continuous enrollment with both higher RA disease activity scores . medical and pharmacy beneﬁts during the Rheumatol Ther 6-month period preceding the index date and Add-on therapy included the ﬁlling of a second for C 1 year post-index. To be considered newly DMARD class while maintaining continuous diagnosed patients, no RA diagnoses were allowed therapy with a previously prescribed DMARD. A during the 6-month pre-index period. Patients maximum of three treatment changes (adding meeting any one of three claims-based criteria on or switching therapy) were captured for each were included: (1) presence of C 1 medical claim patient, resulting in the capture of up to four for an RA diagnosis (ICD-9 714.0x, 714.1x, or lines of DMARD treatment. Medication dosage 714.2x) by a rheumatologist; (2) C 2 medical was not captured in this study. claims with an RA diagnosis by a non-rheuma- Two sensitivity analyses were conducted: (1) tologist occurring at least 30 days apart; or, (3) limiting the analysis to patients identiﬁed during presence of C 1 medical claim with an RA diag- January 1, 2010 through August 31, 2014 to nosis occurring concurrently with or followed by account for the potential evolution in therapeu- a medical or pharmacy claim for any DMARD tic management and treatment recommenda- therapy within 1 year. Patients were followed tions in recent years, and (2) requiring patients to beyond 1 year until they reached the end of the have three distinct RA diagnosis on distinct dates study period, left the health plan, or died, when the diagnostic workups or tests (e.g., whichever occurred ﬁrst. Patients with C 1 claim imaging of joint, laboratory test for C-reactive of a cancer diagnosis (ICD-9 code 140.xx–209.3x, protein, synovial ﬂuid analysis rheumatoid fac- 230.xx–234.xx) during the 6-month pre-index tor, cyclic citrullinated peptide, erythrocyte period were excluded from the study. sedimentation rate) used for diagnosing/ruling out RA were not included, AND visited a rheumatologist at time of ﬁrst diagnosis. Treatment Pattern Deﬁnitions Baseline Characteristics Patterns of DMARD treatment—including csDMARDs, biologic tumor necrosis factor inhibitors (TNFi; etanercept, adalimumab, Prespeciﬁed comorbid conditions were deﬁned inﬂiximab, certolizumab, and golimumab), as the presence of C 1 diagnosis code for the biologic non-TNFi (abatacept, rituximab, tocili- condition of interest during the 6-month pre- zumab, anakinra), and JAKi (tofacitinib)—were index period. The Quan–Charlson Comorbidity assessed during follow-up. Additionally, use of Index (QCI) was calculated to quantify the prescription pain medications (opioids, nons- burden of comorbid disease in this population teroidal anti-inﬂammatory drugs [NSAIDs], . The specialty of the physician associated muscle relaxants, and anticonvulsants) and with the RA diagnosis was identiﬁed on the steroids (oral and injectable glucocorticoids) index date medical claim. If physician infor- were examined. All-cause chronic opioid use mation was not available on the index date, the was deﬁned as receiving C 180 days supply of claim closest to the index date (either pre- or opioid medication during the follow-up period. post-index) with a diagnosis for RA was used to Discontinuation of DMARD treatment was determine diagnosing specialty. considered as the absence of a claim for a DMARD within 90 days after the last day sup- Statistical Analysis plied of the most recent medication claim. The discontinuation date was deﬁned as the last This single-cohort study of RA patients was medication claim date plus the days supplied descriptive in nature, and treatment patterns for that claim. Switching was deﬁned as the were analyzed within the entire cohort with no ﬁlling of a DMARD class that was different from comparison group. Descriptive statistics (means, any medication in the current regimen or most standard deviations, medians, frequencies, and recent prior medication while not continuing percentages) were reported for each outcome of the original therapy; stopping and restarting the interest. All analyses were conducted using SAS same medication was not considered a switch. Enterprise Guide v7.1 (Cary, NC, USA). Rheumatol Ther Fig. 1 Patient selection for the study and the number of patients included in the ﬁnal analytic cohort. *By requiring 6 months pre-index criteria the identiﬁcation period of patients effectively becomes 7/1/2006 through 8/31/2014 were female (Table 1). Most patients were RESULTS enrolled in preferred provider organization (PPO) health plans (72.5%), and 27.8% had Demographics and Baseline Data Medicare Advantage or Medicare Supplemental plus Part D coverage. Nearly half of the study After applying inclusion and exclusion criteria population (44.2%) was diagnosed or treated by (Fig. 1), there were 63,101 RA patients with a rheumatologist at index. Hypertension mean ± SD age of 57 ± 15.6 years, and 72.7% (40.5%), dyslipidemia (34.0%), and Rheumatol Ther Table 1 Baseline patient characteristics (N = 63,101) Table 1 continued Characteristic Mean–SD or Characteristic Mean–SD or n (%) n (%) Female 45,854 (72.7) Respiratory conditions 8790 (13.9) Age at index (years) 57.0±15.6 Fibromyalgia 8768 (13.9) Geographic region Hypothyroidism 8493 (13.5) Northeast 12,181 (19.3) Inﬂammatory polyarthritis 6385 (10.1) Midwest 18,111 (28.7) SD standard deviation, HMO health maintenance orga- nization, PPO preferred provider organization, CDHP South 16,580 (26.3) consumer driven health plan, QCI Quan-Charlson index West 13,467 (21.3) Comorbid conditions counted if a single occurrence of the code appeared, listed are conditions that affected at Unknown 2,762 (4.4) least 10% of the RA population. Comorbid autoimmune Health plan type conditions included lupus erythematous (9.8%), dermatitis (6.2%), psoriatic arthritis (4.1%), psoriasis (3.2%), anky- HMO 14,002 (22.2) losing spondylitis (1.4%), Crohn’s disease (1.3%), and PPO 45,775 (72.5) ulcerative colitis (1.1%) CDHP 3324 (5.3) osteoarthritis (31.1%) were among the most Medicare Advantage, Supplemental 17,543 (27.8) commonly observed comorbid conditions. and Part D Treating/prescribing physician specialty Treatment Patterns Rheumatologist 27,922 (44.2) Follow-up and Time to Treatments Internal medicine 7171 (11.4) Patients were followed for a mean of Family medicine/general practice 6429 (10.2) 3.5 ± 2.1 years after their index date, and 50% of patients were followed for at least 3 years Unknown 2687 (4.3) (Table 2). The mean time from index date to All others 18,892 (29.9) ﬁrst use of each DMARD treatment class was 94 ± 298 days for csDMARD (n = 32,518), QCI 0.9±1.3 315 ± 448 days for biologic TNFi (n = 10,360), QCI categories 757 ± 660 days for non-TNFi (n = 2,228), and 0 29,870 (47.3) 1,254 ± 800 days for JAKi users (n = 274). For patients who received a prescription pain med- 1–2 26,829 (42.5) ication, the mean time to treatment with pain [3 6402 (10.1) medication was 194 ± 346 days. Comorbid conditions DMARD Medication Use Hypertension 25,525 (40.5) Treatments were examined by DMARD class Dyslipidemia 21,464 (34.0) through 90 days from index date and through the entire post-index period (Fig. 2). Many Osteoarthritis 19,642 (31.1) patients (53.8%) received no DMARD therapy Mental illness 12,903 (20.4) within 90 days after the index date; 44.9% of patients never received DMARD therapy during Type 2 diabetes 9075 (14.4) the entire follow-up period. However, examin- Unspeciﬁc arthritis 8919 (14.1) ing rates by medical specialty, 78.9% of patients Rheumatol Ther Table 2 Follow-up time and time to treatment among Table 2 continued patients newly diagnosed with rheumatoid arthritis Mean–SD/n (n = 63,101) (%) Mean–SD/n Number of patients with biologic 7395 (65.7) (%) use diagnosed by rheumatologist Total follow-up time from index to end of 3.5±2.1 Total follow-up time (years) 3.9±2.2 follow-up (years) From index to ﬁrst use(days) 328±462 C2 years of follow-up 44,795 (71.0) From ﬁrst use to end of follow-up 1092±760 C3 years of follow-up 31,293 (49.6) (days) C4 years of follow-up 21,699 (34.4) From csDMARD (in those receiving 388±446 C5 years of follow-up 14,657 (23.2) csDMARD) (days) Treatment characteristics Time to treatment (days from index date treated by a rheumatologist received DMARD to ﬁrst ﬁll) therapy at some point during follow-up, com- csDMARD (n = 32,518) 94±298 pared with just 36.3% of those seen by a non- rheumatologist. Within 90 days of index date, Biologic TNFi (n = 10,360) 315±448 csDMARD therapy was the most prevalent Biologic non-TNFi (n = 2228) 757±660 (41.5%), followed by a biologic TNFi (4.5%), JAKi (n = 274) 1254±800 and a non-TNFi (\ 1%) or JAKi (\ 1%). When examining the entire follow-up period, there Time to pain medication (days) 194±346 was a slight increase in csDMARD use (51.5%), (n =49,123) with larger increases observed for biologic TNFi Opioids (n= 41,148) 302±431 (16.4%) and biologic non-TNFi (3.5%) thera- pies. JAKi use remained rare, occurring in \ 1% NSAIDs (n=32,935) 303±458 of the population. COX 2 inhibitors (n = 17,292) 403±550 Acetaminophen (n = 16) 789±984 Detailed Treatment Patterns Detailed treatment patterns in the newly diag- Muscle relaxants(n = 315) 449±495 nosed RA population are depicted in Fig. 3. Anticonvulsants (n = 12,243) 517±619 Among patients who received DMARD treat- ment (n = 34,795), 67.5% received only csDMARD use csDMARD during follow-up. Initiating a Number of patients with csDMARD 32,518 (51.5) csDMARD and adding a TNFi as second-line use therapy was the most common route to receiv- ing biologic therapy (12.4% of