Abstract
BACKGROUND:
Treatment for inflammatory bowel disease (IBD), composed of Crohn disease (CD) and ulcerative colitis (UC), frequently employs biologics to help control inflammation, reduce symptom burden, and limit disease progression. Because of both the lifelong, progressive nature of disease and potential for loss of therapy response over time, therapy changes are common. Biologic dose escalation is one method to adjust therapy regimens following loss of response.
OBJECTIVE:
To assess the occurrence of dose escalation in CD and UC and its influence on health care costs.
METHODS:
Adults with a diagnosis of CD or UC newly initiating therapy with a biologic (adalimumab, infliximab, ustekinumab, vedolizumab, or tofacitinib [UC only]) from January 1, 2017, to June 30, 2022, were selected in the Merative MarketScan Commercial and Medicare Databases. The first claim for the biologic served as the index date and patients were followed over a 12-month pre-period and a 12-month-or-longer post-period. Dose escalation during the maintenance phase of therapy, discontinuation, and postdiscontinuation switching were assessed in biologic-based subgroups in each of the CD and UC cohorts; per-patient-per-month (PPPM) IBD-related health care costs over the duration of index biologic treatment were also calculated.
RESULTS:
Analyses included 6,056 patients with CD (33.5% adalimumab, 30.6% ustekinumab, 18.8% infliximab, 17.1% vedolizumab) and 4,533 patients with UC (35.1% vedolizumab, 30.6% adalimumab, 16.8% infliximab, 10.0% ustekinumab, 7.5% tofacitinib). Dose escalation occurred in 30.4% of patients with CD (adalimumab 20.8%, infliximab 48.8%, ustekinumab 26.9%, vedolizumab 36.2%) and 30.1% of patients with UC (adalimumab 23.0%, infliximab 44.8%, ustekinumab 25.5%, vedolizumab 31.8%, tofacitinib 24.9%). Patients with evidence of dose escalation had lower rates of discontinuation compared with nonescalators over follow-up but were more likely to switch biologics after discontinuation. PPPM IBD-related health care costs over the course of treatment were substantial and largely driven by index biologic costs. Mean PPPM costs for the CD cohort ranged from $4,543 in the infliximab group to $14,031 in the ustekinumab group; costs were similar in the UC cohort, ranging from $5,213 in the infliximab group to $14,246 in the ustekinumab group. Within both cohorts, dose escalation was a significant predictor of increased IBD-related health care costs.
CONCLUSIONS:
Nearly one-third of patients with IBD in this study escalated their biologic dose, which significantly increased their disease-related health care costs. These results demonstrate current challenges in long-term biologic therapy in IBD and highlight the ongoing need for further research into biologic management strategies to optimize IBD patient care and long-term outcomes, while containing costs.
Plain language summary
This study looked at groups of patients with Crohn disease (CD) or ulcerative colitis (UC) newly starting a biologic therapy and measured biologic dose escalation, discontinuation, and switching along with inflammatory bowel disease–related health care costs over an average of 2 years of treatment. Approximately 30% of patients had dose escalation over follow-up. Escalation risk varied across biologics but was similar between patients with CD and UC. Patients who escalated were less likely than nonescalators to discontinue therapy but incurred increased per-patient-per-month health care costs over follow-up.
Implications for managed care pharmacy
Biologic dose escalation is one method used to personalize therapy regimens in approximately 30% of biologic-treated patients with CD and UC. Although dose escalation patterns are similar between CD and UC, results suggest an association between choice of biologic and dose escalation. Additionally, inflammatory bowel disease–related health care costs were found to be tied to both the specific biologic used and the presence of dose escalation. Thus, therapy choices are likely to not only impact patient outcomes but also influence health care costs.
Inflammatory bowel diseases (IBDs), including Crohn disease (CD) and ulcerative colitis (UC), are inflammatory, autoimmune conditions affecting the gastrointestinal tract. Within the United States, approximately 2.4 million individuals are diagnosed with IBD; diagnosis peaks between 20 and 40 years of age.1–3 Symptoms commonly include abdominal pain, fatigue, and other gastrointestinal issues such as diarrhea, constipation, and bowel urgency; however, patients also experience extraintestinal manifestations, like arthropathy and an increased risk of colorectal cancer.2,3 The burden of IBD is notable, with patients reporting negative quality-of-life impacts and direct costs estimated at $25.4 billion in 2016, owing to high health care resource utilization and high-cost therapies.2–5
CD and UC are both progressive diseases characterized by a relapsing and remitting course; poor control of inflammation and disease symptoms can lead to notable tissue damage. Although mucosal healing is possible with current therapeutics, there is no cure for IBD and individuals require lifelong therapy management; many, especially those with CD, will eventually require surgical intervention (eg, colectomy).2,6 Treatment is aimed at controlling inflammation to help enter and maintain remission. Pharmaceutical therapies span a wide range of agents including aminosalicylates, immunomodulators, steroids, biologics, and small molecule inhibitors (eg, JAK inhibitors).6 Biologic and small molecule therapies have been shown to be effective in managing CD and UC, including achievement of mucosal healing and reductions in hospitalizations and surgeries. However, a number of challenges remain, such as defining the optimal treatment regimen for each individual, long-term therapy approaches, and costs of treatment.4,7 Therapy changes throughout the course of CD and UC are common, as patients with IBD may experience initial nonresponse to specific therapies, loss of response over time, or adverse events. Studies estimate that dose escalation occurs in up to 50% of patients with IBD, while switching is observed in up to one-third; however, rates of dose escalation vary widely, with some analyses reporting rates of dose escalation as low as 5% and differential rates across biologics.6,8,9 Dose escalation can be achieved either through increasing the medication dose at each administration or through reducing the dosing interval. In the United States, dose escalation of advanced therapies is frequently off label as only 1 of the 4 CD therapies (infliximab) and 1 of the 5 UC therapies (tofacitinib) assessed here include a provision for dose escalation in their prescribing information; in both cases, labeled dose escalation is through an increase in dose and not a reduction in the dosing interval. Although multiple investigations have assessed dose escalation, reporting is inconsistent across studies with some analyses reporting by biologic and others reporting overall rates; similarly, some studies report at the IBD level whereas others report separately for patients with CD or UC.10–12 Finally, many analyses take a global perspective, using data from multiple countries; given the differences in biologic prescribing and practice patterns, especially around labeled dose escalation, findings in these studies may not reflect outcomes in the United States.13
Considering the chronic nature of IBD and early age at diagnosis, approaches to help improve the longevity of therapy, including increasing adherence, reducing discontinuation, and limiting unnecessary switches remain important to long-term patient care. This study sought to examine biologic dose escalation in adults, including rates of escalation and time to escalation, as well as associated IBD-related health care costs within patients diagnosed with CD and UC in the United States. Analyses were conducted at the biologic level for each condition to provide a comprehensive assessment of dose escalation in IBD.
