Response to Predictors of Long-Term Sick Leave In The Workplace.

Abstract

W e thank the Dr Tomoyuki Kawada for his relevant remarks, which give us the opportunity to clarify the objectives and conclusions of our article. Our article does not pretend to consider all the different possible interventions that could impact the occurrence of sick leave spells and therefore does not contradict the studies mentioned by Dr Tomoyuki Kawada. Those examples are in fact very relevant and it is worth recalling them. The main objective of our article is to show the interest of random forest methods in the occupational health context, especially in the context of surveys with a wide range of questions. Sick leaves are indeed determined by many processes and usual statistical methods cannot capture all these effects satisfactorily.