Response to Mansoor, et al., “Epidemiology of Inflammatory Bowel Disease in men with high risk homosexual activity”

We were interested to read the letter by Mansoor, et al., “Epidemiology of Inflammatory Bowel Disease in men with high risk homosexual activity”[1], which was in response to the two recently published studies by Agrawal [2] and Cha[3]. However, we would like to point out significant methodological flaws in their design and analysis.

The data used have the potential to misclassify both the exposure and outcomes. The authors inferred sexual behaviors from billing codes, which is not included in the best practices outlined in the 2022 National Academies report on this topic[4]. Sexual orientation and sexual behavior may be misclassified or subject to significant selection bias when relying on billing codes. Up to 98% of sexual minority individuals with documentation in clinic notes of sexual orientation and/or sexual practices do not have identifying billing codes in their records[5]. The imbalance in the prevalence of billing codes for “high-risk sexual activity” between men and women is not consistent with data on sexual practices, such as partner concurrency and condom use,[6] indicating gender-based bias in the use of these codes, which could skew results through differential misclassification. The significant associations reported by race/ethnicity for inflammatory bowel disease (IBD) in men who have sex with men (MSM) and men who have sex with women (MSW) also raise concern for race-based bias in coding.

The outcome definition is also inadequate because of poor positive predictive value, inability to accurately subtype (Crohn’s disease (CD) vs. ulcerative colitis (UC)) based on billing codes alone, and potential misuse of IBD billing codes in MSM. Prior studies have shown that the presence of a single billing code for IBD has poor specificity, with some studies reporting positive predictive values of ≥1 billing code for CD as low as 18%[7]. The incidence of both CD and UC billing codes in the same patient record further complicates subtyping[8]. In MSM, there may be initial misdiagnosis of proctitis, with inaccurate assignment of IBD billing codes in this group[9].

The paper also has analytical shortcomings. It relies on a single measure of sexual behavior without considering timing in relation to IBD diagnosis. There is implication in the discussion of a causal relationship, but this cannot be concluded from the data nor the analysis. There is a lack of methodologic rigor and clarity, as demonstrated by results focused on unadjusted comparisons. Key potential confounders like tobacco use, depression, and other IBD risk factors are not included in the analysis, and it is unclear if the estimates presented are age-adjusted.

Lastly, the language used is stigmatizing to sexual and gender minority people. Though previously a technical term, “homosexual” is now considered derogatory based on its historical use as a clinical diagnosis to pathologize sexual and gender minority individuals. Its use is discouraged by advocacy groups like GLAAD (formerly known as the Gay & Lesbian Alliance Against Defamation), restricted by media sources like the Associated Press and New York Times, and has been critiqued by major medical groups like the American Academy of Family Physicians[10].

Research on stigmatized minority individuals has impacts on the safety and wellbeing of people in those communities; researchers have a responsibility to conduct high-quality work that attempts to limit bias through appropriate data selection and methods. We hope that future studies on this topic will be informed by best practices in research design and communication.