Determining steady-state trough range in vancomycin drug dosing using machine learning

BACKGROUND

Vancomycin is a renally eliminated, nephrotoxic, glycopeptide antibiotic with a narrow therapeutic window, widely used in intensive care units (ICU). We aimed to predict the risk of inappropriate vancomycin trough levels and appropriate dosing for each ICU patient.

METHODS

Observed vancomycin trough levels were categorized into sub-therapeutic, therapeutic, and supra-therapeutic levels to train and compare different classification models. We included adult ICU patients (≥ 18 years) with at least one vancomycin concentration measurement during hospitalization at Mayo Clinic, Rochester, MN, from January 2007 to December 2017.

RESULT

The final cohort consisted of 5337 vancomycin courses. The XGBoost models outperformed other machine learning models with the AUC-ROC of 0.85 and 0.83, specificity of 53% and 47%, and sensitivity of 94% and 94% for sub- and supra-therapeutic categories, respectively. Kinetic estimated glomerular filtration rate and other creatinine-based measurements, vancomycin regimen (dose and interval), comorbidities, body mass index, age, sex, and blood pressure were among the most important variables in the models.

CONCLUSION

We developed models to assess the risk of sub- and supra-therapeutic vancomycin trough levels to improve the accuracy of drug dosing in critically ill patients.

Frontline Providers' and Patients' Perspectives on Improving Diagnostic Safety in the Emergency Department: A Qualitative Study

BACKGROUND

Few studies have described the insights of frontline health care providers and patients on how the diagnostic process can be improved in the emergency department (ED), a setting at high risk for diagnostic errors. The authors aimed to identify the perspectives of providers and patients on the diagnostic process and identify potential interventions to improve diagnostic safety.

METHODS

Semistructured interviews were conducted with 10 ED physicians, 15 ED nurses, and 9 patients/caregivers at two separate health systems. Interview questions were guided by the ED-Adapted National Academies of Sciences, Engineering, and Medicine Diagnostic Process Framework and explored participant perspectives on the ED diagnostic process, identified vulnerabilities, and solicited interventions to improve diagnostic safety. The authors performed qualitative thematic analysis on transcribed interviews.

RESULTS

The research team categorized vulnerabilities in the diagnostic process and intervention opportunities based on the ED-Adapted Framework into five domains: (1) team dynamics and communication (for example, suboptimal communication between referring physicians and the ED team); (2) information gathering related to patient presentation (for example, obtaining the history from the patients or their caregivers; (3) ED organization, system, and processes (for example, staff schedules and handoffs); (4) patient education and self-management (for example, patient education at discharge from the ED); and (5) electronic health record and patient portal use (for example, automatic release of test results into the patient portal). The authors identified 33 potential interventions, of which 17 were provider focused and 16 were patient focused.

CONCLUSION

Frontline providers and patients identified several vulnerabilities and potential interventions to improve ED diagnostic safety. Refining, implementing, and evaluating the efficacy of these interventions are required.

Accurate and interpretable prediction of ICU-acquired AKI

PURPOSE

We developed and validated two parsimonious algorithms to predict the time of diagnosis of any stage of acute kidney injury (any-AKI) or moderate-to-severe AKI in clinically actionable prediction windows.

MATERIALS AND METHODS

In this retrospective single-center cohort of adult ICU admissions, we trained two gradient-boosting models: 1) any-AKI model, predicting the risk of any-AKI at least 6 h before diagnosis (50,342 admissions), and 2) moderate-to-severe AKI model, predicting the risk of moderate-to-severe AKI at least 12 h before diagnosis (39,087 admissions). Performance was assessed before disease diagnosis and validated prospectively.

RESULTS

The models achieved an area under the receiver operating characteristic curve (AUROC) of 0.756 at six hours (any-AKI) and 0.721 at 12 h (moderate-to-severe AKI) prior. Prospectively, both models had high positive predictive values (0.796 and 0.546 for any-AKI and moderate-to-severe AKI models, respectively) and triggered more in patients who developed AKI vs. those who did not (median of 1.82 [IQR 0-4.71] vs. 0 [IQR 0-0.73] and 2.35 [IQR 0.14-4.96] vs. 0 [IQR 0-0.8] triggers per 8 h for any-AKI and moderate-to-severe AKI models, respectively).

CONCLUSIONS

The two AKI prediction models have good discriminative performance using common features, which can aid in accurately and informatively monitoring AKI risk in ICU patients.

Non-invasive Hemoglobin Measurement Predictive Analytics with Missing Data and Accuracy Improvement Using Gaussian Process and Functional Regression Model

Recent use of noninvasive and continuous hemoglobin (SpHb) concentration monitor has emerged as an alternative to invasive laboratory-based hematological analysis. Unlike delayed laboratory based measures of hemoglobin (HgB), SpHb monitors can provide real-time information about the HgB levels. Real-time SpHb measurements will offer healthcare providers with warnings and early detections of abnormal health status, e.g., hemorrhagic shock, anemia, and thus support therapeutic decision-making, as well as help save lives. However, the finger-worn CO-Oximeter sensors used in SpHb monitors often get detached or have to be removed, which causes missing data in the continuous SpHb measurements. Missing data among SpHb measurements reduce the trust in the accuracy of the device, influence the effectiveness of hemorrhage interventions and future HgB predictions. A model with imputation and prediction method is investigated to deal with missing values and improve prediction accuracy. The Gaussian process and functional regression methods are proposed to impute missing SpHb data and make predictions on laboratory-based HgB measurements. Within the proposed method, multiple choices of sub-models are considered. The proposed method shows a significant improvement in accuracy based on a real-data study. Proposed method shows superior performance with the real data, within the proposed framework, different choices of sub-models are discussed and the usage recommendation is provided accordingly. The modeling framework can be extended to other application scenarios with missing values.

A discrete event simulation to evaluate impact of radiology process changes on emergency department computed tomography access

BACKGROUND

Hospitals face the challenge of managing demand for limited computed tomography (CT) resources from multiple patient types while ensuring timely access.

METHODS

A discrete event simulation model was created to evaluate CT access time for emergency department (ED) patients at a large academic medical center with six unique CT machines that serve unscheduled emergency, semi-scheduled inpatient, and scheduled outpatient demand. Three operational interventions were tested: adding additional patient transporters, using an alternative creatinine lab, and adding a registered nurse dedicated to monitoring CT patients in the ED.

RESULTS

All interventions improved access times. Adding one or two transporters improved ED access times by up to 9.8 minutes (Mann-Whitney (MW) CI: [-11.0,-8.7]) and 10.3 minutes (MW CI [-11.5, -9.2]). The alternative creatinine and RN interventions provided 3-minute (MW CI: [-4.0, -2.0]) and 8.5-minute (MW CI: [-9.7, -8.3]) improvements.

CONCLUSIONS

Adding one transporter provided the greatest combination of reduced delay and ability to implement. The projected simulation improvements have been realized in practice.

