Utilization of Colonoscopy Following Treatment Initiation in U.S. Commercially Insured Patients With Inflammatory Bowel Disease, 2013-2019

Abstract

Background

Evaluation of mucosal healing with colonoscopy is recommended for inflammatory bowel disease (IBD) management; however, little is known about real-world use of treat-to-target monitoring following IBD treatment initiation. We aimed to estimate the proportion of U.S. commercially insured IBD patients who receive colonoscopy in the 3 to 15 months after initiating treatment.

Methods

We identified IBD patients, 18 to 64 years of age, in the IBM MarketScan Commercial Claims and Encounters database as those with ≥3 IBD diagnoses prior to initiation of biologic, small molecule, or immunomodulatory treatment. We excluded patients with prior colectomy and with rheumatologic and other indications for these therapies. Colonoscopies were identified using International Classification of Diseases–Ninth Revision, International Classification of Diseases–Tenth Revision, and Current Procedural Terminology procedure codes. We used Kaplan-Meier methods to estimate the proportion of newly treated IBD patients who received colonoscopy in the 3 to 6 months, 3 to 12 months, and 3 to 15 months following treatment initiation, and stratified trends by year, patient age and sex, and region.

Results

From 2013 to 2019, we identified 39 734 initiators of IBD medications (51.9% female, mean age 39.4 years). We observed similar colonoscopy incidence among ulcerative colitis patients (3-6 months: 14.2% [95% confidence interval (CI), 13.6%-14.8%]; 3-12 months: 37.7% [95% CI, 36.8%-38.6%]; 3-15 months: 46.1% [95% CI, 45.2%-47.1%]) and Crohn’s disease patients (3-6 months: 11.2% [95% CI, 10.8%-11.6%]; 3-12 months: 32.2% [95% CI, 31.5%-32.9%]; 3-15 months: CD: 40.1% [95% CI, 39.3%-40.8%]). Overall colonoscopy use was slightly higher among women, patients in the Northeast, and those initiating newer biologic therapies.

Conclusions

Fewer than half of newly treated IBD patients underwent colonoscopy within 3 to 15 months of initiating new treatment, suggesting low uptake of treat-to-target endoscopic disease monitoring in real-world practice.

KEY MESSAGES.

What is already known? Endoscopic remission is recommended as the primary long-term treatment target in inflammatory bowel disease (IBD) and has been associated with improved long-term outcomes.

What is new here? Fewer than half (42%) of U.S. commercially insured adult IBD patients underwent colonoscopy in the 3 to 15 months after initiating new IBD medication; colonoscopy use did not increase considerably over time, and was higher for patients in the Northeast and in those initiating newer medications (vedolizumab, ustekinumab, and tofacitinib).

How can this study help patient care? We highlight the low real-world use of “gold-standard” endoscopic disease monitoring after IBD treatment initiation and identify factors that may be modifying physician and patient decision making to undergo routine colonoscopy.

Introduction

Over the last 2 decades, a number of effective biologic (infliximab, adalimumab, certolizumab pegol, golimumab, vedolizumab, ustekinumab)^1-5 and small molecule (tofacitinib) medications have been approved to treat inflammatory bowel disease (IBD) in the United States. Yet, approaches to assess treatment response have yet to be standardized. Clinical definitions of remission (normalization of symptoms, physical exam findings, and laboratory parameters) can vary between providers, patients, and visits, and often do not correlate with ongoing, subclinical inflammation. Endoscopic assessment may provide a more objective measure of remission and detect subclinical intestinal inflammation,^6-11 which has been associated with progressive bowel damage and further complications, hospitalizations, surgery, and increased cancer risk.^12-14

In 2015, the STRIDE (Selecting Therapeutic Targets in Inflammatory Bowel Disease) working group recommended in the management of IBD a shift from clinical definitions of remission to a more stringent endpoint requiring resolution of subclinical inflammation (ie, endoscopic remission, as detected by colonoscopy).^15-18 Specifically, these treat-to-target (T2T) approaches recommend routinely evaluating for endoscopic remission within 6 to 9 months of treatment initiation, and escalating therapy in patients who have achieved clinical remission but not endoscopic remission, to prevent long-term complications.^7,19 These recommendations were re-emphasized in a 2020 update (STRIDE-II),²⁰ in which endoscopic assessment for mucosal healing was recommended again as the most important long-term treatment target for reducing the risk of progressive bowel damage.

