Hospital food management: a multi-objective approach to reduce waste and costs.

Food waste contributes significantly to greenhouse emissions and represents a substantial portion of overall waste within hospital facilities. Furthermore, uneaten food leads to a diminished nutritional intake for patients, that typically are vulnerable and ill. Therefore, this study developed mathematical models for constructing patient meals in a 1000-bed hospital located in Florida. The objective is to minimize food waste and meal-building costs while ensuring that the prepared meals meet the required nutrients and caloric content for patients. To accomplish these objectives, four mixed-integer programming models were employed, incorporating binary and continuous variables. The first model establishes a baseline for how the system currently works. This model generates the meals without minimizing waste or cost. The second model minimizes food waste, reducing waste up to 22.53 % compared to the baseline. The third model focuses on minimizing meal-building costs and achieves a substantial reduction of 37 %. Finally, a multi-objective optimization model was employed to simultaneously reduce both food waste and cost, resulting in reductions of 19.70 % in food waste and 32.66 % in meal-building costs. The results demonstrate the effectiveness of multi-objective optimization in reducing waste and costs within large-scale food service operations.

Latent Regression Analysis Considering Student, Teacher, and Parent Variables and Their Relationship with Academic Performance in Primary School Students in Chile.

Academic performance in primary students is fundamental to future school success; however, simultaneous analysis of different key individual, family, and teaching factors must be considered to improve understanding and benefit the development of students' potential. This article presents a latent regression analysis model that examines the relationship between the latent variables (self-efficacy, interest in reading, bullying, parental expectations, and discrimination/exclusion, and teacher violence/aggression) and the academic performance of first-cycle primary students. The model investigates the impact of the latent variables on the standardized endogenous variables of SIMCE Mathematics and Language test scores using a quantitative, non-experimental, correlational, and cross-sectional design. The study involved 70,778 students (53.4% female), with an average age of 9.5 years (SD = 0.6), from Chilean public (33.6%) and subsidized (66.4%) schools. The results indicate that the model accounted for 49.8% and 47.7% of the mean variability in SIMCE Mathematics and Language test scores, respectively. The goodness-of-fit indices demonstrated satisfactory fits for both models. In both tests, student self-efficacy emerged as the most significant factor explaining test score variability, followed by parental expectations. Bullying was identified as a relevant factor in reducing mean performance on both tests. The findings suggest that education decision makers should address these issues to improve student outcomes.

A Multistate Decomposition Analysis of Cesarean Rate Variations, Associated Health Outcomes, and Financial Implications in the United States.

OBJECTIVES: Cesarean rates vary widely across the U.S. states; however, little is known about the causes and implications associated with these variations. The objectives of this study were to quantify the contribution of the clinical and nonclinical factors in explaining the difference in cesarean rates across states and to investigate the associated health outcome of cesarean variations. STUDY DESIGN: Using the Hospital Cost and Utilization Project State Inpatient Databases, this retrospective study included all nonfederal hospital births from Wisconsin, Florida, and New York. A nonlinear extension of the Oaxaca-Blinder method was used to decompose the contributions of differences in characteristics to cesarean variations between these states. The risk factors for cesarean delivery were identified using separate multivariable logistic regression analysis for each State. RESULTS: The difference in clinical and nonclinical factors explained a substantial (~46.57-65.45%) proportion of cesarean variations between U.S. states. The major contributors of variation were patient demographics, previous cesareans, hospital markup ratios, and social determinants of health. Cesarean delivery was significantly associated with higher postpartum readmissions and unplanned emergency department visits, greater lengths of stay, and hospital costs across all states. CONCLUSION: Although a proportion of variations in cesarean rates can be explained by the differences in risk factors, the remaining unexplained variations suggest differences in practice patterns and imply potential quality concerns. Since nonclinical factors are likely to play an important role in cesarean variation, we recommend targeted initiatives increasing access to maternal care and improving maternal health literacy. KEY POINTS: · Cesarean rates vary widely almost two folds within U.S. states.. · The difference in risk factors explained substantial (~46.57-65.45%) of the cesarean variations.. · Mother race, hospital factors, and social determinants comprised major proportion of explained variation.. · Adverse outcomes and increased expenditures were associated with cesarean than vaginal delivery.. · Significant potential cost savings for Medicaid if the unnecessary cesarean deliveries are reduced..

