Switching from originator infliximab to biosimilar versus continuing on originator in inflammatory bowel disease: results from the observational Project NORTH study

Abstract

Objective

Project NORTH compared real-world clinical and economic outcomes in Swedish patients with inflammatory bowel disease (IBD) who switched from originator infliximab to its biosimilar.

Materials and methods

Data from electronic medical records and Swedish national registries were linked. Switchers (patients switching from originator infliximab to its biosimilar between 1 April 2014, and 31 December 2017) and non-switchers (patients who received originator infliximab and did not switch to a biosimilar by 31 December 2017) were followed up until 31 October 2019.

Results

Baseline concomitant medication use, disease duration, and inflammatory markers were lower among switchers than non-switchers. At 6 months, the proportion of patients with stable disease was higher among switchers than non-switchers (71/109 [65%] vs 54/107 [50%]; p = .0385); differences were not significant in subsequent follow-ups. At 6 and 24 months, 98% and 93% of switchers, respectively, used concomitant medications versus 96% and 79% of non-switchers. Throughout the study, all-cause treatment discontinuation occurred in 74 (67%) switchers and 105 (95%) non-switchers. At 36-months, mean (SD) number of IBD-related in-patient care days was higher among non-switchers (2.95 [4.71]) than switchers (1.40 [4.20]), as were total medical costs (€16,740 vs €3,872).

Conclusions

No substantial differences in clinical outcomes or healthcare resource utilization were observed between switchers and non-switchers. Several analyses indicate that non-switchers might have more poorly controlled/severe disease than switchers at baseline. Overall, numerous difficulties might arise when executing a high-quality, real-world study, including possible selection bias for patients with better disease control for NMS, limiting the generalizability of the results.

Keywords:

• IBD
• TNF inhibitor
• infliximab
• biosimilar
• non-medical switch
• Crohn’s disease
• ulcerative colitis

Introduction

Inflammatory bowel disease (IBD) is a chronic disease of the gastrointestinal tract classified as ulcerative colitis (UC), Crohn’s disease (CD), or unclassified IBD [1,2]. Prevalence of IBD is higher in developed countries, and most patients are diagnosed in early adulthood [3,4]. Patients with IBD have increased risk of morbidity and mortality compared with the general population [4–6].

Management of IBD has improved over the last two decades with the use of biologics, such as tumor necrosis factor (TNF), integrin and interleukin (IL)–12 and IL-23 inhibitors [7–10]. As the number of approved biosimilars has increased over the years, non-medical switching (NMS) from originator biologics to biosimilars has become more common [11–13]. Although approved biosimilars are highly similar to their originators and expected to produce the same clinical result in any given patient, they are not identical [14,15]. Results on efficacy, safety and immunogenicity from two previous studies raise no major concerns but are limited to observations in a few countries [16,17] and do not meet all the requirements for a robust switching study [12] thus, the potential outcomes of NMS between originators and biosimilars remain incompletely understood, especially over the longer term.

In Swedish patients with IBD, NMS from the originator infliximab (IFX) to biosimilar IFX began in March 2014, preceding the patent expiry of the originator in February 2015. This observational cohort study used patient-level data to compare the clinical and economic outcomes between patients with IBD who switched from originator IFX to biosimilar IFX (switchers) and control patients (non-switchers) in a real-world setting.

Methods

Study design

Project NORTH was a Swedish observational cohort study with study period of 1 January 2008, to 31 October 2019, inclusive. Patients with IBD who underwent NMS from IFX to an IFX biosimilar after 1 April 2014, were compared with a control group comprising patients who either continued treatment with IFX or discontinued IFX treatment without switching to an IFX biosimilar by 31 December 2017 (i.e., non-switchers).