However, little is known about the degree to which expert-recommended T2T approaches are being implemented to assess mucosal healing for treated IBD patients in real-world practice, or the specific factors that may be driving physicians’ decisions to perform T2T colonoscopy after initiating IBD treatment. To describe the real-world trends in colonoscopy use after initiating treatment with IBD-indicated biologic, small molecule, or immunomodulatory therapy, we aimed to (1) estimate the proportion of patients with Crohn’s disease (CD) and ulcerative colitis (UC) who received colonoscopy after treatment initiation and (2) assess whether colonoscopy use patterns varied by geographic region, calendar year, and patient characteristics.

Methods

Data Source

We conducted a retrospective cohort study using data from the IBM MarketScan Commercial Claims and Encounters (CCAE) database from January 1, 2013, to December 31, 2019. We selected our study window to represent a time period during which all anti-tumor necrosis factor-α (anti-TNF) (infliximab, adalimumab, golimumab, certolizumab pegol) and older therapies (methotrexate, azathioprine, or 6-mercaptopurine [6-MP]) were available on the U.S. market, and captured the period of time when newer therapies (vedolizumab, ustekinumab, and tofacitinib) were first approved for treatment in IBD. The selected study window additionally allowed for assessment of colonoscopy use both before and after publication of STRIDE-I recommendations in 2015.¹⁵

The MarketScan data contain routinely collected administrative healthcare claims data for commercially insured individuals in the United States, their spouses, and dependents,²¹ and provides longitudinal data on inpatient and outpatient encounters, as well as prescription medication dispensings, linked to demographic and enrollment information for over 200 million covered individuals across over 300 contributing employers and health plans.²¹ These data are considered to be representative of the U.S. employer-insured population, and individuals are followed continuously while working for the same employer.

Study Population

We identified all individuals, 18 to 64 years of age, with ≥1 claim for either an anti-TNF biologic (infliximab, adalimumab, golimumab, certolizumab pegol), vedolizumab, ustekinumab, tofacitinib, methotrexate, azathioprine, or 6-MP, identified using National Drug Codes or Healthcare Common Procedure Coding System codes for prescription dispensing or administration, respectively (Supplementary Table 1). To restrict to IBD patients, we included only individuals with ≥3 International Classification of Diseases (Ninth Revision, Tenth Revision) IBD diagnoses (CD: 555.xx, K50.xx; UC: 556.xx, K51.xx), on separate dates, in all available claims history prior to the first observed prescription. We differentiated UC and CD using the diagnosis code that was most proximal to the drug prescription or administration date.

To isolate new users of each medication, we additionally required each eligible medication record to be preceded by ≥6 months of continuous enrollment in the MarketScan database, during which we required no prior use of that specific medication (washout period). However, prior use of other IBD medications was allowed; for example, infliximab initiators were required to be infliximab-naïve in the 6 months prior but could have had prior use of adalimumab during that time. Finally, we excluded patients with prior history of colectomies, as well as previous diagnoses for other major indications of the study drugs during the 6-month baseline period, including rheumatoid arthritis, psoriatic arthritis, other arthritis (including plaque psoriasis, juvenile idiopathic arthritis), and ankylosing spondylitis. Additionally, for new users of methotrexate, azathioprine, and 6-MP, we excluded patients with a prior diagnosis of any cancer, including of acute lymphocytic leukemia and chronic myeloid leukemia, during the 6-month baseline period prior to treatment initiation. We defined each patient’s index date for analysis as the first IBD medication initiation date during the study window that met all eligibility criteria described previously (“first eligible”); each individual was allowed to contribute one index date.

Outcomes

Colonoscopies following index date were identified using a combination of Current Procedural Terminology codes (Supplementary Table 2) and International Classification of Diseases–Ninth Revision and International Classification of Diseases–Tenth Revision procedure codes (Supplementary Table 3), for small bowel, small intestine, large intestine, and anorectal endoscopy; colonoscopy (including video capsule endoscopy); and flexible sigmoidoscopy. We specifically considered colonoscopies that were observed in the 3 to 15 months (91-450 days) after treatment initiation to be potentially delivered for T2T monitoring. We assumed the indication for colonoscopies in the first 90 days after treatment initiation was more likely to be for indications other than T2T monitoring (eg, assessment of ongoing symptoms, colorectal cancer surveillance). Acknowledging the lack of specificity of colonoscopy indication on claims data, we additionally conducted a secondary analysis in which we restricted to IBD-specific colonoscopies, defined as those accompanied by an IBD diagnosis in either the primary or secondary positions.

