Introduction

Ulcerative colitis (UC) is a chronic, immune-mediated disorder characterized by colonic inflammation, and a relapsing and remitting course [1]. There were approximately 900,000 patients with UC in the United States (US) in 2011 [2]. Consequently, UC is associated with a substantial economic burden on the US healthcare system, which has been estimated to range from 8 to 15 billion US dollars (US$) annually [3]. Despite the increase in therapeutic options available for UC [4], treatment failure is common [5], resulting in further increased demand on the healthcare system, including hospitalizations and surgery. More data on treatment patterns and outcomes of patients with UC are needed to understand the impact of disease control.

Tofacitinib is an oral small molecule Janus kinase inhibitor for the treatment of UC. The efficacy and safety of tofacitinib for the treatment of moderately to severely active UC were evaluated in phase 2/3 induction studies [6, 7], a phase 3 maintenance study [7], an open-label, long-term extension study [8], and a phase 3b/4 study [9]. Tofacitinib was approved in the US in 2018 for patients with moderately to severely active UC, and the prescribing information was updated in 2019 to be approved for patients with moderately to severely active UC who have had an inadequate response or intolerance to one or more tumor necrosis factor inhibitors (TNFi) [10].

Assessment of real-world data is important to further characterize the experience of patients receiving tofacitinib, and to aid in the interpretation of outcomes derived from the tofacitinib UC clinical trial program. While the efficacy of tofacitinib in a real-world setting has been evaluated previously [11,12,13,14,15,16,17], real-world data on outcomes, such as treatment adherence and persistence, and healthcare resource utilization (HCRU) and costs, in patients with UC in the US, remain limited. Further, adherence and persistence of study drugs within trial data may not reflect real-world outcomes, due to patient selection and controlled interventions. The objective of this analysis was to assess real-world characteristics and outcomes, including adherence, persistence, HCRU, and costs, in patients with UC after initiating tofacitinib treatment, using data from the US IBM® MarketScan® claims database.

Methods

Data Sources and Study Design

The IBM® MarketScan® databases are compiled and maintained by Truven Health Analytics, and contain de-identified, patient-level health data from patients in the US, including inpatient and outpatient claims, outpatient prescription claims, healthcare expenditure, and clinical utilization records [18]. Data are contributed by managed care organizations, hospitals, large employers, electronic medical record providers, Medicare, and Medicaid [19].

Eligible patients initiated tofacitinib between May 2018 and August 2019 in the US, and were ≥ 18 years of age at the index date (defined as the date of the first claim for tofacitinib). Patients were required to have at least one International Classification of Diseases code (9th or 10th revision) claim for UC, related to a clinical encounter during the 12-month pre-index period (defined as 12 months prior to the index date; the earliest date for the pre-index period was June 10, 2017, and the latest date was September 4, 2018) and/or on the index date. Patients with UC diagnostic codes related only to diagnostic tests were excluded. Patients meeting the criteria for a diagnosis of UC, who also had a claim code for a rheumatological condition, were included in this analysis, as there were no means of ascertaining whether tofacitinib was prescribed to these patients for UC or for rheumatological conditions. Exclusion criteria included a prescription for tofacitinib during the 12-month pre-index period and/or for the following UC therapies on the index date: vedolizumab, TNFi, azathioprine, mercaptopurine, or methotrexate.

Study Cohorts

Data were analyzed in two cohorts, which were stratified by prior experience with biologic therapy within the variable pre-index period [defined as the earliest date in the patient’s records before the index date (the earliest possible date was January 1, 2015, and the latest possible date was August 1, 2018)]. The first cohort included all patients initiating tofacitinib who had not previously received biologic therapy (bio-naïve), and the second cohort included all patients who had previously received biologic therapy (bio-experienced).