Methods
DATA SOURCE
Analyses used the Merative MarketScan Commercial and Medicare Databases from January 1, 2016, to June 30, 2023. These closed-system, administrative claims databases contain the full inpatient, outpatient medical, and outpatient pharmacy experience for more than 60 million individuals and dependents with employer-sponsored commercial, Medicare Supplemental, or Medicare Advantage insurance. Because the study used only deidentified patient records and does not involve the collection, use, or transmittal of individually identifiable data, institutional review board approval to conduct this study was not necessary.
STUDY DESIGN
The retrospective cohort study included independent cohorts of patients with CD or UC, defined as at least 1 inpatient or nondiagnostic outpatient claim for either CD (International Classification of Diseases, Tenth Revision, Clinical Modification [ICD-10-CM]: K50.x) or UC (ICD-10-CM: K51.x) and no claims for the other condition (eg, no claims for UC in the CD cohort) from January 1, 2016, through June 30, 2023. Eligible patients also had at least 1 claim for a study biologic, which included adalimumab (or biosimilars), infliximab (or biosimilars), ustekinumab, vedolizumab, or tofacitinib (UC only). The earliest biologic claim served as the study index date; for individuals with claims for more than 1 study biologic, the first claim for the last biologic initiated served as the index date. Each patient had continuous eligibility for 12 months prior to index (pre-period) and 12 or more months after index (post-period). The end of the variable-length post-period was defined as the end of continuous eligibility or the last date of available data (June 30, 2023). The 12-month pre-period was imposed to ensure patients were new initiates to the index biologic. Patients were also aged 18 years or older on index, had at least 1 claim for CD or UC (dependent on cohort) in both the pre- and post-periods, and had at least 1 additional claim for the index biologic at any time during the variable duration of post-index biologic use (index to first 90-day gap in biologic therapy). Patients with at least 2 claims for another autoimmune condition that any study biologic was approved for (ankylosing spondylitis [ICD-10-CM: M08.1, M45.x], atopic dermatitis [ICD-10-CM: L20.x], axial spondyloarthritis [ICD-10-CM: M45A.x], hidradenitis suppurativa [ICD-10-CM: L73], juvenile idiopathic arthritis [ICD-10-CM: M08.8x, M08.9x], psoriasis or psoriatic arthritis [ICD-10-CM: L40.x], rheumatoid arthritis [ICD-10-CM: M05.x, M06.x], and uveitis [ICD-10-CM: D86.83, H20.x, H30.0x-H30.13x, H44.11x, H44.13x]) from the pre- through the post-period were excluded. Figure 1 presents the stepwise cohort selection criteria; Supplementary Figure 1 (418.2KB, pdf) (available in online article) presents a study diagram.
FIGURE 1.
Stepwise Patient Selection Flowchart
CD = Crohn disease; UC = ulcerative colitis.
OUTCOMES
The primary study outcomes were measured over the duration of biologic treatment, defined as the time from the index date until the first discontinuation of the index biologic (based on a 90-day gap in days supply) or the end of the study period. Outcomes included the proportion of patients with evidence of dose escalation during the maintenance phase of biologic therapy, days from index to dose escalation, and total per-patient-per-month (PPPM) IBD-related health care costs. The start of the maintenance phase was defined based on biologic specific dosing or fill patterns in the data (Supplementary Table 1 (418.2KB, pdf) ). The maintenance phase started on or after the index date, with the latter instance occurring in patients with no evidence of an induction phase. Dose escalation was defined as either 2 consecutive medical or pharmacy claims with at least a 20% increase in dose or 2 consecutive medical claims where an administration occurred at least 7 days ahead of the expected schedule (ie, dosing intensification) based on the drug’s prescribing information; 2 sensitivity definitions for dose escalation were also examined (Supplementary Materials (418.2KB, pdf) ). IBD-related health care costs were calculated based on inpatient claims with a diagnosis of IBD in the primary position, outpatient claims with a diagnosis of IBD on any claim line, or outpatient pharmacy claims for an IBD therapy (Supplementary Table 2 (418.2KB, pdf) ); costs included both insurer- and patient-paid components. PPPM IBD-related costs were reported for inpatient, outpatient, and outpatient pharmacy sites of care; costs for the index biologic were also reported. PPPM costs were calculated at the patient level as the total costs incurred over the duration of the variable-length follow-up divided by the duration of follow-up; for patients with at least 30 days of follow-up, total costs were used. Secondary outcomes were reported over the post-period and included index biologic discontinuation (based on a 90-day gap in therapy) and postdiscontinuation switching (reported among discontinuers); these outcomes were reported for subsets of patients with and without evidence of dose escalation (escalators and nonescalators, respectively). Demographics at index and clinical characteristics, including a claims-based Charlson Comorbidity Index (CCI), during the pre-period were also assessed.14
ANALYSES
