highly significantp ≤ 0.001
We deemed predictors as promising if they were not amenable to meta-analysis but single studies reported a highly significant ( p ≤ 0.001) and large association (OR ≥ 2.0 or OR ≤ 0.5) with a study population of at least 1000 patients. 34 To avoid overestimating associations, we imputed an OR of 1 for predictors that were excluded in adjusted analyses because of nonsignificance. 34 , 35 We calculated the absolute risk for each predictor amenable to meta-analysis to facilitate interpretation using baseline risks of 1 in 1000 23 for fatal overdose and 2 in 1000 24 for nonfatal overdose.