Secondary outcomes included imaging studies (magnetic resonance enterography [MRE] and computed tomography enterography [CTE]) as well as laboratory tests (noncardiac C-reactive protein [CRP], fecal calprotectin [fCal], and fecal lactoferrin [fLac]) (Supplementary Table 4). Although not considered gold-standard long-term targets for assessment of mucosal healing as recommended under STRIDE-II,²⁰ we evaluated the prevalence of these laboratory and imaging modalities due to their routine use as intermediate assessments of treatment effectiveness. These secondary outcomes were evaluated using the same 91- to 450-day window after treatment initiation.

Statistical Analysis

Eligible patients were followed from index date to the first occurrence of any of the following: (1) first colonoscopy after 90 days of treatment initiation, (2) 15 months following treatment initiation, (3) disenrollment from a health plan included in the MarketScan CCAE database, or (4) study end date (December 31, 2019). Individuals were right-censored at either the disenrollment date or study end date (similar to an intention-to-treat analysis).

We describe characteristics of patients who initiated different IBD medications, including patient sociodemographics (age, sex, geographic region), IBD subtype (CD vs UC), year of treatment initiation, prior use of other IBD medications during the 6-month baseline period prior to index date, and number of patients who received prior colonoscopies (including flexible sigmoidoscopy), hospitalizations, emergency room visits, and gastroenterologist visits during the 6-month baseline period. Finally, we used Kaplan-Meier methods to estimate the proportion, and 95% confidence interval (CI), of newly treated IBD patients who received colonoscopy in the 3 to 6 months, 3 to 12 months, and 3 to 15 months following treatment initiation (ie, the cumulative incidence of colonoscopy at 6, 12, and 15 months following treatment initiation).

Subgroup and Sensitivity Analyses

In addition to overall colonoscopy use, we also described trends in colonoscopy utilization by calendar year of treatment initiation and by geographic region (Northeast, North Central, South, West). To account for changes in the underlying MarketScan patient population over time, we age- and sex-standardized all calendar year trends to the distribution of eligible patients who initiated IBD treatment in 2016 (midpoint of the study window). We also assessed potential differences in colonoscopy use between select patient subgroups, including (1) patients with diagnosed CD vs UC; (2) men vs women; (3) individuals 18 to 24, 25 to 34, 35 to 44, 45 to 54, or 55 to 64 years of age at the time of IBD treatment initiation; and (4) IBD therapy class on index date, categorized as anti-TNFs (infliximab, adalimumab, golimumab, certolizumab pegol), newer biologics (vedolizumab, ustekinumab) and tofacitinib, or older therapies (methotrexate, azathioprine, 6-MP). Finally, we repeated the primary analysis allowing patients to contribute multiple index dates over the course of the study period, provided they met all inclusion and exclusion criteria for each index period (“all eligible”). In this analysis, variance and 95% CIs were estimated using a robust sandwich estimator to account for patients potentially entering the analysis more than once.

Results

Eligible Population

We identified 39 734 eligible IBD patients who initiated treatment between January 1, 2013, and December 31, 2019 (Figure 1). Fifty-two percent of patients were women (Table 1), with mean age of 39.4 ± 13.3 years. Overall, 61% of new users were CD patients and 39% had UC. The regional distribution of IBD treatment initiators largely reflected that of the overall MarketScan population, with about 42% residing in the U.S. South region. During the 6-month baseline period prior to IBD treatment initiation, 55% of patients received ≥1 colonoscopy (including flexible sigmoidoscopy), 22% had at least 1 hospital admission, 28% had ≥1 emergency department visit, and 79% had ≥1 gastroenterologist visit.

Figure 1.

Population flow chart. AML, acute lymphocytic leukemia; CML, chronic myeloid leukemia; IBD, inflammatory bowel disease.

Table 1.

