Impact of Opioid Use on the Natural History of Inflammatory Bowel Disease: Prospective Longitudinal Follow-up Study

Abstract

Background

Opioid use is increasingly prevalent amongst patients with inflammatory bowel disease (IBD), but whether opioids have deleterious effects, or their use is merely linked with more severe disease, is unclear. We conducted a longitudinal follow-up study examining this issue.

Methods

Data on demographics, gastrointestinal and psychological symptoms, quality of life, and opioid use were recorded at baseline. Data on healthcare use and adverse disease outcomes were obtained from a review of electronic medical records at 12 months. Characteristics at baseline of those using opioids and those who were not were compared, in addition to occurrence of flare, prescription of glucocorticosteroids, treatment escalation, hospitalization, or intestinal resection during the 12 months of follow-up.

Results

Of 1029 eligible participants, 116 (11.3%) were taking opioids at baseline. Medium (odds ratio [OR], 4.67; 95% confidence interval [CI], 1.61-13.6) or high (OR, 8.03; 95% CI, 2.21-29.2) levels of somatoform symptom-reporting and use of antidepressants (OR, 2.54; 95% CI, 1.34-4.84) or glucocorticosteroids (OR, 6.63; 95% CI, 2.26-19.5; P < .01 for all analyses) were independently associated with opioid use. Following multivariate analysis, opioid users were significantly more likely to undergo intestinal resection (hazard ratio,  7.09; 95% CI, 1.63 to 30.9; P = .009), particularly when codeine or dihydrocodeine were excluded (hazard ratio, 42.9; 95% CI, 3.36 to 548; P = .004).

Conclusions

Opioid use in IBD is associated with psychological comorbidity and increased risk of intestinal resection, particularly in stronger formulations. Future studies should stratify the risk of individual opioids, so that robust prescribing algorithms can be developed and assess whether addressing psychological factors in routine IBD care could be an effective opioid avoidance strategy.

Introduction

Crohn’s disease (CD) and ulcerative colitis (UC) are the 2 main types of inflammatory bowel disease (IBD), characterized by recurrent inflammation of the gastrointestinal mucosa. They follow a chronic relapsing and remitting course with symptoms including diarrhea, hematochezia, and urgency. Abdominal pain is experienced by up to 70% of patients during flares of disease activity, which may be linked to underlying mucosal inflammation, extraintestinal manifestations of IBD, or disease complications including strictures, adhesions, or abscess formation.¹

However, persistent abdominal pain is experienced by up to 50% of patients with quiescent IBD, suggesting that a simple linear relationship with disease activity does not explain the underlying etiology of pain in IBD completely.² Psychological factors may contribute to the development of visceral hypersensitivity via complex gut-brain interactions.^2-6 Furthermore, up to 25% of patients with IBD report symptoms compatible with irritable bowel syndrome (IBS), which is also characterized by abdominal pain.^7,8 Pain management in IBD may be complicated by gastrointestinal side effects of many available analgesics.⁹ Nonsteroidal anti-inflammatory drugs can potentiate mucosal injury, and opioids have an array of deleterious effects, including enteric dysmotility, hyperalgesia, and even narcotic bowel syndrome.¹⁰ There are additional concerns that opioids may mask the symptoms of disease activity in IBD, precipitating complications such as toxic megacolon.¹¹

Despite this, data suggest that between 70% and 90% of patients hospitalized due to IBD received opioids during their admission,^12,13 and it has been reported that 1 in 5 patients use them in the outpatient setting.¹⁴ A recent UK-based study reported a 20% increase in the dispensing of opioid prescriptions for patients with IBD between 1990 and 2013.¹⁵ Factors associated with their use include female sex, psychological comorbidity, increased disease severity, and prior gastrointestinal surgery.^12,14-18

In light of the opioid crisis sweeping the United States, concerns are mounting regarding the use of opioids for noncancer related pain.¹⁹ Research into the harmful effects of opioids in patients with IBD remains limited, but there is accumulating evidence that their use in IBD may be associated with reduced quality of life, psychological comorbidity,^16,20 increased healthcare utilization, higher risk of infection,^18,20,21 and premature all-cause mortality.^15,22 However, most studies have been conducted in the United States, which limits generalizability, and few studies detail type of opioid used.^17,22 In addition, opioid use may be associated with more active disease at baseline, making it difficult to determine if these outcomes are a direct consequence of opioids or if opioid use is merely an indicator of more severe disease.¹⁴ We conducted a prospective longitudinal follow-up study of patients in a secondary care setting over 12 months, hypothesizing that use of opioids at baseline would be associated with worse disease outcomes including treatment escalation, hospitalization, intestinal resection, and all-cause mortality.

