A machine learning-based clinical decision support system to identify prescriptions with a high risk of medication error.

OBJECTIVE: To improve patient safety and clinical outcomes by reducing the risk of prescribing errors, we tested the accuracy of a hybrid clinical decision support system in prioritizing prescription checks. MATERIALS AND METHODS: Data from electronic health records were collated over a period of 18 months. Inferred scores at a patient level (probability of a patient's set of active orders to require a pharmacist review) were calculated using a hybrid approach (machine learning and a rule-based expert system). A clinical pharmacist analyzed randomly selected prescription orders over a 2-week period to corroborate our findings. Predicted scores were compared with the pharmacist's review using the area under the receiving-operating characteristic curve and area under the precision-recall curve. These metrics were compared with existing tools: computerized alerts generated by a clinical decision support (CDS) system and a literature-based multicriteria query prioritization technique. Data from 10 716 individual patients (133 179 prescription orders) were used to train the algorithm on the basis of 25 features in a development dataset. RESULTS: While the pharmacist analyzed 412 individual patients (3364 prescription orders) in an independent validation dataset, the areas under the receiving-operating characteristic and precision-recall curves of our digital system were 0.81 and 0.75, respectively, thus demonstrating greater accuracy than the CDS system (0.65 and 0.56, respectively) and multicriteria query techniques (0.68 and 0.56, respectively). DISCUSSION: Our innovative digital tool was notably more accurate than existing techniques (CDS system and multicriteria query) at intercepting potential prescription errors. CONCLUSIONS: By primarily targeting high-risk patients, this novel hybrid decision support system improved the accuracy and reliability of prescription checks in a hospital setting.

Peripheral arterial disease and systematic detection of circulating tumor cells: rationale and design of the DETECTOR prospective cohort study.

BACKGROUND: Smoking is a strong risk factor for cancer and atherosclerosis. Cancer mortality, especially from lung cancer, overtakes cardiovascular (CV) death rate in patients with peripheral arterial disease (PAD). Only a few patients with lung cancer after PAD management may benefit from surgical excision. Circulating tumor cells (CTC) associated with low-dose chest CT (LDCT) may improve early cancer detection. This study focuses on a screening strategy that can address not only lung cancer but all tobacco-related cancers in this high-risk population. METHODS: DETECTOR Project is a prospective cohort study in two French University hospitals. Participants are smokers or former smokers (≥30 pack-years, quitted ≤15 years), aged ≥55 to 80 years, with atherosclerotic PAD or abdominal aortic aneurysm. After the first screening round combining LDCT and CTC search on a blood sample, two other screening rounds will be performed at one-year interval. Incidental lung nodule volume, volume doubling time and presence of CTC will be taken into consideration for adapted diagnostic management. In case of negative LDCT and presence of CTC, a contrast enhanced whole-body PET/CT will be performed for extra-pulmonary malignancy screening. Psychological impact of this screening strategy will be evaluated in population study using a qualitative methodology. Assuming 10% prevalence of smoking-associated cancer in the studied population, a total of at least 300 participants will be enrolled. DISCUSSION: Epidemiological data underline an increase incidence in cancer and related death in the follow-up of patients with PAD, compared with the general population, particularly for tobacco-related cancers. The clinical benefit of a special workup for neoplasms in patients with PAD and a history of cigarette smoking has never been investigated. By considering CTCs detection in this very high-risk selected PAD population for tobacco-induced cancer, we expect to detect earlier pulmonary and extra-pulmonary malignancies, at a potentially curable stage. TRIAL REGISTRATION: The study was registered in the French National Agency for Medicines and Health Products Safety (No N° EUDRACT_ID RCB: 2016-A00657-44) and was approved by the ethics Committee for Persons Protection (IRB number 1072 and n° initial agreement 2016-08-02; ClinicalTrials.gov identifier NCT02849041).

Evaluation of the unified model of information systems continuance (UMISC) in two hospital environments.

