Introduction
Inflammatory bowel disease (IBD) is a group of chronic disorders characterized by gastrointestinal and extra-intestinal inflammation. The primary subtypes, Crohn’s disease (CD) and ulcerative colitis (UC), significantly impair quality of life and contribute to disability. In Canada, the national incidence of IBD was 29.9 per 100 000 in 2023, translating to 11 700 new cases annually.1 An estimated 322 600 individuals (0.8% of the population) live with IBD, with prevalence expected to exceed 1% within a decade.1 Advances in pharmacological treatments, particularly biologics, may help reduce the burden on individuals and the healthcare system.2,3
Ustekinumab (UST) is a monoclonal IgG1K antibody targeting the p40 subunit of interleukin (IL)-12 and IL-23.4,5 Phase III trials demonstrated its safety and efficacy in inducing and maintaining remission in CD and UC, including patients with prior biologic failure.4,5 Health Canada approved UST for moderate-to-severe CD in December 2016 and for moderate-to-severe UC in January 2020. Subsequent real-world studies, including in Canada, have confirmed the clinical efficacy of UST in both indications.6–8
Ontario’s publicly funded healthcare system covers medically necessary services for all eligible residents, including outpatient and inpatient care. However, outpatient medication coverage is limited to specific populations (ie, individuals under 25 or over 65 years of age or patients in exceptional access programs) with most adults relying on private insurance or out-of-pocket payment. As new biologic therapies emerge, understanding their real-world impact on healthcare resource utilization (HCRU) and costs is critical.
Previous studies have examined HCRU and costs before and after initiation of biologics.9,10 Some studies reported significant reductions in outpatient visits, emergency department (ED) visits, and hospitalizations after starting anti-TNF-α biologic therapies (eg, infliximab, adalimumab), though findings have been mixed.9–12 A US study of IBD patients initiating UST demonstrated overall reductions in HCRU and costs.13 A study in Germany determined that nonpersistent patients on UST and other biologic therapies had higher HCRU and costs compared to persistent patients, largely driven by non-biologic drug costs and inpatient treatments.14 A Canadian study also showed that patients with biologic-naïve status had lower HCRU rates and associated annual costs following receiving UST.15
Despite these findings, real-world data on HCRU and associated cost for UST-treated IBD patients in Canada remain limited. Moreover, the impact of treatment persistence and prior biologic exposure on HCRU and costs have not been well-evaluated in this population.
To address these gaps, this study utilized data from the Johnson & Johnson (J&J) BIOADVANCE® national patient support program, linked to administrative medical records from the Institute for Clinical and Evaluative Sciences (ICES) in Ontario, Canada. The primary objective was to assess HCRU and costs among adult patients with CD or UC who initiated UST through J&J’s BIOADVANCE® program in Ontario. In addition, this study also explored outcomes by treatment persistence and prior biologic exposure status (bio-naïve vs. bio-experienced) to provide further insight into how real-world treatment patterns and patient history may influence HCRU following UST initiation.
Methods
Study design
This retrospective, population-based cohort study examined HCRU and healthcare costs (excluding medication costs) in adults (≥18 years) with CD or UC initiating UST in Ontario, Canada. The analysis followed STROBE and RECORD-PE reporting guidelines.
Patients who initiated UST were identified from J&J’s BIOADVANCE® program and linked to the provincial health administrative databases at ICES (Figure S1; Table S1). The study selection periods ensured inclusion of patients who initiated UST after Health Canada approval: (CD: January 1, 2017, to September 30, 2022; UC: February 1, 2020, to September 30, 2022). Outcomes were assessed for each patient 1-year pre-initiation (ie, baseline year) and up to 4 years post-initiation (or until March 31, 2023, whichever occurred first).
Outcomes were presented according to follow-up year and compared with data from 1 year before initiation. Only patients who had a complete history for a specific follow-up year were included; those with less than 12 months of follow-up during that year were excluded.
Data sources
J&J’s BIOADVANCE® is a national patient support program providing medication administration, reimbursement assistance, and case management for UST patients. It is estimated that over 90% of patients with CD and UC treated with UST in Canada were managed through Janssen BIOADVANCE® during the study period.8 Upon enrollment, patients voluntarily consented to data collection for research. This study used data from J&J’s BIOADVANCE® database including patient account, medical order, treatment, and pharmacy fill records.
