Statistics: Detecting a rare adverse drug reaction using spontaneous reports

More than 20,000 patients have been included in studies demonstrating the s afety and efficacy of low-molecular-weight heparins (LMWHs) for postoperati ve venous thromboembolism (1). In May 1993, enoxaparin (Lovenox; Rhone-Poul enc Rorer Pharmaceuticals, Inc., Collegeville, PA) was the first LMWH appro ved by the Food and Drug Administration (FDA) for general use in the U.S. I n December 1997, the FDA issued a public health advisory to alert physician s and others that it had received more than 30 postmarketing reports of pat ients who had developed epidural or spinal hematomas with the use of LMWH a nd spinal or epidural anesthesia or spinal puncture (2). As of April 1998, a total of approximately 40 spinal hematomas were reported (3). The purpose of this report is to describe a statistical approach for identifying adver se drug reactions by using spontaneous reports and to describe the strength s and limitations of statistical analyses by using data from spontaneous ad verse drug reaction reporting systems. The occurrence of spinal hematoma in association with neuraxial blocks in patients who received LMWH is used as an example.