CONTEXT: The deployment and long-term acceptance of clinical information systems (CISs) are faced with multiple difficulties. They include insufficient quality of the systems in place and resistance to the multiple changes they induce in care processes. Permanent evaluation of deployed solutions is a prerequisite to their continuous improvement. OBJECTIVE: The purpose of this study was twofold: (1) To validate the post-adoption unified model of information systems continuance (UMISC) progressively developed at the Georges Pompidou University Hospital (HEGP) in Paris (internal validation); and (2) To compare, using the same evaluation model, the results observed at HEGP with those of the Saint-Joseph Hospital Group (HPSJ), another Paris acute care institution (external validation). METHODS: The UMISC post-adoption model is built around nine dimensions: end-user characteristics, social norm (SN), IS quality (ISQ), facilitating conditions (FC), perceived usefulness (PU), confirmation of expectations (CE), profession-adjusted use (PAU), satisfaction (SAT), and continuance intention (CI). Two semi-quantitative evaluation surveys were performed at HEGP in 2014 and 2015, and one at HPSJ in 2015. Statistical analysis included multiple regression analysis and structural equation modeling (SEM). RESULTS: The analysis concerned 459 responders, 264 at HEGP and 195 at HPSJ. UMISC indicators, with the exception of SN, are superior at HEGP than at HPSJ, which had a shorter CIS anteriority than HEGP. In SEM analysis, the UMISC model explained 25% and 40% of the CIS use, 92% and 93% of health professionals' satisfaction, and 72% and 71% of continuance intention at HEGP and HPSJ, respectively. Seventeen of the 21 tested UMISC hypotheses were supported in at least one of the two sites. CONCLUSION: The UMISC evaluation model can be used as a comparison and explanatory model of CIS use, satisfaction and continuance intention in post-CIS adoption situations that become prevalent in current electronic hospitals.

Accumulation and transport of microbial-size particles in a pressure protected model burn unit: CFD simulations and experimental evidence.

BACKGROUND: Controlling airborne contamination is of major importance in burn units because of the high susceptibility of burned patients to infections and the unique environmental conditions that can accentuate the infection risk. In particular the required elevated temperatures in the patient room can create thermal convection flows which can transport airborne contaminates throughout the unit. In order to estimate this risk and optimize the design of an intensive care room intended to host severely burned patients, we have relied on a computational fluid dynamic methodology (CFD). METHODS: The study was carried out in 4 steps: i) patient room design, ii) CFD simulations of patient room design to model air flows throughout the patient room, adjacent anterooms and the corridor, iii) construction of a prototype room and subsequent experimental studies to characterize its performance iv) qualitative comparison of the tendencies between CFD prediction and experimental results. The Electricité De France (EDF) open-source software Code_Saturne® (http://www.code-saturne.org) was used and CFD simulations were conducted with an hexahedral mesh containing about 300 000 computational cells. The computational domain included the treatment room and two anterooms including equipment, staff and patient. Experiments with inert aerosol particles followed by time-resolved particle counting were conducted in the prototype room for comparison with the CFD observations. RESULTS: We found that thermal convection can create contaminated zones near the ceiling of the room, which can subsequently lead to contaminate transfer in adjacent rooms. Experimental confirmation of these phenomena agreed well with CFD predictions and showed that particles greater than one micron (i.e. bacterial or fungal spore sizes) can be influenced by these thermally induced flows. When the temperature difference between rooms was 7°C, a significant contamination transfer was observed to enter into the positive pressure room when the access door was opened, while 2°C had little effect. Based on these findings the constructed burn unit was outfitted with supplemental air exhaust ducts over the doors to compensate for the thermal convective flows. CONCLUSIONS: CFD simulations proved to be a particularly useful tool for the design and optimization of a burn unit treatment room. Our results, which have been confirmed qualitatively by experimental investigation, stressed that airborne transfer of microbial size particles via thermal convection flows are able to bypass the protective overpressure in the patient room, which can represent a potential risk of cross contamination between rooms in protected environments.

Control and outcome of a large outbreak of colonization and infection with glycopeptide-intermediate Staphylococcus aureus in an intensive care unit.

BACKGROUND: Glycopeptide-intermediate Staphylococcus aureus (GISA) is emerging as a cause of nosocomial infection and outbreaks of infection and colonization in intensive care units (ICUs). We describe an outbreak of GISA colonization/infection and the ensuing control measures in an ICU and investigate outcomes of the affected patients. METHODS: We describe an outbreak of GISA colonization and infection that affected 21 patients in a medical ICU at a tertiary care teaching hospital, as well as the measures taken to eradicate the GISA strain. RESULT: Recognition of the outbreak was difficult. Infections, all of which were severe, were diagnosed in 11 of 21 patients. Patient isolation and barrier precautions failed when used alone. Addition of a stringent policy of restricted admissions, twice daily environmental cleaning, and implementation of hand decontamination with a hydroalcoholic solution led to outbreak termination. This was associated with increases in workload, despite a marked decrease in the number of admissions. CONCLUSION: This first description of a large outbreak of GISA colonization and infection underlines the importance of routine GISA-strain detection when methicillin-resistant S. aureus is isolated. Outbreak control may be difficult to achieve.