This paper reports that warfarin users had lower overall incidence rates of cancer than non-users in a large population-based cohort. The association was present for most cancers.
One major limitations in this analysis, that puts into question the validity of the results, is the lack of adjustment for potential confounders. The reported models seem to only adjust for age and sex. The authors did not consider any other factor that may influence the decision to prescribe or not warfarin. It has been established that frail and sicker individuals are less likely to receive oral anticoagulation with warfarin, even if indicated, and to discontinue treatment (see for example PMID 19151165, 26277091 and 27471198).
In summary, before concluding that a particular medication has a blanket anticancer effect, I would expect a more careful consideration of confounding.
Just a couple of comments. 1. I am somewhat surprised that this is described as a 'case-control study' since it seems to be a cohort study with an exposed group (children from mothers who received a diagnosed of cancer during pregnancy) matched to an unexposed group (children from mothers without a cancer diagnosis), and followed up for outcomes. 2. The results show that, even if the differences were not significant, exposed children had >40% increased risk of low birth weight and higher risk of other outcomes. Based on these data, I do not think the authors have enough evidence to conclude that maternal cancer did not lead to worse outcomes in children.
It is well-known that food-frequency questionnaires (FFQ) are suboptimal tools to assess dietary sodium intake. Thus, not finding an association between FFQ-assessed sodium and the study endpoints is expected, but does not provide any useful information about the health effects of sodium intake.