Data were extracted from electronic medical record (EMR) systems in primary and secondary care, and combined with routinely collected data from four national registries (data extraction period): the Longitudinal Integration Database for Health Insurance and Labour Market Studies (2008 to 2017), registry from Statistics Sweden, National Patient Registry (NPR; 2001 to 2018), Prescribed Drug Registry (2005 to 2019), and Cause of Death Registry (2008 to 2019); EMRs and historical data were extracted as early as possible through 2018. Data from electronic medical records (EMRs) were collected using IQVIA’s extraction tool Customized eXtraction Program (CXP) [18,19].

End of follow-up was defined as the earliest of any of the following events: end of data (the end date for the annual registry dataset in Sweden), death, emigration from Sweden, entry into any clinical trial, pregnancy, breastfeeding or planned pregnancy, and surgeries or adverse events of interest (listed in Supplemental Table 1).

Table 1. Algorithm for Generation of Index Date for Non-switchers.

The index date for non-switchers was generated from the time of distribution for switchers according to the following steps:

Step 1 Estimation of the distribution of elapsed time between the first IFX infusion and the index date for switchers: tests of multiple distributions (normal, log normal, exponential, Weibull and Gamma) for the variable “elapsed time”.

Step 2 The Weibull distribution yielded the highest value of the Cramer-von Mises criterion with the data (time until index date of switchers). From this, time until index date was randomly extracted from the Weibull distribution and assigned to non-switchers to calculate their simulated index date.

Step 3 The actual IFX infusion date among non-switchers closest to the simulated index date was selected as the final index date for non-switchers.

Note If the index date of a non-switcher was the date of their first IFX infusion, the second IFX date was chosen as the index date to avoid comparing non-switchers who just entered treatment (treatment-naive patients) with switchers who were treated with IFX at least once before.

IFX: infliximab.

For switchers, the index date was defined as the first date of switching from originator to biosimilar IFX between 1 April 2014 and 31 December 2017, and did not require that the use of the originator immediately preceded the use of the biosimilar; almost 18% of switchers had not used the originator biologic during the year before the NMS.

The index date for non-switchers was generated from the time of distribution for switchers according to the steps outlined in Table 1. The process for selecting an index date for non-switchers was designed to normalize baseline disease severity in the two groups and to make the distribution of time on originator before index date comparable between the two groups (Supplemental Table 2).

Table 2. Demographics and Baseline Characteristics.

Demographics and Baseline Characteristics	Primary Analysis Population^a		Study Population
	Non-Switchers (n = 111)	Switchers (n = 111)	Non-Switchers (n = 282)	Switchers (n = 125)
Age, y, median (IQR)	30 (21, 42)	27 (19, 42)	30 (22, 45)	28 (19, 42)
Male, n (%)	58 (52.3)	62 (55.9)	147 (52.1)	71 (56.8)
Pediatric patients, n (%)	13 (11.7)	23 (20.7)	35 (12.4)	24 (19.2)
Disease type, n (%)
Indeterminate colitis	21 (18.9)	21 (18.9)	60 (21.3)	22 (17.6)
Ulcerative colitis	36 (32.4)	27 (24.3)	91 (32.3)	29 (23.2)
Crohn's disease	54 (48.6)	63 (56.8)	131 (46.5)	74 (59.2)
Concomitant medication, n (%)	102 (91.9)	97 (87.4)	263 (93.3)	111 (88.8)
Disease duration on index date, years, mean (SD)	6.5 (4.6)	5.9 (4.5)	4.6 (4.3)	6.4 (4.7)
Disease duration ≥ 2 years, n (%)	83 (74.8)	82 (73.9)	157 (55.7)	96 (76.8)
Length of follow-up, years, mean (SD)	4.5 (2.8)	2.75 (0.8)	4.51 (2.7)	2.75 (0.8)
CRP, mg/L, mean (SD)	9.6 (13.3)	6.4 (24.6)	9.5 (16.1)	5.9 (23.3)
	n = 72	n = 110	n = 191	n = 123
Fecal calprotectin, μg/g, mean (SD)	1573 (2260)	554 (770)	2039 (2890)	575 (808)
	n = 26	n = 45	n = 80	n = 49
Medication use within 1 year before or on index date, n (%)
Remicade	111 (100)	94 (84.7)	282 (100)	103 (82.4)
Inflectra	0	68 (61.3)	0	72 (57.6)
Remsima	0	43 (38.7)	0	53 (42.4)
Concomitant medication use within 1 year before or on index date, n (%)
Glucocorticosteroids	74 (66.7)	31 (27.9)	201 (71.3)	36 (28.8)
Antibiotics	35 (31.5)	38 (34.2)	114 (40.4)	43 (34.4)
5-ASA	66 (59.5)	49 (44.1)	169 (59.9)	54 (43.2)
IMMs	104 (93.7)	111 (100)	265 (94.0)	125 (100)
Azathioprine	45 (40.5)	43 (38.7)	121 (42.9)	50 (40.0)

