Use of electroconvulsive therapy for individuals receiving inpatient psychiatric care on a nationwide scale in France: Variations linked to health care supply.

BACKGROUND: A comprehensive understanding of variations in the use of electroconvulsive therapy (ECT) among health care providers in charge of ECT referrals is lacking. OBJECTIVE: Our objectives were to document ECT use and its variations on a nationwide scale in France and to identify the factors that were significantly associated with these variations. METHODS: Administrative health claims data on hospitalization were used to perform a descriptive analysis of ECT use for adult patients receiving inpatient psychiatric care in mainland France in 2019 and its variations across hospitals in charge of ECT referrals. Based on a conceptual framework drawn from the literature on medical practice variations, a multilevel logistic regression was then conducted to identify patients, hospitals and contextual characteristics that were significantly associated with ECT treatment using non-ECT-treated patients receiving inpatient psychiatric care as the reference population. RESULTS: Patients receiving ECT (n = 3288) were older, more frequently female and had more severe diagnoses than other patients seen in inpatient care (n = 295,678). Significant variations were observed in the rate of ECT use across hospitals (n = 468), with a coefficient of variation largely above one. In the multivariable analysis, ECT treatment was associated with patient characteristics (which accounted for 6% of the variations) but also with characteristics of the hospitals and their environments (44% of the variations), including the type of hospital and its distance to the closest facility providing ECT. CONCLUSIONS: Variations in ECT use were strongly linked to health care supply characteristics, which raises questions about access to quality mental health care.

Use of the C4.5 machine learning algorithm to test a clinical guideline-based decision support system.

Well-designed medical decision support system (DSS) have been shown to improve health care quality. However, before they can be used in real clinical situations, these systems must be extensively tested, to ensure that they conform to the clinical guidelines (CG) on which they are based. Existing methods cannot be used for the systematic testing of all possible test cases. We describe here a new exhaustive dynamic verification method. In this method, the DSS is considered to be a black box, and the Quinlan C4.5 algorithm is used to build a decision tree from an exhaustive set of DSS input vectors and outputs. This method was successfully used for the testing of a medical DSS relating to chronic diseases: the ASTI critiquing module for type 2 diabetes.

An ontology of bacteria to help physicians to compare antibacterial spectra.

General practitioners (GPs) may lack specialist microbiological knowledge, making it difficult for them to use documents concerning antibacterial spectra provided by French health authorities. We have developed a tool to help GPs to compare antibacterial spectra, based on an ontology of bacteria generated using OWL-DL language. This tool makes it possible to search for information concerning the antibiotic susceptibility of given bacteria, regardless of the way in which this information is expressed in the document. Applied to the whole document, the tool made 4528 spectra explicit, whereas only 3471 could be understood without microbiological reasoning. A preliminary study showed that the performance of this tool was similar to that of an expert microbiologist (94 to 98% correct responses) and better than that of unassisted GPs (84-90% correct responses).