Baseline Characteristics of New Users of Inflammatory Bowel Disease Therapies in the IBM MarketScan Commercial Claims and Encounters Database, 2013-2019

Patient Characteristics	All Patients Initiating IBD Therapies, 2013-2019
Total	39 734 (100%)
Sex
Female	20 619 (51.9)
Male	19 115 (48.1)
Age, y	39.4 ± 13.3
18-24 y	7051 (17.7)
25-34 y	8797 (22.1)
35-44 y	8976 (22.6)
45-54 y	8111 (20.4)
55-64 y	6799 (17.1)
IBD subtype
CD	24 219 (61.0)
UC	15 490 (39.0)
Both	25 (0.1)
IBD diagnoses before index date	6.5 ± 6.1
Index medication
Adalimumab	12 213 (30.7)
Azathioprine	7518 (18.9)
Infliximab	7152 (18.0)
6-Mercaptopurine	5443 (13.7)
Vedolizumab	3546 (8.9)
Methotrexate	2133 (5.4)
Ustekinumab	1498 (3.8)
Certolizumab pegol	928 (2.3)
Golimumab	546 (1.4)
Tofacitinib	258 (0.6)
IBD medications observed in the 6-mo baseline period
Steroids	27 082 (68.2)
Adalimumab	3238 (8.1)
Infliximab	2844 (7.2)
Azathioprine	2813 (7.1)
6-Mercaptopurine	2091 (5.3)
Methotrexate	658 (1.7)
Certolizumab pegol	649 (1.6)
Vedolizumab	421 (1.1)
Ustekinumab	133 (0.3)
Golimumab	115 (0.3)
Tofacitinib	19 (0.05)
Healthcare utilization in the 6-mo baseline period
Patients with ≥1 hospitalization	8690 (21.9)
Patients with ≥1 emergency department visit	11 265 (28.4)
Patients with ≥1 gastroenterologist visit	31 408 (79.0)
Patients with ≥1 colonoscopya	21 773 (54.8)
Days from last colonoscopy to index date among patients with ≥1 colonoscopy	61.6 ± 48.1
Geographic location at treatment initiation
Northeast	7809 (19.7)
North Central	9260 (23.3)
South	16 619 (41.8)
West	5658 (14.2)
Unknown	388 (1.0)
Year of treatment initiation
2013	5114 (12.9)
2014	6911 (17.4)
2015	5323 (13.4)
2016	6388 (16.1)
2017	5284 (13.3)
2018	5261 (13.2)
2019	5453 (13.7)

Abbreviations: CD, Crohn’s disease; IBD, inflammatory bowel disease; UC, ulcerative colitis.

^aIncludes Current Procedural Terminology codes for flexible sigmoidoscopy.

Adalimumab represented almost a third of new use episodes (n = 12 213, 30.7%); azathioprine (n = 7518, 18.9%), infliximab (n = 7152, 18.0%), and 6-MP (n = 5443, 13.7%) accounted for the majority of the remaining new users. Approximately 4% of eligible patients were dispensed either 2 (n = 1475) or 3 (n = 13) different therapies on the same index date. Among eligible patients, about 78% (n = 31 024) received >1 consecutive therapy, with a median of 4 (interquartile range [IQR], 2-9) total consecutive therapy claims observed. When assessing trends by individual therapy, we observed a higher proportion of female certolizumab initiators (67%) and a higher proportion of male methotrexate initiators (52%) (Supplementary Tables 5 and 6). Mean age ranged from 38.0 (infliximab) to 41.3 (tofacitinib) years at index date.

Colonoscopy Use

Overall, 12.4% (95% CI, 12.0%-12.7%) of all IBD patients received a colonoscopy in the 3 to 6 months after treatment initiation; utilization rose to 34.3% (95% CI, 33.8%-34.9%) at 12 months and 42.4% (95% CI, 41.8%-43.0%) at 15 months. Among patients who received colonoscopy, the median time from treatment initiation to first colonoscopy was 232 (IQR, 158-324) days. Colonoscopy use after treatment initiation appeared to be similar for both CD and UC patients (Figure 2). Between 3 to 6 months after treatment initiation, 11.2% (95% CI, 10.8%-11.6%) of newly treated CD patients and 14.2% (95% CI, 13.6%-14.8%) of newly treated UC patients had received colonoscopy. Colonoscopy use increased at 12 months (CD: 32.2% [95% CI, 31.5%-32.9%]; UC: 37.7% [95% CI, 36.8%-38.6%]) and 15 months (CD: 40.1% [95% CI, 39.3%-40.8%]; UC: 46.1% [95% CI, 45.2%-47.1%]) following treatment initiation. Finally, we observed notably lower IBD-specific colonoscopy incidence (33.2% [95% CI, 32.7%-33.8%]) in the 3 to 15 months after treatment initiation; colonoscopy use was again comparable between CD (31.2%) and UC (36.4%) patients by 15 months after treatment initiation (Figure 2).