Improving Non-invasive Hemoglobin Measurement Accuracy Using Nonparametric Models

Uncontrolled hemorrhage is a leading cause of preventable death among patients with trauma. Early recognition of hemorrhage can aid in the decision to administer blood transfusion and improve patient outcomes. To provide real-time measurement and continuous monitoring of hemoglobin concentration, the non-invasive and continuous hemoglobin (SpHb) measurement device has drawn extensive attention in clinical practice. However, the accuracy of such a device varies in different scenarios, so the use is not yet widely accepted. This article focuses on using statistical nonparametric models to improve the accuracy of SpHb measurement device by considering measurement bias among instantaneous measurements and individual evolution trends. In the proposed method, the robust locally estimated scatterplot smoothing (LOESS) method and the Kernel regression model are considered to address those issues. Overall performance of the proposed method was evaluated by cross-validation, which showed a substantial improvement in accuracy with an 11.3% reduction of standard deviation, 23.7% reduction of mean absolute error, and 28% reduction of mean absolute percentage error compared to the original measurements. The effects of patient demographics and initial medical condition were analyzed and deemed to not have a significant effect on accuracy. Because of its high accuracy, the proposed method is highly promising to be considered to support transfusion decision-making and continuous monitoring of hemoglobin concentration. The method also has promise for similar advancement of other diagnostic devices in healthcare.

Extended Patient Alone Time in Emergency Department Leads to Increased Risk of 30-Day Hospitalization

OBJECTIVES

This study was conducted to describe patients at risk for prolonged time alone in the emergency department (ED) and to determine the relationship between clinical outcomes, specifically 30-day hospitalization, and patient alone time (PAT) in the ED.

METHODS

An observational cohort design was used to evaluate PAT and patient characteristics in the ED. The study was conducted in a tertiary academic ED that has both adult and pediatric ED facilities and of patients placed in an acute care room for treatment between May 1 and July 31, 2016, excluding behavioral health patients. Simple linear regression and t tests were used to evaluate the relationship between patient characteristics and PAT. Logistic regression was used to evaluate the relationship between 30-day hospitalization and PAT.

RESULTS

Pediatric patients had the shortest total PAT compared with all older age groups (86.4 minutes versus 131 minutes, P < 0.001). Relationships were seen between PAT and patient characteristics, including age, geographic region, and the severity and complexity of the health condition. Controlling for Charlson comorbidity index and other potentially confounding variables, a logistic regression model showed that patients are more likely to be hospitalized within 30 days after their ED visit, with an odds ratio (95% confidence interval) of 1.056 (1.017-1.097) for each additional hour of PAT.

CONCLUSIONS

Patient alone time is not equal among all patient groups. Study results indicate that PAT is significantly associated with 30-day hospitalization. This conclusion indicates that PAT may affect patient outcomes and warrants further investigation.

Incorporating RTLS-Based Spatiotemporal Information in Studying Physical Activities of Clinical Staff. ANNUAL INTERNATIONAL CONFERENCE OF THE IEEE ENGINEERING IN MEDICINE AND BIOLOGY SOCIETY. IEEE ENGINEERING IN MEDICINE AND BIOLOGY SOCIETY. ANNUAL INTERNATIONAL CONFERENCE 2021;2021:2386-2391. [PMID: 34891762 DOI: 10.1109/embc46164.2021.9630597] [Citation(s) in RCA: 0] [Impact Index Per Article: 0]

Clinicians and staff who work in intense hospital settings such as the emergency department (ED) are under an extended amount of mental and physical pressure every day. They may spend hours in active physical pressure to serve patients with severe injuries or stay in front of a computer to review patients' clinical history and update the patients' electronic health records (EHR). Nurses on the other hand may stay for multiple consecutive days of 9-12 working hours. The amount of pressure is so much that they usually end up taking days off to recover the lost energy. Both of these extreme cases of low and high physical activities are shown to affect the physical and mental health of clinicians and may even lead to fatigue and burnout.In this study Real-Time location systems (RTLS) are used for the first time, to study the amount of physical activity exerted by clinicians. RTLS systems have traditionally been used in hospital settings for locating staff and equipment, whereas our proposed method combines both time and location information together to estimate the duration, length, and speed of movements within hospital wards such as the ED. It is also our first step towards utilizing non-wearable devices to measure sedentary behavior inside the ED. This information helps to assess the workload on the care team and identify means to reduce the risk of performance compromise, fatigue, and burnout.We used one year worth of raw RFID data that covers movement records of 38 physicians, 13 residents, 163 nurses, 33 staff in the ED. We defined a walking path as the continuous sequences of movements and stops and identified separate walking paths for each individual on each day. Walking duration, distance, and speed, along with the number of steps and the duration of sedentary behavior, are then estimated for each walking path. We compared our results to the values reported in the literature and showed despite the low spatial resolution of RTLS, our non-invasive estimations are closely comparable to the ones measured by Fitbit or other wearable pedometers.Clinical Relevance- Adequate assessment of workload in a dynamic care delivery space plays an important role in ensuring safe and optimal care delivery [7]. Systems capable of measuring physical activities on a continuous basis during daily work can provide precious information for a variety of purposes including automated assessment of sedentary behaviors and early detection of work pressure. Such systems could help facilitate targeted changes in the number of staff, duration of their working shifts leading to a safer and healthier environment for both clinicians and patients.

Estimation of Baseline Serum Creatinine with Machine Learning

INTRODUCTION

Comparing current to baseline serum creatinine is important in detecting acute kidney injury. In this study, we report a regression-based machine learning model to predict baseline serum creatinine.

METHODS

We developed and internally validated a gradient boosting model on patients admitted in Mayo Clinic intensive care units from 2005 to 2017 to predict baseline creatinine. The model was externally validated on the Medical Information Mart for Intensive Care III (MIMIC III) cohort in all ICU admissions from 2001 to 2012. The predicted baseline creatinine from the model was compared with measured serum creatinine levels. We compared the performance of our model with that of the backcalculated estimated serum creatinine from the Modification of Diet in Renal Disease (MDRD) equation.

RESULTS

Following ascertainment of eligibility criteria, 44,370 patients from the Mayo Clinic and 6,112 individuals from the MIMIC III cohort were enrolled. Our model used 6 features from the Mayo Clinic and MIMIC III datasets, including the presence of chronic kidney disease, weight, height, and age. Our model had significantly lower error than the MDRD backcalculation (mean absolute error [MAE] of 0.248 vs. 0.374 in the Mayo Clinic test data; MAE of 0.387 vs. 0.465 in the MIMIC III cohort) and higher correlation (intraclass correlation coefficient [ICC] of 0.559 vs. 0.050 in the Mayo Clinic test data; ICC of 0.357 vs. 0.030 in the MIMIC III cohort).

DISCUSSION/CONCLUSION

Using machine learning models, baseline serum creatinine could be estimated with higher accuracy than the backcalculated estimated serum creatinine level.

Mapping of critical events in disease progression through binary classification: Application to amyotrophic lateral sclerosis

BACKGROUND

The progression of many degenerative diseases is tracked periodically using scales evaluating functionality in daily activities. Although estimating the timing of critical events (i.e., disease tollgates) during degenerative disease progression is desirable, the necessary data may not be readily available in scale records. Further, analysis of disease progression poses data challenges, such as censoring and misclassification errors, which need to be addressed to provide meaningful research findings and inform patients.

METHODS

We developed a novel binary classification approach to map scale scores into disease tollgates to describe disease progression leveraging standard/modified Kaplan-Meier analyses. The approach is demonstrated by estimating progression pathways in amyotrophic lateral sclerosis (ALS). Tollgate-based ALS Staging System (TASS) specifies the critical events (i.e., tollgates) in ALS progression. We first developed a binary classification predicting whether each TASS tollgate was passed given the itemized ALSFRS-R scores using 514 ALS patients' data from Mayo Clinic-Rochester. Then, we utilized the binary classification to translate/map the ALSFRS-R data of 3,264 patients from the PRO-ACT database into TASS. We derived the time trajectories of ALS progression through tollgates from the augmented PRO-ACT data using Kaplan-Meier analyses. The effects of misclassification errors, condition-dependent dropouts, and censored data in trajectory estimations were evaluated with Interval Censored Kaplan Meier Analysis and Multistate Model for Panel Data.