5-ASA: 5-aminosalicylic acid; CRP: C-reactive protein; IBD: inflammatory bowel disease; IMM: immunomodulatory medication; IQR: interquartile range.

^aPrimary analysis population consists of patients after propensity score matching; study population consists of all patients in the regional IBD population (n = 407).

Participants

Inclusion criteria included ≥2 diagnoses of IBD in the NPR, available and complete EMR, ≥2 infusions of IFX, and no treatment with an IFX biosimilar before 1 April 2014. Patients diagnosed with unresolved malignancy or hepatitis B or C infection, those with colectomy within 5 years before the index date, or diagnosed with uncontrolled infection or organ failure within 180 days before index date were excluded.

Cohort 1 comprised Swedish patient-level data from the NPR starting from 2001 and was used to describe patient demographics, disease history, and drug utilization and to assess whether Cohort 2 was representative of the broader Swedish patient population. Cohort 2 was a subset of Cohort 1 that included regional IBD populations based on patient-level data from EMRs and included all patients treated for IBD in the health regions of Stockholm (data from Uppsala, Dalarna, and Västerbotten were excluded owing to lack of data coverage on drug dosage).

This study was approved by the Regional Ethics Board, Uppsala, on 1 November 2017 and was conducted according to the ethical principles of the Declaration of Helsinki.

Objectives

The primary endpoint was time to all-cause treatment discontinuation, defined as the time interval from the index date to discontinuation of originator IFX or biosimilar IFX. Patients who did not discontinue treatment at the time of analysis were censored at the last date of follow-up.

Secondary study objectives (assessed post–index date) included: proportion of patients who discontinued treatment (all causes), reasons for discontinuation, concomitant drug use, changes in prescribed IBD treatment dosage, changes in disease activity during follow-up (reflecting 90 days before and 90 days after each time point), adherence to treatment measured as medication possession ratio, and healthcare resource utilization (HCRU). Total medical costs (including cost to the patient and total cost of prescribed IBD treatments, including IFX and concomitant medications, and IBD-related surgical treatment) were also assessed.

Exploratory objectives included patient-reported number of sick leave days, proportion of patients who switched to other biologics and proportion of switchers who switched back to originator IFX.

Comparisons of study characteristics pre- and post-switch among switchers were made for the following variables: C-reactive protein (CRP), erythrocyte sedimentation rate (ESR), fecal calprotectin (FC), and white blood cell (WBC) count at index date and 3, 6, 9, and 12 months post–index date, and HCRU 1 year before and 1 year after the switch.

Statistical analyses

Propensity score matching was used to match switchers and non-switchers (Supplemental Table 3). The primary endpoint was analyzed using the Kaplan-Meier method. Descriptive statistics are presented, including for comparisons of pre- and post-switch among switchers. For analyses requiring statistical inference, model-based point estimates were presented together with their 95% CIs and P values. Statistical Analysis Software, version 9.3 or higher (SAS Institute, Cary, NC, USA) were used for data management and statistical analyses.

