Pharmaceutical Decision Support System Using Machine Learning to Analyze and Limit Drug-Related Problems in Hospitals.

The health product circuit corresponds to the chain of steps that a medicine goes through in hospital, from prescription to administration. The safety and regulation of all the stages of this circuit are major issues to ensure the safety and protect the well-being of hospitalized patients. In this paper we present an automatic system for analyzing prescriptions using Artificial Intelligence (AI) and Machine Learning (ML), with the aim of ensuring patient safety by limiting the risk of prescription errors or drug iatrogeny. Our study is made in collaboration with Lille University Hospital (LUH). We exploited the MIMIC-III (Medical Information Mart for Intensive Care) a large, single-center database containing information corresponding to patients admitted to critical care units at a large tertiary care hospital.

Escitalopram-induced hepatitis: A case report.

BACKGROUND: The antidepressant escitalopram is widely prescribed for the treatment of depression. It is generally well-tolerated, and cholestasis is not mentioned in its summary of product characteristics (SmPC). We present a case of cholestatic and cytolysis liver injury due to escitalopram and a VigiBase® study. CASE SUMMARY: A 68-year-old man was admitted to our emergency unit due to clinical jaundice associated with hepatitis, pruritus and dark urine. We tested the patient for the most common etiologies of jaundice, including hemolysis, viral hepatitis, cirrhosis, carcinoma, cholangitis, cholelithiasis and intrahepatic or extrahepatic obstruction. The etiological study was negative, and an adverse drug reaction was the sole possible explanation. The patient was receiving treatment with escitalopram. Two days after its withdrawal, pruritus was resolved. Ten days after withdrawal, clinical jaundice disappeared. It took a month and three weeks after withdrawal for the patient to have normalized liver function tests. To our knowledge, this is the first reported case of cholestasis where treatment with escitalopram was the only possible cause, with a highly probable causality. In addition, we determined whether escitalopram is associated with hepatotoxicity and cholestasis by performing a disproportionality analysis. All cases of hepatobiliary disorders induced by escitalopram and reported in the World Health Organization pharmacovigilance database (VigiBase®) were analyzed to characterize this toxicity. We found that patients treated with escitalopram had an increased risk of hepatitis [odds ratio (OR) = 1.938(1.186-3.166)] and cholestasis [OR = 1.866(1.279-2.724)] [OR (95% confidence interval)]. The median duration between the introduction of escitalopram and the occurrence of acute hepatitis and/or cholestasis was ten days +/- seven days. CONCLUSION: Although extremely rare, this case report, the review of the literature and the pharmacovigilance update confirm that escitalopram can cause drug-induced hepatotoxicity and cholestasis, generally within a week after initiation. Thus, escitalopram should be withdrawn immediately if an iatrogenic cause cannot be excluded. If its responsibility is ascertained, escitalopram should be consequently contraindicated. In addition, serotoninergic antidepressants in patients with non-severe depression are ineffective and harmful. Finally, the SmPC of escitalopram should be updated to alert for this risk and give clear clinical guidelines.

Effect of automated drug distribution systems on medication error rates in a short-stay geriatric unit.

RATIONALE, AIMS AND OBJECTIVES: To assess the impact of an automated drug distribution system on medication errors (MEs). METHODS: Before-after observational study in a 40-bed short stay geriatric unit within a 1800 bed general hospital in Valenciennes, France. Researchers attended nurse medication administration rounds and compared administered to prescribed drugs, before and after the drug distribution system changed from a ward stock system (WSS) to a unit dose dispensing system (UDDS), integrating a unit dose dispensing robot and automated medication dispensing cabinet (AMDC). RESULTS: A total of 615 opportunities of errors (OEs) were observed among 148 patients treated during the WSS period, and 783 OEs were observed among 166 patients treated during the UDDS period. ME [medication administration error (MAE)] rates were calculated and compared between the two periods. Secondary measures included type of errors, seriousness of errors and risk reduction for the patients. The implementation of an automated drug dispensing system resulted in a 53% reduction in MAEs. All error types were reduced in the UDDS period compared with the WSS period (P<0.001). Wrong dose and wrong drug errors were reduced by 79.1% (2.4% versus 0.5%, P=0.005) and 93.7% (1.9% versus 0.01%, P=0.009), respectively. CONCLUSION: An automated UDDS combining a unit dose dispensing robot and AMDCs could reduce discrepancies between ordered and administered drugs, thus improving medication safety among the elderly.

P-glycoprotein and cytochrome P450 3A4 involvement in risperidone transport using an in vitro Caco-2/TC7 model and an in vivo model.

The possible involvement of P-glycoprotein (P-gp) and cytochrome P450 (CYP) 3A4 in risperidone transport was investigated using in vitro and in vivo models. Firstly, uptake studies were performed on a Caco-2/TC7 cell monolayer; the effects of 1 microg ml(-1) risperidone on apparent permeability were determined for secretory and absorptive directions, in the presence or absence of various P-gp and CYP3A4 inhibitors (verapamil, ketoconazole, erythromycin), and of an associated multidrug-resistant protein inhibitor (indomethacin). Secondly, on a conscious rat model, risperidone pharmacokinetic parameters, notably absorption parameters, were determined using compartmental and deconvolution methods. Three groups of seven rats received respectively an IV risperidone dose, an oral risperidone dose (PO group) and the same oral risperidone dose after verapamil administration (POV group). No formation of 9-hydroxyrisperidone was observed on Caco-2 cells after risperidone administration; there was no evidence that intestinal CYP3A4 is involved in risperidone metabolising. Risperidone secretory permeation was higher than absorptive permeation. Verapamil increased risperidone absorption permeation and decreased its secretory permeation. Indomethacin did not modify these permeation values. In rats, verapamil led to an increase in both risperidone and 9-hydroxyrisperidone plasmatic concentrations. The fraction absorbed in the verapamil group was 3.18 times higher than in the oral group (65.9% and 20.7% for POV group and PO group). The absorption rate constant was lower in the verapamil group. Our results indicate that P-gp decreases the intestinal absorption of risperidone and that intestinal CYP3A4 is not involved in risperidone metabolism.