RESULTS

The approach using Mayo Clinic data accurately estimated tollgate-passed states of patients given their itemized ALSFRS-R scores (AUCs > 0.90). The tollgate time trajectories derived from the augmented PRO-ACT dataset provide valuable insights; we predicted that the majority of the ALS patients would have modified arm function (67%) and require assistive devices for walking (53%) by the second year after ALS onset. By the third year, most (74%) ALS patients would occasionally use a wheelchair, while 48% of the ALS patients would be wheelchair-dependent by the fourth year. Assistive speech devices and feeding tubes were needed in 49% and 30% of the patients by the third year after ALS onset, respectively. The onset body region alters some tollgate passage time estimations by 1-2 years.

CONCLUSIONS

The estimated tollgate-based time trajectories inform patients and clinicians about prospective assistive device needs and life changes. More research is needed to personalize these estimations according to prognostic factors. Further, the approach can be leveraged in the progression of other diseases.

Monitoring Diagnostic Safety Risks in Emergency Departments: Protocol for a Machine Learning Study

BACKGROUND

Diagnostic decision making, especially in emergency departments, is a highly complex cognitive process that involves uncertainty and susceptibility to errors. A combination of factors, including patient factors (eg, history, behaviors, complexity, and comorbidity), provider-care team factors (eg, cognitive load and information gathering and synthesis), and system factors (eg, health information technology, crowding, shift-based work, and interruptions) may contribute to diagnostic errors. Using electronic triggers to identify records of patients with certain patterns of care, such as escalation of care, has been useful to screen for diagnostic errors. Once errors are identified, sophisticated data analytics and machine learning techniques can be applied to existing electronic health record (EHR) data sets to shed light on potential risk factors influencing diagnostic decision making.

OBJECTIVE

This study aims to identify variables associated with diagnostic errors in emergency departments using large-scale EHR data and machine learning techniques.

METHODS

This study plans to use trigger algorithms within EHR data repositories to generate a large data set of records that are labeled trigger-positive or trigger-negative, depending on whether they meet certain criteria. Samples from both data sets will be validated using medical record reviews, upon which we expect to find a higher number of diagnostic safety events in the trigger-positive subset. Machine learning will be used to evaluate relationships between certain patient factors, provider-care team factors, and system-level risk factors and diagnostic safety signals in the statistically matched groups of trigger-positive and trigger-negative charts.

RESULTS

This federally funded study was approved by the institutional review board of 2 academic medical centers with affiliated community hospitals. Trigger queries are being developed at both organizations, and sample cohorts will be labeled using the triggers. Machine learning techniques such as association rule mining, chi-square automated interaction detection, and classification and regression trees will be used to discover important variables that could be incorporated within future clinical decision support systems to help identify and reduce risks that contribute to diagnostic errors.

CONCLUSIONS

The use of large EHR data sets and machine learning to investigate risk factors (related to the patient, provider-care team, and system-level) in the diagnostic process may help create future mechanisms for monitoring diagnostic safety.

INTERNATIONAL REGISTERED REPORT IDENTIFIER (IRRID)

DERR1-10.2196/24642.

NATURAL LANGUAGE PROCESSING BASED MACHINE LEARNING MODEL USING CARDIAC MRI REPORTS TO IDENTIFY HYPERTROPHIC CARDIOMYOPATHY PATIENTS

Hypertrophic Cardiomyopathy (HCM) is the most common genetic heart disease in the US and is known to cause sudden death (SCD) in young adults. While significant advancements have been made in HCM diagnosis and management, there is a need to identify HCM cases from electronic health record (EHR) data to develop automated tools based on natural language processing guided machine learning (ML) models for accurate HCM case identification to improve management and reduce adverse outcomes of HCM patients. Cardiac Magnetic Resonance (CMR) Imaging, plays a significant role in HCM diagnosis and risk stratification. CMR reports, generated by clinician annotation, offer rich data in the form of cardiac measurements as well as narratives describing interpretation and phenotypic description. The purpose of this study is to develop an NLP-based interpretable model utilizing impressions extracted from CMR reports to automatically identify HCM patients. CMR reports of patients with suspected HCM diagnosis between the years 1995 to 2019 were used in this study. Patients were classified into three categories of yes HCM, no HCM and, possible HCM. A random forest (RF) model was developed to predict the performance of both CMR measurements and impression features to identify HCM patients. The RF model yielded an accuracy of 86% (608 features) and 85% (30 features). These results offer promise for accurate identification of HCM patients using CMR reports from EHR for efficient clinical management transforming health care delivery for these patients.

Including urinary output to define AKI enhances the performance of machine learning models to predict AKI at admission

PURPOSE

Acute kidney injury (AKI) is a prevalent and detrimental condition in intensive care unit patients. Most AKI predictive models only predict creatinine-triggered AKI (AKI_Cr) and might underperform when predicting urine-output-triggered AKI (AKI_UO). We aimed to describe how admission AKI_Cr prediction models perform in all AKI patients.

MATERIALS AND METHODS

Three types of models were trained: 1) pAKI_any, predicting AKI based on creatinine or urine output, 2) pAKI_UO, predicting AKI based only on urine output, and 3) pAKI_Cr, predicting AKI based only on creatinine. We compared model performance and predictive features.

RESULTS

The pAKI_any models had the best overall performance (AUROC 0.673-0.716) and the most consistent performance across three patient cohorts grouped by type of AKI trigger (min AUROC of 0.636). The pAKI_Cr models had fair performance in predicting AKI_Cr (AUROCs 0.702-0.748) but poor performance predicting AKI_UO (AUROCs 0.581-0.695). The predictive features for the pAKI_Cr models and pAKI_UO models were distinct, while top features for the pAKI_any models were consistently a combination of those for the pAKI_Cr and pAKI_UO models.

CONCLUSION

Ignoring urine output in the outcome during model training resulted in models that are unlikely to predict AKI_UO adequately and may miss a substantial proportion of patients in practice.

Measuring Sensitivity and Precision of Real-Time Location Systems (RTLS): Definition, Protocol and Demonstration for Clinical Relevance

The ability of a Real Time Location System (RTLS) to provide correct information in a clinical environment is an important consideration in evaluating the effectiveness of the technology. While past efforts describe how well the technology performed in a lab environment, the performance of such technology has not been specifically defined or evaluated in a practice setting involving workflow and movement. Clinical environments pose complexity owing to various layouts and various movements. Further, RTL systems are not equipped to provide true negative information (where an entity is not located). Hence, this study defined sensitivity and precision in this context, and developed a simulation protocol to serve as a systematic testing framework using actors in a clinical environment. The protocol was used to measure the sensitivity and precision of an RTL system in the emergency department space of a quaternary care medical center. The overall sensitivity and precision were determined to be 84 and 93% respectively. These varied for patient rooms, staff area, hallway and other rooms.

Using operative features to identify surgical complexity: a case in breast surgery practice

Increasing workload is one of the main problems that surgical practices face. This increase is not only due to the increasing demand volume but also due to increasing case complexity. This raises the question on how to measure and predict the complexity to address this issue. Predicting surgical duration is critical to parametrize surgical complexity, improve surgeon satisfaction by avoiding unexpected overtime, and improve operation room utilization. Our objective is to utilize the historical data on surgical operations to obtain complexity groups and use this groups to improve practice.Our study first leverages expert opinion on the surgical complexity to identify surgical groups. Then, we use a tree-based method on a large retrospective dataset to identify similar complexity groups by utilizing the surgical features and using surgical duration as a response variable. After obtaining the surgical groups by using two methods, we statistically compare expert-based grouping with the data-based grouping. This comparison shows that a tree-based method can provide complexity groups similar to the ones generated by an expert by using features that are available at the time of surgical listing. These results suggest that one can take advantage of available data to provide surgical duration predictions that are data-driven, evidence-based, and practically relevant.