Impact of Statewide Mandatory Medicaid Managed Care (SMMC) Programs on Hospital Obstetric Outcomes.

The state of Florida implemented mandatory managed care for Medicaid enrollees via the Statewide Medicaid Managed Care (SMMC) program in April of 2014. The objective of this study was to examine the impact of the implementation of the SMMC program on the access to care and quality of maternal care for Medicaid enrollees, as measured by several hospital obstetric outcomes. The primary data source for this retrospective observational study was the Hospital Cost and Utilization Project (HCUP) all-payer State ED (SED) visit and State Inpatient Databases (SIDs) from 2010 to 2017. The primary health outcomes for obstetric care were primary cesarean, preterm birth, postpartum preventable ED visits, postpartum preventable readmissions, and vaginal delivery after cesarean (VBAC) rates. Using difference-in-differences (DID) estimation, selected health outcomes were examined for Florida residents with Medicaid beneficiaries (treatment) and the commercially insured population (comparison), before and after the implementation of SMMC. Improvement in disparities for racial/ethnic minority Medicaid enrollees was estimated relative to whites, compared to the relative change among commercially insured patients. From the DID estimation, the findings showed that SMMC is statistically significantly associated with a higher reduction in primary cesarean rates, preterm births, preventable postpartum ED visits, and readmissions among Medicaid beneficiaries relative to their commercially insured counterparts. However, this study did not find any significant reduction in racial/ethnic disparities in obstetric outcomes. In general, this study highlights the impact of SMMC implementation on obstetric outcomes in Florida and provides important insights and potential scope for improvement in obstetric care quality and associated racial/ethnic disparities.

Impact of the COVID-19 Pandemic on the Pediatric Hospital Visits: Evidence from the State of Florida.

Although early evidence reported a substantial decline in pediatric hospital visits during COVID-19, it is unclear whether the decline varied across different counties, particularly in designated Medically Underserved Areas (MUA). The objective of this study is to explore the state-wide impact of COVID-19 on pediatric hospital visit patterns, including the economic burden and MUA communities. We conducted a retrospective observational study of pediatric hospital visits using the Florida State all-payer Emergency Department (ED) and Inpatient dataset during the pandemic (April-September 2020) and the same period in 2019. Pediatric Treat-and-Release ED and inpatient visit rates were compared by patient demographics, socioeconomic, diagnosis, MUA status, and hospital characteristics. Pediatric hospital visits in Florida decreased by 53.7% (62.3% in April-June, 44.2% in July-September) during the pandemic. The Treat-and-Release ED and inpatient visits varied up to 5- and 3-fold, respectively, across counties. However, changes in hospital visits across MUA counties were similar compared with non-MUA counties except for lower Treat-and-Release ED volume in April-May. The disproportional decrease in visits was notable for the underserved population, including Hispanic and African American children; Medicaid coverages; non-children's hospitals; and diagnosed with respiratory diseases, appendicitis, and sickle-cell. Florida Hospitals experienced a USD 1.37 billion (average USD 8.3 million) decline in charges across the study period in 2020. Disproportionate decrease in hospital visits, particularly in the underserved population, suggest a combined effect of the persistent challenge of care access and changes in healthcare-seeking behavior during the pandemic. These findings suggest that providers and policymakers should emphasize alternative interventions/programs ensuring adequate care during the pandemic, particularly for high-risk children.

The Waiting Game - How Cooperation Between Public and Private Hospitals Can Help Reduce Waiting Lists.