Table 3. Persistence Rate^a Among Non-switchers and Switchers Throughout the Study.

Follow-Up	Non-switchers	Switchers
	(n = 111)	(n = 111)
6 months
Persistence rate (95% CI)	0.30 (0.22–0.39)	0.83 (0.74–0.89)
Still at risk, n	32	81
12 months
Persistence rate (95% CI)	0.25 (0.18–0.34)	0.74 (0.64–0.81)
Still at risk, n	27	63
18 months
Persistence rate (95% CI)	0.20 (0.13–0.28)	0.55 (0.43–0.65)
Still at risk, n	21	35
24 months
Persistence rate (95% CI)	0.16 (0.10–0.24)	0
Still at risk, n	17	0

^aPersistence rate defined as the proportion who had not discontinued treatment owing to any cause.

Results

Cohort 1 comprised 96,654 patients and Cohort 2 10,619 patients; the regional IFX study population comprised 407 patients (125 switchers and 282 non-switchers; Figure 1). Switchers (n = 111) were matched 1:1 with non-switchers (n = 111) by propensity score (Supplemental Table 3), resulting in a primary analysis population of 222 patients.

Figure 1. Project NORTH study population. Propensity score matching was used to match 111 switchers with 111 non-switchers (also see Supplemental Table 3). IBD: inflammatory bowel disease; IFX: infliximab.

Median (interquartile range) age at baseline was 27 (19–42) years among switchers and 30 (21–42) years among non-switchers (Table 2). The most frequent IBD diagnosis was CD among both switchers (57%) and non-switchers (49%). At baseline, switchers had lower use of concomitant medications (glucocorticosteroids; 5-aminosalicylic acid [5-ASA]), lower CRP and FC levels, and shorter disease duration than non-switchers. IFX treatment was stable (uninterrupted for a period of ≥120 days between infusions during follow-up) among 48% of switchers and 78% of non-switchers. Time from the first IFX dose to index date was 447 days for non-switchers versus 954 days for switchers. The most common reason for end of follow-up in the study was “end of data” for both non-switchers (n = 76 [68%]) and switchers (n = 101 [91%]), followed by IBD-related abdominal surgery, which was reported more frequently among non-switchers (n = 31 [28%] vs n = 7 [6%]).

Primary endpoint

Median time to IFX discontinuation was considerably longer among switchers than non-switchers (559 vs 56 days; Figure 2). During the initial follow-up period (up to day 56), the rate of treatment discontinuation was higher among non-switchers than switchers. Conversely, after day 56, a steeper rate of discontinuation was observed among switchers than non-switchers. Median follow-up time among switchers was much shorter than among non-switchers (2.7 vs 4.7 years).

Figure 2. Primary endpoint of time to infliximab treatment discontinuation due to all causes for non-switchers vs switchers among patients with IBD. All-cause discontinuation was defined as the following events: the last observable infusion of originator infliximab (non-switchers) or biosimilar infliximab (switchers) occurring ≥120 days before the end of available data plus 56 days or date of an infusion or subcutaneous injection of a non-infliximab medication for IBD (adalimumab, vedolizumab, or ustekinumab), or date of an infusion of originator infliximab (applicable only for the switchers). IBD: inflammatory bowel disease.

Secondary endpoints

All-cause treatment discontinuation occurred in 74 (67%) switchers and 105 (95%) non-switchers (Figure 2). Among patients who did not discontinue treatment, the persistence rate was consistently higher among switchers than non-switchers up to 18 months (Table 3). Treatment was discontinued in six patients after an adverse event of interest in each cohort (5.4%).