A new optimization algorithm for HDR brachytherapy that improves DVH-based planning: Truncated Conditional Value-at-Risk (TCVaR)

Purpose:To introduce a new optimization algorithm that improves DVH results and is designed for the type of heterogeneous dose distributions that occur in brachytherapy.Methods:The new optimization algorithm is based on a prior mathematical approach that uses mean doses of the DVH metric tails. The prior mean dose approach is referred to as conditional value-at-risk (CVaR), and unfortunately produces noticeably worse DVH metric results than gradient-based approaches. We have improved upon the CVaR approach, using the so-called Truncated CVaR (TCVaR), by excluding the hottest or coldest voxels in the structure from the calculations of the mean dose of the tail. Our approach applies an iterative sequence of convex approximations to improve the selection of the excluded voxels. Data Envelopment Analysis was used to quantify the sensitivity of TCVaR results to parameter choice and to compare the quality of a library of 256 TCVaR plans created for each of prostate, breast, and cervix treatment sites with commercially-generated plans.Results:In terms of traditional DVH metrics, TCVaR outperformed CVaR and the improvements increased monotonically as more iterations were used to identify and exclude the hottest/coldest voxels from the optimization problem. TCVaR also outperformed the Eclipse-Brachyvision TPS, with an improvement in PTVD95% (for equivalent organ-at-risk doses) of up to 5% (prostate), 3% (breast), and 1% (cervix).Conclusions:A novel optimization algorithm for HDR treatment planning produced plans with superior DVH metrics compared with a prior convex optimization algorithm as well as Eclipse-Brachyvision. The algorithm is computationally efficient and has potential applications as a primary optimization algorithm or quality assurance for existing optimization approaches.

Unnecessary wait hours: a novel ED system-level metric

BACKGROUND

Emergency department (ED) operations leaders are under increasing pressure to make care delivery more efficient. Publicly reported ED efficiency metrics are traditionally patient centred and do not show situational or facility-based improvement opportunities. We propose the consideration of a novel metric, the 'Number of Unnecessary Waits (NUW)' and the corresponding 'Unnecessary Wait Hours (UWH)', to measure space efficiency, and we describe how we used NUW to evaluate operational changes in our ED.

METHODS

UWH summarises the relationship between the number of available rooms and the number of patients waiting by returning a value equal to the number of unnecessary patient waits. We used this metric to evaluate reassigning a clinical technician assistant (CTA) to the new role of flow CTA.

RESULTS

We retrospectively analysed 3.5 months of data from before and after creation of the flow CTA. NUW metric analysis suggested that the flow CTA decreased the amount of unnecessary wait hours, while higher patient volumes had the opposite effect.

CONCLUSIONS

Situational system-level metrics may provide a new dimension to evaluating ED operational efficiencies. Studies focussed on system-level metrics to evaluate an ED practice are needed to understand the role these metrics play in evaluation of a department's operations.

Quantifying the Impact of Resuscitation-Team Activation in Hospital Emergency Departments

Hospital emergency department (ED) operations are affected when critically ill or injured patients arrive. Such events often lead to the initiation of specific protocols, referred to as Resuscitation-team Activation (RA), in the ED of Mayo Clinic, Rochester, MN where this study was conducted. RA events lead to the diversion of resources from other patients in the ED to provide care to critically ill patients; therefore, it has an impact on the entire ED system. This paper presents a data-driven and flexible statistical learning model to quantify the impact of RA on the ED. The model learns the pattern of operations in the ED from historical patient arrival and departure timestamps and quantifies the impact of RA by measuring the deviation of the departure of patients during RA from normal processes. The proposed method significantly outperforms baseline methods based on measuring the average time patients spend in the ED.

Prospective Implementation and Evaluation of a Decision-Tree Algorithm for Route of Hysterectomy

OBJECTIVE

To evaluate the rate of vaginal hysterectomy and outcomes after initiation of a prospective decision-tree algorithm to determine the optimal surgical route of hysterectomy.

METHODS

A prospective algorithm to determine optimal route of hysterectomy was developed, which uses the following factors: history of laparotomy, uterine size, and vaginal access. The algorithm was implemented at our institution from November 24, 2015, to December 31, 2017, for patients requiring hysterectomy for benign indications. Expected route of hysterectomy was assigned by the algorithm and was compared with the actual route performed to identify compliance compared with deviation. Surgical outcomes were analyzed.

RESULTS

Of 365 patients who met inclusion criteria, 202 (55.3%) were expected to have a total vaginal hysterectomy, 57 (15.6%) were expected to have an examination under anesthesia followed by total vaginal hysterectomy, 52 (14.2%) were expected to have an examination under anesthesia followed by robotic-assisted total laparoscopic hysterectomy, and 54 (14.8%) were expected to have an abdominal or robotic-laparoscopic route of hysterectomy. Forty-six procedures (12.6%) deviated from the algorithm to a more invasive route (44 robotic, two abdominal). Seven patients had total vaginal hysterectomy when robotic-assisted total laparoscopic hysterectomy or abdominal hysterectomy was expected by the algorithm. Overall, 71% of patients were expected to have a vaginal route of hysterectomy per the algorithm, of whom 81.5% had a total vaginal hysterectomy performed; more than 99% of the total vaginal hysterectomies attempted were successfully completed.

CONCLUSION

Vaginal surgery is feasible, carries a low complication rate with excellent outcomes, and should have a place in gynecologic surgery. National use of this prospective algorithm may increase the rate of total vaginal hysterectomy and decrease health care costs.

Technology Implementation and Associated Pharmacy Interruptions

This study focuses on interruptions in an inpatient pharmacy setting and the impact of CPOE implementation on the types, frequency, and duration of interruptions. A cross-sectional observation study of pharmacy employees in an inpatient pharmacy was conducted. The independent variables included day of week, time of day, job position of the person interrupted, and description of each interruption. A total of 552 interruptions were observed with a mean frequency of 10.6 interruptions per hour lasting a mean (SD) duration of 1.34 (1.43) minutes. Incoming calls were the most frequent interruption type across all phases. Pharmacy employees spend almost a quarter of their time on interruptions, and pharmacists have longer interruptions than technicians. Immediately after CPOE implementation, durations tend to be one-and-a-half times longer than before. CPOE implementation did not affect the frequency of interruptions. Recommendations included redesign of work processes and job responsibilities.

Patient-Care Team Contact Patterns Impact Treatment Length of Stay in the Emergency Department

Real-time location systems (RTLS) has found extensive application in the healthcare setting, that is shown to improve safety, save cost, and increase patient satisfaction. More specifically, some studies have shown the efficacy of RTLS leading to an improved workflow in the emergency department. However, due to substantial implementation costs of such technologies, hospital administrators show reluctance in RTLS adoption. Our previous preliminary studies with RFID data in the emergency department (ED) demonstrated for the first time the quantification of `patient alone time' and its relationship to outcomes such as 30-day hospitalization. In this study, we use ED RTLS data to analyze patient-care team contact time (PCTCT) and its relationship to the total treatment length of stay (LOS) in ED. An observational cohort study was performed in the ED using RTLS data from Jan 17 - Sep 17, 2017, which included a total of 51,697 patients. PCTCT within the first hour of a patient's placement in a treatment bed was calculated and its relationship to treatment LOS was analyzed while controlling for confounding factors affecting treatment LOS. Results show that treatment LOS is highly correlated with the ED crowding captured by the patient-perprovider ratio, negatively correlated to the physician and resident visit frequency, and positively correlated to nurse visit frequency. The results can inform designing new guidelines for ideal patient-care team interactions and be used to determine optimal ED staffing levels and care team composition for effective care delivery.

