Knowledge of uncommon adverse drug reactions ordinarily derives from case reports based on clinical observations. Size limitations of clinical trials preclude identification of rare adverse reactions.1 Yet individual case reports have two problems. First, an observed clinical event may be coincidental to a particular drug exposure, be linked to the disease being treated, or be due to other drugs. Only in the unusual instance when rechallenge and reappearance of the event are described can causality be clearly inferred from a case report. Second, individual case reports cannot generally provide a basis for incidence estimates.1 For prospective clinical decisions, the issue is not whether a drug might cause a reaction, but how often this happens.
Thus, the extraordinary study of the relationship between analgesics and aplastic anemia and agranulocytosis in this issue of The JOURNAL is most welcome and important.2 Because these conditions are rare, a case-control study