treated patients), Number of patients with csDMARD 20,850 (64.1) while treatment with a TNFi without ever use diagnosed by rheumatologist receiving a csDMARD was the next most com- Total follow-up time (years) 3.7±2.1 mon (5.5%). While uncommon (1.2% of all DMARD-treated patients), non-TNFi use was From index to ﬁrst use (days) 94±298 most likely to occur after starting on a From ﬁrst use to end of follow-up (days) 1244±765 csDMARD, adding a TNFi, and then switching to a non-TNFi while remaining on a csDMARD. Biologic use Of patients who did not receive DMARD Number of patients with biologic use 11,250 (17.8) therapy (n = 28,306) during follow-up, many received symptomatic treatment; 56.5% Rheumatol Ther Fig. 2 DMARD medication use by class within 90 days of the index date (blue bars) and at any time during the post-index period (orange bars)(n = 63,101) received opioids, 19.4% had evidence of chronic and 15.9% discontinued. As for the group of opioid use, and more than a third (37.9%) patients who initiated a non-TNFi as ﬁrst-line received steroids. In contrast, among those who therapy (n = 339), 33.0% had no change, 27.7% received DMARD therapy (n = 34,795), 72.3% discontinued, and 39.2% patients either added received opioids, 24.8% were chronic opioid or switched to a new therapy. In non-TNFi ﬁrst- users, and 72.9% received steroids at any time line users moving to a second-line therapy during follow-up. (n = 133), 35.3% added on or switch therapy. Table 3 describes treatment line characteris- Occurrence of a second treatment change tics based on ﬁrst DMARD treatment. The two was relatively infrequent, as 62.2% of those on most common ﬁrst treatment changes were second-line therapy had no additional changes moving from a csDMARD to TNFi therapy either observed. The most common treatment change by adding on the TNFi therapy to the csDMARD was discontinuation of therapy (15.0% of all regimen or switching from csDMARD therapy second-line patients); thus just 22.8% of sec- to TNFi monotherapy. Of 31,217 patients who ond-line patients moved on to a third line started csDMARD as ﬁrst-line therapy, 42.6% (n = 2,078, 6.0% of all treated patients). Three- had no change, 32.8% discontinued, and 24.5% quarters of third-line therapies contained a (n = 7656) either added on or switched to a new csDMARD, while TNFi (32.7%) and non-TNFi DMARD class. Among these 7,656 csDMARD- (47.4%) therapies were also observed. JAKi use initiating patients who had second-line ther- rose above 1%, occurring in 5.7% of third-line apy, 62.2% had no additional therapy change, treatment regimens. 14.5% discontinued, and 23.3% added or swit- ched to the new therapy (i.e., moved to a third DISCUSSION treatment line). In patients who started a TNFi as ﬁrst-line This study provides insight into treatment pat- therapy (n = 3,303), 41.6% had no change in terns among newly diagnosed RA patients in a therapy class, 18.1% discontinued, and 40.3% large managed-care setting and highlights either added or switched DMARD class. In 1,332 important potential gaps in treatment. Treat- TNFi ﬁrst-line users who had second-line ther- ment rates in our study population indicate that apy, 65.5% remained on their second-line a large proportion of patients (nearly half of this therapy, 18.5% switched or added-on therapy, Rheumatol Ther Fig. 3 Treatment patterns in patients with rheumatoid arthritis. a Treatment patterns within rheumatoid arthritis patients who received DMARD therapy (n = 34,795). Medications linked by a ‘?’ symbol indicates the medication classes were received simultaneously. b Non-DMARD treatments in RA patients (n = 63,101) cohort of patients with RA) did not receive one treatment line, but the proportion guideline-recommended DMARD treatment. In remained quite low. However, this DMARD the course of the study, only slightly more than class was introduced into the market as a treat- one-half of the cohort received a csDMARD, less ment option late in this study observation than one-ﬁfth were treated with a biologic TNFi, period. and very few were treated with a biologic non- Prior published studies have reported pat- TNFi or a JAKi. terns of DMARD drug utilization for RA treat- Among patients who received DMARD ther- ment similar to what was observed in our apy, the use of csDMARD mono-class therapy study—csDMARDs were the most commonly was the most commonly initiated ﬁrst-line used treatment, and while fewer patients initi- treatment. Although the use of biologics was ated treatment with biologics as expected not prevalent during the ﬁrst-line treatment, [23, 24], their use increased over time;  fur- the proportion of biologics increased among thermore, while our study showed a surpris- patients who had second, third, and fourth lines ingly low rate of DMARD use, other studies