Getting to the Heart of the Matter: A Triage Model to Improve Utilization of Cardiology Consultative Services

Objective

Patient and Methods

Results

Conclusion

CCMapper: An adaptive NLP-based free-text chief complaint mapping algorithm

OBJECTIVE

Chief complaint (CC) is among the earliest health information recorded at the beginning of a patient's visit to an emergency department (ED). We propose a heuristic methodology for automatically mapping the free-text data into a structured list of CCs.

METHODS

A comprehensive structured list categorizing CCs was developed by experienced Emergency Medicine (EM) physicians. Using this list, we developed a natural language processing-based algorithm, referred to as Chief Complaint Mapper (CCMapper), for automatically mapping a CC into the most appropriate category (ies). We trained and validated CCMapper using free-text CC data from the Mayo Clinic ED in Rochester, MN. We developed a consensus-based validation approach to handle both indifferences and disagreements between the two EM physicians who manually mapped a random sample of free-text CCs into categories within the structured list.

RESULTS

The kappa statistic demonstrated a high level of agreement (κ = 0.958) between the two physicians with less than 2% human error. CCMapper achieved a total sensitivity of 94.2% with a specificity of 99.8% and F-score of 94.7% on the validation set. The sensitivity of CCMapper when mapping free-text data with multiple CCs was 82.3% with a specificity of 99.1% and total F-score of 82.3%.

CONCLUSION

Due to its simplicity, high performance, and capability of incorporating new free-text CC data, CCMapper can be readily adopted by other EDs to support clinical decision making. CCMapper can facilitate the development of predictive models for the type and timing of important events in ED (e.g., ICU admission).

Pattern-based strategic surgical capacity allocation

Strategic allocation of limited operating room (OR) capacity to surgeons is crucial for the coordination of surgical work flow, including planning of consultation and surgery days, and staff assignment to perioperative teams. However, it is a challenging problem in practice, since the capacity allocation needs to be cyclic for schedule predictability and surgical team coordination, and also needs to satisfy surgeons' preferences. It is further complicated by the practice of surgeons sharing ORs. In this study, we propose a mathematical optimization model to coordinate capacity allocation among surgeons in order to improve the utilization of surgical capacity. We introduce the concept of capacity allocation patterns to account for schedule cyclicity and surgeons' preferences. Further, we develop a data-driven approach to coordinate OR sharing among surgeons based on their historical OR usage. The proposed methodology is applied to a case study with data from a surgical division at Mayo Clinic. Compared with the state-of-the-practice, the proposed approach shows a substantial potential in reducing the maximum number of ORs allocated daily to the division with little overtime. With a solution time of less than 0.5 s, the proposed methodology can be readily used as a decision support tool in surgical practice.

Application of Constrained Optimization Methods in Health Services Research: Report 2 of the ISPOR Optimization Methods Emerging Good Practices Task Force

BACKGROUND

Constrained optimization methods are already widely used in health care to solve problems that represent traditional applications of operations research methods, such as choosing the optimal location for new facilities or making the most efficient use of operating room capacity.

OBJECTIVES

In this paper we illustrate the potential utility of these methods for finding optimal solutions to problems in health care delivery and policy. To do so, we selected three award-winning papers in health care delivery or policy development, reflecting a range of optimization algorithms. Two of the three papers are reviewed using the ISPOR Constrained Optimization Good Practice Checklist, adapted from the framework presented in the initial Optimization Task Force Report. The first case study illustrates application of linear programming to determine the optimal mix of screening and vaccination strategies for the prevention of cervical cancer. The second case illustrates application of the Markov Decision Process to find the optimal strategy for treating type 2 diabetes patients for hypercholesterolemia using statins. The third paper (described in Appendix 1) is used as an educational tool. The goal is to describe the characteristics of a radiation therapy optimization problem and then invite the reader to formulate the mathematical model for solving it. This example is particularly interesting because it lends itself to a range of possible models, including linear, nonlinear, and mixed-integer programming formulations. From the case studies presented, we hope the reader will develop an appreciation for the wide range of problem types that can be addressed with constrained optimization methods, as well as the variety of methods available.

CONCLUSIONS

Constrained optimization methods are informative in providing insights to decision makers about optimal target solutions and the magnitude of the loss of benefit or increased costs associated with the ultimate clinical decision or policy choice. Failing to identify a mathematically superior or optimal solution represents a missed opportunity to improve economic efficiency in the delivery of care and clinical outcomes for patients. The ISPOR Optimization Methods Emerging Good Practices Task Force's first report provided an introduction to constrained optimization methods to solve important clinical and health policy problems. This report also outlined the relationship of constrained optimization methods relative to traditional health economic modeling, graphically illustrated a simple formulation, and identified some of the major variants of constrained optimization models, such as linear programming, dynamic programming, integer programming, and stochastic programming. The second report illustrates the application of constrained optimization methods in health care decision making using three case studies. The studies focus on determining optimal screening and vaccination strategies for cervical cancer, optimal statin start times for diabetes, and an educational case to invite the reader to formulate radiation therapy optimization problems. These illustrate a wide range of problem types that can be addressed with constrained optimization methods.

Improving Accuracy of Noninvasive Hemoglobin Monitors: A Functional Regression Model for Streaming SpHb Data

OBJECTIVE

The purpose of this paper is to develop a method for improving the accuracy of SpHb monitors, which are noninvasive hemoglobin monitoring tools, leading to better critical care protocols in trauma care.

METHODS

The proposed method is based on fitting smooth spline functions to SpHb measurements collected over a time window and then using a functional regression model to predict the true HgB value for the end of the time window.

RESULTS

The accuracy of the proposed method is compared to traditional methods. The mean absolute error between the raw SpHb measurements and the gold standard hemoglobin measurements was 1.26 g/Dl. The proposed method reduced the mean absolute error to 1.08 g/Dl. [1] Conclusion: Fitting a smooth function to SpHb measurements improves the accuracy of Hgb predictions.

SIGNIFICANCE

Accurate prediction of current and future HgB levels can lead to sophisticated decision models that determine the optimal timing and amount of blood product transfusions.

Consequences of the 48-h rule: A lens into the psychiatric patient flow through an emergency department

OBJECTIVE

Psychiatric patient boarding in emergency department (ED) is a severe and growing problem. In July 2013, Minnesota implemented a law requiring jailed persons committed to state psychiatric facilities be transferred within 48-h of commitment. This study aims to quantify the effect of this law on a large ED's psychiatric patient flow.

METHODS

A pre- and post- comparison of 2011-2015 ED length of stay (LOS) for adult psychiatric patients was performed using electronic medical record data. Comparisons of the median LOS were assessed using a segmented regression model with time series error, and risk differences (RD) were used to determine changes in the proportion of patients with LOS ≥3 and ≥5days. Changes in patient disposition proportions were assessed using risk ratios.