Prolonged waiting to access health care is a primary concern for nations aiming for comprehensive effective care, due to its adverse effects on mortality, quality of life, and government approval. Here, we propose two novel bargaining frameworks to reduce waiting lists in two-tier health care systems with local and regional actors. In particular, we assess the impact of 1) trading patients on waiting lists among hospitals, the 2) introduction of the role of private hospitals in capturing unfulfilled demand, and the 3) hospitals' willingness to share capacity on the system performance. We calibrated our models with 2008-2018 Chilean waiting list data. If hospitals trade unattended patients, our game-theoretic models indicate a potential reduction of waiting lists of up to 37%. However, when private hospitals are introduced into the system, we found a possible reduction of waiting lists of up to 60%. Further analyses revealed a trade-off between diagnosing unserved demand and the additional expense of using private hospitals as a back-up system. In summary, our game-theoretic frameworks of waiting list management in two-tier health systems suggest that public-private cooperation can be an effective mechanism to reduce waiting lists. Further empirical and prospective evaluations are needed.

Characteristics and Outcomes of Pediatric Nonindex Readmission: Evidence From Florida Hospitals.

OBJECTIVES: Increasing pediatric care regionalization may inadvertently fragment care if children are readmitted to a different (nonindex) hospital rather than the discharge (index) hospital. Therefore, this study aimed to assess trends in pediatric nonindex readmission rates, examine the risk factors, and determine if this destination difference affects readmission outcomes. METHODS: In this retrospective cohort study, we use the Healthcare Cost and Utilization Project State Inpatient Database to include pediatric (0 to 18 years) admissions from 2010 to 2017 across Florida hospitals. Risk factors of nonindex readmissions were identified by using logistic regression analyses. The differences in outcomes between index versus nonindex readmissions were compared for in-hospital mortality, morbidity, hospital cost, length of stay, against medical advice discharges, and subsequent hospital visits by using generalized linear regression models. RESULTS: Among 41 107 total identified readmissions, 5585 (13.6%) were readmitted to nonindex hospitals. Adjusted nonindex readmission rate increased from 13.3% in 2010% to 15.4% in 2017. Patients in the nonindex readmissions group were more likely to be adolescents, live in poor neighborhoods, have higher comorbidity scores, travel longer distances, and be discharged at the postacute facility. After risk adjusting, no difference in in-hospital mortality was found, but morbidity was 13% higher, and following unplanned emergency department visits were 28% higher among patients with nonindex readmissions. Length of stay, hospital costs, and against medical advice discharges were also significantly higher for nonindex readmissions. CONCLUSIONS: A substantial proportion of children experienced nonindex readmissions and relatively poorer health outcomes compared with index readmission. Targeted strategies for improving continuity of care are necessary to improve readmission outcomes.

Identifying Children at Readmission Risk: At-Admission versus Traditional At-Discharge Readmission Prediction Model.

The timing of 30-day pediatric readmissions is skewed with approximately 40% of the incidents occurring within the first week of hospital discharges. The skewed readmission time distribution coupled with delay in health information exchange among healthcare providers might offer a limited time to devise a comprehensive intervention plan. However, pediatric readmission studies are thus far limited to the development of the prediction model after hospital discharges. In this study, we proposed a novel pediatric readmission prediction model at the time of hospital admission which can improve the high-risk patient selection process. We also compared proposed models with the standard at-discharge readmission prediction model. Using the Hospital Cost and Utilization Project database, this prognostic study included pediatric hospital discharges in Florida from January 2016 through September 2017. Four machine learning algorithms-logistic regression with backward stepwise selection, decision tree, Support Vector machines (SVM) with the polynomial kernel, and Gradient Boosting-were developed for at-admission and at-discharge models using a recursive feature elimination technique with a repeated cross-validation process. The performance of the at-admission and at-discharge model was measured by the area under the curve. The performance of the at-admission model was comparable with the at-discharge model for all four algorithms. SVM with Polynomial Kernel algorithms outperformed all other algorithms for at-admission and at-discharge models. Important features associated with increased readmission risk varied widely across the type of prediction model and were mostly related to patients' demographics, social determinates, clinical factors, and hospital characteristics. Proposed at-admission readmission risk decision support model could help hospitals and providers with additional time for intervention planning, particularly for those targeting social determinants of children's overall health.