Outcome Measures

Patient demographics and the clinical characteristics of patients who initiated tofacitinib in each cohort were collected, including age, sex, geographic region, and comorbidities. Tofacitinib adherence during the post-index period (defined as a fixed length of 12 months after initiation of tofacitinib) was evaluated using proportion of days covered (PDC; mean, median, and the proportions of patients achieving a PDC threshold of 80%). Persistence was evaluated by tofacitinib discontinuation (defined as either a gap in therapy of a maximum 60 days [sensitivity analysis] or 90 days) and time to tofacitinib discontinuation up to 12 months post-index. The proportions of patients receiving other medications, including oral corticosteroids (OCS), during the 12-month pre-index period, and up to 12 months post-index, were calculated. In addition, use of OCS according to tofacitinib persistence was assessed after a tapering period of 16 weeks from the index date, to allow patients sufficient time to taper off the use of OCS. Discontinuation of OCS was defined as a lack of prescription for OCS after the 16-week tapering period up to month 12. Tofacitinib dose during the 12 months post-index was evaluated, including proportions receiving immediate-release (IR) versus extended-release (XR) formulations at index, and those experiencing dose escalation or de-escalation. Differences in HCRU and healthcare costs were compared before and after the index date. HCRU was calculated for inpatient, outpatient, and emergency room (ER) visits, and it was defined as UC-related if the claim was for a provider-delivered service or administered treatment associated with a UC diagnostic code. Healthcare costs in the 12-month pre- and post-index periods were calculated as all-cause costs and UC-related costs, in line with previous publications [20, 21]. All-cause total costs included pharmacy costs (prescriptions) and medical costs (inpatient, outpatient, and ER visits). Given the variability in the capture of Healthcare Common Procedure Coding System codes in inpatient settings across databases, UC-related treatment costs were limited to outpatient visits only. UC-related costs included pharmacy costs (prescriptions plus administration costs for UC-related treatments) and medical costs (non-administration costs for inpatient, outpatient, and ER visits) for patients with a UC diagnosis in any claim position. Costs were adjusted to 2020 US$ using the annual medical care component of the Consumer Price Index to reflect inflation since 2017.

Statistical Analysis

Kaplan–Meier time-to-event analysis was performed to estimate time to treatment discontinuation in bio-naïve and bio-experienced patients, and the log-rank test was used to compare the resultant curves; p < 0.05 denoted statistical significance. The remaining outcome measures were analyzed descriptively in each cohort; only general trends were described, and no formal statistical analyses were carried out. Categorical variables were reported as numbers and percentages, and continuous variables as means, medians, and standard deviations (SDs). All analyses were conducted using SAS® statistical software.

Results

Patient Demographics and Clinical Characteristics

In total, 276 patients with UC who initiated tofacitinib during the study period were included in this analysis. Of these, during the variable pre-index period, 68/276 (24.6%) patients were bio-naïve (range for variable period: January 1, 2015 to April 1, 2018; median 1,197.5 days), and 208/276 (75.4%) were bio-experienced (range for variable period: January 1, 2015 to August 1, 2018; median 1311.5 days).

Patient demographics and clinical characteristics were generally similar across cohorts (Table 1). In total, 58.7%, 78.4%, and 26.4% of bio-experienced patients and 76.5%, 66.2%, and 11.8% of bio-naïve patients received 5-aminosalicylic acid (5-ASA), OCS, or thiopurines, respectively, during the 12-month pre-index period. The most common extra-intestinal immune-mediated inflammatory diseases (IMIDs) experienced by patients with UC in this analysis were arthralgia (33.7% of patients) and arthropathy (7.3% of patients); 8.0% of patients had at least one IMID for which tofacitinib is an approved therapy (i.e. either rheumatoid arthritis, psoriatic arthritis, or ankylosis spondylitis).

Table 1 Patient demographics and clinical disease characteristics

Medications Received During the 12 Months Post-index Period

The proportions of patients receiving conventional medications up to 12 months post-index are shown in Table 2. The proportions of patients receiving OCS, thiopurines, and 5-ASA decreased for all cohorts in the 12-month post-index period [all patients; difference vs pre-index: − 24.6%, − 13.8% and − 19.2%, respectively (Table 2)].

Table 2 Proportion of patients receiving conventional medications up to 12 months post-index

Adherence and Persistence

Tofacitinib adherence at 12 months post-index is shown in Table 3. Overall median (interquartile range) and mean (SD) tofacitinib adherence (PDC) was 0.82 (0.54) and 0.68 (0.32), respectively. The overall proportion of patients with PDC ≥ 0.80 was 51.8%. Adherence to tofacitinib was generally similar across cohorts.

