We appreciate the interest that Lai and colleagues have shown in our population-based case–control study on systemic glucocorticoids and the risk of colorectal cancer.[1, 2] In their letter, they report an increased risk of colorectal cancer among ever users of glucocorticoids [IRR = 1.43 (95% CI: 1.30–1.58)] compared with never users based on Taiwanese insurance claims data. This increased risk persisted in analyses stratified by age and gender.
Although the available information on material and methods in the Lai letter is sparse, several factors may have contributed to the apparent discrepancies between our findings.
The Lai Taiwanese study population also included patients using topical glucocorticoids. These drugs are prescribed for the treatment of skin disorders affecting all age groups. Consequently, the indication for glucocorticoid use may vary between the study populations. In addition, Lai and colleagues did not adjust for underlying comorbid diseases, which may influence their findings (i.e. confounding by indication).
The binary glucocorticoid exposure definition (ever use vs. no use) in the Lai study does not accommodate for differences in routes of administration,[3, 4] duration of use, or cumulative dose. Estimating risk according to these factors would have provided more detailed insight into the reported association. We examined the use of systemic glucocorticoids according to duration of use and cumulative dose. We also investigated exposure to combined glucocorticoids (systemic glucocorticoids combined with inhaled and/or intestinal acting glucocorticoids or the latter two alone) and risk of colorectal cancer. However, our results were consistently close to the null.
In conclusion, uncontrolled confounding may likely explain the Lai Taiwanese findings rather than true causality between glucocorticoids and colorectal cancer incidence.