According to a study published on the pages of Clinical Pharmacology and Therapeutics, a "mining" activity of clinical notes, written in free format by doctors on electronic prescriptions, allows the first signs of adverse effects of therapies to be identified. They are authors Paea LePendu and a group of researchers from Stanford University in California, who point out that other studies had already analyzed the information contained in prescriptions registered in electronic format. «However – they write – the previous searches had mainly taken into consideration the codified diagnoses, the notes of resignation or the information for the insurance companies. However, some experts believe that more than 90% of the adverse effects that are actually recorded may be missing from the encoded data. Free-form text could instead play an important role in pharmacovigilance, "especially - the authors continue - if we are able to transform the notes written daily by doctors, nurses and other professionals into data that is easier to identify for data mining programs ». With their analysis, the California researchers were able to reproduce the known association between rofecoxib and myocardial infarction, calculating an odds ratio of 1.31, whereas a search carried out on the coded data had failed to establish any correlation. Overall, the method developed by LePendu and his colleagues was able to identify an association t! between a single drug and its adverse effects with an Auc (area under the curve) of 75.3%, which rose to 80.4% after adjusting for confounders such as age, gender, ethnicity, or the presence of comorbidity. According to the authors, of nine drugs on which there had been an alarm from the FDA and for which sufficient data was available, in six cases the method would have identified adverse effects before official reporting.
April 12, 2013 – DoctorNews33