Methods

Participants and Setting

Ambulatory patients aged 18 years or over with an established diagnosis of CD, UC, or IBD unclassified (IBD-U) according to histological, endoscopic, or radiological criteria seen in the IBD clinic from 2017 to 2020 at Leeds Teaching Hospitals were sent invitations to participate by post. We are the sole provider of care for all these patients through the UK NHS. This study was nested within a longitudinal survey that aimed to examine the impact of the trajectories of symptoms of common mental disorders on the natural history of IBD.²³ The invitation enabled access to an online patient information leaflet and questionnaire, as well as a consent form via online link with a personalized uniform resource locator. Paper versions of these documents were available, if preferred. Patients with any form of stoma or ileo-anal pouch were excluded, due to difficulties in assessing clinical disease activity. We obtained approval for this study from the Wales research ethics committee in February 2020 (REC ref: 20/WA/0044).

Data Collection and Synthesis

Baseline data, including sex, age, marital status, educational level, and lifestyle (tobacco and alcohol consumption) were recorded, in addition to clinical disease activity, psychological health, quality of life, Rome IV IBS-type symptoms, and medication use. We assessed clinical disease using a modified Harvey-Bradshaw index (HBI) for CD, excluding the examination for abdominal mass,^24,25 and the simple clinical colitis activity index (SCCAI) for UC.²⁶ We used a threshold of <5 to indicate clinical remission for both, as recommended.^27,28 We used the hospital anxiety and depression scale (HADS) to screen for anxiety or depression symptoms.²⁹ The combined total score for both anxiety and depression ranges from 0 to 21. Scores are classed as normal (score 0-7), borderline (8-10), or abnormal (≥11), as recommended.²⁹ Somatoform symptom-reporting data were collected using the patient health questionnaire-12 (PHQ-12),³⁰ derived from the PHQ-15.³¹ The total score ranges from 0 to 24, and we categorized these as high (total score ≥13), medium (8-12), low (4-7), or minimal (≤3). The PHQ-12 has been previously validated in other patients with chronic gastrointestinal symptoms.³⁰ In this study, the upper limit of the interquartile range in healthy people reached 13, hence our selection of this cutoff to define high levels of somatoform symptoms. The median value in those with IBS was 8, hence our choice of this to define medium levels of somatoform symptoms. Quality of life was assessed using the short IBD questionnaire (SIBDQ) health survey.³² We screened for IBS-type symptoms using the Rome IV questionnaire,³³ which requires abdominal pain to be present on a weekly basis to meet criteria for Rome IV IBS. Finally, we collected data concerning current use of opioids at baseline, asking patients whether they were regularly using opioids for pain control, as well as current use of antidepressants at baseline.

Electronic medical records were reviewed by one investigator (K.M.F.), who was blinded to the gastrointestinal and psychological symptom data. Inflammatory bowel disease type, extent and location of disease, previous intestinal surgery for IBD, and current medications for IBD (eg, 5-aminosalicylates, immunosuppressants, biologic therapies, or glucocorticosteroids) were collected. This investigator also recorded the occurrence of any of the following clinical outcomes during a minimum of 12 months follow-up, along with the date they occurred: flare of disease activity based on a physician’s global assessment; glucocorticosteroid prescription; escalation of medical therapy due to uncontrolled IBD activity; hospitalization for active IBD; or intestinal resection for active IBD. Changes to therapy that were not for activity of IBD (eg, based on therapeutic drug monitoring) or surgery for perianal CD were not included.

Statistical Analysis

We compared characteristics of those using opioids at baseline with those not using these drugs. For between group comparisons, we used a Pearson’s χ² test for categorical data and an independent samples t test for continuous data. Multivariate logistic regression was performed to assess for factors independently associated with opioid use at baseline, with results expressed as odds ratios (ORs) with 95% confidence intervals (CIs). To assess for an association between opioid use at baseline and each of the disease activity outcomes of interest during longitudinal follow-up, we compared their rates between users and nonusers. We also assessed for an association between opioid use and occurrence of composite outcomes, including one or more of the events of interest. We used a Pearson χ² test for categorical data and univariate Cox regression analysis, with results of the latter expressed as hazard ration (HRs) with 95% CIs. Factors independently associated with occurrence of each of the outcomes of interest were determined by performing multivariate Cox regression analysis, controlling for all baseline characteristics including prior intestinal resection, disease activity at baseline, self-report of a flare at baseline, and presence of Rome IV IBS-type symptoms at baseline; these were reported as HRs with 95% CIs. Due to multiple comparisons, a 2-tailed P value of < .01 was selected to denote statistical significance for all analyses, which were performed using SPSS for Windows version 26.0.

