Goal: Make it possible to spot falsified prescriptions which were delivered and reimbursed.
Method: 1- Use health insurance administration data bases to single patients liable to suspicion of fraud (repeated presentations of the same prescription to several pharmacies the same day) 2- Attribution of a diagnosis of fraud, by the CEIP (Center of Evaluation and Information about Pharmacodependence) based on transferred data, and by the medical department analysis of the complete file, and convocation when applicable 3- Calculation of quantitative performance of the tool.
Results: We have singled 2030 patients. Among them 25 were able to get 553 deliveries out of 288 pharmacies from 114 prescriptions in one year. Our tool specificity is 99.5% and its sensitivity 69.4%.
Conclusion: Creating this tool which can become automatic and used on a large scale, brings a new light in the evaluation, by making hindsight fraud detection possible.