It turns out we in fact have tests to determine statistical significance - and the fact that this study was peer reviewed and published means the results were indeed statistically significant!

> We included 1,781 participants (mean age 74.9 ± 12.5 years, mean education 4.8 ± 4.0 years, 49.6% women, and 64.1% White). Compared with participants who never consumed alcohol, moderate (odds ratio [OR] 1.60, 95% CI 1.19–2.15, p = 0.001), heavy (OR 2.33, 95% CI 1.50–3.63, p < 0.001), and former heavy (OR 1.89, 95% CI 1.41–2.54, p < 0.001) alcohol consumptions were associated with hyaline arteriolosclerosis while only heavy (OR 1.41, 95% CI 1.10–2.30, p = 0.012) and former heavy (OR 1.31, 95% CI 1.02–1.68, p = 0.029) alcohol consumptions were associated with neurofibrillary tangles. Former heavy drinking was associated with a lower brain mass ratio (β −4.45, 95% CI −8.55 to −0.35, p = 0.033) and worse cognitive abilities (β 1.31, 95% CI 0.54–2.09, p < 0.001). The association between impaired cognitive abilities and alcohol consumption was fully mediated by hyaline arteriolosclerosis (β 0.13, 95% CI 0.02–0.22, p = 0.012).

Being statistically significant is about how good your measurement is. (In one particular way.)

It's entirely unrelated to being practically significant.