Results

In total, 4823 patients who were seen in the IBD clinic between January 2017 and June 2020 were contacted. Of these, 1119 (23.2%) responded to the baseline questionnaire. However, 88 (7.9%) had a stoma or ileo-anal pouch and were excluded, and 2 more (0.2%) individuals did not provide complete data regarding opioid use at baseline. This meant 1029 (92.0%) of 1119 responders (mean age 52.6 years [SD 16.9 years], 563 [54.7%] female, 459 [44.6%] CD) were included and provided baseline data. In total, 116 (11.3%) patients were taking opioids at baseline. These consisted of codeine in 70 (60.3%) patients, tramadol in 23 (19.8%) patients, dihydrocodeine in 7 (6.0%) patients, oramorph in 5 (4.3%) patients, buprenorphine in 4 (3.4%) patients, fentanyl in 3 (2.6%) patients, morphine in 3 (2.6%) patients, and oxycodone in 1 (0.9%) patient. We included data from all patients in our primary analysis, but given that codeine and dihydrocodeine may be used as antidiarrheal agents by some patients with IBD, we excluded these 2 drugs in a sensitivity analysis.

Characteristics of Patients Using Opioids at Baseline

After univariate analysis, those reporting use of opioids at baseline were significantly more likely to be older, to have CD, to have had a previous intestinal resection, to report a flare at baseline, to exhibit clinical disease activity according to the HBI or SCCAI, to be taking antidepressants, to have abnormal HADS-anxiety or HADS-depression scores, to have higher somatoform symptom-reporting scores, and to meet the Rome IV criteria for IBS (Table 1). They were significantly less likely to have attained university or postgraduate level of education and had significantly lower quality of life scores. After multivariate logistic regression, factors independently associated with use of opioids included use of antidepressants (OR, 2.54; 95% CI, 1.34-4.84; P = .004) or glucocorticosteroids (OR, 6.63; 95% CI, 2.26-19.5; P = .001) and exhibiting medium (OR, 4.67; 95% CI, 1.61-13.6; P = .005) or high (OR, 8.03; 95% CI, 2.21-29.2; P = .002) levels of somatoform symptom-reporting.

Clinical Outcomes of Patients According to Opioid Use

The use of opioids at baseline was significantly associated with intestinal resection during follow-up (HR, 4.44; 95% CI, 1.64-12.0; P = .003; Table 2). After multivariate Cox regression analysis, those using opioids were significantly more likely to undergo intestinal resection (HR, 7.09; 95% CI, 1.63-30.9; P = .009; Table 2 and Figure 1). This association persisted in multivariate Cox regression analysis when patients with a prior history of intestinal resection (HR, 8.16; 95% CI, 1.32-50.3,;P = .24) or patients self-reporting a current flare of disease activity (HR, 23.9; 95% CI, 1.75-326; P = .17) were excluded from the analysis.

Clinical Outcomes of Patients According to Opioid Use Other Than Codeine or Dihydrocodeine

After excluding the 77 patients taking either codeine or dihydrocodeine, the significant association between use of opioids at baseline and intestinal resection (HR, 6.75; 95% CI, 1.88 to 24.2, P = .003) remained on univariate analysis (Table 3). After multivariate analysis, those using opioids other than codeine or dihydrocodeine were significantly more likely to undergo intestinal resection (HR, 42.9; 95% CI, 3.36-548; P = .004; Table 3).