Figure 2.

Cumulative incidence (Kaplan-Meier) curves of treat-to-target colonoscopy following inflammatory bowel disease (IBD) treatment initiation. CD, Crohn’s disease; CI, confidence interval; UC, ulcerative colitis.

Subgroup and Sensitivity Analyses

Colonoscopy use in the 3 to 15 months following treatment initiation increased only modestly over time (Figure 3), with age- and sex-standardized 15-month colonoscopy incidence ranging from 41.4% in 2014 to 44.5% in 2018. At 15 months following treatment initiation, we observed slightly higher colonoscopy use among female patients (43.1% vs 41.8% among male patients) and patients located in the Northeast (45.7%), relative to other major geographic regions (41.1% to 42.6%), while slightly lower colonoscopy use was observed among patients 18 to 24 years of age at time of treatment initiation (40.8%), compared with older patients (42.3% to 43.1%) (Figure 4). Patterns were similar for 3 to 6 months and 3 to 12 months and are presented in Supplementary Figure 1.

Figure 3.

Age- and sex-standardized calendar-year trends in colonoscopy use in the 3 to 6 months, 3 to 12 months, and 3 to 15 months following inflammatory bowel disease treatment initiation, from the IBM MarketScan Commercial Claims and Encounters Database, 2013-2019. Standardized to the age and sex distributions of eligible patients who initiated treatment in 2016 (study midpoint).

Figure 4.

Proportion of inflammatory bowel disease patients receiving colonoscopy in the 3 to 15 months after treatment initiation, by patient subgroups. F, female; M, male.

Initiation of newer biologics or tofacitinib was most commonly accompanied by colonoscopy (Figure 5) by 15 months (50.6%), followed by initiation of anti-TNF biologics (42.3%), and was lowest in patients initiating methotrexate, azathioprine, or 6-MP (40.1%). When stratified by individual drugs, vedolizumab initiators exhibited the highest colonoscopy use (52.2%) by 15 months; otherwise, colonoscopy use was comparable between individual drugs within each of the previously mentioned strata (data not shown). Finally, when expanding the analysis to all eligible new use periods from 2013 to 2019, we observed similar colonoscopy use trends, with 45.1% (95% CI, 44.5%-45.7%) of all IBD new use periods accompanied by a colonoscopy in the 3 to 15 months after medication initiation. In this analysis, we observed a small increase in the proportion of all-eligible new use periods attributed to vedolizumab (11.3% vs 8.9% in first-eligible analysis) and ustekinumab (5.3% vs 3.8% in first-eligible analysis).

Figure 5.

Proportion of inflammatory bowel disease patients receiving colonoscopy after treatment initiation, by inflammatory bowel disease treatment class initiated, from the IBM MarketScan Commercial Claims and Encounters Database, 2013-2019. 6-MP, 6-mercaptopurine; TNF, tumor necrosis factor α.

Secondary Outcomes

In addition to colonoscopy, we evaluated the use of other modalities for disease activity monitoring following treatment initiation, namely imaging studies (MRE, CTE) and laboratory parameters (noncardiac CRP, fCal, fLac) in the 3 to 15 months following treatment initiation (Supplementary Figure 2). We observed that approximately 19.9% (95% CI, 19.5%-20.4%) of patients received either colonoscopy or imaging in the 3 to 6 months after treatment initiation, which increased to 46.4% (95% CI, 45.8%-47.0%) at 12 months and 54.7% (95% CI, 54.1%-55.3%) at 15 months. Among patients who received either MRE or CTE, median time to first imaging was 208 (IQR, 139-304) days.

