Post-Surgical Opioid Initiation and Diverticulitis Risk: a Multi-jurisdictional Population-Based Cohort Study

Background: Opioids may increase the risk of infection of small pouches in the colon (i.e., diverticulitis) via constipation. Previous case-control studies suggested increased diverticulitis risk with opioids, but these findings may have been influenced by protopathic bias and did not estimate the absolute increase in severe diverticulitis events.

Objectives: To compare diverticulitis incidence rates between new users and non-users of opioids after surgery using incidence rate ratios (IRRs) and differences (IRDs) .

Methods: We identified adult patients from administrative health data in the Canadian provinces of Ontario and Manitoba who underwent a variety of surgeries warranting opioid therapy from 2012 to 2020 (Ontario) or 2004 to 2020 (Manitoba). Patients filling prescriptions for opioids within 7 days of their surgery with no opioid prescriptions in the 365 days prior were classified as opioid new users and those who did not fill any prescriptions for opioids within 7-days were classified as opioid non-users. Follow-up started at the end of the 7-day period. We censored new users after 30 days with no opioid supply and non-users when they received opioid prescriptions. We used odds weights based on key demographic and clinical predictors of diverticulitis and elements of the Charlson comorbidity score, combined with inverse probability of censoring weights based on select baseline covariates. We estimated IRRs and IRDs for hospitalized diverticulitis (based on inpatient diagnosis codes) in the weighted populations, comparing opioid new users and non-users. Confidence limits were calculated via bootstrap and combined via random effects meta-analysis. We also estimated IRDs in some surgical subgroups (hip and knee replacement, C-section, excision, hernia, and other).

Results: The Ontario cohort had 625,946 opioid new users and 813,111 non-users. The Manitoba cohort had a similar exposure distribution, with 161,550 new users and 235,933 non-users. After odds weighting, new users and non-users had similar covariate distributions with standardized mean differences for confounders below 0.10. The pooled IRR for diverticulitis was 2.75 (95% CI: 1.52, 4.96) with a pooled IRD of 5.2 per 10,000 person-years (95% CI: 0.9, 9.5). The hip and knee surgery subgroup had the largest IRD among surgery categories (12.0 per 10,000 person-years, 95% CI: 4.7, 19.4).

Conclusions: We observed an increased hospitalized diverticulitis risk with sustained opioid use, though this association may be biased due to time-varying confounders of diverticulitis and continued opioid use. While the absolute magnitude of this association was small compared to other known adverse effects of sustained opioid use, it did increase in some subgroups susceptible to diverticulitis.