ICES is a nonprofit research institute authorized under Ontario’s privacy laws to collect and analyze health care for health system evaluation. It maintains deidentified, longitudinal health records for almost 21 million people residing in Ontario since 1986, including Ontario Health Insurance Plan (OHIP) claims, hospitalizations, and long-term care services.16 Individual records are linked across datasets using encoded identifiers.
This study used the Ontario Crohn’s and Colitis Cohort (OCCC)—a validated, population-based dataset of IBD patients in Ontario created and maintained at ICES—to identify patients with Crohn’s or colitis in the Ontario health administrative data.17 Details of all ICES datasets used in this study are provided in Table S2.
Study objectives
The primary objective was to describe changes in HCRU and non-medication costs from 1 year before and up to 4 years after UST initiation. The secondary objectives included: describing changes in HCRU and non-medication costs among patients who initiated and were persistent on UST treatment in each analysis year (sub-analysis 1); assessing the association of persistence at 1 year with HCRU events and non-medication costs (sub-analysis 2); and, assessing the association of previous biologic experience with HCRU events and non-medication costs (sub-analysis 3).
Study population and eligibility criteria
The study included adult patients with CD or UC who initiated UST during the selection period, were enrolled in J&J’s BIOADVANCE® program, and successfully linked to ICES databases.
The index date was defined as the first recorded UST pharmacy fill (“UST initiation”) during the selection period. Patients were included if they were ≥18 years old, had ≥1 medical order and UST pharmacy fill, and the first medical order included an intravenous induction. CD or UC must have been the primary indication.
Patients were excluded if they had both CD and UC listed as primary and/or secondary indication, had fistulizing CD (excluded due to small patient numbers), had <1 year of data available at ICES prior to UST initiation, had <1 year of data available at ICES following UST initiation, or had a recorded date of death before UST initiation.
Eligible patients were identified in J&J’s BIOADVANCE® and linked to individuals in the OCCC database using a deterministic record linkage methodology using an exact match for date of birth, sex, and postal code as recorded in the ICES databases.18 For instances where eligible patients could be linked to more than 1 OCCC record, a second step was applied to incorporate concordance of IBD indication (CD or UC) between J&J’s BIOADVANCE® and the OCCC database to resolve ambiguities. Only patients who could be uniquely linked on a one-to-one basis to a single OCCC record were included. No analytic matching was performed; linkage variables were used solely to identify individual records across data sources.
The results of the linkage process and exclusions are detailed in Figure S2.
Study outcomes
Patient demographic and clinical characteristics at index or 1 year prior to index, including age, sex, residence, neighborhood income quintile, previous biologic usage, comorbidities (using diagnostic code-based definitions), and complications, were also reported (Tables S3-S5). HCRU outcomes included outpatient visits (general practitioner [GP] and specialist visits differentiated using OHIP physician specialty codes); ED visits; hospitalizations; hospital length of stay; same-day surgeries (SDS, including endoscopy); inpatient rehabilitation; home care; and long-term care days (Table S3). Cost outcomes included all-cause and IBD-specific costs associated with these HCRU events, excluding medication costs.
IBD-specific HCRU and costs were defined by diagnostic codes (ICD-9: 555, 556; ICD-10: K50, K51) recorded as the most responsible diagnosis. HCRU and associated costs were reported annually during the baseline and analysis periods. “Other GP visits” and “Other specialist visits” reflect costs for physician services captured in hospital, emergency, or other undefined settings.
Drug costs were not included because the ICES databases only capture medication costs for drugs reimbursed through the Ontario Drug Benefit (ODB) program. Since not all study participants were eligible for ODB and UST was not covered under this program at the time of the study, drug costs for all included patients could not be ascertained. Moreover, ICES does not contain concomitant or other drug cost information for medications paid for by private insurance or out-of-pocket.
Due to these limitations, the total drug cost for patients could not be assessed and the overall healthcare costs assessed should be interpreted as non-medication costs. In addition, drug insurance coverage (private or compassionate access) was not available for inclusion as an analytic variable.
Persistence on UST was defined as patients being on therapy from first UST pharmacy fill date until discontinuation rule, which was a treatment interruption of >20 weeks (ie, approximately 2.5 cycles). Discontinued patients are defined as those with a treatment interruption of >20 weeks. Patients reaching the end of the analysis period were censored as of this date, without having a discontinuation event.