Reporting was conducted in independent cohorts of patients with CD and UC indexing on each of the study biologics and was descriptive in nature with no statistical testing between CD and UC cohorts or biologic groups. Cox proportional hazards models were developed to examine factors associated with the time to first dose escalation in each of the CD and UC cohorts; patient follow-up was censored at the first escalation. As dose escalation trajectories for the index drugs varied over time, the proportional hazard assumption was violated; thus, hazard ratios (HRs) were modeled with a log-linear function of time, defined as log(HR) = log(intercept) + log(slope)*log(t-1), where t indexes month of follow-up starting at 1 full month post-index date. Intercept estimates for index biologics represent the initial hazard ratio of escalation at the first full month of follow-up compared with adalimumab (reference); index biologic slopes provide an estimate of change in the HR of dose escalation over time. General linear models (GLMs) with γ error distribution and logarithmic link function were developed to model mean PPPM IBD-related total health care costs; index drug was treated as a categorical variable with adalimumab as the reference. Other model covariates for both Cox models and GLMs included days from index to maintenance (Cox models only), index year (reference: 2017), baseline biologic use, baseline immunomodulator use, baseline surgery, age, sex (reference: female), mean CCI score, and baseline IBD-related inpatient admission. The GLMs also included a variable for evidence of dose escalation (no escalation, escalation <6 months, escalation ≥6 months) with no escalation as the reference. In addition to the GLMs, ensemble models were also developed to model IBD-related health care costs. These models include the GLMs as well as random forest, xgboost, and adaptive spline algorithms to include the best features of each individual model to optimize cost predictions. Ensemble models were trained and validated across 5 folds of cross-validation to prevent overfitting and false positives. Percent reduction in mean square error (MSE) was reported for the ensemble models to identify covariates with the greatest influence on IBD-related health care costs, with greater MSE percent reduction being associated with a greater cost impact.
Statistical Analysis Software was used for data management; R was used for statistical analyses.
Results
CD AND UC COHORTS
The study included a total of 6,056 patients with CD and 4,533 patients with UC (Figure 1). Within the CD cohort, 33.5% indexed on adalimumab, 30.6% on ustekinumab, 18.8% on infliximab, and 17.1% on vedolizumab. The mean (SD) age in the CD cohort ranged from 38 (15) years in infliximab-treated patients to 45 (15) years in vedolizumab-treated patients (Supplementary Table 3 (418.2KB, pdf) ). Duration of index therapy was 24 (17) months for the adalimumab group, 24 (16) months for the infliximab and vedolizumab groups, and 26 (17) months for the ustekinumab group. Within the UC cohort, 35.1% of patients indexed on vedolizumab, followed by 30.6% on adalimumab, 16.8% on infliximab, 10.0% on ustekinumab, and 7.5% on tofacitinib (Supplementary Table 3 (418.2KB, pdf) ). Mean age in the UC cohort ranged from 42 (14) years in the adalimumab group to 44 (14) years in the tofacitinib group (Supplementary Table 3 (418.2KB, pdf) ). Duration of index therapy treatment for UC ranged from 21 (10) months for ustekinumab to 25 (16) months for vedolizumab (Supplementary Table 3 (418.2KB, pdf) ). Sex distributions were similar in all biologic groups in the CD and UC cohorts (Supplementary Table 3 (418.2KB, pdf) ).
BASELINE CHARACTERISTICS
Within the CD cohort, the mean CCI score ranged from 0.44 (0.89) in the infliximab cohort to 0.72 (1.33) in the vedolizumab cohort (Supplementary Table 3 (418.2KB, pdf) ). More adalimumab-treated (83.9%) and infliximab-treated (71.9%) patients were biologic naive at index compared with ustekinumab-treated (34.1%) and vedolizumab-treated (47.0%) patients. Despite differences in pre-index biologic use, all 4 biologic groups in the CD cohort had similar proportions of patients with baseline use of corticosteroids (65.7% of adalimumab users to 69.1% of ustekinumab users) and immunomodulators (41.2% of ustekinumab users to 49.3% of infliximab users). Fewer than 15% of patients in each biologic group had evidence of baseline surgery (Supplementary Table 3 (418.2KB, pdf) ).
Across the 5 biologic groups in the UC cohort, the mean CCI score ranged from 0.41 (0.91) in the adalimumab group to 0.48 (1.03) in the vedolizumab group (Supplementary Table 3 (418.2KB, pdf) ). Differences in prior biologic use were observed, with the greatest proportions of biologic-naive patients in the adalimumab (85.8%) and infliximab (71.2%) groups and the lowest proportions of biologic-naive patients in the tofacitinib (27.7%) and ustekinumab (19.9%) groups; just over half (54.2%) of vedolizumab-treated patients were biologic naive at index. The vast majority of UC patients had baseline use of corticosteroids (80.1% of ustekinumab users to 86.4% of infliximab users) and immunomodulators (67.8% of ustekinumab users to 85.3% of adalimumab users). Rates of baseline surgery were extremely low, with fewer than 2% of patients in each biologic group with evidence of surgery (Supplementary Table 3 (418.2KB, pdf) ).