Osteonecrosis in individuals with systemic lupus erythematosus: a predictive model / Osteonecrosis en individuos con lupus eritematoso sistémico: un modelo predictivo

OBJECTIVE: This work attempts to provide a model to predict the development of osteonecrosis (ON) in individuals with systemic lupus erythematosus (SLE) using pharmacological, demographic, and psychoactive factors. METHOD: A review of the literature was conducted to construct a survey administered across Chile to individuals with SLE during a period of three weeks. This work used a sample size of 46 de-identified data records. Two Bayesian logistic regression models were created, with non-informative prior and informative prior distributions, and a random forest model was done for comparison. All models were cross-validated. RESULTS: The significant variables used were mean corticosteroids per day (mg) and tobacco use. The random forest model provided good accuracy and sensitivity, but low specificity. Bayesian logistic regression with prior information increased the specificity. CONCLUSIONS: This work determined that the use of corticosteroids and tobacco are significant variables to predict ON. Using prior information provides good accuracy, specificity, and sensitivity to the prediction. Further studies need to be conducted to validate the model using a testing set

OBJETIVO: Este trabajo busca determinar un modelo predictivo de desarrollo de osteonecrosis (ON) en individuos diagnosticados con lupus eritematoso sistémico (LES) utilizando factores farmacológicos, demográficos y psicoactivos. MÉTODO: Se realizó una revisión bibliográfica para construir una encuesta, la cual fue administrada a individuos con LES a lo largo de Chile durante un periodo de 3 semanas. En este trabajo se utilizó una muestra de 46 registros de datos no identificados. Se desarrollaron 2 modelos de regresión logística bayesiana con información a priori no informativa e informativa, y también se desarrolló un modelo comparativo utilizando bosques aleatorios. Los modelos fueron validados usando validación cruzada. RESULTADOS: Se usaron las variables significativas promedio de corticosteroides por día (mg) y consumo de tabaco. Bosques aleatorios provee una precisión y sensibilidad alta, pero una baja especificidad. La regresión logística bayesiana con información a priori incrementó el valor de la especificidad. CONCLUSIONES: Este trabajo ha determinado que el uso de corticosteroides y tabaco son variables significativas para predecir ON. Usando información a priori arroja buenos resultados en precisión, especificidad y sensibilidad en la predicción. Se requieren realizar más estudios aumentando el tamaño de la muestra para validar el modelo usando un conjunto de prueba

Prediction of Chronic Disease-Related Inpatient Prolonged Length of Stay Using Machine Learning Algorithms.

OBJECTIVES: The study aimed to develop and compare predictive models based on supervised machine learning algorithms for predicting the prolonged length of stay (LOS) of hospitalized patients diagnosed with five different chronic conditions. METHODS: An administrative claim dataset (2008-2012) of a regional network of nine hospitals in the Tampa Bay area, Florida, USA, was used to develop the prediction models. Features were extracted from the dataset using the International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) codes. Five learning algorithms, namely, decision tree C5.0, linear support vector machine (LSVM), k-nearest neighbors, random forest, and multi-layered artificial neural networks, were used to build the model with semi-supervised anomaly detection and two feature selection methods. Issues with the unbalanced nature of the dataset were resolved using the Synthetic Minority Over-sampling Technique (SMOTE). RESULTS: LSVM with wrapper feature selection performed moderately well for all patient cohorts. Using SMOTE to counter data imbalances triggered a tradeoff between the model's sensitivity and specificity, which can be masked under a similar area under the curve. The proposed aggregate rank selection approach resulted in a balanced performing model compared to other criteria. Finally, factors such as comorbidity conditions, source of admission, and payer types were associated with the increased risk of a prolonged LOS. CONCLUSIONS: Prolonged LOS is mostly associated with pre-intraoperative clinical and patient socioeconomic factors. Accurate patient identification with the risk of prolonged LOS using the selected model can provide hospitals a better tool for planning early discharge and resource allocation, thus reducing avoidable hospitalization costs.

Osteonecrosis in individuals with systemic lupus erythematosus: A predictive model.