have of treatment. The use of JAKi increased during reported comparable rates of DMARD treatment follow-up among patients who had more than [1, 25–27]. Rheumatol Ther Table 3 Treatment line characteristics based on ﬁrst DMARD treatment in patients newly diagnosed with rheumatoid arthritis and treated with a DMARD therapy (n = 34,795) First DMARD received Overall csDMARD TNFi non-TNFi (n 5 34,795) (n 5 31,217) (n 5 3,303) (n 5 339) Mean–SD/n(%) Mean – SD/n(%) Mean – SD/n(%) Mean – SD/n(%) First line characteristics 34,795 (100.0) 31,217 (100.0) 3303 (100.0) 339 (100.0) Treatment line contains csDMARD 31,217 (89.7) 31,217 (100.0) 56 (1.7) 12 (3.5) TNFi 3303 (9.5) 56 (0.2) 3303 (100.0) 0 (0.0) non-TNFi 339 (1.0) 12 (0.0) 0 (0.0) 339 (100.0) JAKi 4 (\0.1) 0 (0.0) 0 (0.0) 0 (0.0) Length of ﬁrst-line therapy 724 ± 693 730 ± 694 700 ± 694 466 ± 528 (days) First treatment change characteristics No change 14,750 (42.4) 13,313 (42.6) 1373 (41.6) 112 (33.0) Switch therapy 2587 (7.4) 2096 (6.7) 399 (12.1) 91 (26.8) Add medication(s) 6529 (18.8) 5560 (17.8) 933 (28.2) 42 (12.4) Discontinue treatment 10,929 (31.4) 10,248 (32.8) 598 (18.1) 94 (27.7) Second line characteristics 9116 (100.0) 7656 (100.0) 1332 (100.0) 133 (100.0) Treatment line contains csDMARD 6888 (75.6) 5612 (73.3) 1197 (89.9) 84 (63.2) TNFi 7888 (86.5) 6901 (90.1) 936 (70.3) 52 (39.1) non-TNFi 885 (9.7) 704 (9.2) 143 (10.7) 44 (33.1) JAKi 61 (0.7) 56 (0.7) 4 (0.3) 1 (0.8) Length of second-line therapy 748 ± 680 746 ± 672 795 ± 729 447 ± 582 (days) Second treatment change characteristics No change 5673 (62.2) 4764 (62.2) 873 (65.5) 41 (30.8) Switch therapy 877 (9.6) 700 (9.1) 141 (10.6) 36 (27.1) Add medication(s) 1201 (13.2) 1084 (14.2) 106 (8.0) 11 (8.3) Discontinue treatment 1365 (15.0) 1108 (14.5) 212 (15.9) 45 (33.8) Third line characteristics 2078 (100.0) 1784 (100.0) 247 (100.0) 47 (100.0) Treatment line contains csDMARD 1552 (74.7) 1416 (79.4) 121 (49.0) 15 (31.9) TNFi 679 (32.7) 524 (29.4) 138 (55.9) 17 (36.2) Rheumatol Ther Table 3 continued First DMARD received Overall csDMARD TNFi non-TNFi (n 5 34,795) (n 5 31,217) (n 5 3,303) (n 5 339) Mean–SD/n(%) Mean – SD/n(%) Mean – SD/n(%) Mean – SD/n(%) non-TNFi 984 (47.4) 860 (48.2) 100 (40.5) 25 (53.2) JAKi 119 (5.7) 109 (6.1) 7 (2.8) 3 (6.4) Length of third-line therapy 537 ± 546 528 ± 539 606 ± 598 535 ± 486 (days) Third treatment change characteristics No change 1237 (59.5) 1072 (60.1) 137 (55.5) 28 (59.6) Switch therapy 303 (14.6) 251 (14.1) 38 (15.4) 14 (29.8) Add medication(s) 278 (13.4) 229 (12.8) 47 (19.0) 2 (4.3) Discontinue treatment 260 (12.5) 232 (13.0) 25 (10.1) 3 (6.4) Fourth line characteristics 581 (100.0) 480 (100.0) 85 (100.0) 16 (100.0) Treatment line contains csDMARD 359 (61.8) 285 (59.4) 65 (76.5) 9 (56.3) TNFi 242 (41.7) 199 (41.5) 38 (44.7) 5 (31.3) non-TNFi 218 (37.5) 184 (38.3) 30 (35.3) 4 (25.0) JAKi 61 (10.5) 56 (11.7) 4 (4.7) 1 (6.3) Length of fourth-line therapy 447 ± 474 425 ± 454 567 ± 541 460 ± 602 (days) Fourth treatment change characteristics No change 319 (54.9) 262 (54.6) 50 (58.8) 7 (43.8) Switch therapy 116 (20.0) 96 (20.0) 14 (16.5) 6 (37.5) Add medication(s) 278 (47.8) 229 (47.7) 47 (55.3) 2 (12.5) Discontinue treatment 60 (10.3) 49 (10.2) 10 (11.8) 1 (6.3) Data stratiﬁed by ﬁrst line JAKi use not shown due to insufﬁcient sample size for analysis (n = 4) ‘‘Treatment line contains’’: among patients receiving a ﬁrst-, second-, third-, or fourth-line of therapy, the following tabulates the numbers of each DMARD received during the line of therapy. Multiple treatments may be present during any given treatment line Our results showed that RA patients diag- rheumatologist is consistent with data from the nosed by rheumatologists had higher rates of ACR’s RISE (Rheumatology Informatics System DMARDs prescribed (79%) compared with only for Effectiveness) registry of rheumatology slightly more than a third of the patients diag- practices, which showed that 91% of RA nosed by non-rheumatologists. The higher rate patients were on a DMARD at their last clinical of DMARD use among those treated by a encounter . This could help to explain some Rheumatol Ther of the inconsistencies observed in the DMARD While symptomatic management alone is treatment patterns in this study relative to inconsistent with treatment recommendations treatment recommendations, as less than half and quality metrics for RA management, it of the RA patients were diagnosed by a could well be part of a broader trend in which rheumatologist. 