When we further considered laboratory parameters, the cumulative incidence of any monitoring modality (colonoscopy or imaging or laboratory test) was observed to be about 42.0% (95% CI, 41.5%-42.6%) in the 3 to 6 months after treatment initiation, 69.9% (95% CI, 69.4%-70.5%) at 12 months, and 76.5% (95% CI, 75.9%-77.0%) at 15 months (Supplementary Figure 2). Among patients who received at least 1 CRP, fCal, or fLac test, median time to first laboratory test was 163 (IQR, 118-246) days. Similar to the primary analysis (colonoscopy only), we observed lower incidence of disease monitoring tests when restricting to only IBD-specific procedures (those with CD or UC diagnosis on the same claim or hospitalization), with 15-month monitoring incidence reaching 42.0% (95% CI, 41.4%-42.6%) when considering colonoscopy or imaging, and 61.8% (95% CI, 61.2%-62.3%) when further adding laboratory tests (Supplementary Figure 2).

Discussion

Among U.S. commercially insured IBD patients, 18 to 64 years of age, who initiated new IBD treatment between 2013 and 2019 (infliximab, adalimumab, certolizumab pegol, golimumab, vedolizumab, ustekinumab, tofacitinib, methotrexate, azathioprine, 6-MP), we observed use of colonoscopy in only about 12% of patients in the 3 to 6 months after treatment initiation. Colonoscopy use increased to 34% at 12 months and to 42% by 15 months after treatment initiation. These findings indicate that fewer than half of treated IBD patients in the United States may be followed under a “gold-standard” T2T monitoring strategy, as recommended by the STRIDE working group.^15,20 Observed colonoscopy use was slightly higher among patients who were women or located in the Northeast (relative to other major geographic regions). Finally, colonoscopy use was slightly lower among patients 18 to 24 years of age (40.8%) compared with older patients (42.3% to 43.1%), although this difference was not considered to be clinically meaningful.

Patients who initiated newer therapies (51%; vedolizumab, ustekinumab, or tofacitinib) exhibited higher observed colonoscopy use within 3 to 15 months, compared with patients who initiated anti-TNF biologics (42%; infliximab, adalimumab, certolizumab pegol, or golimumab) and immunomodulators (40%; methotrexate, azathioprine, or 6-MP). It is possible that immunomodulators were used to treat patients with milder disease, leading to a potential preference for clinical evaluation over close endoscopic monitoring for mucosal healing. Further research should examine the potential for differential medication, and subsequent colonoscopy, use by IBD severity, which is not readily and directly available in administrative claims databases.

When further stratifying by individual drugs, we observed that vedolizumab initiators had the highest observed proportion of colonoscopy use in this period (52%), although vedolizumab initiators accounted for only 9% of all new users. The relatively low proportion of patients initiating vedolizumab may be partly related to our design choice to include only the first-eligible new use period for each patient. When we repeated the analysis to include all-eligible new use periods, we observed similar results, with only slightly increased vedolizumab use (11% vs 9% in primary analysis) and comparable colonoscopy use in the 3 to 15 months after treatment initiation (45% vs 42% in primary analysis).

Current evidence remains sparse regarding the real-world uptake of colonoscopy assessment for mucosal healing in clinical practice. A U.S. study in 2019 by Limketkai et al²² found that over half of IBD patients initiating biologic therapy did not receive any form of mucosal disease activity assessment (endoscopy, fCal, or cross-sectional imaging) in the 12 months after biologic therapy initiation. Outside of the United States, Bryant et al²³ reported lower use of endoscopic targets (47%), compared with traditional clinical markers such as CRP (75%), in a cross-section of 97 UC patients with clinically active disease in South Australia. In the same study, Bryant et al also reported longer time to endoscopy (median 84 [IQR, 29-180] days) compared with CRP (median 13 [IQR, 4-40] days).

Our study extends this prior work by specifically quantifying the low utilization of gold-standard colonoscopy to assess mucosal healing following treatment initiation, as recommended by the STRIDE working group.^15,20 Although fewer than half (42%) of treated IBD patients received a colonoscopy within 3 to 15 months following treatment initiation, a greater proportion of patients were monitored by imaging (55% colonoscopy or imaging) or laboratory tests (77% colonoscopy or imaging or lab). Our findings suggest that (1) a sizeable proportion of those who are receiving routine monitoring may not be receiving STRIDE-recommended endoscopic assessment for mucosal healing and (2) up to a quarter of treated IBD patients may not be receiving any routine monitoring following treatment initiation.

