Margaux Lafaurie1, Liselotte Pochard1, Clémence Lotiron1, Laurent Molinier2,3, Maryse Lapeyre-Mestre1,2, Emilie Jouanjus4,5. 1. CEIP-Addictovigilance, Service de Pharmacologie Médicale et Clinique, CHU Toulouse, 37 allées Jules Guesde, 31000, Toulouse, France. 2. UMR1027 Inserm-Université Paul Sabatier Toulouse III, 37 allées Jules Guesde, 31000, Toulouse, France. 3. Département d'Information Médicale, Hôtel-Dieu Saint-Jacques, CHU Toulouse, 2 rue Viguerie, 31059, Toulouse, France. 4. CEIP-Addictovigilance, Service de Pharmacologie Médicale et Clinique, CHU Toulouse, 37 allées Jules Guesde, 31000, Toulouse, France. emilie.jouanjus@univ-tlse3.fr. 5. UMR1027 Inserm-Université Paul Sabatier Toulouse III, 37 allées Jules Guesde, 31000, Toulouse, France. emilie.jouanjus@univ-tlse3.fr.
Abstract
BACKGROUND AND OBJECTIVE: Studies have explored hospital records to identify serious complications related to use of psychoactive drugs, but this approach is time consuming with a high rate of false positives. We propose a method to improve the detection of these somatic complications from an inpatient database. METHODS: Hospitalisations in Toulouse University Hospital (France) between 1 July and 31 December 2013 with at least one International Classification of Diseases, Tenth Edition (ICD-10) code related to possible abuse/addiction (F11-F19: "mental and behavioural disorder due to psychoactive substance use", T40-T43: "poisoning", or X61-X62: "self-poisoning") and at least another ICD-10 code unrelated to abuse/addiction were extracted. Hospital discharge summaries (HDS) were reviewed using two strategies: in Strategy 1, all HDS were reviewed, whereas in Strategy 2, associated ICD-10 codes unrelated to abuse/addiction were firstly assessed to preselect some HDS. Positive predictive values (PPVs) were calculated to evaluate their performance. RESULTS: With Strategy 1, we found 58 psychoactive drug-related somatic complications among the 578 hospitalisations extracted (PPV = 10.0%), including three cases spontaneously reported to the French Addictovigilance Network. Strategy 2 retained 94.8% of the hospitalisations identified with Strategy 1, while the number of reviewed HDS was reduced by half (PPV = 20.1%). Cannabis (56.9%), cocaine (27.6%) and prescription opioids (22.4%) were mainly involved. Complications mainly corresponded to nervous (25.9%) and respiratory and circulatory (22.4%) system disorders. CONCLUSIONS: Combining extraction of ICD-10 codes and a focused review of a preselection of relevant hospitalisations appears to be efficient and time-saving. This method should be applied in other hospital settings before considering the exploration of inpatient data on a wider scale.
BACKGROUND AND OBJECTIVE: Studies have explored hospital records to identify serious complications related to use of psychoactive drugs, but this approach is time consuming with a high rate of false positives. We propose a method to improve the detection of these somatic complications from an inpatient database. METHODS: Hospitalisations in Toulouse University Hospital (France) between 1 July and 31 December 2013 with at least one International Classification of Diseases, Tenth Edition (ICD-10) code related to possible abuse/addiction (F11-F19: "mental and behavioural disorder due to psychoactive substance use", T40-T43: "poisoning", or X61-X62: "self-poisoning") and at least another ICD-10 code unrelated to abuse/addiction were extracted. Hospital discharge summaries (HDS) were reviewed using two strategies: in Strategy 1, all HDS were reviewed, whereas in Strategy 2, associated ICD-10 codes unrelated to abuse/addiction were firstly assessed to preselect some HDS. Positive predictive values (PPVs) were calculated to evaluate their performance. RESULTS: With Strategy 1, we found 58 psychoactive drug-related somatic complications among the 578 hospitalisations extracted (PPV = 10.0%), including three cases spontaneously reported to the French Addictovigilance Network. Strategy 2 retained 94.8% of the hospitalisations identified with Strategy 1, while the number of reviewed HDS was reduced by half (PPV = 20.1%). Cannabis (56.9%), cocaine (27.6%) and prescription opioids (22.4%) were mainly involved. Complications mainly corresponded to nervous (25.9%) and respiratory and circulatory (22.4%) system disorders. CONCLUSIONS: Combining extraction of ICD-10 codes and a focused review of a preselection of relevant hospitalisations appears to be efficient and time-saving. This method should be applied in other hospital settings before considering the exploration of inpatient data on a wider scale.
Authors: A Daveluy; G Miremont-Salamé; A Kostrzewa; A Couret; L Lacoin; C Lecomte; N Moore; V Gilleron; F Haramburu Journal: Pharmacoepidemiol Drug Saf Date: 2012-10-30 Impact factor: 2.890