OBJECTIVE: This work attempts to provide a model to predict the development of osteonecrosis (ON) in individuals with systemic lupus erythematosus (SLE) using pharmacological, demographic, and psychoactive factors. METHOD: A review of the literature was conducted to construct a survey administered across Chile to individuals with SLE during a period of three weeks. This work used a sample size of 46 de-identified data records. Two Bayesian logistic regression models were created, with non-informative prior and informative prior distributions, and a random forest model was done for comparison. All models were cross-validated. RESULTS: The significant variables used were mean corticosteroids per day (mg) and tobacco use. The random forest model provided good accuracy and sensitivity, but low specificity. Bayesian logistic regression with prior information increased the specificity. CONCLUSIONS: This work determined that the use of corticosteroids and tobacco are significant variables to predict ON. Using prior information provides good accuracy, specificity, and sensitivity to the prediction. Further studies need to be conducted to validate the model using a testing set.

Comparison of Machine Learning Algorithms for the Prediction of Preventable Hospital Readmissions.

A diverse universe of statistical models in the literature aim to help hospitals understand the risk factors of their preventable readmissions. However, these models are usually not necessarily applicable in other contexts, fail to achieve good discriminatory power, or cannot be compared with other models. We built and compared predictive models based on machine learning algorithms for 30-day preventable hospital readmissions of Medicare patients. This work used the same inclusion/exclusion criteria for diseases used by the Centers for Medicare and Medicaid Services. In addition, risk stratification techniques were implemented to study covariate behavior on each risk strata. The new models resulted in improved performance measured by the area under the receiver operating characteristic curve. Finally, factors such as higher length of stay, disease severity index, being discharged to a hospital, and primary language other than English were associated with increased risk to be readmitted within 30 days. In the future, better predictive models for 30-day preventable hospital readmissions can point to the development of systems that identify patients at high risk and lead to the implementation of interventions (e.g., discharge planning and follow-up) to those patients, providing consistent improvement in the quality and efficiency of the healthcare system.

A strategic gaming model for health information exchange markets.

Current market conditions create incentives for some providers to exercise control over patient data in ways that unreasonably limit its availability and use. Here we develop a game theoretic model for estimating the willingness of healthcare organizations to join a health information exchange (HIE) network and demonstrate its use in HIE policy design. We formulated the model as a bi-level integer program. A quasi-Newton method is proposed to obtain a strategy Nash equilibrium. We applied our modeling and solution technique to 1,093,177 encounters for exchanging information over a 7.5-year period in 9 hospitals located within a three-county region in Florida. Under a set of assumptions, we found that a proposed federal penalty of up to $2,000,000 has a higher impact on increasing HIE adoption than current federal monetary incentives. Medium-sized hospitals were more reticent to adopt HIE than large-sized hospitals. In the presence of collusion among multiple hospitals to not adopt HIE, neither federal incentives nor proposed penalties increase hospitals' willingness to adopt. Hospitals' apathy toward HIE adoption may threaten the value of inter-connectivity even with federal incentives in place. Competition among hospitals, coupled with volume-based payment systems, creates no incentives for smaller hospitals to exchange data with competitors. Medium-sized hospitals need targeted actions (e.g., outside technological assistance, group purchasing arrangements) to mitigate market incentives to not adopt HIE. Strategic game theoretic models help to clarify HIE adoption decisions under market conditions at play in an extremely complex technology environment.

Activating clinical trials: a process improvement approach.

BACKGROUND: The administrative process associated with clinical trial activation has been criticized as costly, complex, and time-consuming. Prior research has concentrated on identifying administrative barriers and proposing various solutions to reduce activation time, and consequently associated costs. Here, we expand on previous research by incorporating social network analysis and discrete-event simulation to support process improvement decision-making. METHODS: We searched for all operational data associated with the administrative process of activating industry-sponsored clinical trials at the Office of Clinical Research of the University of South Florida in Tampa, Florida. We limited the search to those trials initiated and activated between July 2011 and June 2012. We described the process using value stream mapping, studied the interactions of the various process participants using social network analysis, and modeled potential process modifications using discrete-event simulation. RESULTS: The administrative process comprised 5 sub-processes, 30 activities, 11 decision points, 5 loops, and 8 participants. The mean activation time was 76.6 days. Rate-limiting sub-processes were those of contract and budget development. Key participants during contract and budget development were the Office of Clinical Research, sponsors, and the principal investigator. Simulation results indicate that slight increments on the number of trials, arriving to the Office of Clinical Research, would increase activation time by 11 %. Also, incrementing the efficiency of contract and budget development would reduce the activation time by 28 %. Finally, better synchronization between contract and budget development would reduce time spent on batching documentation; however, no improvements would be attained in total activation time. CONCLUSION: The presented process improvement analytic framework not only identifies administrative barriers, but also helps to devise and evaluate potential improvement scenarios. The strength of our framework lies in its system analysis approach that recognizes the stochastic duration of the activation process and the interdependence between process activities and entities.

