Dear Editor,
We thank Katkuri et al. for their careful reading of our article and for their constructive engagement with our work [Reference Pradhan1]. We appreciate their recognition of the importance of adherence research in childhood diarrhoea and of the methodological strengths of our review. We address their specific methodological concerns below.
Inclusion of interventional design
The authors presented their concern regarding the inclusion of a study by Nuzhat et al., described by its authors as a ‘prospective open-label interventional study’ [Reference Nuzhat2]. However, a detailed methodological appraisal indicates that, despite this terminology, the study does not meet the defining criteria of an interventional trial. The study did not involve randomization, allocation to alternative interventions, or the testing of any adherence-enhancing strategies; rather, all participants received zinc as part of routine diarrhoeal case management, with the investigators solely documenting completion of therapy. The outcome of interest was descriptive and behavioural (treatment adherence) rather than comparative efficacy, and no protocol-driven manipulation of exposure was undertaken. In epidemiological terms, this design corresponds to a prospective observational cohort study conducted in a real-world setting instead of an experimental trial [Reference Soumerai, Starr and Majumdar3]. As our meta-analysis synthesized prevalence of adherence, not treatment effects, inclusion of such real-world observational data is consistent with established guidance for prevalence meta-analysis. We analysed proportions rather than effect sizes and conducted sensitivity analyses, which showed that excluding this study did not materially change the pooled estimates. While we did not modify or reinterpret the study design as reported by the original authors, our synthesis was based solely on the observational adherence data presented in the study [Reference Shrier4]. Its inclusion was therefore methodologically appropriate for estimating adherence prevalence and does not compromise the validity or interpretability of our findings.
Use and scoring of the NHLBI Risk-of-Bias Tool
The authors have also commented regarding our application of the National Heart, Lung, and Blood Institute (NHLBI) Quality Assessment Tool. We acknowledge that the tool’s developers caution against creating psychometric summary scores and against assuming equal weighting across domains. Accordingly, our intention was not to use numerical scoring to weight studies or influence pooled estimates, but to provide a transparent overview of methodological limitations [Reference Mamikutty, Aly and Marhazlinda5]. We have explicitly reported the selected domains (D1–D6) and their coding in the Methods section, with each item scored dichotomously (presence of bias = 1; absence of bias = 0). The overall categorization (low risk, some concern, high risk) was based on the cumulative number of domains indicating potential bias, such that higher scores reflected a greater number of identified methodological limitations rather than a psychometric weighting of study quality. Importantly, no study was excluded or down-weighted based on this classification. The guidance site on the NHLBI Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies explicitly states that items not applicable to certain designs (e.g., cross-sectional studies) may be marked as ‘Not Applicable’, and that ‘researchers may determine their own parameters for making judgments’ [6]. Thus, while the tool does not prescribe a mandatory scoring framework, it permits flexible, study-specific application. Our approach, therefore, aligns with the NHLBI guidance and was used solely to contextualize internal validity. Furthermore, as explained earlier, the study described as ‘interventional’ lacks randomization or experimental manipulation and is analytically observational, supporting the appropriateness of domain-level assessment.
High heterogeneity and interpretation of pooled estimates
The authors further highlighted the extremely high heterogeneity (I 2 ≈ 99%), and we explicitly cautioned against over-interpretation of the pooled estimates. The aim of our meta-analysis was not to derive a single ‘definitive’ prevalence of adherence, but to quantify variability and highlight operational challenges in zinc programme implementation across heterogeneous settings. To address this, we conducted separate analyses for 10-day and 14-day regimens, performed leave-one-out sensitivity analyses, and interpreted findings in conjunction with contextual determinants reported in the primary studies [Reference Schroll, Moustgaard and Gøtzsche7]. Our results, therefore, emphasize consistent patterns of suboptimal adherence, particularly for longer regimens rather than reliance on a single summary estimate. This approach is consistent with current methodological guidance that meta-analysis in the presence of substantial heterogeneity can remain informative for hypothesis generation and policy signalling rather than precise estimation [Reference Kang and Choi8].
Variable definitions of adherence
Finally, the authors also mentioned that adherence was treated as a homogeneous construct, despite variability in its definition across studies, including completion of 10-day or 14-day regimens and caregiver-reported duration of use. To address this conceptual heterogeneity, we explicitly stratified analyses by treatment duration, thereby avoiding inappropriate pooling across fundamentally different adherence definitions. Moreover, our narrative synthesis mapped determinants such as early discontinuation due to symptom resolution, vomiting, caregiver perceptions, and provider counselling, factors that reflect distinct behavioural mechanisms underlying non-adherence. To enhance transparency, we have also presented the definitions of diarrhoea and zinc adherence used in the included studies in Appendix D of the original article. Rather than obscuring these differences, our approach highlights the multidimensional nature of adherence in community diarrhoea management [Reference Bentley9].
Conclusion
We appreciate the thoughtful critique offered by Katkuri et al. and agree that adherence research in childhood diarrhoea demands careful methodological consideration. Our study provides the first structured synthesis of adherence patterns, demonstrating consistently suboptimal completion, particularly for 14-day regimens, across diverse LMIC settings. Importantly, by analysing 10-day and 14-day courses separately, our work also seeks to clarify the programmatic ambiguity arising from the WHO recommendation of ‘10–14 days’ of zinc therapy, highlighting how regimen duration itself influences adherence. We believe that, when interpreted appropriately, our findings contribute important evidence for programme design, formulation strategies, and caregiver-provider communication, and we welcome continued scholarly dialogue to strengthen the evidence base for zinc implementation in child health.
Data availability statement
No new data were generated or analysed for this author response. The content is based solely on previously published studies and the methodological interpretation of the authors’ original work.
Author contribution
Conceptualization: S.K.P., A.P., S.P., J.S.K.; Methodology: S.K.P., A.P., J.S.K.; Writing - original draft: S.K.P., J.S.K.; Writing - review & editing: A.P., S.P., J.S.K.
Funding statement
This author response did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
Competing interests
The authors declare that they have no competing interests related to this response.
Ethical standard
Ethical approval was not required for this author response, as it did not involve human participants, primary data collection, or the use of individual-level data.
Declaration of generative AI and AI-assisted technologies in the writing process
During the preparation of this author response, the authors did not use generative AI or AI-assisted technologies for writing, editing, or content development. The authors take full responsibility for the integrity and originality of the content.
Open access