Table 3 Tofacitinib adherence at 12 months post-index

During the 12-month post-index period, the mean treatment duration was 268.6, 293.2, and 260.6 days, in all patients, bio-naïve patients, and bio-experienced patients, respectively; 43.8% of all patients, 38.2% of bio-naïve patients, and 45.7% of bio-experienced patients discontinued tofacitinib for a maximum of 90 days (Fig. 1a). The corresponding proportions for tofacitinib discontinuation for a maximum of 60 days (sensitivity analysis) were 49.3%, 44.1%, and 51.0% in all patients, bio-naïve patients, and bio-experienced patients, respectively. Overall time to treatment discontinuation is shown in Fig. 1b (maximum gap of 90 days) and Fig. S1 (Online Resource; maximum gap of 60 days). In general, time to tofacitinib discontinuation was longer among bio-naïve patients, compared with bio-experienced patients; however, this was not significant.

Fig. 1
figure 1

a Tofacitinib discontinuation at 12 months post-index, and b Time to treatment discontinuation through 12 months (90-day gap). aMean duration of tofacitinib therapy post-index (days). N number of patients in each group, n number of patients in each category

OCS Use

Of 94 patients who received ≥ 1 OCS during the 16-week tapering period from the index date, 38 patients (40.4%) discontinued OCS, and 56 patients (59.6%) continued to receive OCS after the 16-week tapering period (Fig. 2a). Similar results were observed for bio-naïve and bio-experienced patients (Fig. 2a). Of 182 patients who did not receive OCS during the 16-week tapering period, 120 patients (65.9%) had no OCS use, and 62 patients (34.1%) received ≥ 1 OCS after the 16-week tapering period.

Fig. 2
figure 2

OCS use after the 16-week tapering period up to 12 months post-index, a in patients who received ≥ 1 OCS during the 16-week tapering period, and b overall and by tofacitinib use. aDiscontinuation of tofacitinib was defined as a gap in therapy of a maximum of 90 days. number of patients in each group, n number of patients in each category, N1 number of patients who received ≥ 1 OCS in the 16-week tapering period in each cohort, OCS oral corticosteroid

Overall, after the 16-week tapering period, 118 patients (42.8%) were receiving ≥ 1 OCS (including 56 patients who received ≥ 1 OCS and 62 patients who did not receive OCS in the 16-week tapering period). The proportion of patients who received OCS after the 16-week tapering period was higher among those who discontinued tofacitinib, compared with those who continued receiving tofacitinib (59.5% vs 29.7%, respectively; Fig. 2b). Similar results were observed for bio-naïve and bio-experienced patients (Fig. 2b).

Tofacitinib Dosing

Data for tofacitinib dosing at index and 12 months post-index are shown in Table 4. In total, 96.4% of patients received tofacitinib as an IR formulation, while 3.6% received an XR formulation. The majority of patients (84.1%) received tofacitinib 20 mg/day at the index date. Of 171 patients who were receiving an average tofacitinib dose of 20 mg/day at index, 102 (59.6%) patients remained on that dose during the 12-month post-index period, 59 (34.5%) patients had a reduction in dose, 9 (5.3%) patients had a reduction in dose and a subsequent escalation in dose, and 1 (0.6%) patient had an escalation in dose.

Table 4 Tofacitinib dosing at index and 12 months post-index

HCRU and Costs

Across all cohorts, the mean numbers of all-cause and UC-related outpatient visits per month were lower in the 12-month post-index period compared with the 12-month pre-index period (Table 5). All-cause total costs and UC-related total costs pre- and post-index are shown in Fig. S2 (Online Resource). Patients who were bio-naïve had greater all-cause and UC-related total costs post-index versus pre-index (US$69,463 vs US$21,707 and US$56,189 vs US$9,037 per year, respectively), which were mainly driven by increased pharmacy costs in both cases. Patients who were bio-experienced had lower all-cause total costs and UC-related total costs post-index versus pre-index (US$85,148 vs US$94,613 and US$69,585 vs US$82,033 per year, respectively), which were driven by decreases in pharmacy costs. Slight increases in all-cause and UC-related medical costs were observed for patients who were bio-experienced (bio-naïve difference [post-index minus pre-index]: US$2301 and − US$1100 per year, respectively; bio-experienced difference: US$4,524 and US$3,380 per year, respectively).