Discussion

This prospective study has examined factors associated with opioid use in patients with IBD and the impact of their use on prognosis using objective markers of disease activity. The extent of opioid exposure is consistent with a recent multicenter audit of UK IBD patients.³⁴ Factors that were independently associated with opioid use included antidepressant or glucocorticosteroid prescription and higher somatoform symptom-reporting scores. Opioid use was significantly associated with intestinal resection during longitudinal follow-up in univariate analysis. After multivariate Cox regression, use of opioids was associated with a 7 times higher risk of intestinal resection. When we excluded patients with a prior history of intestinal resection, patients self-reporting a flare of disease activity at baseline, or patients taking codeine or dihydrocodeine, which may be used as antidiarrheal agents and are classed as weaker opioids by the British National Formulary,³⁵ similar results were observed. After multivariate Cox regression, use of opioids was still associated with intestinal resection during longitudinal follow-up, with a more than 30 times higher risk, although 95% CIs around this estimate were wide due to a smaller sample size, as more than 60% of the patients were taking these drugs.

Our study included a relatively large cohort of patients with IBD. This means the results are likely to be able to generalized to patients seen in other secondary and tertiary IBD centers. The fact that enrollment was predominantly online is likely to have reduced the likelihood of there being missing data among participants. We assessed the impact of opioid use on disease outcomes via electronic medical record review, and this was undertaken by a single blinded investigator, which is likely to have avoided any potential for interobserver variation.

However, there are some limitations of this study. Although we are the sole provider of care for all these patients, we cannot rule out that a small number of patients may have experienced one of the events of interest in another center, and these data would not have been captured by our review of electronic medical records. The sample size was large, and rates of hospitalization and intestinal resection were numerically higher among those using opioids, but only 1 in 4 of those invited chose to participate, and the relatively short study duration meant that event rates for these some of these end points were low. Longer follow-up in this patient group would, therefore, be useful. There were fewer patients taking strong opioids, meaning that making definitive conclusions regarding the impact of these drugs on the prognosis of IBD are inappropriate. We did not assess for other health conditions or extraintestinal manifestations of IBD, which may be related to opioid use, for example inflammatory or degenerative joint disease. Nevertheless, the number of extraintestinal manifestations currently experienced is included within both the HBI and SCCAI, and disease activity at baseline was controlled for in multivariate analysis. Likewise, we did not confirm the indication for opioid use, or the exact drug, from the medical records but relied on patient-report. Nor did we assess whether prescriptions for opioids had been filled or whether patients were adherent. Only 23% of patients we contacted responded to our questionnaire, which may have introduced volunteer bias. The study was nested within a longitudinal follow-up study assessing the impact of trajectories of symptoms of anxiety or depression on the natural history of IBD.²³ However, as the rates of reporting of symptoms of anxiety or depression were similar to those reported in the literature among patients with IBD, we suspect volunteer bias on this basis is unlikely.³⁶ In addition, the impact of opioids on disease end points in IBD was not mentioned at all in the participant information sheet. We therefore feel this is also unlikely to have had a huge impact on our results. We could not assess the characteristics of patients choosing not to take part in the study. This means that we cannot exclude the potential for volunteer bias among patients enrolled, with those using opioids more likely to participate. However, as this was not the main aim of this study,²³ we suspect this is unlikely. Finally, we did not use objective markers of active disease, such as endoscopic assessment, fecal calprotectin, or C-reactive protein, to assess IBD activity at baseline. Instead, we used clinical disease activity scores. As recruitment took place during the COVID-19 pandemic, asking patients to attend a face-to-face appointment to provide stool or blood for analysis was undesirable, and performing an endoscopic assessment in over 1000 patients would have been impractical. This could have led to residual confounding if mucosal inflammation, rather than disease activity scores, is linked to opioid prescribing, although we are not aware of any data to support that. However, this will not have affected our findings during longitudinal follow-up, as these were based on a blinded assessment of medical records.

The prevalence of opioid use among patients in this study was more than double that of the general population in the UK³⁷ and is in line with findings from other studies of patients with IBD.^17,34 In our study, opioid users were 3 times more likely to be taking antidepressants than nonopioid users, which is comparable with other studies examining opioid use in IBD.^13-15 Although a recent meta-analysis by Niccum et al suggested depression is independently associated with opioid use,¹⁷ psychological comorbidity is more prevalent in patients with IBD than among healthy people.³⁶ The relationship between depression or prescribed antidepressants and increased opioid use is, therefore, probably more complex than a simple cause and effect relationship, with opioid use in this subgroup of patients perhaps a marker for a higher burden of chronic pain and coexisting psychological comorbidity.^8,22 In addition, higher levels of somatoform symptom-reporting in patients taking opioids was also demonstrated in our study. This is perhaps unsurprising given that pain is often a symptom of somatoform disorders, is linked with irritable bowel syndrome, which occurs in around 25% of patients with IBD,^7,8 and that other chronic painful conditions often coexist in patients with IBD. Furthermore, healthcare professionals may use antidepressants like amitriptyline to manage irritable bowel syndrome-type symptoms occurring in some patients with IBD. Along with antidepressant use, the higher level of somatoform symptom reporting associated with opioid use in this study underlines that addressing psychological factors in IBD is important, which is not considered by the current management paradigm in most countries.⁴