DOSE CHANGES AND DISCONTINUATION
Biologic dose during the maintenance period was examined among the 5,587 patients with CD who had evidence of a maintenance period without changing between routes of administration (eg, switch between subcutaneous and intravenous administration during maintenance; Supplementary Materials). Dose escalation was observed in 30.4% of patients with CD; the highest rates of dose escalation were in infliximab-treated patients (48.8%), whereas adalimumab-treated patients (20.8%) had the lowest (Figure 2). Similar findings were observed with less stringent sensitivity definitions of dose escalation that required a single claim with a 20% or 30% increase in dose (Supplementary Figure 2 (418.2KB, pdf) ). Among the 1,699 patients with dose escalation, mean time from index to first dose escalation was 413 days; time to dose escalation was 386 days for infliximab-treated patients, 392 days for adalimumab-treated patients, and 444 days for both vedolizumab- and ustekinumab-treated patients. Discontinuation during the maintenance phase was observed in 21.2% of escalators and 37.7% of nonescalators. Adalimumab-treated patients were most likely to discontinue within both escalator and nonescalator samples (Figure 3). Among the 361 escalators who discontinued their index biologic, 45.4% switched to another biologic; the switch rate was only 28.9% among the 1,466 nonescalators who discontinued their index biologic. Trends for higher switching rates in escalators vs nonescalators persisted within the biologic groups (Supplementary Figure 3 (418.2KB, pdf) ).
FIGURE 2.
Percentage of Patients With an Index Biologic Dose Escalation During the Period of Index Biologic Use
Mean (SD) months of index biologic use per cohort: adalimumab CD 24.4 (16.8), adalimumab UC 22.9 (17.2); infliximab CD 23.9 (15.9), infliximab UC 22.8 (16.7); ustekinumab CD 26.0 (16.5), ustekinumab UC 20.5 (10.3); vedolizumab CD 24.3 (16.3), vedolizumab UC 25.1 (15.7); tofactinib UC 22.5 (15.2).
CD = Crohn disease; UC = ulcerative colitis.
FIGURE 3.
Percentage of Dose Escalators and Nonescalators Discontinuing Their Index Biologic During Follow-Up
Mean (SD) months of total post-period follow-up: adalimumab CD 32.3 (17.2), adalimumab UC 31.2 (17.4); infliximab CD 29.9 (15.9), infliximab UC 29.4 (16.2); ustekinumab CD 31.6 (16.3), ustekinumab UC 24.6 (9.5); vedolizumab CD 30.7 (15.5), vedolizumab UC 30.4 (15.6); tofacitinib UC 32.0 (13.6).
CD = Crohn disease; UC = ulcerative colitis.
Dose characteristics during the maintenance phase were examined within the 4,119 patients with UC who had evidence of a maintenance phase and did not change biologic route of administration. Of these, 1,238 patients (30.1%) had a dose escalation; the highest (44.8%) and lowest (23.0%) rates of dose escalation were observed for infliximab-treated patients and adalimumab-treated patients, respectively (Figure 2). Mean time from index to first dose escalation was 384 days; the shortest time to dose escalation was observed in adalimumab-treated patients at 300 days followed by ustekinumab-treated patients at 328 days, tofacitinib-treated patients at 361 days, infliximab-treated patients at 369 days, and vedolizumab-treated patients at 468 days. Discontinuation was observed in 23.7% of the 1,238 escalators and 37.1% of the 2,881 nonescalators; adalimumab-treated patients were most likely to discontinue in both escalator and nonescalator subgroups (Figure 3). Among the 293 escalators who discontinued, 53.2% switched to another biologic; among the 1,070 nonescalators who discontinued, 38.0% switched to another biologic. Switching in UC was more common among escalators compared with nonescalators in all biologic groups except infliximab (Supplementary Figure 3 (418.2KB, pdf) ).
Cox proportional hazard models revealed varying trajectories of dose escalation over follow-up (Supplementary Figures 4 and 5 (418.2KB, pdf) ). Within both the CD and UC models, all index biologics, except tofacitinib in UC, were associated with a lower initial risk of dose escalation at month 1 compared with adalimumab, adjusting for selected covariates (Table 1; P < 0.05). Over time, all index biologics except tofacitinib were associated with an increased risk of dose escalation relative to adalimumab, leading to a higher rate of dose escalation over the study period (Table 1). For patients with CD, the dose escalation risk in infliximab surpassed adalimumab at month 3, ustekinumab at month 6, and vedolizumab at month 12. Among patients with UC, the dose escalation risk for infliximab, ustekinumab, and vedolizumab surpassed adalimumab at months 3, 8, and 12, respectively. Other variables significantly associated with a lower probability of dose escalation were a longer duration between index and the start of maintenance and older age (per 10-year increment) (Table 1). Indexing later in the study period (2020-2022 vs 2017) and use of biologics in the baseline period were factors significantly associated with a higher probability of dose escalation in both models (Table 1). Within the CD model, use of immunomodulators in baseline was associated with a higher probability of dose escalation, whereas the presence of a CD-related surgery in baseline was associated with a lower probability of dose escalation (Table 1). In the UC models, a higher baseline CCI score and presence of IBD-related inpatient admissions in baseline were associated with a higher probability of dose escalation (Table 1).
TABLE 1.