Table 5 HCRU (12-month post-index versus 12-month pre-index)

Discussion

To our knowledge, this analysis of real-world US claims data assessing early tofacitinib treatment experience is the largest cohort studied so far in patients with UC initiating tofacitinib. The majority of patients in this analysis (75.4%) had previously received biologic therapy. This is consistent with a previous real-world study of a US claims database, in which 80.9% of patients had prior biologic use [22].

In this study, 56.2% of patients remained on tofacitinib treatment at 12 months post-index (maximum gap of 90 days). The discontinuation rates observed here were consistent with those reported in previous real-world studies of patients receiving tofacitinib with UC who had high levels of prior biologic exposure. Rates of tofacitinib persistence were previously estimated to be 71% at 26 weeks in a UK-based multi-center study of 134 patients [23], 60% at 24 weeks in a Dutch study of 123 patients [24], and 66% and 54% at 24 and 52 weeks, respectively, in a Spanish study of 113 patients [12]. In a recent, real-world study of patients with UC treated with tofacitinib in US claims data, higher rates of persistence (84.9%) were reported than those observed in this analysis, which is likely due to a shorter follow-up time of 4 months. However, tofacitinib adherence in this previous study at 6 months (median [mean]: 0.8 [0.7]) was similar to that reported here at 12 months (median [mean]: 0.82 [0.68]) [22].

OCS use after the 16-week tapering period, which included those patients remaining on and initiating OCS, was higher among those patients who discontinued tofacitinib, compared with those who continued receiving tofacitinib (59.5% vs 29.7%, respectively). In addition, the use of conventional medications decreased in the 12 months post-index; 50.7% of patients used OCS after day 30 up to 12 months post-index versus 75.4% of patients in the pre-index period. In a previous real-world study of US claims data from patients with UC, 41.3% of patients used OCS during a period of 6 months post-tofacitinib initiation versus 76.9% of patients during a 12-month baseline period [22].

In this analysis, the greatest reductions in all-cause and UC-related HCRU in patients initiating tofacitinib were observed in outpatient visits, and UC-related pharmacy costs and all-cause costs decreased post-index for bio-experienced patients. In a prior, retrospective study, all-cause pharmacy costs were increased post-index for bio-experienced patients initiating biologics for UC, but all-cause total costs were decreased at 12 months [25]. The important observations on HCRU and cost in this analysis have potential impact in terms of healthcare costs. Further analysis is required to understand the drivers of HCRU and costs, and the difference across studies.

Limitations of this study include the fact that the presence of diagnostic codes used to identify UC was not validated by a review of the medical records, allowing the possibility of misclassification of disease. However, this algorithm has been used in previous studies [26]. In addition, approximately 8.0% of patients had a diagnosis code for another indication for which tofacitinib is approved (i.e. rheumatoid arthritis, psoriatic arthritis, or ankylosis spondylitis); therefore, it is possible that some of the patients in this analysis received tofacitinib for non-UC conditions. Some important considerations, such as quality of life, disease extent, and activity, could not be assessed in this analysis, and compliance with treatment was assumed based on prescription refills. Further limitations include a potential underestimation of prior use of biologic therapies, plus the use of over-the-counter medication or free samples not being captured in the health plan data. In addition, the reasons for discontinuation (e.g. side effects or lack of efficacy) could not be evaluated in this study. Lastly, these analyses were descriptive only, and no hypothesis testing or multivariable modeling was performed.

Conclusion

In this analysis of real-world US claims data, the majority of patients with UC initiating tofacitinib treatment had previously received ≥ 1 biologic therapy and remained on tofacitinib at 12 months. There was a reduction in OCS use in those who continued tofacitinib treatment versus those who discontinued tofacitinib. Decreased HCRU was observed for all patients, and decreased costs were observed for bio-experienced but not bio-naïve patients, mainly driven by pharmacy costs. This analysis provides a valuable insight into the real-world use and outcomes of tofacitinib in patients with UC.