Risk of drug-induced agranulocytosis:: the case of calcium dobesilate

Background: In the last 20 years, some cases of agranulocytosis associated with calcium dobesilate consumption in Spain have been reported. A high risk of dobesilate-associated agranulocytosis (121 cases per million per year) calculated using both a case-control and a case-population strategy has been published. However few spontaneous reports have been noted in the same period of time. No explanation exists for this disagreement. Methods: Estimated incidence rates of agranulocytosis in the IAAAS study and the calculated risk of dobesilate-associated agranulocytosis were used as background risks in a Poisson-based methodology, to calculate the number of coincidental reports of agranulocytosis among patients treated with dobesilate. The influence of treatment duration, notification rate and population characteristics were calculated. Results: During the period 1978-2000, a total of 23 cases would have taken place if the background risk of agranulocytosis were 4.7 per million per year (IAAAS's risk); however, only 9 spontaneous cases of agranulocytosis associated to dobesilate were noted. A simulation showed that with notification rates equal to or higher than 17%, it was not possible to exclude that the 9 cases were false-positives. With notification rates equal or inferior to 16%, it would be unlikely that cases of agranulocytosis were noted in this population with a risk of 4.7 per million per year; therefore, it is necessary to assume a higher agranulocytosis risk. More than 1 case per year could be a false-positive if the background risk of agranulocytosis is 9.5 per million per year, this being the appropriate risk for a population of patients older than 60 years. The duration of treatment beyond 30 days increases the probability of a random coincidence of the intake of drug and an agranulocytosis event. Conclusions: The disagreement between calculated dobesilate-associated agranulocytosis risk and the number of noted spontaneous reports may be explained by at least three different factors: under-reporting, duration of treatment and age of patients. It is possible, with the methodology presented, to estimate the influence of these factors to avoid confusion with possible false-positive cases and then to design the correct prospective trial that can provide the true agranulocytosis risk.