Cox Proportional Hazards Models for Risk of Dose Escalation in the CD and UC Cohorts
| CD cohort | UC cohort | |||||||
|---|---|---|---|---|---|---|---|---|
| Estimate | 95% CI lower bound | 95% CI upper bound | P value | Estimate | 95% CI lower bound | 95% CI upper bound | P value | |
| Time from index to maintenance start (months, log scale) | 0.694 | 0.584 | 0.825 | <0.001 | 0.772 | 0.635 | 0.938 | 0.018 |
| Index drug (reference adalimumab) | ||||||||
| Infliximab intercept | 0.364 | 0.226 | 0.588 | <0.001 | 0.089 | 0.049 | 0.161 | <0.001 |
| Infliximab slope | 2.617 | 2.179 | 3.144 | <0.001 | 4.304 | 3.396 | 5.454 | <0.001 |
| Ustekinumab intercept | 0.317 | 0.199 | 0.505 | <0.001 | 0.156 | 0.071 | 0.346 | <0.001 |
| Ustekinumab slope | 1.758 | 1.457 | 2.121 | <0.001 | 2.394 | 1.702 | 3.367 | <0.001 |
| Vedolizumab intercept | 0.006 | 0.003 | 0.013 | <0.001 | 0.004 | 0.002 | 0.009 | <0.001 |
| Vedolizumab slope | 8.108 | 6.338 | 10.371 | <0.001 | 9.426 | 7.274 | 12.213 | <0.001 |
| Tofacitinib intercept | 1.108 | 0.514 | 2.391 | 0.833 | ||||
| Tofacitinib slope | 1.038 | 0.724 | 1.488 | 0.839 | ||||
| Index year (reference 2017) | ||||||||
| 2018 | 0.951 | 0.813 | 1.111 | 0.524 | 1.055 | 0.869 | 1.282 | 0.693 |
| 2019 | 1.074 | 0.920 | 1.254 | 0.386 | 1.081 | 0.886 | 1.319 | 0.581 |
| 2020 | 1.308 | 1.116 | 1.533 | 0.002 | 1.322 | 1.086 | 1.610 | 0.011 |
| 2021 | 1.417 | 1.205 | 1.667 | <0.001 | 1.634 | 1.341 | 1.991 | <0.001 |
| 2022 | 1.385 | 1.106 | 1.734 | 0.007 | 1.484 | 1.148 | 1.920 | 0.006 |
| Biologic use in baseline | 1.436 | 1.291 | 1.597 | <0.001 | 1.154 | 1.017 | 1.309 | 0.040 |
| Immunomodulator use in baseline | 1.178 | 1.070 | 1.298 | 0.002 | 1.156 | 0.996 | 1.341 | 0.078 |
| Surgery in baseline | 0.823 | 0.688 | 0.983 | 0.044 | 0.842 | 0.397 | 1.788 | 0.724 |
| Age, 10-year increment | 0.955 | 0.923 | 0.989 | 0.014 | 0.952 | 0.913 | 0.993 | 0.035 |
| Sex (female) | 0.936 | 0.851 | 1.030 | 0.209 | 0.970 | 0.867 | 1.085 | 0.693 |
| CCI score | 1.033 | 0.985 | 1.083 | 0.209 | 1.077 | 1.017 | 1.141 | 0.019 |
| IBD-related inpatient admission in baseline | 1.095 | 0.957 | 1.254 | 0.209 | 1.310 | 1.119 | 1.534 | 0.002 |
Time from index to maintenance start was defined as the time from the index date (first fill for the index biologic) until the initiation of the maintenance phase of therapy (Supplementary Materials).
Dose escalation trajectories for the index drugs varied over time and violated the proportional hazard assumption; thus, hazard ratios were modeled with a log-linear function of time, defined as log(HR) = log(intercept) + log(slope)*log(month of follow-up). On the exponentiated scale, this is HR = intercept*(slope^log(month of follow-up)). Intercepts reflect the relative risk of dose escalation (adalimumab reference) in the first month of follow-up. Risk in subsequent months of follow-up can be calculated using the slope. Using infliximab in the CD model as an example: at the first month of follow-up, infliximab had a lower relative rate of escalation of 0.306 (0.226, 0.588) of that of adalimumab. By month 12, infliximab has an estimated risk ratio of 3.97 to adalimumab. The point at which infliximab switches to higher estimated risk is estimated at 3 months (found by setting log(HR) to 0 and solving for months).
CCI = Charlson Comorbidity Index; CD = Crohn disease; HR = hazard ratio; IBD = inflammatory bowel disease; UC = ulcerative colitis.
IBD-RELATED HEALTH CARE COSTS
Over the duration of index biologic treatment, mean unadjusted PPPM IBD-related health care costs were similar in the CD and UC biologic groups (Supplementary Table 4 (418.2KB, pdf) ). The lowest mean PPPM IBD-related costs were in the infliximab groups (CD: $4,543, UC: $5,123), whereas the highest costs were in the ustekinumab groups (CD: $14,301, UC $14,246). PPPM IBD-related costs in the remaining biologic groups were lower for the tofacitinib (UC: $5,925) group and vedolizumab groups (CD: $6,908, UC: $6,800) compared with the adalimumab groups (CD: $8,260; UC: $8,744). Index biologic costs were the primary cost driver in all biologic groups.
GLMs identified several significant predictors of IBD-related health care costs. Within both the CD and UC models, dose escalation was associated with increased costs compared with no escalation; escalation within 6 months was associated with the highest costs (Table 2). In addition to dose escalation, use of ustekinumab was associated with significantly higher costs compared with adalimumab (Table 2). Conversely, use of the other study drugs was associated with significantly lower costs compared with adalimumab (Table 2). Biologic use and presence of IBD-related inpatient admissions in the pre-period were both associated with higher costs in the CD and UC models, whereas older age (per 10-year increment) was associated with lower costs. There was a slight influence of index year, with indexing in 2019 (CD model) or 2019-2020 (UC model) being associated with lower costs. Finally, baseline IBD surgery was associated with slightly lower costs in the CD model, whereas baseline use of immunomodulators was associated with slightly higher costs (Table 2). The ensemble model results were in line with the GLM and identified index drug category as the primary cost driver with an MSE reduction of 52.0% in the CD model and 41.1% in the UC model. Dose escalation (CD: 3.2%, UC: 2.6%) and index year (CD: 1.7%, UC: 2.5%) were identified as moderate cost drivers; all other MSE percent reductions were less than 1% (Supplementary Table 4 (418.2KB, pdf) ).
TABLE 2.