3–4% of adult Americans are treated with opi- This difference in DMARD treatment rates by oids for pain control . While clinical studies medical specialty may be driven by three fac-  and surveys [30–32] have reported short- tors: ﬁrst, rheumatologists are likely more term pain relief associated with opioid use, few familiar and up to date with RA treatment rec- studies have investigated longer-term outcomes ommendations and thus more likely to pre- for opioid analgesics for pain therapy . scribe appropriate therapy; second, claims- While pain management is a key aspect to suc- based diagnoses coming from a rheumatologist cessfully treating RA, current data inform of for RA may be more accurate than those coming considerable death and addiction-related risks— from non-rheumatologists in which rule-out more than 165,000 people died from opioid diagnoses may be more common; and third, overdosing related to pain management in the non-rheumatologists may exhibit less comfort US during 1999–2014,  an estimated 1.9 with prescribing or managing RA patients on million people abused or were dependent on DMARD therapies. Further commentary on the opioid pain medicines during 20 [13, 33] and in limitations of using administrative claims to 2014, more than 10 million people in the US identify RA patients can be found in the reported using prescription opioids for non- ‘Strengths and Limitations’ section below. medical reasons . The use of steroids and pain medications, Also concerning is the relatively frequent use including chronic opioid use, in the DMARD of steroids in this patient cohort. Even at low untreated population, while not as high as in doses, steroids such as prednisone are associated the DMARD-treated patients, is problematic. with elevations in blood pressure, blood glu- Symptomatic treatment alone, without use of cose, and lipid levels, as well as increased DMARDs, is not effective in preventing further osteoporosis risk . An assessment of steroid joint damage in RA. Khanna et al. reported use in a cohort mirroring the US population’s similar results in a Medicare population: nearly proportion of women suggests that care must be 68 and 48% of these RA patients had C 1 pre- exercised to ensure that treatments do not scription ﬁll for opioid analgesics and steroids, exacerbate conditions like osteoporosis and respectively . In the overall study popula- attendant bone fracture risks [1, 5]. tion, more than two-thirds received opioids, These results may help to inform decision- with slightly more than one-ﬁfth considered making by stakeholders engaged in the man- chronic users, and more than a half of all agement of RA. Future treatment patterns may patients received steroids. The use of opioids respond to an increased awareness of the treat- and steroids was more common in those ment recommendations and the introduction patients receiving DMARD therapy at any time of new therapies and therapeutic indications on during follow-up compared with those who formulary decisions and payment structures. never received a DMARD during the study per- DMARDs are recommended for all newly iod. Unfortunately, it is a limitation of claims diagnosed patients according to ACR treatment data that reasons for use of medications cannot recommendations [9, 36, 37]. In addition, be explicitly determined. For instance, the DMARD use among patients diagnosed with RA presence of opioid use does not necessarily is one of the quality metrics established by the mean they were being used for treatment of RA National Quality Forum , the Center for symptoms, but may have been used for other Medicare and Medicaid (CMS) Physician Qual- causes of pain unrelated to their RA, such as low ity Reporting System , and the National back pain or surgery. The claims data do not Committee for Quality Assurance Healthcare specify a diagnosis for which a medication was Effectiveness Data and Information Set , prescribed, only that it was dispensed. which require reporting of the percentage of Rheumatol Ther adults with RA treated with DMARD therapy—a patients, possibly under-representing elderly metric which also factors into CMS Star ratings patients enrolled in Medicare. . The results observed in this study raise an Furthermore, diagnosis codes are good indi- important concern of a potential care quality cators but not guarantees of the presence of gap in RA management. disease as they may be subject to miscoding or may be included as a rule-out criterion. As noted, the reliance on administrative claims (a Strengths and Limitations combination of ICD-9 codes from medical claims and claims for DMARD therapy), without The use of a large, geographically diverse, access to medical records, is potentially limiting administrative claims database provided this in identifying RA patients. Studies examining study with several key strengths. First, the data the validity of administrative claims to identify allowed us to retrospectively observe the longi- RA patients have found mixed results depend- tudinal treatment patterns in newly diagnosed ing on the algorithm used. Typically