The limitations of prospective observational studies will generally be the same regardless of the topic, there are few controls, a lack of randomization, and me

The limitations of prospective observational studies will generally be the same regardless of the topic, there are few controls, a lack of randomization, and mechanisms of action are only proposed and not directly tested for (in this case the paper doesn't suggest any mechanism of action at all). This paper simply states that there is a correlation with alcohol abstinence and all-cause dementia. That claim alone is probably true, but without randomization or controls dementia can not be directly attributable to alcohol consumption because there could be all types of confounding factors. As an extreme hypothetical imagine if in a fictional society only the wealthiest .01% could afford alcohol, nearly any observational study will show better health outcomes for alcohol drinkers if you don't control for wealth. In reality observational studies have to control for many, many more factors to be considered strong evidence of causation. From the paper: "It is difficult to provide evidence of causality in observational research, but triangulation of evidence from alternative study designs and across different research settings has been identified as a promising avenue ... Among current older drinkers, the current study failed to find a consistent relationship between alcohol use and dementia across different contexts, some of which are likely to be subject to quite different confounding structures Meanwhile, Mendelian randomisation studies to date have generally failed to demonstrate a causal relationship between alcohol use and dementia, except perhaps with respect to age of onset for dementia. While there are limitations associated with the current study, as well as the studies based on Mendelian randomisation, strong evidence for the causal link between alcohol use and dementia remains elusive."