The specific drivers of low uptake of T2T endoscopic monitoring are yet unclear. Our study explored possible differences in colonoscopy delivery by patient subgroups, including age, sex, and region, which have not previously been described in this patient population. Notably, we also observed no major increases in colonoscopy use over calendar time during our study window, which suggests that the STRIDE recommendations may not have drastically changed clinical practice following initial publication in 2015. Further studies are needed to better understand potential provider (and other healthcare system–level) drivers of T2T monitoring uptake. For example, Bryant et al²³ surveyed 61 practicing gastroenterologists about their perceived challenges to implementing T2T in real-world practice and observed a discrepancy between the proportion who were familiar with the T2T concept (80%) and the proportion who considered it relevant to local clinical practice (64%). In that study, duration of clinical practice (10+ years) was associated with lower odds of using T2T approaches for management of IBD (odds ratio, 0.27; 95% CI, 0.05-1.25); no other provider-level drivers were identified. Improved understanding of the possible sources of hesitancy or lack of access to monitoring colonoscopy in future work may help to guide continuing education and training around T2T monitoring guidelines.

Our study had several strengths, key of which is the use of a large administrative claims database of United States commercially insured IBD patients. This allowed for a description of real-world colonoscopy use in a large and geographically diverse population of patients followed in a variety of care settings. We were able to analyze colonoscopy utilization by specific patient subgroups, which has not previously been described in this patient population. Additionally, we used a relatively high-specificity study population definition to identify eligible IBD patients (≥3 IBD diagnosis on separate dates and no diagnoses for other indications of study drugs and treatment initiation determined using a 6-month washout). We believe that this approach minimized inclusion of false positive IBD patients attributed to rule-out diagnoses or treatment for rheumatoid and other autoimmune conditions, and thus yields a study cohort of patients for whom T2T colonoscopy assessment should be most relevant.

The first limitation of the study is the lack of data on important potential sources of health disparities, including race or ethnicity and socioeconomic status, as well as clinical data on IBD phenotype, severity, and symptoms, which may be important determinants or modifiers of colonoscopy use in this population. Moreover, data on factors such as colonoscopy costs and copays, which may play a role in contributing to the low observed adherence to T2T endoscopic monitoring, are only captured among patients who undergo these procedures. This precludes comparisons with patients who do not receive colonoscopy, as these data would not be available for such patients. Second, although we included and identified patients who may have been on combination therapy, we did not analyze these patients as a separate group because the STRIDE guidelines do not currently make distinct recommendations for patients on monotherapy vs combination therapy. Third, the MarketScan CCAE database is limited to patients under 65 years of age, so our findings may not generalize to the older United States adult population (≥65 years of age). Finally, we cannot definitively determine colonoscopy indication in claims data, although in practice an IBD-specific billing code is often deemed less important than the fact that the procedure and mucosal assessment were performed. Additionally, we believe that it is evident that patients without any colonoscopy during this interval are not being followed under an optimal long-term T2T paradigm, and we expect our estimates to represent a potential upper bound of T2T colonoscopy use.

Conclusions

Fewer than half of newly treated IBD patients underwent colonoscopy within 3 to 15 months of initiating a new treatment. These findings indicate low uptake of T2T endoscopic disease monitoring in real-world clinical practice and suggest that endoscopic mucosal healing is not yet a feasible endpoint for real-world comparative effectiveness research in IBD. More data on routine use of these endoscopic assessment endpoints are needed to support future studies regarding the real-world comparative effectiveness of T2T monitoring plans on long-term clinical outcomes.

Funding

This research was funded, in part, by a grant from the National Institutes of Health (T32 DK007634) and by the University of North Carolina at Chapel Hill (UNC) Royster Society of Fellows. The database infrastructure used for this project was funded by the Center for Pharmacoepidemiology, Department of Epidemiology; the UNC Gillings School of Global Public Health; the Cecil G. Sheps Center for Health Services Research at UNC; the CER Strategic Initiative of UNC's Clinical and Translational Science Award (UL1TR001111); and the UNC School of Medicine.

Conflicts of Interest

J.Y.Y. has no conflicts to disclose. J.L.L. has received grants from AbbVie and Roche Genentech to the University of North Carolina, Chapel Hill; and owns stock in GlaxoSmithKline, outside the submitted work. V.P. has no conflicts to disclose. M.D.K. has received grants from Janssen, AbbVie, Pfizer, Takeda, Lilly, Bristol Myers Squibb, Celltrion, Arena Pharmaceuticals, Boehringer Ingelheim, and Roche Genentech; has received personal fees from Janssen, AbbVie, Pfizer, Takeda, and Lilly; and owns stock in Janssen.