Biologic-naïve status was defined as the patient self-reported status recorded at enrollment in J&J’s BIOADVANCE® program. All eligible individuals were studied for the primary objective irrespective of UST persistence. Additionally, 3 sub-analyses were conducted as a part of the secondary objectives.
Statistical methods
Descriptive analyses included counts and percentages for categorical variables and means (standard deviation [SD]), medians interquartile range [IQR]), minimums, and maximums for continuous variables. HCRU events (from hereon, HCRU) was reported as the count of each occurrence or as the total duration (days) of all events, depending on the variable. Non-medication costs were calculated using the ICES person-level costing methodology previously developed by Wodchis et al standardized to 2022 Canadian dollars.19 Percentage changes are presented for interpretability and were derived from model-based ratios.
HCRU counts were compared between pre-/post-UST initiation using a negative binomial mixed model. Mean non-medication costs were compared pre- and post-UST initiation using a gamma mixed model. For each follow-up year, outcomes of patients with full history in that specific follow-up year were compared with baseline pre-UST initiation.
Univariate models were employed to describe changes in HCRU and non-medication costs in the main cohort (primary objective) and among persistent patients (sub-analysis 1). Multivariate models were used to estimate associations between 1-year UST persistence and HCRU and non-medication costs (sub-analysis 2), and associations between prior biologic experience HCRU and non-medication costs (sub-analysis 3), and included age, sex, rural residence, time since OCCC entry, treatment initiation year, and biologic-naïve status.
Missing data were reported as their own category, where possible and no imputation was performed. Analyses were conducted using SAS® (version 9.2 or higher). Results were aggregate and suppressed, where necessary, to reduce the risk of patient re-identification in accordance with ICES reporting standards.
Ethical considerations
At the time of J&J’s BIOADVANCE® enrollment, patients provided voluntary consent for deidentified information to be collected for use in future studies and scientific research. The protocol for this study was reviewed and approved by the Advarra Institutional Review Board (IRB; Pro00072328) (registered with the Office of Human Research Protections and FDA under IRB#00000971).
Results
Patient demographics and clinical characteristics
Of the 25 858 patients with a primary indication of CD or UC in J&J’s BIOADVANCE® program, 2279 patients with CD (Year 1: n = 2279; Year 2: n = 1659; Year 3: n = 1057; Year 4: n = 453) and 626 patients with UC (Year 1: n = 626; Year 2: n = 317; Year 3: n = 49) in Ontario met study eligibility criteria. The mean (SD) age was 45.7 (16.2) years for CD patients and 45.6 (17.2) for UC patients, with the distribution of age groups being similar between CD and UC patients (Table 1). More than half of patients with CD (53.6%) and 44.1% of patients with UC were female and most resided in urban areas (CD: 87.2%, UC: 88.8%).
UST initiation was evenly distributed among income quintiles for patients with CD, whereas patients with UC had a higher proportion in the 2 highest income quintiles. Prior biologic therapy was more common in UC (77.3%) than CD (64.3%).
HCRU and non-medication costs for main study cohort
Among patients with CD, mean all-cause HCRU events decreased in each year following UST initiation compared to baseline (Year 1: −11.1%, Year 2: −13.1%, Year 3: −19.4%, and Year 4: − 24.7%; P < .001) (Figure 1). Patients with UC experienced significant HCRU reductions in Year 2 (−20.8%, P < .001) and Year 3 (−43.1%, P < .01) (Figure 2). Significant decreases were observed across both cohorts for GP visits, specialist visits, hospitalizations, ED visits, and SDS (Table S6A and B).
IBD-specific HCRU significantly decreased in both cohorts in all follow-up years (Table S7A and B).
Mean all-cause non-medication costs for patients with CD decreased significantly in Years 1–3 (Year 1: −11.2%, Year 2: −27.0%, Year 3: −22.0%; P < .001 in Year 1–2, P < .01 in Year 3) with a non-significant 8.8% reduction in Year 4 compared to baseline (Figure 1). Non-medication cost savings were largely driven by reductions in GP visits (Years 1–2), hospitalizations (Year 2), and specialist visits, ED visits, and SDS across all years (Table S6A). Patients with UC experienced non-medication cost reductions in all years (Year 1: −9.2%, Year 2: −35.8%, Year 3: −33.9%), though significance was only reached in Year 2 (P < .01).
