Dear Editor
Beaufort and colleagues found that acetic acid chromoendoscopy (AAC) increased suspicious-lesion recognition without improving Barrett’s neoplasia detection. Their publicly shared statistical code raises an important question about the denominator for targeted-biopsy yield.¹,²
Table 3 reports an adjusted odds ratio of 1.05 (95% confidence interval 0.49 to 2.24) among 2267 endoscopies. However, the shared script defines the targeted-biopsy outcome, ‘ndrtar’, as missing unless a targeted-lesion grade is recorded. The subsequent mixed-effects logistic regression therefore retains 196 patients, not 2267. We reconstructed the published cohort from the deposited dataset: 95 AAC examinations and 101 controls had a coded targeted-lesion outcome, with neoplasia identified in 14 and 29, respectively. Fitting the stated model to these 196 patients yielded an adjusted odds ratio of approximately 0.35 (0.16 to 0.80). Conversely, coding endoscopies without targeted detection as zero restored the 2267-person denominator and reproduced the published estimate: 1.05 (0.49 to 2.25).
These estimates address distinct questions. Per-endoscopy targeted detection was 2.1% with AAC versus 1.8% without it, whereas neoplasia among examinations with suspicious lesions was 14.7% versus 28.7%. Because AAC increased ascertainment of suspicious lesions (14.0% versus 6.4%), the conditional comparison is subject to post-intervention selection and should not be interpreted as a randomised treatment effect. Nevertheless, it is informative about the diagnostic workload generated by additional suspicious findings.
We ask the authors to provide an executable script consistent with Table 3 and to distinguish per-endoscopy detection from conditional positive yield. This clarification would strengthen interpretation of AAC’s implementation burden without altering the study’s primary null finding.
References
1. Beaufort IN, Boer LS, Milne AN, et al. Acetic acid chromoendoscopy for detection of neoplastic Barrett’s oesophagus: stepped wedge cluster randomised controlled trial (ACID-trial). BMJ. 2026;395:e100872. doi:10.1136/bmj-2026-100872.
2. Beaufort I, Boer L, Milne A, et al. Data and code from: Acetic acid chromoendoscopy for the detection of neoplastic Barrett esophagus: a stepped wedge cluster randomized controlled trial (ACID-trial) [dataset]. Dryad. 2026. doi:10.5061/dryad.cfxpnvxp2.
Two denominators for targeted-biopsy yield in the ACID trial