Reply by Authors.

Abstract

You have accessJournal of UrologyLetters to the Editor/Errata1 Oct 2020Reply by Authorsis a reply to letterRe: Delayed Radical Prostatectomy is Not Associated with Adverse Oncological Outcomes: Implications for Men Experiencing Surgical Delay Due to the COVID-19 Pandemic K. B. Ginsburg, G. L. Curtis, R. E. Timar, A. K. George, and M. L. Cher K. B. GinsburgK. B. Ginsburg More articles by this author , G. L. CurtisG. L. Curtis More articles by this author , R. E. TimarR. E. Timar More articles by this author , A. K. GeorgeA. K. George More articles by this author , and M. L. CherM. L. Cher More articles by this author View All Author Informationhttps://doi.org/10.1097/JU.0000000000001193AboutPDF ToolsAdd to favoritesDownload CitationsTrack CitationsPermissionsReprints ShareFacebookTwitterLinked InEmail Rosen et al raise important statistical considerations that we indeed took into account while designing this analysis and interrogating the models included in the article. Regarding collinearity, as suggested, a pairwise correlation matrix was used to assess for correlation of variables included in the models. The risk of collinearity rises with a correlation coefficient above 0.8. None of the combinations of variables included in our models approached this threshold, thus reducing the risk. Rosen et al also cite a very useful publication by Bujang et al4 regarding general rules for logistic regression models while raising the issue of overfitting the models.4 They mention that 10 to 15 events are needed per independent variable included in the model and an estimated sample size of 100 + 50 × (number of independent variables) patients to support an adequately powered logistic regression model. According to this formula, the minimum number of patients needed for a model with 37 predictors would be 1,950, with approximately 370 to 555 events. The sample size afforded to us by the National Cancer Database allowed for an adequately powered analysis of all of the models presented in the article, as our smallest subgroup contained more than 17,000 patients with more than 2,000 events. Regarding the relationship between the other predictors in the models and the outcomes of interest, we purposely omitted this information for brevity. We presented a large volume of data, and we did not want to distract from the relationship between treatment delay and oncologic outcomes by creating multiple lengthy tables including the ORs, 95% CIs and p values for all of the other covariates. We believe the data presented in our study are congruent with previously published reports and are intuitive regarding what we know about the slow natural history of prostate cancer. A common mantra of many academics, which is credited to George Box, comes to mind: “All models are wrong, but some are useful.”5 Acknowledging the limitations of these data, we believe our work may be beneficial to patients and clinicians alike while discussing the implications of unintentional treatment delay and optimal timing for treatment of prostate cancer amid the current COVID-19 environment and waves or surges in the future. References 4. : Sample size guidelines for logistic regression from observational studies with large population: emphasis on the accuracy between statistics and parameters based on real life clinical data.Malays J Med Sci 2018; 25: 122. Google Scholar 5. : Science and statistics. J Am Stat Assoc 1976; 71: 791. Google Scholar © 2020 by American Urological Association Education and Research, Inc.FiguresReferencesRelatedDetailsRelated articlesJournal of Urology1 Jul 2020Re: Delayed Radical Prostatectomy is Not Associated with Adverse Oncological Outcomes: Implications for Men Experiencing Surgical Delay Due to the COVID-19 Pandemic Volume 204Issue 4October 2020Page: 845-845 Advertisement Copyright & Permissions© 2020 by American Urological Association Education and Research, Inc.MetricsAuthor Information K. B. Ginsburg More articles by this author G. L. Curtis More articles by this author R. E. Timar More articles by this author A. K. George More articles by this author M. L. Cher More articles by this author Expand All Advertisement Advertisement PDF downloadLoading ...