Systematic assessment of non‐medical use of prescription drugs using doctor‐shopping indicators: A nation‐wide, repeated cross‐sectional study

Abstract

AimsThe aim of this study was to present the first nation‐wide, systematic, repeated assessment of doctor‐shopping (i.e. visiting multiple physicians to be prescribed the same drug) during 10 years for more than 200 psychoactive prescription drugs in the 67 million inhabitants in France.DesignThis was a nation‐wide, repeated cross‐sectional study.Setting and ParticipantsData are from the French National Health Data System in 2010, 2015 and 2019 for 214 psychoactive prescription drugs (i.e. anaesthetics, analgesics, antiepileptics, anti‐Parkinson drugs, psycholeptics, psychoanaleptics, other nervous system drugs and antihistamines for systemic use).MeasurementsThe detection and quantification of doctor‐shopping relied upon an algorithm that detects overlapping prescriptions from repeated visits to different physicians. We used two doctor‐shopping indicators aggregated at population level for each drug dispensed to more than 5000 patients: (i) the quantity doctor‐shopped, expressed in defined daily doses (DDD), which measures the total quantity doctor‐shopped by the study population for a given drug; and (ii) the proportion doctor‐shopped, expressed as a percentage, which standardizes the quantity doctor‐shopped according to the use level of the drug.FindingsThe analyses included approximately 200 million dispensings to approximately 30 million patients each year. Opioids (e.g. buprenorphine, methadone, morphine, oxycodone and fentanyl), benzodiazepines and non‐benzodiazepine hypnotics (Z‐drugs) (e.g. diazepam, oxazepam, zolpidem and clonazepam) had the highest proportions doctor‐shopped during the study period. In most cases, the proportion and the quantity doctor‐shopped increased for opioids and decreased for benzodiazepines and Z‐drugs. Pregabalin had the sharpest increase in the proportion doctor‐shopped (from 0.28 to 1.40%), in parallel with a sharp increase in the quantity doctor‐shopped (+843%, from 0.7 to 6.6 DDD/100 000 inhabitants/day). Oxycodone had the sharpest increase in the quantity doctor‐shopped (+1000%, from 0.1 to 1.1 DDD/100 000 inhabitants/day), in parallel with a sharp increase in the proportion doctor‐shopped (from 0.71 to 1.41%). Detailed results for all drugs during the study period can be explored interactively at: https://soeiro.gitlab.io/megadose/.ConclusionsIn France, doctor‐shopping occurs for many drugs from many pharmacological classes, and mainly involves opioid maintenance drugs, some opioids analgesics, some benzodiazepines and Z‐drugs and pregabalin.

Use and misuse of psychoactive medicines: a descriptive cross-sectional study in a densely populated region of Portugal

“…The doctor shopping analysis performed in our study has shown that this does not seem to be a cause for concern in ARSLVT. In France the scenario is different, with several studies (Ponté et al, 2018 ; Pradel et al, 2010 ; Soeiro et al, 2023 ) pointing to higher DSI, and for more PMed, than in ARSLVT. In our results, although the DSQ for BZD (14.0 DDD/100,000 inhabitants/day) was almost 2-fold higher than for OA (corrected DSQ = 8.6 DDD/100,000 inhabitants/day), their doctor shopping indicator (0.3%) was one fifth that of OA (1.5%).…”

“…excessive use) of a given medicine. Doctor shopping parameters were calculated and analysed according to the methodology described by several authors (Frauger et al, 2011 , 2016 ; Micallef et al, 2015 ; Ponté et al, 2018 ; Pradel et al, 2004 , 2009 , 2010 ; Soeiro et al, 2023 ), both at INN and at therapeutic group level. A detailed description of the method, that considers the number of overlaps of prescriptions of a given medicine or therapeutic group issued by different prescribers, is provided in Supplemental Material 1 .…”

“…As such, for PMed and therapeutic groups with DSI > 1%, the corrected DSI (DSI c : DSI minus 1%) and the corrected DSQ (DSQ c = Qto t *DSI c ) were estimated. Because the quantities involved in this study are low, results were expressed in DDD/100,000 inhabitants/day, instead of DDD/1000 inhabitants/day (Ponté et al, 2018 ; Soeiro et al, 2023 ).…”