I have the paper and I don't think that the "propensity matching" takes into account the degree of obesity (e.g., BMI), just as it fails to account for the severity of T2D. Thus, there is the possibilty -- indeed, likelihood -- that the patients receiving Ozempic were more obese and that this increases the risk of NAION; you would not jump to give Ozempic to someone with BMI of 25.1, I assume.