Concomitant medication use was 98% and 96% at month 6 and 93% and 79% at month 24 among switchers and non-switchers, respectively (Figure 3). The most commonly used concomitant medication was immunomodulatory medications among switchers (97%) and non-switchers (84%) throughout the 36-month study (p = .0009). Glucocorticosteroid use was reported in 57% of non-switchers and 34% of switchers throughout the study (p = .0012). The proportion of patients using concomitant glucocorticosteroids decreased in both groups from month 6 to month 24: from 29% to 21% among non-switchers and from 14% to 10% among switchers (Figure 3).

Figure 3. Proportion of patients with (A) overall concomitant medication use, (B) concomitant glucocorticosteroid use, (C) concomitant 5-ASA use, and (D) concomitant azathioprine use. 5-ASA, 5-aminosalicylic acid; mo: months.

The proportion of patients reporting a 6-month run-in of stable disease (i.e., no increase in disease activity or changes to IBD treatment) was higher among switchers than non-switchers (65% vs 51%; p = .0385) at 6 months. No significant differences between groups was noted in subsequent follow-up (Figure 4).

Figure 4. Proportion of patients switching medications (A), mean number of sick leave days per year (B), proportion of patients with 6-month stable disease (C), and mean total medical costs throughout the study (D) among non-switchers and switchers. IL: interleukin; n/a: not applicable; TNF: tumor necrosis factor. TNF inhibitors were adalimumab, adhesion blocker was vedolizumab, and anti-IL12/23 was ustekinumab (for patients with Crohn’s disease; relevant from 2017 onward).

Total medical costs were significantly higher among non-switchers (€16,740) versus switchers (€3,872) throughout the study (p = .0000; Figure 4). However, patients in both groups reported a similar mean (SD) number of outpatient clinic visits because of their IBD (switchers, 8.82 [6.14]; non-switchers, 9.20 [9.56]) whereas a slightly lower median number of outpatient visits to gastroenterologists (non-switchers, 6; switchers, 8). The mean (SD) number of in-patient care days post–index date was 2-fold higher among non-switchers versus switchers (2.95 [4.71] vs 1.40 [4.20]; p = .6536).

Exploratory endpoints

Number of sick leave days among employed patients were available for 97 (87%) switchers and 110 (99%) non-switchers. The mean (95% CI) number of sick leave days per year post–index date was slightly higher among non-switchers (25.2 [13.7–36.7] vs 21.2 [8.7–33.8]; Figure 4). Among switchers, 16 patients switched to another TNF inhibitor (14%), 10 switched to biologic adhesion blockers (9%), and <5 switched to anti-IL12/23 medication post–index date (Figure 4).

Following NMS, 8 (7%) patients switched back to originator IFX. Although 28% of switchers reported using 2 IFX biosimilars during follow-up, the 2 biosimilars were the same compound (CT-P13) using different brand names. Among non-switchers, 49 (44%) patients switched to another TNF inhibitor, 24 (22%) to adhesion blockers, and 8 (7%) to anti-IL12/23 medication (Figure 4). Among the 5 (5%) non-switchers who switched to an IFX biosimilar, all reported using only one IFX biosimilar during follow-up.

Among switchers, CRP was numerically higher at baseline than during follow-up, whereas WBC concentration was relatively consistent between baseline and follow-up (Supplemental Figure 1). There was more variability in levels of FC and ESR, but interpretation is limited by the small number of patients evaluated for these variables at baseline.

A significantly lower mean (SD) number of in-patient care days was observed among switchers in the year following the switch (0.84 [3.1] days) compared with the year before the switch (1.96 [5.7] days; p = .029; Supplemental Figure 2). However, the mean total medical costs were significantly higher 1 year after the switch (€1,452) versus 1 year before the switch (€802; p< .0196). Otherwise, HCRU was comparable 1 year before and 1 year after the switch measured by mean number of outpatient visits, total or primary care visits and patients’ medical costs (Supplemental Figure 2).