General Linear Model for IBD-Related Health Care Costs in the CD and UC Cohorts
| CD cohort | UC cohort | |||||||
|---|---|---|---|---|---|---|---|---|
| Estimate | 95% CL lower bound | 95% CI upper bound | P value | Estimate | 95% CL lower bound | 95% CI upper bound | P value | |
| Index drug (reference adalimumab) | ||||||||
| Infliximab | 0.485 | 0.466 | 0.505 | <0.001 | 0.463 | 0.439 | 0.488 | <0.001 |
| Ustekinumab | 1.642 | 1.580 | 1.706 | <0.001 | 1.557 | 1.451 | 1.671 | <0.001 |
| Vedolizumab | 0.769 | 0.736 | 0.803 | <0.001 | 0.720 | 0.690 | 0.753 | <0.001 |
| Tofacitinib | 0.604 | 0.557 | 0.656 | <0.001 | ||||
| Dose escalation (reference no escalation) | ||||||||
| Escalation within 6 months | 1.539 | 1.443 | 1.644 | <0.001 | 1.417 | 1.315 | 1.530 | <0.001 |
| Escalation after 6 months | 1.151 | 1.113 | 1.189 | <0.001 | 1.119 | 1.075 | 1.166 | <0.001 |
| Index year (reference 2017) | ||||||||
| 2018 | 0.967 | 0.924 | 1.011 | 0.183 | 0.957 | 0.901 | 1.016 | 0.204 |
| 2019 | 0.932 | 0.891 | 0.975 | 0.003 | 0.922 | 0.869 | 0.978 | 0.011 |
| 2020 | 0.981 | 0.937 | 1.027 | 0.419 | 0.881 | 0.830 | 0.934 | <0.001 |
| 2021 | 0.973 | 0.930 | 1.018 | 0.261 | 0.978 | 0.923 | 1.035 | 0.521 |
| 2022 | 1.025 | 0.970 | 1.084 | 0.405 | 1.011 | 0.947 | 1.080 | 0.781 |
| Biologic use in baseline | 1.090 | 1.056 | 1.125 | <0.001 | 1.095 | 1.053 | 1.139 | <0.001 |
| Age, 10-year increment | 0.986 | 0.976 | 0.996 | 0.007 | 0.982 | 0.970 | 0.995 | 0.008 |
| Sex (female) | 0.982 | 0.955 | 1.010 | 0.250 | 0.973 | 0.940 | 1.006 | 0.166 |
| Mean CCI score (continuous) | 1.013 | 0.998 | 1.028 | 0.118 | 1.001 | 0.983 | 1.020 | 0.932 |
| Immunomodulator use in baseline | 1.046 | 1.017 | 1.076 | 0.003 | 1.021 | 0.977 | 1.067 | 0.455 |
| IBD-related inpatient admission in baseline | 1.101 | 1.058 | 1.146 | <0.001 | 1.148 | 1.089 | 1.210 | <0.001 |
| Surgery in baseline | 0.921 | 0.876 | 0.970 | 0.003 | 1.084 | 0.880 | 1.356 | 0.521 |
CCI = Charlson Comorbidity Index; CD = Crohn disease; IBD = inflammatory bowel disease; UC = ulcerative colitis.
Discussion
This retrospective, claims-based analysis examined dose escalation, biologic treatment characteristics, and IBD-related health care costs among individuals with CD or UC newly initiating adalimumab, infliximab, tofacitinib (UC only), ustekinumab, or vedolizumab. Overall, this analysis identified similar patterns of dose escalation between CD and UC biologic initiators, with approximately 30% of patients evidencing dose escalation. More notable differences in dose escalation were observed across the various biologic groups than between the CD and UC cohorts. This finding indicates that dose escalation may be more closely related to the specific biologic initiated than the underlying condition being treated. Consistent with these findings, analyses around IBD-related health care costs also identified choice in the specific biologic, which differed between the CD and UC cohorts, as the primary cost driver. The presence of dose escalation, especially early dose escalation, was also significantly associated with increased costs. Additionally, analyses identified varying trajectories of dose escalation over follow-up. Initial rates of dose escalation were generally highest among adalimumab-treated patients. However, over the course of treatment, dose escalation in the adalimumab groups stagnated while dose escalation rates in the infliximab, ustekinumab, and vedolizumab groups increased. Trends for tofacitinib in UC largely followed that of adalimumab.
Multiple groups have previously examined dose escalation in IBD populations, and the results of this study are generally in line with those publications.8-12,15-20 US-based studies have reported dose escalation in IBD to range between 8% and 45%; our finding that approximately 30% of patients with CD and UC dose escalated fits in this range.9,10,12,17-20 Further, US studies conducted using administrative claims specifically have reported slightly higher rates of dose escalation ranging from 14% to 45%, again consistent with the results of this analysis.10,12,18-20 Another study examining the impact of biologic dose escalation on costs in patients with UC similarly observed escalation to be associated with increased UC-related costs, notably because of increased drug costs.12 One strength of our analysis is our reporting of dose escalation by biologic in the CD and UC cohorts. Although prior analyses have provided this detailed level of reporting, they have primarily been systematic reviews; thus, their conclusions have likely been affected by differences in methods among the primary studies.11 Our analysis used a single methodology and data source, therefore allowing for more direct examination of rates of dose escalation across various condition cohorts and biologic groups.
Biologic dose escalation is only one method used to try and improve therapy response among patients receiving inadequate benefit from the initial labeled dose; other therapy changes, including biologic switches and augmentation with other medications, are commonly used. To investigate other therapy revisions, rates of discontinuation and subsequent switching were assessed among patients who did or did not escalate their index biologic. As with dose escalation rates, results were similar between the CD and UC cohorts. Nonescalators were more likely to discontinue compared with escalators, suggesting that escalation may be one method to extend use of the index therapy. Among individuals who discontinued, escalators were also more likely than nonescalators to switch. Differences in patterns of discontinuation and switching between escalators and nonescalators may point to inherent differences between groups; specifically, escalators appear more willing to test alternative therapy options compared with nonescalators. It is also possible that prescriber preferences or managed care policies may influence choices in therapy. Further analysis into the characteristics of escalators and nonescalators is warranted to determine how factors like disease severity or treatment history influence the decision to escalate vs switch and their association with long-term health care costs.
