Introduction

Patients with inflammatory bowel disease (IBD) have a 2- to 3-fold increased risk of colorectal cancer death [1, 2]; therefore, colorectal cancer surveillance via colonoscopy is recommended for IBD patients [3]. Despite its potential benefit to reduce colorectal cancer deaths among IBD patients [4, 5], prior studies suggest that the colonoscopy procedure, or bowel preparation, may exacerbate IBD symptoms [6–8]. IBD patients that experience flare-ups following colonoscopy may be less likely to comply with colonoscopy surveillance regimens [9]. Thus, it is imperative to understand the association between colonoscopy and symptom exacerbation in IBD patients.

The frequency of colonoscopy-induced IBD flare-ups and the types of patients that are most susceptible to adverse events following colonoscopy are unknown. It is also unclear if specific interventions, such as medications or mild bowel preparation agents, may reduce IBD flare-ups following colonoscopy. One prior study suggested that immunomodulator medications may reduce the risk of colonoscopy-induced flare-ups in IBD patients, but this study was small and has not yet been replicated [7].

We conducted a study of over 200,000 IBD patients to determine the risk of adverse events associated with colonoscopy and to evaluate the hypothesis that immunomodulator medications may reduce the risk of flare-ups. We also explored differences in the association between colonoscopy and adverse events in IBD patients by demographics, disease characteristics, and bowel preparation type.

Materials and methods

Self-controlled risk interval design

The self-controlled risk-interval design is particularly useful for studying acute outcomes (e.g., ER visit) following transient exposures (e.g., colonoscopy), especially when an appropriate comparison group does not exist [12]. Only those who received a colonoscopy and had an ER visit during the risk and/or control interval are included in analyses. Risk and control intervals are defined in relation to the exposure. The risk interval is the time period in which the outcome is likely to occur if it is related to the exposure. The control interval is the time period that occurs outside of the risk interval that represents the baseline event rate. There is also a run-in interval and wash out period.

We defined the risk interval as the first 4 weeks after the index date of each colonoscopy, and control interval as weeks 7–10 following each colonoscopy. The run-in interval was the 10-weeks preceding each colonoscopy. All colonoscopies with an ER visit during the run-in interval were excluded from the analysis. Such exclusions help to limit the possibility that a severe increase in IBD-related symptoms prompted the colonoscopy. The wash-out period included the time between the risk and control intervals, weeks 5–6 following colonoscopy. If an ER-visit occurred during the wash out period, it was not counted in either the risk or control interval. Fig 1 illustrates the time intervals used in this project in relation to the timing of the colonoscopy.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Diagram of self-controlled risk-interval study design in relation to colonoscopy.

https://doi.org/10.1371/journal.pone.0210262.g001

Results

In the MPCD, there were 227,292 patients with IBD and at least 1 colonoscopy from 2007–2010; among these, 12,456 with prior or prevalent colorectal cancer, 2,522 with prior colectomy, and 109 having > 8 colonoscopies were excluded, resulting in 212,205 eligible patients. Most of the study population was >40 years old (80%), female (57%), Non-Hispanic White (83%), and had ulcerative colitis (60%) (Table 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Baseline demographics and disease characteristics, n (%).

https://doi.org/10.1371/journal.pone.0210262.t001

There were 3,699 patients that had an ER visit during the risk and/or control interval after 1 or more of their colonoscopies, resulting in 4,405 ER visits included in the analyses in Table 2. Overall, risk of an ER visit was higher in the 4-week risk interval following colonoscopy compared to the control interval (RR = 1.24; 95% CI: 1.17–1.32). Sensitivity analyses using the 2-week and 1-week interval following colonoscopy suggested a stronger effect during the time period closest to colonoscopy. The effect was also strongest in those who were ages <41 years old (RR = 1.60; 95% CI: 1.21–2.11), and stronger in women (RR = 1.32; 95% CI: 1.21–1.44) than in men (RR = 1.13; 95% CI: 1.02–1.26). There was no increased risk of ER visits following colonoscopy in IBD patients with Crohn’s disease that involved only the small bowel (RR = 0.98; 95% CI: 0.68, 1.42). IBD patients using immunomodulator maintenance medications prior to their colonoscopy also had no increased risk of ER visits (0.75; 95% CI: 0.35–1.59). Those with a sodium phosphate bowel preparation (e.g., Visicol, OsmoPrep, Fleet Phospho-Soda) had an increased risk of ER visits following colonoscopy (RR = 2.09; 95% CI: 1.02–4.29).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Relative risk of ER visits after colonoscopy comparing risk and control intervals among IBD patients in the MPCD 2007–2010.