RESULTS

The median ED LOS for patients admitted for psychiatric care increased by 5.22h from 2011 to 2015 (95% CI: (4.33, 7.15)), while the frequency of patient encounters remained constant. Although no significant difference in the rate of ED LOS increase was found pre- and post- implementation, the proportion of adults with LOS ≥3days and ≥15days increased (RD 0.017 (95% CI: (0.013, 0.021)); 0.002 (95% CI: (0.001,0.004)), respectively).

CONCLUSIONS

The proportion of ED adult psychiatric patients experiencing prolonged LOS increased following the implementation of a statewide law requiring patients committed through the criminal justice system be transferred to a state psychiatric hospital within 48h. Identifying characteristics of subsets of psychiatric patients disproportionally affected could suggest focused healthcare system improvements to improve ED psychiatric care.

A work systems analysis approach to understanding fatigue in hospital nurses

Occupational fatigue is an important challenge in improving health and safety in health care systems. A secondary analysis of cross-sectional data from a survey sample comprised 340 hospital nurses was conducted to explore the relationships between components of the nursing work system (person, tasks, tools and technology, environment, organisation) and nurse fatigue and recovery levels. All components of the work system were significantly associated with changes in fatigue and recovery. Results of a tree-based classification method indicated significant interactions between multiple work system components and fatigue and recovery. For example, the relationship between a task variable of 'excessive work' and acute fatigue varied based on an organisation variable related to 'time to communicate with managers/supervisors'. A work systems analysis contributes to increased understanding of fatigue, allowing for a more accurate representation of the complexity in health care systems to guide future research and practice to achieve increased nurse health and safety. Practitioner Summary: This paper explored the relationships between nursing work system components and nurse fatigue. Findings revealed significant interactions between work system components and nurses' fatigue and recovery. A systems approach allows for a more accurate representation of complexity in work systems and can guide interventions to improve nurse health and safety.

Physician, Interrupted: Workflow Interruptions and Patient Care in the Emergency Department

BACKGROUND

It is unclear how workflow interruptions impact emergency physicians at the point of care.

OBJECTIVES

Our study aimed to evaluate interruption characteristics experienced by academic emergency physicians.

METHODS

This prospective, observational study collected interruptions during attending physician shifts. An interruption is defined as any break in performance of a human activity that briefly requires attention. One observer captured interruptions using a validated tablet PC-based tool that time stamped and categorized the data. Data collected included: 1) type, 2) priority of interruption to original task, and 3) physical location of the interruption. A Kruskal-Wallis H test compared interruption priority and duration. A chi-squared analysis examined the priority of interruptions in and outside of the patient rooms.

RESULTS

A total of 2355 interruptions were identified across 210 clinical hours and 28 shifts (means = 84.1 interruptions per shift, standard deviation = 14.5; means = 11.21 interruptions per hour, standard deviation = 4.45). Physicians experienced face-to-face physician interruptions most frequently (26.0%), followed by face-to-face nurse communication (21.7%), and environment (20.8%). There was a statistically significant difference in interruption duration based on the interruption priority, χ²(2) = 643.98, p < 0.001, where durations increased as priority increased. Whereas medium/normal interruptions accounted for 53.6% of the total interruptions, 53% of the interruptions that occurred in the patient room (n = 162/308) were considered low priority (χ² [2, n = 2355] = 78.43, p < 0.001).

CONCLUSIONS

Our study examined interruptions over entire provider shifts and identified patient rooms as high risk for low-priority interruptions. Targeting provider-centered interventions to patient rooms may aid in mitigating the impacts of interruptions on patient safety and enhancing clinical care.

Constrained Optimization Methods in Health Services Research-An Introduction: Report 1 of the ISPOR Optimization Methods Emerging Good Practices Task Force

Providing health services with the greatest possible value to patients and society given the constraints imposed by patient characteristics, health care system characteristics, budgets, and so forth relies heavily on the design of structures and processes. Such problems are complex and require a rigorous and systematic approach to identify the best solution. Constrained optimization is a set of methods designed to identify efficiently and systematically the best solution (the optimal solution) to a problem characterized by a number of potential solutions in the presence of identified constraints. This report identifies 1) key concepts and the main steps in building an optimization model; 2) the types of problems for which optimal solutions can be determined in real-world health applications; and 3) the appropriate optimization methods for these problems. We first present a simple graphical model based on the treatment of "regular" and "severe" patients, which maximizes the overall health benefit subject to time and budget constraints. We then relate it back to how optimization is relevant in health services research for addressing present day challenges. We also explain how these mathematical optimization methods relate to simulation methods, to standard health economic analysis techniques, and to the emergent fields of analytics and machine learning.

A Data-Driven Approach For Better Assignment Of Clinical And Surgical Capacity In An Elective Surgical Practice

This work analyzes strategies for better allocation of surgeon resources in an elective surgical practice. Among the metrics considered to evaluate the assignment of tasks are OR-to-Clinic ratio per provider, OR-to-Clinic ratio per day, patient access to clinic, and patient access to surgery. In addition, a simulation model is used to evaluate the clinical and surgical capacity of the calendar to identify potential inefficiencies and propose strategic changes to the calendar.

Determining Optimal Route of Hysterectomy for Benign Indications: Clinical Decision Tree Algorithm

OBJECTIVE

To evaluate practice change after initiation of a robotic surgery program using a clinical algorithm to determine the optimal surgical approach to benign hysterectomy.

METHODS

A retrospective postrobot cohort of benign hysterectomies (2009-2013) was identified and the expected surgical route was determined from an algorithm using vaginal access and uterine size as decision tree branches. We excluded the laparoscopic hysterectomy route. A prerobot cohort (2004-2005) was used to evaluate a practice change after the addition of robotic technology (2007). Costs were estimated.

RESULTS

Cohorts were similar in regard to uterine size, vaginal parity, and prior laparotomy history. In the prerobot cohort (n=473), 320 hysterectomies (67.7%) were performed vaginally and 153 (32.3%) through laparotomy with 15.1% (46/305) performed abdominally when the algorithm specified vaginal hysterectomy. In the postrobot cohort (n=1,198), 672 hysterectomies (56.1%) were vaginal; 390 (32.6%) robot-assisted; and 136 (11.4%) abdominal. Of 743 procedures, 38 (5.1%) involved laparotomy and 154 (20.7%) involved robotic technique when a vaginal approach was expected. Robotic hysterectomies had longer operations (141 compared with 59 minutes, P<.001) and higher rates of surgical site infection (4.7% compared with 0.2%, P<.001) and urinary tract infection (8.1% compared with 4.1%, P=.05) but no difference in major complications (P=.27) or readmissions (P=.27) compared with vaginal hysterectomy. Algorithm conformance would have saved an estimated $800,000 in hospital costs over 5 years.

CONCLUSION

When a decision tree algorithm indicated vaginal hysterectomy as the route of choice, vaginal hysterectomy was associated with shorter operative times, lower infection rate, and lower cost. Vaginal hysterectomy should be the route of choice when feasible.

Hours and Miles: Patient and Health System Implications of Transfer for Psychiatric Bed Capacity

Introduction

An increasing number of behavioral health (BH) patients are presenting to the emergency department (ED) while BH resources continue to decline. This situation-may lead to more external transfers to find care.

Methods

This is a retrospective cohort study of consecutive patients presenting to a tertiary care academic ED from February 1, 2013, through January 31, 2014. Patients were identified through electronic health record documentation of psychiatric consultation during ED evaluation. We reviewed electronic health records for demographic characteristics, diagnoses, payer source, ED length of stay, ED disposition, arrival method, and distance traveled to an external facility for inpatient admission. Univariable and multivariable associations with transfer to an external facility in comparison with patients admitted internally were evaluated with logistic regression models and summarized with odds ratios (OR).