Model findings may partially address the role of prior treatment history in dose escalation, as several factors that frequently serve as a proxy for disease severity were associated with an increased risk of dose escalation. Evidence of prior biologic use was a significant predictor of dose escalation in both models, whereas prior immunomodulator use and surgery were risk factors in the CD model and higher CCI scores and the presence of a baseline IBD-related inpatient admission were risk factors in the UC models. All of these factors are generally associated with more severe, or at least longer-tenured, disease. It is possible that the increased risk of dose escalation in biologic-experienced patients is an attempt by patients and their physicians trying to extend the duration of biologic efficacy to reduce switches and cycling through available therapies used to manage these lifelong conditions. Younger age was also associated with an increased risk of dose escalation in both models; although this variable may be expected to be associated with a shorter tenure of disease, it may point to the need to reserve treatments for the future and thus make patients and clinicians more likely to attempt treatment alterations that allow them to stay on the current therapy.
LIMITATIONS
Limitations of this analysis include generalizability stemming from the study sample being composed of individuals with employer-sponsored insurance with at least 24 months of continuous eligibility. The potential for misclassification of CD or UC status and other study outcomes is also present as administrative claims data are collected for billing as opposed to research purposes and thus may be lacking clinical detail or subject to coding errors. Toward this same point, assessment of disease severity or inflammation in claims is limited, as clinical information (eg, laboratory test results, imaging results) are not available and diagnosis information is limited to the level of detail in the ICD-10 taxonomy. This study therefore relied on claims-based proxies (eg, prior biologic use, surgeries) to estimate disease severity. A 12-month baseline period was imposed to help establish new medication initiation; however, it is possible that patients used the index medication prior to that 12-month period. Cox models and GLMs were used to assess dose escalation and IBD-related costs, respectively, while adjusting for differences in demographics and baseline characteristics; however, confounding due to unmeasured variables, such as disease severity, may still be present.
Dose changes and treatment patterns in this study were assessed via administrative claims data, and thus, some assumptions were made. First, it was assumed that patients took medications consistent with fill patterns in the outpatient pharmacy. Early refills to provide patients with a constant supply of medication were also permitted; therefore, outpatient pharmacy fill dates were adjusted using the days supply on the claim to account for early fills. Dose increases were based on dose changes as calculated via the metric quantity, strength, and days supply on pharmacy claims as well as administration dates and estimated dose, based on paid amounts, for medical claims for physician-administered drugs. For these reasons, 2 consecutive claims with a dose change was required. Further, analyses were not able to account for patient weight in the cases of medications with weight-based dosing. It is also worth noting that although biosimilars were included in the analysis, the timing of the study largely predated their use, especially for adalimumab. Increased use of biosimilars would be expected to influence IBD-related health care costs as index biologic costs were identified as one of the primary cost drivers.
Conclusions
Biologic therapy is more common in CD compared with UC, with approximately 30% of patients with CD using biologics vs 12% of patients with UC. However, among biologic users in the United States, patterns of biologic use were similar among cohorts of patients with CD and UC. Dose escalation occurred in approximately 30% of patients over an average of approximately 2 years. The rate and trajectory of dose escalation varied across biologics, with adalimumab-treated patients most likely to escalate early in the first months of therapy but least likely to escalate over the subsequent duration of follow-up (average of 2 years). Patients treated with infliximab, ustekinumab, or vedolizumab escalated later in therapy, leading to overall higher escalation rates. Although this study did not assess clinical outcomes following escalation, escalators demonstrated reduced rates of discontinuation compared with nonescalators, suggesting that escalation may be one method to help extend the duration of biologic therapy. However, this approach is not without drawbacks, as dose escalation was associated with increased IBD-related health care costs. Further investigation into outcomes associated with different treatment modifications, predictors of effective treatment, and associated cost implications for various management strategies are needed to help further optimize the management of IBD in the United States.
Disclosures
Casey Chapman reports consulting fees from AbbVie, Janssen, and Eli Lilly as well as honoraria from AbbVie. Aisha Vadhariya, Nicholas Bires, Thomas L. Creveling, Tommaso Panni, Faye W. Chan-Diehl, and Kim McGinnis are all employees of Eli Lilly and hold stock in the company. Brenna L. Brady, Ryan Ross, and Kristin A. Evans are all employees of Merative who received funding from Eli Lilly to conduct this work.
This study was funded by Eli Lilly; Eli Lilly had input into the initial study design and decision to publish, but no influence on study execution.
Acknowledgments
The authors would like to thank Helen Varker of Merative for her work on the study programming.
All database records are de-identified and fully compliant with United States patient confidentiality requirements, including the Health Insurance Portability and Accountability Act (HIPAA) of 1996. The databases have been evaluated and certified by an independent third party to be in compliance with the HIPAA statistical de-identification standard. The databases were certified to satisfy the conditions set forth in Sections 164.514 (a)-(b)1ii of the HIPAA privacy rule regarding the determination and documentation of statistically de-identified data. Because the study used only de-identified patient records and does not involve the collection, use, or transmittal of individually identifiable data, Institutional Review Board (IRB) approval to conduct this study was not necessary.
The data that support the findings of this study are available from Merative. Restrictions apply to the availability of these data, which were used under license for this study.