https://doi.org/10.1371/journal.pone.0210262.t002

Discussion

Our large analysis of IBD patients suggest that there is an increased risk of ER visits following colonoscopy among IBD patients. One prior study by Menees, et al, included 55 IBD patients and also reported an increased risk of ER visits following colonoscopy [7]. Similar to our results, Menees reported that the increased risk in ER visits was greatest in the 1^st week following colonoscopy and that immunomodulators may reduce the risk of ER visits following colonoscopy in IBD patients. Because of our large sample size, we were also able to conduct stratified analyses which suggested that IBD patients who are young, women, with IBD types that involve the colon (rather than small bowel only), and those who receive sodium phosphate bowel preparations may be particularly high-risk for colonoscopy-induced symptom exacerbation.

The potential mechanism for colonoscopy-induced symptom exacerbation in IBD patients is unclear. Colonic biopsy collection during colonoscopy is standard practice for patients with IBD, and it results in mild trauma to colonic mucosa [13]. This may set off an inflammatory response and trigger IBD flare-ups. Another possible mechanism is that certain bowel preparation agents, including those containing sodium phosphate, can promote inflammation [14], metabolic changes [15], and ulcerative abnormalities [16, 17] in colonic mucosa. These effects could occur in those with, and those without, IBD, but prior studies suggest that IBD patients may be more sensitive to the adverse effects of bowel preparation agents [18] and difficulty with bowel preparation was the most common reason cited by IBD patients for non-adherence to colonoscopy surveillance [13]. Thus, the use of mild bowel preparation agents, and in particular agents that do not contain sodium phosphate, should be further evaluated for IBD patients.

Strengths of this study include: a large sample size which allowed for stratified analyses and precise risk estimates, use of a risk interval design in IBD patients where the general population would not serve as adequate controls, and comprehensive claims data. Despite these strengths, there are several limitations. First, our study is observational, so we did not randomize patients to type of bowel preparation or IBD maintenance medication. Therefore, unmeasured confounding may limit interpretation of these analyses. Nonetheless, our analyses provide preliminary evidence to support the need for future randomized trials to assess the potential role for certain bowel preparations and IBD maintenance medications to mitigate risk of IBD flare-ups associated with colonoscopy. Another study limitation is that the MPCD lacked the granularity necessary to conduct analyses according to reason for colonoscopy or ER visit. Despite this, documentation of ER visit incidence is reliable in claims data, and ER visits following colonoscopy have been used and validated in prior studies of colonoscopy adverse events [7, 11]. Finally, even though our sample size was more than 60 times larger than the only other prior study of the association between colonoscopy and IBD symptom exacerbation, power was limited for some stratified analyses.

Conclusion

In conclusion, our study results suggest that immunomodulators and mild bowel preparation agents may mitigate the risk of colonoscopy-induced IBD flare-ups. These results are promising but warrant further evaluation in clinical trial settings. Given the increased risk of colorectal cancer death among IBD patients [1, 2], additional work to address barriers to colonoscopy surveillance in IBD patients is crucial to improving outcomes for IBD patients.

Acknowledgments

The authors acknowledge the Multi-Payer Claims Database (MPCD) as the source of all data for this study; access to this data was granted by the US Department of Health & Human Services, Office of the Assistant Secretary for Planning and Evaluation, and the Data Use Agreement prohibits the authors from sharing individual-level data to a 3^rd party. However, we have shared data in our manuscript tables that allow for the calculation of the relative risks and confidence intervals that the study conclusions are based on. Also, public data sources consisting of medical claims data, such as the Center for Medicare and Medicaid Services data, could be used to reproduce these analyses.