Preventable Readmission Risk Factors for Patients With Chronic Conditions.

Evidence indicates that the largest volume of hospital readmissions occurs among patients with preexisting chronic conditions. Identifying these patients can improve the way hospital care is delivered and prioritize the allocation of interventions. In this retrospective study, we identify factors associated with readmission within 30 days based on claims and administrative data of nine hospitals from 2005 to 2012. We present a data inclusion and exclusion criteria to identify potentially preventable readmissions. Multivariate logistic regression models and a Cox proportional hazards extension are used to estimate the readmission risk for 4 chronic conditions (congestive heart failure [CHF], chronic obstructive pulmonary disease [COPD], acute myocardial infarction, and type 2 diabetes) and pneumonia, known to be related to high readmission rates. Accumulated number of admissions and discharge disposition were identified to be significant factors across most disease groups. Larger odds of readmission were associated with higher severity index for CHF and COPD patients. Different chronic conditions are associated with different patient and case severity factors, suggesting that further studies in readmission should consider studying conditions separately.

A user needs assessment to inform health information exchange design and implementation.

BACKGROUND: Important barriers for widespread use of health information exchange (HIE) are usability and interface issues. However, most HIEs are implemented without performing a needs assessment with the end users, healthcare providers. We performed a user needs assessment for the process of obtaining clinical information from other health care organizations about a hospitalized patient and identified the types of information most valued for medical decision-making. METHODS: Quantitative and qualitative analysis were used to evaluate the process to obtain and use outside clinical information (OI) using semi-structured interviews (16 internists), direct observation (750 h), and operational data from the electronic medical records (30,461 hospitalizations) of an internal medicine department in a public, teaching hospital in Tampa, Florida. RESULTS: 13.7 % of hospitalizations generate at least one request for OI. On average, the process comprised 13 steps, 6 decisions points, and 4 different participants. Physicians estimate that the average time to receive OI is 18 h. Physicians perceived that OI received is not useful 33-66 % of the time because information received is irrelevant or not timely. Technical barriers to OI use included poor accessibility and ineffective information visualization. Common problems with the process were receiving extraneous notes and the need to re-request the information. Drivers for OI use were to trend lab or imaging abnormalities, understand medical history of critically ill or hospital-to-hospital transferred patients, and assess previous echocardiograms and bacterial cultures. About 85 % of the physicians believe HIE would have a positive effect on improving healthcare delivery. CONCLUSIONS: Although hospitalists are challenged by a complex process to obtain OI, they recognize the value of specific information for enhancing medical decision-making. HIE systems are likely to have increased utilization and effectiveness if specific patient-level clinical information is delivered at the right time to the right users.

Bounceback branchpoints: using conditional inference trees to analyze readmissions.

BACKGROUND: We sought to identify risks for 30-day readmission in patients undergoing colorectal surgery. METHODS: We reviewed 2011 American College of Surgery National Surgical Quality Improvement Program data to identify patients readmitted after colorectal surgery. We found 3,228 readmissions from 30,412 records. Using statistically suggestive variables from logistic regression (P < .1), we built conditional inference trees (CTREES) with subsampled records to identify combined risks. RESULTS: Logistic regression identified 27 potentially significant variables. Using these in new logistic regression and CTREES, we found classification accuracies of 0.70 and 0.63, respectively. CTREES predicted that the majority of patients who required reoperation during their hospitalization were predicted to require readmission (n = 496), along with the majority of patients who developed organ space infection (n = 671). Patients with deep infections discharged ≤10 days from their index operation required readmission in 443 of 459 of records; this approach predicted a >99% risk of readmission in patients with these infections who were discharged ≤5 days (220/222). Additionally, >90% (253/271) of patients who returned to the operating room and were discharged ≤8 days from the first operation are predicted to return. CONCLUSION: Subgroups identified using the CTREES model demonstrate that patients with deep space infections or who return to the operating room have a greater readmission rate if they are discharged early. Modeled patients found to have organ space infections and who returned to the operating room had 30-day readmission risks of >50%, with those discharged a rate of >90%. We show herein that CTREES can be used with retrospective data on surgery populations to bring hidden patterns into relief.