Discussion

This study assessed the potential clinical and economic outcomes of NMS from originator IFX to its biosimilars among patients with IBD in Sweden. Median time to IFX discontinuation was shorter among non-switchers versus switchers and the rate of discontinuation was higher among non-switchers during the first 56 days post-switch. After day 56, the discontinuation rate in the non-switcher group stabilized. These observations likely reflect the analytical creation of an index date among non-switchers, the definition of discontinuation (set at 56 days among non-switchers, with their last IFX dose at the index date), and the switchers’ shorter follow-up compared with non-switchers, which resulted in an artificial cliff on day 56, with many non-switchers meeting the definition for discontinuation after the expected IFX dosing interval of 8 weeks.

Several analyses indicated that non-switchers had more poorly controlled/severe disease than switchers at index date and accordingly may not have been selected for NMS (e.g., higher CRP, higher IBD-related hospitalizations and surgeries, higher use of concomitant glucocorticosteroids and more switching to another anti-TNF or other biologics). The higher use of glucocorticosteroids among non-switchers may explain the relatively higher proportion of patients reporting a run-in with stable disease between the index date and 6-month follow-up.

Although there was evidence of lower HCRU associated post-NMS, no meaningful differences in HCRU between switchers and non-switchers were observed except for the higher total medical costs among non-switchers. No clinically meaningful differences in HCRU between pre- and post-switch were observed among switchers, with the exception of mean number of in-patient care days, which decreased by approximately 50% in the year after the switch (another marker indicating that the switchers may have had more well-controlled disease at baseline); however, total medical costs were higher 1 year after versus 1 year before the switch.

The results are broadly consistent with previous studies on NMS. A recent literature review concluded that NMS was not associated with notable efficacy, safety, or immunogenicity concerns, but acknowledged the limitations of the available study designs including lack of control arms [20]. Our data are also commensurate with findings from the NOR-SWITCH study, which showed no difference in efficacy and safety between continuing on originator and switching to a biosimilar [16,21,22]. Similar findings were shown in another international non-inferiority study [17] and two observational studies [23,24]. However, some studies have indicated poorer outcomes after a switch. A recent study of patients with chronic inflammatory diseases (including CD and UC) showed that patients who switched from the IFX originator to a biosimilar were more likely to switch again (either to a different biosimilar or back to originator) or to discontinue treatment entirely than non-switchers [25]. A double-blind, randomized IBD study showed that 11% more patients who switched to a biosimilar lost response or had to stop therapy compared with non-switchers [26], whereas a retrospective multicenter cohort study demonstrated that a significantly higher rate of relapse was observed among switchers (14% vs 5%) [27]. Although not assessed in this study, the nocebo effect may play a role for loss of efficacy or adverse events following a switch and has been reviewed in detail previously across NMS studies [28]. Although results vary, real-world NMS studies are valuable for contextualizing the safety and efficacy of switching than pivotal clinical trials that cannot target the broader IBD population [29].

Our study has a number of strengths. First, the length of the follow-up period after switching was, at 36 months, longer than reported in most of the NMS studies (≤6 months in nearly half of the studies analyzed in a recent review)[12]. Second, this study used real-world data that better reflects the true clinical outcomes of NMS in patients with IBD, than randomized phase 3 or phase 4 clinical trials [16,21,29]. Third, by using multiple real-world data sources and by examining a wide range of baseline characteristics, diagnostic tests, and clinical outcomes plus within-patient comparisons before and after switching, we could describe a range of real-world results due to NMS in patients with IBD.

Limitations of this study included the small size and its restriction to a single country (Sweden); some analyses (e.g., ESR, FC) included only a few patients, and results may not be generalizable to other geographic regions. Furthermore, limited data collection (such as specific reasons for discontinuation) and use of proxies (i.e., concomitant medications, use of glucocorticosteroids, laboratory values) to describe disease activity are limitations, and some of the EMR-based variables were derived from free-text searches, potentially leading to information bias. No measurements of serum trough levels or anti-drug antibodies were presented but no differences were expected in immunogenicity based on previous studies [30,31].

