Performance of the standardised MedDRA® queries for case retrieval in the French spontaneous reporting database.

OBJECTIVE

The objective of this study was to evaluate the performance of Standardised MedDRA® Queries (SMQs) in adverse drug reaction (ADR) identification.

METHODS

ADR cases included in the last complete year of the French Pharmacovigilance database for research were used to test four SMQs (narrow and broad): agranulocytosis, demyelination, osteonecrosis and psychosis. Reference cases were identified by free-text search and validated by two authors. Sensitivity (Se), specificity (Sp), positive predictive value (PPV) and negative predictive value (NPV) of narrow and broad searches of each SMQ were calculated and reported as proportions with 95% exact confidence interval (CI).

RESULTS

Among 20,830 cases reported in 2009, 337 validated cases of agranulocytosis, 17 of demyelination, 52 of osteonecrosis and 230 of psychosis were included in the reference sets. Estimations of SMQ narrow search performance were as follows: Se 62.9% (95% CI 57.5-68.1) and PPV 46.8% (95% CI 42.1-51.5) for agranulocytosis; Se 88.2% (95% CI 63.6-98.5) and PPV 34.1% (95% CI 20.5-49.9) for demyelination; Se 94.2% (95% CI 84.1-98.8) and PPV 74.2% (95% CI 62.0-84.2) for osteonecrosis; and Se 75.1% (95% CI 69.0-80.6) and PPV 87.8% (95% CI 82.3-92.0) for psychosis. Results obtained using the broad search were similar except for PPV concerning osteonecrosis (52.7% [95% CI 42.1-63.1]) and psychosis (61.4% [95% CI 55.7-66.8]). For all selected SMQs, Sp and NPV were greater than 98% for both narrow and broad searches.

CONCLUSION

Heterogeneous performance of SMQs for case retrieval was found and seems to be related to the characteristics of the condition of interest. It could therefore be expected that for other SMQs, performance may be affected in the same manner.