productive. A second important limitation is our inability to evaluate whether the individuals included in our analyses suffered from mild versus severe pain. Although high genetic predisposition for chronic pain may itself be a risk factor for OUD, this concern was partially addressed by our mtCOJO analysis in which we conditioned on pain. That analysis revealed that the association between POU and OUD persevered even after correcting for pain, suggesting that POU was not correlated with OUD via its ability to capture genetic predisposition to pain. Lastly, the 23andme participant base was not ascertained for OUD and is more educated and has a higher socioeconomic status than the broader population; therefore, it is possible that a similar misuse phenotype captured in other higher risk populations may yield different results.