References
- 1.Lewis JD, Parlett LE, Jonsson Funk ML, et al. Incidence, prevalence, and racial and ethnic distribution of inflammatory bowel disease in the United States. Gastroenterology. 2023;165(5):1197-205.e2. doi: 10.1053/j.gastro.2023.07.003 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Ramos GP, Papadakis KA. Mechanisms of disease: Inflammatory bowel diseases. Mayo Clin Proc. 2019;94(1):155-65. doi: 10.1016/j.mayocp.2018.09.013 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Mak WY, Zhao M, Ng SC, Burisch J. The epidemiology of inflammatory bowel disease: East meets west. J Gastroenterol Hepatol. 2020;35(3):380-9. doi: 10.1111/jgh.14872 [DOI] [PubMed] [Google Scholar]
- 4.Mao EJ, Hazlewood GS, Kaplan GG, Peyrin-Biroulet L, Ananthakrishnan AN. Systematic review with meta-analysis: Comparative efficacy of immunosuppressants and biologics for reducing hospitalisation and surgery in Crohn’s disease and ulcerative colitis. Aliment Pharmacol Ther. 2017;45(1):3-13. doi: 10.1111/apt.13847 [DOI] [PubMed] [Google Scholar]
- 5.Singh S, Qian AS, Nguyen NH, et al. Trends in U.S. health care spending on inflammatory bowel diseases, 1996-2016. Inflamm Bowel Dis. 2022;28(3):364-72. doi: 10.1093/ibd/izab074 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Farrell RJ. Biologics and beyond anti-TNF agents for ulcerative colitis – Efficacy, safety, and cost? N Engl J Med. 2019;381(13):1279-81. doi: 10.1056/NEJMe1910742 [DOI] [PubMed] [Google Scholar]
- 7.Cholapranee A, Hazlewood GS, Kaplan GG, Peyrin-Biroulet L, Ananthakrishnan AN. Systematic review with meta-analysis: Comparative efficacy of biologics for induction and maintenance of mucosal healing in Crohn’s disease and ulcerative colitis controlled trials. Aliment Pharmacol Ther. 2017;45(10):1291-302. doi: 10.1111/apt.14030 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Gemayel NC, Rizzello E, Atanasov P, Wirth D, Borsi A. Dose escalation and switching of biologics in ulcerative colitis: A systematic literature review in real-world evidence. Curr Med Res Opin. 2019;35(11):1911-23. doi: 10.1080/03007995.2019.1631058 [DOI] [PubMed] [Google Scholar]
- 9.Khan S, Rupniewska E, Neighbors M, Singer D, Chiarappa J, Obando C. Real-world evidence on adherence, persistence, switching and dose escalation with biologics in adult inflammatory bowel disease in the United States: A systematic review. J Clin Pharm Ther. 2019;44(4):495-507. doi: 10.1111/jcpt.12830 [DOI] [PubMed] [Google Scholar]
- 10.Ehrenberg R, Griffith J, Theigs C, McDonald B. Dose escalation assessment among targeted immunomodulators in the management of inflammatory bowl disease. J Manag Care Spec Pharm. 2020;26(6):758-65. doi: 10.18553/jmcp.2020.19388 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Panaccione R, Lee WJ, Clark R, et al. Dose escalation patterns of advanced therapies in Crohn’s disease and ulcerative colitis: A systemic literature review. Adv Ther. 2023;40(5):2051-81. doi: 10.1007/s12325-023-02457-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Long MD, Cohen RD, Smith TW, et al. Retrospective database analysis: Dose escalation and adherence in patients initiating biologics for ulcerative colitis. Dig Dis. 2022;40(5):553-64. doi: 10.1159/000521299 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Chapman TP, Gomes CF, Louis E, Colombel JF, Satsangi J. De-escalation of immunomodulator and biological therapy in inflammatory bowel disease. Lancet Gastroenterol Hepatol. 2020;5(1):63-79. doi: 10.1016/S2468-1253(19)30186-4 [DOI] [PubMed] [Google Scholar]
- 14.Beyrer J, Manjelievskaia J, Bonafede M, et al. Validation of an International Classification of Disease, 10th revision coding adaptation for the Charlson Comorbidity Index in United States healthcare claims data. Pharmacoepidemiol Drug Saf. 2021;30(5):582-93. doi: 10.1002/pds.5204 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Thomas PWA, Smits LJT, Te Groen M, et al. ; IBDREAM. De-escalation of biological therapy in inflammatory bowel disease patients following prior dose escalation. Eur J Gastroenterol Hepatol. 2022;34(5):488-95. doi: 10.1097/MEG.0000000000002336 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Baert F, Glorieus E, Reenaers C, et al. ; BIRD (Belgian IBD Research and Development). Adalimumab dose escalation and dose de-escalation success rate and predictors in a large national cohort of Crohn’s patients. J Crohns Colitis. 2013;7(2):154-60. doi: 10.1016/j.crohns.2012.03.018 [DOI] [PubMed] [Google Scholar]
- 17.Ollech JE, Normatov I, Peleg N, et al. Effectiveness of usetekinumab dose escalation in patients with Crohn’s disease. Clin Gastroenterol Hepatol. 2021;19(1):104-10. doi: 10.1016/j.cgh.2020.02.035 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Teeple A, Sah J, Mallampati R, Adams C, Waters D, Muser E. Persistence, dosing, and other treatment patterns among Crohn’s disease patients initiating biologics in United States. Crohns Colitis 360. 2021;3(4):otab076. doi: 10.1093/crocol/otab076 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Ghosh S, Kathe N, Umashankar K, et al. Dose escalation of biologics in biologic-naïve patients with ulcerative colitis: outcomes from the ODESSA-UC study. Crohns Colitis 360. 2023;5(4):otad061. doi: 10.1093/crocol/otad061 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Cleveland NK, Ghosh S, Kathe N, et al. Dose escalation of biologics in biologic-naive patients with Crohn’s disease: Outcomes from the ODESSA-CD study. J Manag Care Spec Pharm. 2024;30(11):1276-87. doi: 10.18553/jmcp.2024.30.11.1276 [DOI] [PMC free article] [PubMed] [Google Scholar]