the use of RA patients over a period of 1–9 years (mean, multiple claims with RA diagnoses improved 3.5 years) after their initial RA diagnosis. Sec- the positive predictive value of the deﬁnitions, ond, a sensitivity analysis limiting the observa- and requiring a diagnosis from a rheumatologist tion period to 2010 onward showed consistent improves the algorithm even further [42–44]. results, and thus the ﬁndings are a reﬂection of Studies that utilized a more focused code list of historical treatment patterns over the past dec- RA diagnoses, such as the one used in this ade as well as recent treatment practices. study, tended to have higher validity compared Finally, these data provide real-world evidence with those that used all ICD-9 codes starting of disease management rather than being lim- with 714 . While our deﬁnition of RA may ited to the strict guidelines of a clinical trial or include some patients who do not truly have other controlled experiment. This study RA, the algorithm was developed with the goal includes a contemporary patient population of minimizing this occurrence. We also con- and presents a view of the current treatment ducted a sensitivity analysis using a much more landscape. restrictive (i.e., speciﬁc) deﬁnition of RA, and While there were clear strengths of this study results were not fundamentally different. study, these ﬁndings must be viewed against In general, claims databases are subject to some important limitations. First, we identiﬁed common limitations including the inability to patients newly initiating RA medication after a account for medications from other sources 6-month period in which there were no claims such as over-the-counter purchases or samples for RA medications or RA diagnoses before the from providers. Claims data provide limited index RA medication claim/RA diagnosis code. insight into the clinical status of patients, par- This approach may not have captured the true ticularly their severity or activity of their disease incident RA population, i.e., prevalent RA cases and the process surrounding moving onto may be included, and as a result, our results may DMARD therapy—for example, whether be conservatively biased. Second, the minimum DMARD was offered then declined; prescribed continuous enrollment required was and never ﬁlled; ﬁlled and not used; or not 12 months; however, patients were followed offered at all. Similarly, it is unknown whether beyond the 12-month timeframe until the end patients not seen by a rheumatologist were of the study period, and their patterns later into referred but declined, or not referred. Reasons enrollment may be affected by loss to follow-up. for initiation of one DMARD class over another, The mean observation period of 3.5 years from such as the selection of a non-TNFi or a JAKi as index may offer insufﬁcient time to fully the index DMARD therapy, are unknown, and observe the actual use of more advanced these patients may differ from the general RA DMARD therapies as biologics or JAKi. Finally, population. our analysis focused on commercially insured Rheumatol Ther Lilly and Company. Cynthia J Larmore is a CONCLUSIONS former employee and stock owner of Eli Lilly and Company. Natalie N Boytsov is a Nearly half (45%) of the RA patients in our employee/stock owner of Eli Lilly and Com- study did not receive DMARD therapy at any pany. Andre B Araujo is an employee/stock time during the follow-up period, while the use owner of Eli Lilly and Company. Ralph A of pain medications and steroids was high. The Quimbo is an employee of HealthCore, Inc. most common DMARD treatment pattern was Joseph Singer is an employee of HealthCore, receiving only csDMARD monotherapy with a Inc. Sze-jung Wu is an employee of HealthCore, much smaller portion receiving biologic ther- Inc. John T Hinton is an employee of Anthem, apy at any time. Less than half of patients were Inc. diagnosed with RA by a rheumatologist and is the most likely reason for the variance in Compliance with Ethics Guidelines. Study adherence to treatment recommendations. data were handled in compliance with all Future research efforts may build upon this applicable Health Insurance Portability and work to understand why this treatment gap Accountability Act (HIPAA) rules. Patient exists, to develop solutions to address this anonymity and conﬁdentiality were preserved unmet need, and to ultimately understand how with data de-identiﬁcation throughout. Inves- treatment patterns in RA patients are associated tigational Review Board (IRB) informed consent with disease activity, quality of life, and clinical requirements were waived for this non-experi- and economic outcomes. mental study, which was conducted under the research exception provisions of the Privacy Rule, 45 CFR 164.514(e). ACKNOWLEDGEMENTS