before we go overboard.... i agree there is some need for monitoring, but...
it appears that this retrospective study is almost to the point of data mining.
the incidence of hypoglycemia in the patients who were given tramadol, requiring hospitalization, is 0.7 per 1000 person years. fatal overdose was 10% - but that would mean 0.07 per 1000 person years (a number that is not mentioned).
the article states that the groups are equal with all respects except recent surgery - yet 34% of the tramadol patients were on insulin, and 1.3% of non-tramadol, and 27.5% of tramadol patients were on metformin, compared to 3.4% of non-tramadol, and 28.8% were on sulfonylureas with tramadol vs 2.7% for non-tramadol. how are these groups equal? (this is from the Table 1 provided).