Authors’ reply to Pentapati et al.

We are grateful to Pentapati et al.[1] for their careful review of our article[2] and for their insightful comments. We acknowledge that the National Family Health Survey (NFHS) data are limited to women below 50 years of age, thus excluding women beyond the reproductive age group who remain at risk for cervical cancer. However, given that cervical cancer incidence in India peaks among women aged 50-60 years, our findings underscore the critical need for the early initiation of cancer screening.[3] We also agree that the NFHS-5 estimates for cervical cancer screening may represent a significant underestimation. This potential underestimation is likely due to the use of the technical term “cervical cancer” in the interview schedules instead of local vernacular terms, which may have reduced accurate ascertainment. We wish to clarify that our analysis utilized Poisson regression with robust standard errors. This approach is appropriate for analyzing binary outcomes in cross-sectional studies when adjusted prevalence ratios (aPRs) are the desired measure of association. We regret that the manuscript text incorrectly stated the use of negative binomial regression (NBR); therefore, we have issued a correction in the form of an erratum to the original article[4] and have also provided an explanation here. We specifically chose the Poisson method because our outcome, cervical cancer screening, had a low prevalence (approximately 2%). In such scenarios, logistic regression, while commonly used, can yield odds ratios (ORs) that may overestimate the true relative effect when interpreted as prevalence ratios, particularly when the outcome is not rare. Although it is mathematically true that the OR approximates the PR when the outcome is rare, as widely acknowledged in the literature, this discrepancy becomes more pronounced with increasing prevalence, potentially leading to exaggerated interpretations of effect size.[5,6] To avoid misinterpretation and provide directly interpretable measures of association, we therefore employed Poisson regression with robust standard errors to estimate aPRs for cervical screening uptake. We appreciate the opportunity to clarify this point and confirm that, despite the inaccurate reference to NBR in the original manuscript, the analytical approach employed was methodologically sound, statistically valid, and appropriate for the nature of the data and the research question addressed. Financial support and sponsorship Nil. Conflicts of interest There are no conflicts of interest.