Selection bias potentially occurred owing to several study design elements. To balance the distribution of disease duration from the first IFX dose to the index date between groups, an analytical index date for non-switchers was used. However, many patients had received their last treatment before their simulated index date, causing the time between first IFX dose and index date to be much smaller among non-switchers than switchers; index dates for non-switchers were, on average, 2 years earlier than for switchers. Furthermore, lower percentage of patients with disease duration of ≥2 years on index date was observed in non-switchers and time from the first IFX dose to the index date was much shorter for non-switchers, possibly introducing shorter exposure time and, thus underestimating outcomes among some switchers.

Because fewer therapeutic options were available during the earlier time periods of our study, patients may have continued on IFX longer with suboptimal clinical outcomes, leading to potential channeling bias. In addition, a higher proportion of non-switchers had UC, which typically has a higher risk of more severe disease than CD [32], which may have affected the decision regarding NMS. It is feasible that the higher cost of originator treatment may have resulted in originator being reserved for the relatively sicker patients, resulting in patients with less severe disease being selected for NMS. The definition of switcher did not require that the use of the originator immediately preceded the use of the biosimilar.

Finally, the possibility of immortal time bias should be considered (a period of participant observation during which the outcome of interest [or death] cannot occur because of the definitions of exposure and/or start of study follow-up used) [33]. To reduce this bias, we applied an analytical matching process to balance the distribution of time from first IFX treatment to index date between switchers and non-switchers. After defining the analytical index date, 38% of non-switchers were found to have discontinued before 1 April 2014. A remaining potential bias in follow-up was the study design requirement to have switchers continue treatment until the date of NMS for inclusion in the switcher group. This aspect of the study design may have contributed to the faster discontinuation observed among non-switchers during the first 2 months of follow-up compared with switchers.

In conclusion, the findings of our study indicate that there are no substantial differences in clinical outcomes or HCRU, with the exception of total medical costs, associated with NMS from originator IFX to its biosimilar in Swedish patients with IBD. However, the selection of patients with less severe disease or better disease control for NMS may limit the generalizability of these results. This study shows that numerous design and analytical difficulties exist when trying to execute a high-quality real-world study for complex biologics. Accordingly, ongoing investigations of the clinical and economic/HCRU implications of NMS for IBD treatment for each originator-biosimilar combination seems warranted.

Acknowledgments

Medical writing support was provided by Maria Hovenden, PhD, and Moira Hudson, PhD, of ICON (PA, USA) and was funded by AbbVie Inc.

Disclosure statement

F. Faccin and Z. Xue are full-time AbbVie employees and may own AbbVie stock and/or options. E. Gemmen, H. Koo, and H. A. Ward are full-time IQVIA employees and were hired by AbbVie to run this study. P. M. Hellström and P. Malmborg report economical study compensation from IQVIA.

Data availability statement

AbbVie Inc. is committed to responsible data sharing regarding the clinical trials we sponsor. This includes access to anonymized, individual and trial-level data (analysis data sets), as well as other information (e.g., protocols and Clinical Study Reports), as long as the trials are not part of an ongoing or planned regulatory submission. This includes requests for clinical trial data for unlicensed products and indications.

These clinical trial data can be requested by any qualified researchers who engage in rigorous, independent scientific research and will be provided following review and approval of a research proposal and Statistical Analysis Plan (SAP) and execution of a Data Sharing Agreement (DSA). Data requests can be submitted at any time and the data will be accessible for 12 months, with possible extensions considered. For more information on the process, or to submit a request, visit the following link: https://www.abbvie.com/our-science/clinical-trials/clinical-trials-data-and-information-sharing/data-and-information-sharing-with-qualified-researchers.html.

Additional information

Funding

AbbVie Inc. funded the study, contributed to its design, and was involved in the collection, analysis, and interpretation of the data and in the writing, review, and approval of the manuscript.