IBD-specific non-medication costs decreased significantly in Years 1–3 for both cohorts (Table S7A and B). In some categories, the number of patients with a ≥ 1 recorded event may differ from the number of patients with non-zero costs, reflecting the ICES person-level costing outputs for those encounters.
HCRU and non-medication costs for patients persistent on UST (sub-analysis 1)
CD and UC patients who initiated and persisted on UST had significantly lower all-cause HCRU events in all follow-up years compared to baseline. Among patients with CD, reductions were observed in all follow-up years (CD: Year 1: −18.6%, Year 2: −23.8%, Year 3: −36.9%, Year 4: −35.0%, P < .001). Patients with UC had a similar pattern, with significant reductions in Years 1–3 (Year 1: −23.9%, Year 2: −35.4%, Year 3: −53.5%, P < .001 for Years 1–2, P < .05 for Year 3) (Table 2).
Patients with CD who initiated and persisted on UST had significant non-medication cost reductions in all follow-up years (−31.1% to −48.3%; P < .05 to P < .001), while patients with UC had significant non-medication cost decreases in Years 1 and 2 (P < .001). IBD-specific HCRU and non-medication costs were consistently lower in both cohorts compared to baseline (Table S8A and B).
HCRU and non-medication costs for persistent vs discontinued UST patients (sub-analysis 2)
Demographic and clinical characteristics of patients with CD and UC by persistence status in Year 1 are presented in Table S9A and B. Patients with CD who were persistent on UST in Year 1 had significantly lower all-cause HCRU events in Years 2–4 compared to non-persistent patients (Year 2: −30.9%, Year 3: −32.8%, Year 4: −31.5%, P < .001 in Year 2–3, P < .05 in Year 4) (Table 3). Patients with UC who were persistent had significantly lower HCRU in Year 2 (–35.2%, P < .001), though reductions in Year 3 were not significant.
All-cause non-medication costs in persistent patients with CD were significantly lower in Years 2–3 (−37.5% and −40.8%, P < .001), but the difference in Year 4 (−2.0%) was not significant. In patients with UC, significant non-medication cost reductions were observed in Year 2 (−5.8%, P < .05), though non-medication costs trended higher for persistent patients in Year 3 (80.2%, P > .05). IBD-specific non-medication costs followed a similar pattern (Table S10A and B).
HCRU and non-medication costs for biologic-naïve vs biologic-experienced (sub-analysis 3)
Demographic and clinical characteristics of patients with CD and UC stratified by biologic experience are presented in Table S9A and B. Among CD patients, bio-naïve individuals had significantly greater reductions in all-cause HCRU events and non-medication costs than bio-experienced patients (Year 1: −30.5%, Year 2: −37.5%, Year 3: −32.1%, Year 4: −28.8%; P < .001) (Table 4). In contrast, UC, patients had inconsistent findings: all-cause HCRU was lower in Years 1 (−17.6%) and 3 (−31.6%), while non-medication cost reductions were only observed in Years 2 (−18.0%) and 3 (−7.3%), though none reached statistical significance.
IBD-specific HCRU and non-medication costs were significantly lower in bio-naïve patients with CD in all follow-up years (Table S11A). For UC, bio-naïve patients had lower HCRU and non-medication costs in Year 1, but in subsequent years, non-medication costs were numerically higher, but not statistically significant (Table S11B).
Discussion
This study is the first in Canada to report real-world HCRU and non-medication costs following UST initiation in adult patients with CD or UC. Using data from a retrospective patient support program linked to provincial health administrative datasets, we found significant reductions in all-cause HCRU events and non-medication costs in patients with CD and UC following UST initiation. The largest reductions compared to the baseline year were observed in patients who initiated and were persistent on UST and may reflect treatment response and better disease control.
The real-world trends observed in this study may demonstrate benefit among patients treated with UST as ED visits and hospitalizations were lower in periods after treatment initiation, potentially reducing the overall burden on the healthcare system and supporting the role of UST as a therapeutic agent in CD and UC. Given the observational pre/post design and lack of a concurrent comparator group, these findings should be interpreted as associations that may also reflect improved disease control or other time-varying factors rather than a causal effect of UST alone.
