Low rates of agreement between algorithms and GI were found (average 43%). Total agreement was not found for any algorithm. None of the adverse events was imputed equally by all the algorithms. Algorithms presented high rates of sensitivity (average of 93%) and low rates of specificity (average of 7%).