Detection and incidence of drug-induced agranulocytosis in hospital: a prospective analysis from laboratory signals

AIMS: Our objectives were to assess the detection and incidence of drug-induced agranulocytosis in two university hospitals using hematology laboratory data. METHODS: A prospective study was undertaken at Toulouse University Hospital (France) and Navarra University Hospital (Spain) for 1 year (from 1 May 2004 to 30 April 2005). Using a computerized process and hematology laboratory data, all neutrophil counts with a value less than 500/mm(3) were registered, allowing identification of inpatients suffering from agranulocytosis during the period of the study. Medical records of all selected patients were then consulted. Cytostatic drugs were excluded from this study. RESULTS: During the period of the study, 225,659 neutrophil counts were performed in both hospitals, of which 2,835 (1.26%) had a neutrophil count less than 500/mm(3), corresponding to 739 patients. Seventeen patients were excluded because of lack of data, and 20 cases of infants younger than 3 months were excluded. Among the remaining patients (n = 702), 23 cases of drug-induced agranulocytosis (excluding cytostatic drugs) were suspected. All cases were classified as "serious" since they led to death in 2 cases, hospitalization or prolongation of hospitalization in 19 cases and threatening of vital prognosis in 2 cases. Withdrawal of suspected drugs was done in all cases with regression of neutropenia in 21 cases. According to hospitalization data, the annual incidence of drug-induced agranulocytosis was 1.62 (1.0-2.6) per 10,000 inpatients in Toulouse University hospital (based on 534 cases) and 3.24 (0.9-8.3) per 10,000 inpatients in Navarra University Hospital (based on 168 cases). The involved drugs were mainly antibacterial (30.4%), immunosuppressive (17.4%), antithyroid (13.0%), antiplatelet (8.7%) and nonsteroidal anti-inflammatory (8.7%) ones. Only seven cases from Toulouse University Hospital were spontaneously reported by physicians during the same period. Thus, the underreporting coefficient (U) was 2.71 (63.2%) in France. CONCLUSION: Our survey allowed us to identify the suspected drug-induced agranulocytosis through a prospective study in a large sample of inpatients using only laboratory data analysis. We also note an important underreporting rate of this serious adverse drug reaction (ADR) to the official French pharmacovigilance system. Laboratory data analysis could be used for identifying serious ADRs.