Our findings align with previous analyses of biologic therapies in IBD.9,10 For instance, a US study reported reductions in all-cause and CD-related HCRU, including ED visits by 20.5% and hospitalizations by 30.3%, in the 1-year following UST initiation.13 Our study showed similar decreases—at least 11.1% for HCRU and 8.8% for non-medication costs in patients with CD, and at least 5.1% for HCRU and 9.2% for non-medication costs in patients with UC. The JUSTify study in Canada, focusing on bio-naïve patients with CD, also found reduced HCRU following UST initiation, particularly for ER visits and hospitalizations.15 A US claims study of bio-experienced patients with CD found significant reductions in all-cause hospitalization by 47.6% and costs by 49.2% over 24 months.20 Our study sub-analyses revealed that bio-naïve patients had lower all-cause HCRU compared to bio-experienced patients and may be suggestive of other findings that bio-naïve patients often experience greater benefits from biologic treatments.
In our study, patients with CD and UC who remained persistent on UST had significant reductions in both HCRU and non-medication costs, which have been observed elsewhere. For example, a German study of biologic therapies, including UST, found lower outpatient visits and treatment costs (excluding medication) among persistent IBD patients.15 Previous research showed that sustained treatment, which is a surrogate marker for satisfactory treatment response and tolerance in observational studies, has been reported to be associated with lower healthcare costs.15,21,22 Taken together, our findings may suggest the value of treatment persistence in managing these chronic conditions as well as lowering the burden of illness to the public payer.21 However, subgroup findings by persistence and prior biologic exposure should be interpreted cautiously, as these groups may differ systematically in baseline disease severity and care pathways, and the study was not designed to support causal comparisons between groups.
Our study’s results are strengthened by the large sample size and linkage of J&J’s BIOADVANCE® program with provincial health administrative data at ICES, with the longitudinal design supporting comprehensive capture of healthcare touchpoints and billing.
This study has several limitations. As mentioned previously, the pre/post design of this study and lack of a comparator treatment group restricts the ability to make causal inferences about the effect of UST. Regression to the mean may have contributed to reductions in HCRU and non-medication costs, as patients may have initiated therapy during periods of exacerbated disease activity with high HCRU and non-medication costs.23 Secular changes in clinical practice or healthcare delivery may have further influenced utilization patterns.
Without a comparator treatment group, we cannot discern the extent to which reductions in HCRU and non-medication costs are attributed to UST initiation. Another limitation is that this study did not capture important clinical variables, such as baseline disease severity, phenotype, non-IBD comorbidities, concomitant medication use, treatment response and escalation, and reasons for discontinuing treatment. Our association estimates for persistence may be confounded by the fact that persistent patients tend to be better responders to treatment.
Inclusion of patients from the BIOADVANCE® program may have introduced selection bias, although BIOADVANCE® participants represent more than 90% of the treated population. Further, ICES data reflect Ontario residents with public healthcare coverage, and generalizability may be limited in other Canadian and international jurisdictions. Linkage of BIOADVANCE® records to ICES records may introduce misclassification bias, and retrospective administrative data are subject to measurement error and variation in coding practices.
A major limitation was the exclusion of medication costs, including UST and other medications, which hinder our interpretation of the impact on overall healthcare costs. Medication costs were not available because UST was not publicly reimbursed for CD or UC under Ontario’s drug benefit program during the study period, and private/compassionate medication costs are not captured in the ICES databases; therefore, overall costs are likely to be underestimated. Further, hospital costs were calculated using a weighted average resource intensity method that could potentially underestimate outliers.
Despite these limitations, the consistent trends and associations of reduced HCRU events and non-medication costs following UST initiation across all study outcomes support the real-world benefits of UST in managing IBD. Although medication costs were not included, UST has established efficacy and may reduce HCRU and other costs through improved disease control, which may help offset the significant contribution of biologics to healthcare costs. Medication costs may be further mitigated by the availability of UST biosimilars.
Conclusion
Our study showed that CD and UC patients treated with UST had decreased HCRU events and healthcare non-medication costs in the years following initiation compared to baseline in Ontario, Canada. Moreover, CD patients who were bio-naïve or persistent on treatment experienced greater reductions in HCRU and non-medication costs compared to patients who were bio-experienced or discontinued treatment. Despite the lack of a comparator group, these findings further our current understanding about the impact of UST on disease burden in individuals with CD and UC.
Future research including patients from jurisdictions outside of Ontario and leveraging a prospective cohort design, comparator treatment group, and data on patient-reported outcome would better characterize the benefits of UST on reducing disease burden in real-world settings across Canada.
Supplementary Material