Clustering-based methodology for analyzing near-miss reports and identifying risks in healthcare delivery.

Near-miss reports are qualitative descriptions of events that could have harmed patients but did not due to a timely intervention or a convenient evolution of the circumstances. Near-miss reporting has increasingly become a relevant tool to support patient safety efforts since they provide some evidence of risk in the system before patients suffer adverse consequences. Near-misses are usually classified into pre-specified categories that correspond to sources of risk in the system or its processes. Their analysis often consists of tallying classified near-misses to determine risk priorities based on frequency within each pre-specified risk category. Our research aims to use different combinations of near-miss reports to find potential sources of risk. We propose an unsupervised bisecting k-prototypes algorithm for clustering coded near-miss reports to identify relationships between events that would not otherwise have been easily identified. Subsequent study of resulting clusters will lead to the identification of potentially dangerous, but unsuspected system interactions. We illustrate or methodology with preliminary results of its implementation at the University of South Florida Health clinics.

Evaluation of an error-reduction training program for surgical residents.

PURPOSE: To reduce errors in surgery using a resident training program based on a taxonomy that highlights three kinds of errors: judgment, inattention to detail, and problem understanding. METHOD: The training program module at the University of South Florida incorporated a three-item situational judgment test, video training (which included a lecture and behavior modeling), and role-plays (in which residents participated and received feedback from faculty). Two kinds of outcome data were collected from 33 residents during 2006-2007: (1) behaviors during the training and (2) on-the-job surgical complication records 12 months before and 6 months after training. For the data collected during training, participants were assigned to a condition (19 video condition, 13 control condition); for the data collected on the job, an interrupted time series design was used. RESULTS: Data from 32 residents were analyzed (one resident's data were excluded). One of the situational judgment items improved significantly over time (d = 0.45); the other two did not (d = 0.36, 0.25). Surgical complications and errors decreased over the course of the study (the correlation between complications and time in months was r = -0.47, for errors and time, r = -0.55). Effects of video behavior modeling on specific errors measured during role-plays were not significant (effect sizes for binary outcomes were phi = -0.05 and phi = 0.01, and for continuous outcomes, d ranged from -0.02 to 0.34). CONCLUSIONS: The training seemed to reduce errors in surgery, but the training had little effect on the specific kinds of errors targeted during training.

Human error, not communication and systems, underlies surgical complications.

OBJECTIVE: This study prospectively assesses the underlying errors contributing to surgical complications over a 12-month period in a complex academic department of surgery using a validated scoring template. BACKGROUND: Studies in "high reliability organizations" suggest that systems failures are responsible for errors. Reports from the aviation industry target communication failures in the cockpit. No prior studies have developed a validated classification system and have determined the types of errors responsible for surgical complications. METHODS: A classification system of medical error during operation was created, validated, and data collected on the frequency, type, and severity of medical errors in 9,830 surgical procedures. Statistical analysis of concordance, validity, and reliability were performed. RESULTS: Reported major complications occurred in 332 patients (3.4%) with error in 78.3%: errors in surgical technique (63.5%), judgment errors (29.6%), inattention to detail (29.3%), and incomplete understanding (22.7%). Error contributed more than 50% to the complication in 75%. A total of 13.6% of cases had error but no injury, 34.4% prolongation of hospitalization, 25.1% temporary disability, 8.4% permanent disability, and 16.0% death. In 20%, the error was a "mistake" (the wrong thing), and in 58% a "slip" (the right thing incorrectly). System errors (2%) and communication errors (2%) were infrequently identified. CONCLUSIONS: After surgical technique, most surgical error was caused by human factors: judgment, inattention to detail, and incomplete understanding, and not to organizational/system errors or breaks in communication. Training efforts to minimize error and enhance patient safety must address human factor causes of error.