Results

We identified 2,585 BH visits, of which 1,083 (41.9%) resulted in discharge. A total of 1,502 patient visits required inpatient psychiatric admission, and of these cases, 177 patients (11.8%; 95% CI = [10.2–13.5]) required transfer to an external facility. The median ED length of stay for transferred patients was 13.9 hours (interquartile range [IQR], 9.3–20.2 hours; range, 3.0–243.0 hours). The median distance for transport was 83 miles (IQR, 42–111 miles; range, 42–237 miles). In multivariable analysis, patients with suicidal or homicidal ideation had increased risk of transfer (odds ratio [OR] [95% CI], 1.93 [1.22–3.06]; P=0.005). Children younger than 18 years (OR [95% CI], 2.34 [1.60–3.40]; P<0.001) and adults older than 65 years (OR [95% CI], 3.46 [1.93–6.19]; P<0.001) were more likely to require transfer and travel farther to access care.

Conclusion

Patients requiring external transfer for inpatient psychiatric care were found to have prolonged ED lengths of stay. Patients with suicidal and homicidal ideation as well as children and adults older than 65 years are more likely to require transfer.

Prediction and detection models for acute kidney injury in hospitalized older adults

Background

Acute Kidney Injury (AKI) occurs in at least 5 % of hospitalized patients and can result in 40–70 % morbidity and mortality. Even following recovery, many subjects may experience progressive deterioration of renal function. The heterogeneous etiology and pathophysiology of AKI complicates its diagnosis and medical management and can add to poor patient outcomes and incur substantial hospital costs. AKI is predictable and may be avoidable if early risk factors are identified and utilized in the clinical setting. Timely detection of undiagnosed AKI in hospitalized patients can also lead to better disease management.

Methods

Data from 25,521 hospital stays in one calendar year of patients 60 years and older was collected from a large health care system. Four machine learning models (logistic regression, support vector machines, decision trees and naïve Bayes) along with their ensemble were tested for AKI prediction and detection tasks. Patient demographics, laboratory tests, medications and comorbid conditions were used as the predictor variables. The models were compared using the area under ROC curve (AUC) evaluation metric.

Results

Logistic regression performed the best for AKI detection (AUC 0.743) and was a close second to the ensemble for AKI prediction (AUC ensemble: 0.664, AUC logistic regression: 0.660). History of prior AKI, use of combination drugs such as ACE inhibitors, NSAIDS and diuretics, and presence of comorbid conditions such as respiratory failure were found significant for both AKI detection and risk prediction.

Conclusions

The machine learning models performed fairly well on both predicting AKI and detecting undiagnosed AKI. To the best of our knowledge, this is the first study examining the difference between prediction and detection of AKI. The distinction has clinical relevance, and can help providers either identify at risk subjects and implement preventative strategies or manage their treatment depending on whether AKI is predicted or detected.

Transforming Healthcare Delivery: Integrating Dynamic Simulation Modelling and Big Data in Health Economics and Outcomes Research

In the era of the Information Age and personalized medicine, healthcare delivery systems need to be efficient and patient-centred. The health system must be responsive to individual patient choices and preferences about their care, while considering the system consequences. While dynamic simulation modelling (DSM) and big data share characteristics, they present distinct and complementary value in healthcare. Big data and DSM are synergistic-big data offer support to enhance the application of dynamic models, but DSM also can greatly enhance the value conferred by big data. Big data can inform patient-centred care with its high velocity, volume, and variety (the three Vs) over traditional data analytics; however, big data are not sufficient to extract meaningful insights to inform approaches to improve healthcare delivery. DSM can serve as a natural bridge between the wealth of evidence offered by big data and informed decision making as a means of faster, deeper, more consistent learning from that evidence. We discuss the synergies between big data and DSM, practical considerations and challenges, and how integrating big data and DSM can be useful to decision makers to address complex, systemic health economics and outcomes questions and to transform healthcare delivery.

Intelligent Emergency Department: Validation of Sociometers to Study Workload

Sociometers are wearable sensors that continuously measure body movements, interactions, and speech. The purpose of this study is to test sociometers in a smart environment in a live clinical setting, to assess their reliability in capturing and quantifying data. The long-term goal of this work is to create an intelligent emergency department that captures real-time human interactions using sociometers to sense current system dynamics, predict future state, and continuously learn to enable the highest levels of emergency care delivery. Ten actors wore the devices during five simulated scenarios in the emergency care wards at a large non-profit medical institution. For each scenario, actors recited prewritten or structured dialogue while independent variables, e.g., distance, angle, obstructions, speech behavior, were independently controlled. Data streams from the sociometers were compared to gold standard video and audio data captured by two ward and hallway cameras. Sociometers distinguished body movement differences in mean angular velocity between individuals sitting, standing, walking intermittently, and walking continuously. Face-to-face (F2F) interactions were not detected when individuals were offset by 30°, 60°, and 180° angles. Under ideal F2F conditions, interactions were detected 50 % of the time (4/8 actor pairs). Proximity between individuals was detected for 13/15 actor pairs. Devices underestimated the mean duration of speech by 30-44 s, but were effective at distinguishing the dominant speaker. The results inform engineers to refine sociometers and provide health system researchers a tool for quantifying the dynamics and behaviors in complex and unpredictable healthcare environments such as emergency care.

Surgical Duration Estimation via Data Mining and Predictive Modeling: A Case Study

Operating rooms (ORs) are one of the most expensive and profitable resources within a hospital system. OR managers strive to utilize these resources in the best possible manner. Traditionally, surgery durations are estimated using a moving average adjusted by the scheduler (adjusted system prediction or ASP). Other methods based on distributions, regression and data mining have also been proposed. To overcome difficulties with numerous procedure types and lack of sufficient sample size, and avoid distributional assumptions, the main objective is to develop a hybrid method of duration prediction and demonstrate using a case study.

Surgical never events and contributing human factors

INTRODUCTION

We report the first prospective analysis of human factors elements contributing to invasive procedural never events by using a validated Human Factors Analysis and Classification System (HFACS).

METHODS

From August 2009 to August 2014, operative and invasive procedural "Never Events" (retained foreign object, wrong site/side procedure, wrong implant, wrong procedure) underwent systematic causation analysis promptly after the event. Contributing human factors were categorized using the 4 levels of error causation described by Reason and 161 HFACS subcategories (nano-codes).

RESULTS

During the study, approximately 1.5 million procedures were performed, during which 69 never events were identified. A total of 628 contributing human factors nano-codes were identified. Action-based errors (n = 260) and preconditions to actions (n = 296) accounted for the majority of the nano-codes across all 4 types of events, with individual cognitive factors contributing one half of the nano-codes. The most common action nano-codes were confirmation bias (n = 36) and failed to understand (n = 36). The most common precondition nano-codes were channeled attention on a single issue (n = 33) and inadequate communication (n = 30).

CONCLUSION

Targeting quality and interventions in system improvement addressing cognitive factors and team resource management as well as perceptual biases may decrease errors and further improve patient safety. These results delineate targets to further decrease never events from our health care system.