Open Access. This article is distributed under the terms of the Creative Commons Funding. The study and article processing Attribution-NonCommercial 4.0 International charges was funded by Eli Lilly & Company. All License (http://creativecommons.org/licenses/ authors had full access to all of the data in this by-nc/4.0/), which permits any non- study and take complete responsibility for the commercial use, distribution, and reproduction integrity of the data and accuracy of the data in any medium, provided you give appropriate analysis. credit to the original author(s) and the source, provide a link to the Creative Commons license, Medical Writing and/or Editorial Assis- and indicate if changes were made. tance. Bernard Tulsi and Mukul Singhal, employees of HealthCore, Inc., provided writ- ing support for this manuscript. Anthony J. Zagar, an employee of Eli Lilly and Company, REFERENCES provided statistical support for the study design. 1. Crane MM, Juneja M, Allen J, et al. Epidemiology Authorship. All named authors meet the and treatment of new-onset and established International Committee of Medical Journal rheumatoid arthritis in an insured US population. Editors (ICMJE) criteria for authorship for this Arthritis Care Res (Hoboken). 2015;67:1646–55. article, take responsibility for the integrity of 2. Helmick CG, Felson DT, Lawrence RC, et al. Esti- the work as a whole, and have given their mates of the prevalence of arthritis and other approval for this version to be published. rheumatic conditions in the United States. Part I. Arthritis Rheum. 2008;58:15–25. Disclosures. David M Kern is a former 3. Salafﬁ F, Carotti M, Gasparini S, Intorcia M, Grassi employee of HealthCore, Inc. Kalyani Son- W. The health-related quality of life in rheumatoid awane is a former employee of HealthCore, Inc. arthritis, ankylosing spondylitis, and psoriatic Lawrence Chang is a former employee of Eli arthritis: a comparison with a selected sample of Rheumatol Ther healthy people. Health Qual Life Outcomes. disease-modifying anti-rheumatic drugs. Rheuma- 2009;7:25. tology (Oxford). 2012;51(Suppl 5):v12–21. 4. Barrett EM, Scott DG, Wiles NJ, Symmons DP. The 17. Birnbaum H, Pike C, Kaufman R, Marynchenko M, impact of rheumatoid arthritis on employment Kidolezi Y, Cifaldi M. Societal cost of rheumatoid status in the early years of disease: a UK commu- arthritis patients in the US. Curr Med Res Opin. nity-based study. Rheumatology (Oxford). 2010;26:77–90. 2000;39:1403–9. 18. Ward MM, Javitz HS, Yelin EH. The direct cost of 5. Greenapple R. Trends in biologic therapies for rheumatoid arthritis. Value Health. 2000;3:243–52. rheumatoid arthritis: results from a survey of payers and providers. Am Health Drug Beneﬁts. 19. Ward MM, Lubeck D, Leigh JP. Long-term health 2012;5:83–92. outcomes of patients with rheumatoid arthritis treated in managed care and fee-for-service practice 6. Agarwal SK. Core management principles in settings. J Rheumatol. 1998;25:641–9. rheumatoid arthritis to help guide managed care professionals. J Manag Care Pharm. 2011;17:S03–8. 20. Hanly JG, Thompson K, Skedgel C. The use of administrative health care databases to identify 7. Cutolo M, Kitas GD, van Riel PL. Burden of disease patients with rheumatoid arthritis. Open Access in treated rheumatoid arthritis patients: going Rheumatol Res Rev. 2015;7:69–75. beyond the joint. Semin Arthritis Rheum. 2014;43:479–88. 21. Kim SY, Servi A, Polinski JM, et al. Validation of rheumatoid arthritis diagnoses in health care uti- 8. Singh JA. American College of Rheumatology lization data. Arthritis Res Ther. 2011;13:R32. guideline for the treatment of rheumatoid arthritis 2015. Arthritis Care Res. 2015;68:1–25. 22. Quan H, Li B, Couris CM, et al. Updating and val- idating the Charlson comorbidity index and score 9. Singh JA, Saag KG, Bridges SL Jr, et al. 2015 Amer- for risk adjustment in hospital discharge abstracts ican College of Rheumatology Guideline for the using data from 6 countries. Am J Epidemiol. Treatment of Rheumatoid Arthritis. Arthritis 2011;173:676–82. Rheumatol. 2016;68:1–26. 23. Khanna R, Smith MJ. Utilization and costs of 10. Agarwal SK. Biologic agents in rheumatoid arthritis: medical services and prescription medications for an update for managed care professionals. J Manag rheumatoid arthritis among recipients covered by a Care Pharm. 2011;17:S14–8. state Medicaid program: a retrospective, cross-sec- tional, descriptive, database analysis. Clin Ther. 11. Genovese MC, Kremer J, Zamani O, et al. Baricitinib 2007;29:2456–67. in patients with refractory rheumatoid arthritis. N Engl J Med. 2016;374:1243–52. 24. Zhang HF, Gauthier G, Hiscock R, Curtis JR. Treat- ment patterns in psoriatic arthritis patients newly 12. O’Shea JJ, Laurence A, McInnes IB. Back to the initiated on oral nonbiologic or biologic disease- future: oral targeted therapy for RA and other modifying antirheumatic drugs. Arthritis Res Ther. autoimmune diseases. Nat Rev Rheumatol. 