Drug-induced thrombocytopenia: a population study.

<bold>Background: </bold>Drug-induced immune thrombocytopenia, excluding heparin-induced thrombocytopenia, is a rare adverse drug reaction for which the evidence about frequency, relative risk and risk factors mainly originates from case reports and case studies. This study aims to quantify the risk for thrombocytopenia following exposure to drugs that are most often reported to cause thrombocytopenia in the general population.<bold>Methods: </bold>A retrospective, case-control study was conducted within the PHARMO record linkage system. Cases were defined as patients hospitalised for thrombocytopenia in the period 1 January 1990 to 31 December 2002. For each case, up to four controls were matched based on age, sex and geographical area. Exposure on the index date to anticonvulsants, beta-lactam antibacterials, cinchona alkaloids, disease modifying antirheumatic drugs (DMARDs), diuretics, NSAIDs, sulfonamide antibacterials and tuberculostatics was assessed and categorised into mutually exclusive groups of current, recent, past and non-use. The risk was quantified with multivariate conditional logistic regression analysis.<bold>Results: </bold>The study population comprised 705 cases and 2658 controls. Current use of beta-lactam antibacterials was associated with an increased risk for thrombocytopenia (adjusted odds ratio 7.4, 95% CI 1.8, 29.6). Increased risk estimates, although not significant, were found for current exposure to DMARDs and the sulfonamide antibacterial cotrimoxazole (trimethoprim/sulfamethoxazole). No increased risk was found for anticonvulsants, cinchona alkaloids, diuretics, NSAIDs or tuberculostatics.<bold>Conclusion: </bold>More evidence for an increased risk for thrombocytopenia in current use of beta-lactam antibacterials in the general population was provided. The expected increase in risk could not be confirmed for the other drugs investigated, which is possibly a result of the limited statistical power. Future studies including more patients and with laboratory data should confirm our findings before drawing definite conclusions. [ABSTRACT FROM AUTHOR]