Selecting a dynamic simulation modeling method for health care delivery research-part 2: report of the ISPOR Dynamic Simulation Modeling Emerging Good Practices Task Force

In a previous report, the ISPOR Task Force on Dynamic Simulation Modeling Applications in Health Care Delivery Research Emerging Good Practices introduced the fundamentals of dynamic simulation modeling and identified the types of health care delivery problems for which dynamic simulation modeling can be used more effectively than other modeling methods. The hierarchical relationship between the health care delivery system, providers, patients, and other stakeholders exhibits a level of complexity that ought to be captured using dynamic simulation modeling methods. As a tool to help researchers decide whether dynamic simulation modeling is an appropriate method for modeling the effects of an intervention on a health care system, we presented the System, Interactions, Multilevel, Understanding, Loops, Agents, Time, Emergence (SIMULATE) checklist consisting of eight elements. This report builds on the previous work, systematically comparing each of the three most commonly used dynamic simulation modeling methods-system dynamics, discrete-event simulation, and agent-based modeling. We review criteria for selecting the most suitable method depending on 1) the purpose-type of problem and research questions being investigated, 2) the object-scope of the model, and 3) the method to model the object to achieve the purpose. Finally, we provide guidance for emerging good practices for dynamic simulation modeling in the health sector, covering all aspects, from the engagement of decision makers in the model design through model maintenance and upkeep. We conclude by providing some recommendations about the application of these methods to add value to informed decision making, with an emphasis on stakeholder engagement, starting with the problem definition. Finally, we identify areas in which further methodological development will likely occur given the growing "volume, velocity and variety" and availability of "big data" to provide empirical evidence and techniques such as machine learning for parameter estimation in dynamic simulation models. Upon reviewing this report in addition to using the SIMULATE checklist, the readers should be able to identify whether dynamic simulation modeling methods are appropriate to address the problem at hand and to recognize the differences of these methods from those of other, more traditional modeling approaches such as Markov models and decision trees. This report provides an overview of these modeling methods and examples of health care system problems in which such methods have been useful. The primary aim of the report was to aid decisions as to whether these simulation methods are appropriate to address specific health systems problems. The report directs readers to other resources for further education on these individual modeling methods for system interventions in the emerging field of health care delivery science and implementation.

Applying dynamic simulation modeling methods in health care delivery research-the SIMULATE checklist: report of the ISPOR simulation modeling emerging good practices task force

Health care delivery systems are inherently complex, consisting of multiple tiers of interdependent subsystems and processes that are adaptive to changes in the environment and behave in a nonlinear fashion. Traditional health technology assessment and modeling methods often neglect the wider health system impacts that can be critical for achieving desired health system goals and are often of limited usefulness when applied to complex health systems. Researchers and health care decision makers can either underestimate or fail to consider the interactions among the people, processes, technology, and facility designs. Health care delivery system interventions need to incorporate the dynamics and complexities of the health care system context in which the intervention is delivered. This report provides an overview of common dynamic simulation modeling methods and examples of health care system interventions in which such methods could be useful. Three dynamic simulation modeling methods are presented to evaluate system interventions for health care delivery: system dynamics, discrete event simulation, and agent-based modeling. In contrast to conventional evaluations, a dynamic systems approach incorporates the complexity of the system and anticipates the upstream and downstream consequences of changes in complex health care delivery systems. This report assists researchers and decision makers in deciding whether these simulation methods are appropriate to address specific health system problems through an eight-point checklist referred to as the SIMULATE (System, Interactions, Multilevel, Understanding, Loops, Agents, Time, Emergence) tool. It is a primer for researchers and decision makers working in health care delivery and implementation sciences who face complex challenges in delivering effective and efficient care that can be addressed with system interventions. On reviewing this report, the readers should be able to identify whether these simulation modeling methods are appropriate to answer the problem they are addressing and to recognize the differences of these methods from other modeling approaches used typically in health technology assessment applications.

Effect of obesity and clinical factors on pre-incision time: study of operating room workflow

As the obese population is increasing rapidly worldwide, there is more interest to study the different aspects of obesity and its impact especially on healthcare outcomes and health related issues. Targeting non-surgical times in the operating room (OR), this study focuses on the effect of obesity along with clinical factors on pre-incision times in OR. Specifically, both the individual and combined effect of clinical factors with obesity on pre-incision times is studied. Results show that with the confidence of 95%, pre-incision time in the OR of obese patients is significantly higher than those for non-obese patients by approximately five percent. Findings also show that more complex cases do not exhibit significant differences between these patient subgroups.

Relationships between Wellness, Fatigue, and Intershift Recovery in Hospital Nurses

Fatigue and recovery have been associated with health, safety, and performance outcomes in hospital nurses. There is a growing emphasis on fatigue risk management and health promotion programs to support nurses and promote quality in hospitals. However, little is known about the relationships between fatigue and wellness measures in nurses. This study used a tree-based classification method to identify associations between self-reported fatigue, recovery and wellness measures in a survey study of hospital nurses. Significant relationships between multiple wellness measures and acute fatigue, chronic fatigue, and intershift recovery levels were identified. Specifically, the findings include critical levels of wellness measures where fatigue and recovery change. These findings have implications for ongoing efforts to develop effective fatigue management programs in this population.

Multi-disciplinary communication networks for skin risk assessment in nursing homes with high IT sophistication

BACKGROUND

The role of nursing home (NH) information technology (IT) in quality improvement has not been clearly established, and its impacts on communication between care givers and patient outcomes in these settings deserve further attention.

OBJECTIVES

In this research, we describe a mixed method approach to explore communication strategies used by healthcare providers for resident skin risk in NH with high IT sophistication (ITS).

METHODS

Sample included NH participating in the statewide survey of ITS. We incorporated rigorous observation of 8- and 12-h shifts, and focus groups to identify how NH IT and a range of synchronous and asynchronous tools are used. Social network analysis tools and qualitative analysis were used to analyze data and identify relationships between ITS dimensions and communication interactions between care providers.

RESULTS

Two of the nine ITS dimensions (resident care-technological and administrative activities-technological) and total ITS were significantly negatively correlated with number of unique interactions. As more processes in resident care and administrative activities are supported by technology, the lower the number of observed unique interactions. Additionally, four thematic areas emerged from staff focus groups that demonstrate how important IT is to resident care in these facilities including providing resident-centered care, teamwork and collaboration, maintaining safety and quality, and using standardized information resources.

CONCLUSION

Our findings in this study confirm prior research that as technology support (resident care and administrative activities) and overall ITS increases, observed interactions between staff members decrease. Conversations during staff interviews focused on how technology facilitated resident centered care through enhanced information sharing, greater virtual collaboration between team members, and improved care delivery. These results provide evidence for improving the design and implementation of IT in long term care systems to support communication and associated resident outcomes.

Impact of fatigue on performance in registered nurses: data mining and implications for practice

Performance of nurses has a direct effect on the quality and safety of care that is delivered. Fatigue has been identified as a factor that leads to performance decrements in healthcare workers, especially nurses. Determining associations between dimensions of fatigue and performance is imperative to better understanding fatigue in nurses and the potential implications for both patient and provider safety. This article identifies associations between ranges of fatigue levels and significant differences in perceived performance, and analyzes interactions between fatigue dimensions in relation to perceived performance scores. Overall, mental fatigue tended to have higher perceived performance decrements than physical and total fatigue in the highest fatigue ranges. As physical fatigue begins to develop in nurses, physical exertion rather than discomfort is more critical to perceived performance. As acute fatigue levels increase, perceived performance levels continue to decrease, whereas the role of chronic fatigue is relatively constant. Minimizing the development of acute fatigue may help in maintaining higher performance levels. The findings from this study provide valuable information in quantifying the changes in perceived performance with regard to specific fatigue levels, as well as an initial understanding of how the individual dimensions and states of fatigue vary in their association with perceived performance decrements.