Until a few years ago, moderate alcohol consumption was thought to have (mild) beneficial effects on health. However, some recent studies have suggested that “there is no safe level” of alcohol intake. Consequently, public health institutions have responded by advising against any level of alcohol use and suggesting governments a number of policies to reduce the overall alcohol consumption. Nonetheless, medical studies suffer from a variety of methodological issues that could undermine the reliability of the findings, especially when focusing on low-intake levels. We apply a search algorithm to extract 19,981 Confidence Intervals (CIs) from 6,763 medical abstracts, and show the existence of a clear publication bias which appears to have even increased and not decreased, in recent years. Further, we assess the quality of a sample of articles, showing the presence of several limitations such as omitted variable bias, miscalculation of alcohol intake, use of linear in place of non-linear models, and lack of validation of Mendelian Randomization (MR) assumptions. We conclude that the methodological limitations of the literature preclude us from claiming that ”there is no safe level” of alcohol intake.
 

---

Download from REPEC

Download from SSRN

---