Authors’ reply to Labos and Brophy, Kerr, Hill, Hettinga, Sundar, and Bonneux

We thank all authors for their views on our article and would like to clarify some points.1

Labos and Brophy suggest that our results are a consequence of reverse causation.2 We discussed this possibility in our article. Our analyses found no association between higher milk consumption and a family history of osteoporotic fracture. Moreover, in women, we examined whether a fracture or a diagnosis of a disease captured by Charlson’s comorbidity index between the first food frequency questionnaire and the second influenced the reporting of dairy products. We found no such indication. Interaction analysis by sex would not be meaningful given the different study designs of the male and female cohorts. In addition, our response to Bonneux below provides explanations for sex differences in our study.

In response to Kerr, few people in Sweden use raw milk because pasteurisation of commercially sold milk has been mandatory since 1937.3 We could therefore not compare risk estimates for raw and pasteurised milk. The fat content of milk did not seem to explain our associations. In fact, we found higher rates of mortality with low fat …