Commentary on methodological rigor in “Predictors of exclusive breastfeeding: a systematic review and meta-analysis” Authors: Arun Kumar, Ankur Sharma, Saumya Das, Preeti Dnyandeo Sonje, Dhanya Dedeepya Published: 2025 ABSTRACT AbstractKalhor et al. (2025) conducted a systematic review and meta-analysis on predictors of exclusive breastfeeding, offering valuable insights but facing methodological challenges. Despite a robust search strategy, PROSPERO registration, and use of the Newcastle–Ottawa Scale, heterogeneity remained high (I² >85%), with limited exploration through prediction intervals. Combining diverse designs and minimal subgroup justification reduced interpretability. Future reviews should incorporate prediction intervals, sensitivity analyses, and clearer subgroup rationale to enhance methodological transparency and reproducibility. Dear editor, We read with interest the article by Kalhor et al. (2025) entitled “Predictors of Exclusive Breastfeeding: A Systematic Review and Meta-Analysis” [1]. The study provides valuable insight into the multifactorial determinants of exclusive breastfeeding (EBF) by synthesizing evidence across diverse populations. While the review offers an extensive synthesis and follows PRISMA guidelines, several methodological aspects warrant further consideration to enhance interpretability and reproducibility. The authors’ comprehensive search strategy across nine major databases and inclusion of both English and Persian literature are commendable steps toward reducing selection bias. Registering the protocol in PROSPERO strengthens transparency and prospective planning. Moreover, the application of the Newcastle–Ottawa Scale (NOS) for assessing study quality demonstrates methodological diligence. However, the adaptation of NOS for cross-sectional studies remains debated for its limited discrimination power between moderate and high-quality studies; thus, complementary quality appraisal tools such as the JBI checklist may have provided more nuanced evaluation [2]. A critical point relates to the handling and interpretation of heterogeneity. The authors quantified heterogeneity using the I² statistic, but the reported values were exceedingly high (often >85%), suggesting substantial between-study variability. As highlighted by Borenstein (2020), I² reflects the proportion of total variance attributable to heterogeneity rather than the magnitude of effect size dispersion, and thus should not be interpreted as a direct measure of inconsistency [3]. In such contexts, the use of prediction intervals—which estimate the range in which the true effect of a new study is expected to fall—would have provided more clinically meaningful context [4, 5]. Without these intervals, readers may overestimate the precision of pooled odds ratios, especially in the presence of diverse populations and study designs. The choice of the restricted maximum likelihood (REML) random-effects model is statistically sound for highly heterogeneous data. Nonetheless, the study could have benefited from sensitivity analyses comparing REML with alternative estimators (e.g., Knapp–Hartung adjustment or Bayesian hierarchical approaches), as recommended in contemporary meta-analytic [2]. Furthermore, subgroup analyses were briefly mentioned but insufficiently detailed regarding the rationale for grouping or the handling of overlapping covariates, which may have introduced residual confounding. Publication bias assessment through Begg’s and Egger’s tests was appropriate, though the authors’ reliance on symmetrical funnel plots despite small subgroup counts (< 10 studies) might have yielded underpowered results. The article could also have discussed the influence of small-study effects and selective reporting biases, which often accompany social and behavioral outcomes such as breastfeeding practices. Another methodological concern arises from combining heterogeneous study designs (cross-sectional, cohort, descriptive) into unified pooled estimates. While this approach increases sample size, it may compromise causal inference and inflate between-study variance. Distinguishing pooled effects by design type could have mitigated this risk and allowed readers to discern consistency across analytical frameworks. In summary, this meta-analysis makes a substantial contribution by identifying socio-economic and psychosocial predictors of exclusive breastfeeding. However, interpretative caution is advised due to high heterogeneity, limited exploration of prediction intervals, and the aggregation of diverse observational designs. Incorporating more robust heterogeneity modeling, explicit subgroup rationale, and prediction interval reporting in future updates would considerably enhance the study’s methodological transparency and applicability. Acknowledgements None. Author contributions Arun Kumar: Conceptualisation; Writing-Original Draft; Writing-Review & Editing. Ankur Sharma: Conceptualisation; Writing-Original Draft, Writing-Review & Editing. Saumya Das: Writing-Original Draft, Writing-Review & Editing. Preeti Dnyandeo Sonje: Writing-Original Draft, Writing-Review & Editing. Dhanya Dedeepya: Validation, Writing-Review & Editing. Funding No funding. Data availability Not applicable. Declarations Ethics, consent to participate, and consent to publish declarations Not applicable. Approval of the research protocol by an institutional reviewer board Not applicable. Informed consent Not applicable. Registry and the registration No. of the study/trial Not applicable. Animal studies Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement Not applicable.