2014;16:420. 2013;9:173–82. 25. Bonafede MM, Fox KM, Johnson BH, Watson C, 13. Yamaoka K. Janus kinase inhibitors for rheumatoid Gandra SR. Factors associated with the initiation of arthritis. Curr Opin Chem Biol. 2016;32:29–33. disease-modifying antirheumatic drugs in newly diagnosed rheumatoid arthritis: a retrospective 14. Kamal KM, Madhavan SS, Hornsby JA, Miller LA, claims database study. Clin Ther. 2012;34:457–67. Kavookjian J, Scott V. Use of tumor necrosis factor inhibitors in rheumatoid arthritis: a national survey 26. Makol A, Davis JM, Crowson CS, Therneau TM, of practicing United States rheumatologists. Joint Gabriel SE, Matteson EL. Time trends in glucocor- Bone Spine 2006;73. ticoid use in rheumatoid arthritis: results from a population-based inception cohort 1980–1994 vs. 15. Wolfe F, Michaud K. Resistance of rheumatoid 1995–2007. Arthritis Care Res. 2014;66:1482–8. arthritis patients to changing therapy: discordance between disease activity and patients’ treatment 27. Schmajuk G, Trivedi AN, Solomon DH, et al. choices. Arthritis Rheum. 2007;56:2135–42. Receipt of disease-modifying antirheumatic drugs among patients with rheumatoid arthritis in 16. Pavelka K, Kavanaugh AF, Rubbert-Roth A, Ferrac- Medicare managed care plans. JAMA. cioli G. Optimizing outcomes in rheumatoid 2011;305:480–6. arthritis patients with inadequate responses to Rheumatol Ther 28. Yazdany J, Bansback N, Clowse M, et al. The for the use of nonbiologic and biologic disease- rheumatology informatics system for effectiveness modifying antirheumatic drugs in rheumatoid (RISE): a national informatics-enabled registry for arthritis. Arthritis Rheum. 2008;59:762–84. quality improvement. Arthritis Care Res (Hoboken) 2016. 37. Singh JA, Furst DE, Bharat A, et al. 2012 update of the 2008 American College of Rheumatology rec- 29. Boudreau D, Von Korff M, Rutter CM, et al. Trends ommendations for the use of disease-modifying in long-term opioid therapy for chronic non-cancer antirheumatic drugs and biologic agents in the pain. Pharmacoepidemiol Drug Saf. treatment of rheumatoid arthritis. Arthritis Care 2009;18:1166–75. Res (Hoboken). 2012;64:625–39. 30. Anastassopoulos KP, Chow W, Tapia CI, Baik R, 38. (NQF) MSC. NQF-Endorsed Measures for Muscu- Moskowitz B, Kim MS. Reported side effects, bother, loskeletal Conditions National Quality Forum; 2015 satisfaction, and adherence in patients taking 1/2015. hydrocodone for non-cancer pain. J Opioid Manag. 2013;9:97–109. 39. Services CfMM. 2016 Physician Quality Reporting System (PQRS) Measures Groups Speciﬁcations 31. Gregorian RS Jr, Gasik A, Kwong WJ, Voeller S, Manual 2015. Kavanagh S. Importance of side effects in opioid treatment: a trade-off analysis with patients and 40. NCQA. Continuous Improvement and the Expan- physicians. J Pain. 2010;11:1095–108. sion of Quality Measurement 2011. 32. Thielke SM, Turner JA, Shortreed SM, et al. Do 41. Services CfMM. Medicare 2015 Part C & D Star patient-perceived pros and cons of opioids predict Rating Technical Notes 2014 09/03/2014. sustained higher-dose use? Clin J Pain. 2014;30:93–101. 42. Greenlee R, Drahos J, VanWormer J, Landgren O, Koshiol J. PS2-22: Accuracy of Diagnostic Codes to 33. Chou R, Deyo R, Devine B, et al. The effectiveness Identify Rheumatoid Arthritis in Archived Elec- and risks of long-term opioid treatment of chronic tronic Health System Data: Support for Future pain [Evidence Report/Technology Assessment No. Cancer Research Network Studies of Lymphoma 218]. Agency for Healthcare Research and Quality. Risk Pathways. Clinical Medicine & Research http://www.effectivehealthcare.ahrq.gov/ehc/prod 2013;11:154. ucts/557/1971/chronic-pain-opioid-treatment-repo rt-141007.pdf. Accessed 19 Sep 2016. 43. Ng B, Aslam F, Petersen NJ, Yu H-J, Suarez-Almazor ME. Identiﬁcation of rheumatoid arthritis patients 34. Multiple cause of death data. CDC WONDER. using an administrative database: a Veterans Affairs Centers for Disease Control and Prevention. http:// study. Arthritis Care Res. 2012;64:1490–6. wonder.cdc.gov/mcd.html. Accessed 19 Sep 2016. 44. Chung CP, Rohan P, Krishnaswami S, McPheeters 35. Moghadam-Kia S, Werth VP. Prevention and treat- ML. A systematic review of validated methods for ment of systemic glucocorticoid side effects. Int J identifying patients with rheumatoid arthritis using Dermatol. 2010;49:239–48. administrative or claims data. Vaccine. 2013;31(Suppl 10):K41–61. 36. Saag KG, Teng GG, Patkar NM, et al. American College of Rheumatology 2008 recommendations
Rheumatology and Therapy – Springer Journals
Published: May 30, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera