But many studies do account for that scenario of "sick quitters". The statistical correlation still showed that light moderate drinkers had lower mortality rates.
The study itself may still be flawed but the point is that the researchers were aware of "sick quitters" being categorized as "non-drinkers" and thus distorting the results.
Excerpt from 90+ Study publication https://pubmed.ncbi.nlm.nih.gov/17350977/ :
>The greater proportion of surviving participants in our study who reported not drinking at follow-up compared to baseline is probably due to a combination of the cohort’s becoming older and drinking less, secular changes in alcohol consumption with time, and survival selection effects. Decline in consumption is also associated with the development of ill health, particularly cardiovascular disease. This accumulation of ill health and higher mortality risk in ‘sick quitters’ increases the observed benefit in drinkers if non-drinkers include these ex-drinkers [6, 24, 26]. In addition, regular light drinkers tend to have characteristics advantageous to health [26, 27]. Thus, controlling for baseline disease is important, particularly if alcohol history is unknown. With adjustment for baseline medical history as well as body mass index, exercise, and smoking, the lower risks of death among light and moderate alcohol drinkers observed in our study did not change substantially and remained statistically significant although those associated with heavier drinking were reduced.