Detection of adverse drug reactions using hospital databases-a nationwide study in Portugal.

PURPOSE: This study aimed to detect and characterize adverse drug reactions (ADRs) that occurred during hospitalization (ADRIn ) and ADRs associated with admission (ADRAd ) in Portugal from 2000 to 2009. We also intended to compare the results of this methodology with spontaneous reporting.
METHODS: We conducted a nationwide study using a hospital administrative database that included all acute care public hospitalizations in Portugal, from 2000 to 2009. We used International Classification of Diseases-9(th) Revision-Clinical Modification coding data for the detection of ADRs. Codes searched included "E" codes (E930 to E949.9, codes that exclude poisonings and errors) and five groups of diagnoses codes associated with high prevalence of ADR as found in a previous study: hypoglycemia, drug-induced neutropenia, hepatitis unspecified, anaphylactic shock due to drugs, and shock due to anesthesia.
RESULTS: From 9 271 122 hospitalizations within that period, 116 720 ADRs were detected through the database methodology, representing 1.26% from all hospitalizations. Of the ADRs, 97.3% occurred during hospitalization (ADRIn ), whereas 2.7% were associated with admission. Age, female sex, and comorbidities such as pneumonia, heart failure, diabetes, and malignancies were associated with ADRs (all with differences statistically significant). There were 13 562 spontaneous reports from 2000 to 2009.
CONCLUSIONS: Several methods have been used for the detection of ADRs, but they are difficult to apply at a national level. Spontaneous reporting is widely used but grossly underestimates the frequency of ADRs. The database methodology can be very useful to estimate ADRs frequency and to perform a simple characterization of ADRs nationwide.