“…Because the outcome was common, and in such situations the odds ratios generated by logistic regression analyses overestimate the relative risk, we used Poisson regression models with log-link function and robust standard errors to calculate relative risks [ 27 ]. Potential risk factors were included in the multivariable model based on univariable analyses, previous studies [ 8 , 19 , 20 ], and clinical relevance, resulting in the variables gender, age at diagnosis, H&N RT, treatment with vinca alkaloids, treatment with epipodophyllotoxins, and the cyclophosphamide equivalent dose (CED) [ 28 ]. Additionally, we evaluated the contribution of other chemotherapy classes (anthracyclines, platinum compounds, and antimetabolites), but those were not included in the final model, as they were not significantly associated with DDDs in univariable analyses.…”