Commentary: Healthy Eating Index-2020 and bowel habits: a cross-sectional analysis of NHANES

constipating drugs. These steps would help to determine whether the reported diet-bowel habit associations are robust to adjustment for pharmacologic influences.Chen et al. appropriately noted the cross-sectional nature of their analysis [1], but we emphasize the risk of temporal bias (reverse causality) because of the timing of exposure and outcome measurement. Dietary exposure was derived from two non-consecutive 24-hour recalls (HEI averaged from DR1 and DR2), whereas bowel symptoms were queried over the prior 30 days; this mismatch makes it plausible that symptoms preceded and modified short-term intake (for example, increased fiber intake in response to constipation), producing reverse causation or temporally biased associations. Yuan et al. describe how misaligned exposure/outcome windows may bias effect estimates [5], and Savitz & Wellenius discuss limits on causal inference from cross-sectional studies [6]. We therefore recommend that Chen et al. explicitly state exposure/outcome windows in the Methods and add a brief discussion naming reverse causality as a plausible alternative explanation; prospective or longitudinal measurement would help support causal claims.Chen et al. began with the NHANES sample and, after exclusions, arrived at the final analytic N [1]. Specifically, the study started with 31,034 NHANES participants; participants with missing Bowel Health Questionnaire or HEI-2020 data (n = 18,318) and those with inflammatory bowel disease or colorectal cancer (n = 129) were excluded, resulting in a final analytic sample of 11,590. Excluding 19,444 participants (approximately 62.7% of the initial sample) is substantial and raises selection-bias concerns. Although the authors describe missing data handling, they do not present diagnostics comparing included versus excluded participants or sensitivity analyses using principled methods. We recommend (a) a table comparing excluded vs included participants on key sociodemographic and clinical variables, and (b) application of multiple imputation (or combined MI/IPW) with comparison to complete-case results, as multiple imputation can reduce bias relative to listwise deletion in many practical settings [7].Outcome ascertainment relied on self-report questionnaires and the Bristol Stool Form Scale (BSFS) without clinician adjudication, a limitation Chen et al. acknowledge [1]. We reiterate this because studies show only modest concordance between self-reported BSFS and objective stool consistency measures (for example, stool water content) in some clinical groups, implying potential misclassification of constipation/diarrhoea/incontinence [8]. Where feasible, validation