The 4-times higher risk of intestinal resection in opioid users identified in this study is noteworthy, as intestinal resection is a significant cause of morbidity in IBD,³⁸ and there is limited evidence in the literature relating to this particular association. One study by Wren et al found that long-term use of opioids was associated with more gastrointestinal surgeries in adolescents;¹⁴ and in a retrospective cohort study of adult patients with IBD, those undergoing a colectomy were more likely to have used opioids for >60 days in the preceding year.²⁰ Whether the higher risk of intestinal resection is causally related to opioid exposure or whether patients requiring opioid analgesia represent a more severe phenotype requires further study, although disease activity and prior intestinal resection were controlled for in our multivariate analysis.

Pain management represents a substantial unmet need for patients with IBD; however, pain is associated with lower health-related quality of life and functional disability.^1,6 This makes a blanket recommendation for clinicians to cease the prescriptions of all opioids in patients with IBD unrealistic, particularly in view of the limited alternative analgesics available to manage pain in these patients.⁹ However, altering prescribing practices may be achievable with more detailed knowledge of effects of individual opioids on outcomes in IBD. Studies in non-IBD populations demonstrate that stronger opioid formulations are associated with increased morbidity and mortality compared with weaker formulations.³⁹ In one study that examined the association between weak and strong opioid formulations and mortality in IBD, stronger formulations were associated with an increased mortality.¹⁵ Our study examined the effects of different opioid formulations on specific IBD outcomes, demonstrating that effect sizes for intestinal resection were even higher when individuals using codeine or dihydrocodeine, which may be taken for their antidiarrheal effects, were excluded from the analysis. This suggests that it is other, stronger opioids that are associated with worse disease outcomes in IBD. However, further studies will be needed to replicate our findings. This could help stratify individual risk of these drugs on disease outcomes, allowing more nuanced recommendations on their prescribing to be made.

In summary, the high prevalence of opioid use coupled with its association with increased healthcare utilization and intestinal resection in IBD means that it is imperative that prescribers understand the risks associated with use of these drugs in a potentially vulnerable group of patients. As seen in other studies, we demonstrated that psychological comorbidity was associated with opioid use in patients with IBD, suggesting that addressing psychological factors in routine IBD care is essential going forward. This may be one strategy to reduce opioid prescribing. In addition, there may be a prominent association between worsening disease outcomes and the use of certain stronger opioid formulations, which may be useful to guide future prescribing recommendations and warrants further research.

Abbreviations:

Abbreviations:
 
• CD

  Crohn’s disease

 
• CI

  confidence interval

 
• HADS

  hospital anxiety and depression scale

 
• HBI

  Harvey-Bradshaw index

 
• HR

  hazard ratio

 
• IBD

  inflammatory bowel disease

 
• IBD-U

  inflammatory bowel disease unclassified

 
• IBS

  irritable bowel syndrome

 
• OR

  odds ratio

 
• PHQ

  patient health questionnaire

 
• SCCAI

  simple clinical colitis activity index

 
• UC

  ulcerative colitis

Acknowledgments

Authors are grateful to the patients who took the time to participate in this study.

Author Contributions

K.M.F., D.J.G., and A.C.F. conceived and drafted the study. K.M.F. collected all data. C.R. and A.C.F. analyzed and interpreted the data. C.R. and A.C.F. drafted the manuscript. All authors contributed to and approved the final draft of the manuscript.

Funding

None.

Conflicts of Interest

C.R.: none. K.F.: none. C.S.: has received unrestricted research grants from Warner Chilcott, Janssen, and AbbVie, has provided consultancy to Warner Chilcott, Dr Falk, AbbVie, Takeda, Fresenius Kabi, Galapagos, Arena, and Janssen, and had speaker arrangements with Warner Chilcott, Dr Falk, AbbVie, MSD, Pfizer, Celltrion, and Takeda. D.G.: none. A.F.: none.

References

Author notes