Data Omission by Physician Trainees on ICU Rounds

Association of Sociodemographic Factors With Overtriage, Undertriage, and Value of Care After Major Surgery

Objective

To determine whether certain patients are vulnerable to errant triage decisions immediately after major surgery and whether there are unique sociodemographic phenotypes within overtriaged and undertriaged cohorts.

Background

In a fair system, overtriage of low-acuity patients to intensive care units (ICUs) and undertriage of high-acuity patients to general wards would affect all sociodemographic subgroups equally.

Methods

This multicenter, longitudinal cohort study of hospital admissions immediately after major surgery compared hospital mortality and value of care (risk-adjusted mortality/total costs) across 4 cohorts: overtriage (N = 660), risk-matched overtriage controls admitted to general wards (N = 3077), undertriage (N = 2335), and risk-matched undertriage controls admitted to ICUs (N = 4774). K-means clustering identified sociodemographic phenotypes within overtriage and undertriage cohorts.

Results

Compared with controls, overtriaged admissions had a predominance of male patients (56.2% vs 43.1%, P < 0.001) and commercial insurance (6.4% vs 2.5%, P < 0.001); undertriaged admissions had a predominance of Black patients (28.4% vs 24.4%, P < 0.001) and greater socioeconomic deprivation. Overtriage was associated with increased total direct costs [$16.2K ($11.4K-$23.5K) vs $14.1K ($9.1K-$20.7K), P < 0.001] and low value of care; undertriage was associated with increased hospital mortality (1.5% vs 0.7%, P = 0.002) and hospice care (2.2% vs 0.6%, P < 0.001) and low value of care. Unique sociodemographic phenotypes within both overtriage and undertriage cohorts had similar outcomes and value of care, suggesting that triage decisions, rather than patient characteristics, drive outcomes and value of care.

Conclusions

Postoperative triage decisions should ensure equality across sociodemographic groups by anchoring triage decisions to objective patient acuity assessments, circumventing cognitive shortcuts and mitigating bias.

Using ChatGPT-4 to Create Structured Medical Notes From Audio Recordings of Physician-Patient Encounters: Comparative Study

BACKGROUND

Medical documentation plays a crucial role in clinical practice, facilitating accurate patient management and communication among health care professionals. However, inaccuracies in medical notes can lead to miscommunication and diagnostic errors. Additionally, the demands of documentation contribute to physician burnout. Although intermediaries like medical scribes and speech recognition software have been used to ease this burden, they have limitations in terms of accuracy and addressing provider-specific metrics. The integration of ambient artificial intelligence (AI)-powered solutions offers a promising way to improve documentation while fitting seamlessly into existing workflows.

OBJECTIVE

This study aims to assess the accuracy and quality of Subjective, Objective, Assessment, and Plan (SOAP) notes generated by ChatGPT-4, an AI model, using established transcripts of History and Physical Examination as the gold standard. We seek to identify potential errors and evaluate the model's performance across different categories.

METHODS

We conducted simulated patient-provider encounters representing various ambulatory specialties and transcribed the audio files. Key reportable elements were identified, and ChatGPT-4 was used to generate SOAP notes based on these transcripts. Three versions of each note were created and compared to the gold standard via chart review; errors generated from the comparison were categorized as omissions, incorrect information, or additions. We compared the accuracy of data elements across versions, transcript length, and data categories. Additionally, we assessed note quality using the Physician Documentation Quality Instrument (PDQI) scoring system.

RESULTS

Although ChatGPT-4 consistently generated SOAP-style notes, there were, on average, 23.6 errors per clinical case, with errors of omission (86%) being the most common, followed by addition errors (10.5%) and inclusion of incorrect facts (3.2%). There was significant variance between replicates of the same case, with only 52.9% of data elements reported correctly across all 3 replicates. The accuracy of data elements varied across cases, with the highest accuracy observed in the "Objective" section. Consequently, the measure of note quality, assessed by PDQI, demonstrated intra- and intercase variance. Finally, the accuracy of ChatGPT-4 was inversely correlated to both the transcript length (P=.05) and the number of scorable data elements (P=.05).

CONCLUSIONS

Our study reveals substantial variability in errors, accuracy, and note quality generated by ChatGPT-4. Errors were not limited to specific sections, and the inconsistency in error types across replicates complicated predictability. Transcript length and data complexity were inversely correlated with note accuracy, raising concerns about the model's effectiveness in handling complex medical cases. The quality and reliability of clinical notes produced by ChatGPT-4 do not meet the standards required for clinical use. Although AI holds promise in health care, caution should be exercised before widespread adoption. Further research is needed to address accuracy, variability, and potential errors. ChatGPT-4, while valuable in various applications, should not be considered a safe alternative to human-generated clinical documentation at this time.

Examining Pediatric Resident Electronic Health Records Use During Prerounding: Mixed Methods Observational Study

BACKGROUND

Electronic health records (EHRs) play a substantial role in modern health care, especially during prerounding, when residents gather patient information to inform daily care decisions of the care team. The effective use of the EHR system is crucial for efficient and frustration-free prerounding. Ideally, the system should be designed to support efficient user interactions by presenting data effectively and providing easy navigation between different pages. Additionally, training on the system should aim to make user interactions more efficient by familiarizing the users with best practices that minimize interaction time while using the full potential of the system's capabilities. However, formal training on EHR systems often falls short of providing residents with all the necessary EHR-related skills, leading to the adoption of inefficient practices and the underuse of the system's full range of capabilities.

OBJECTIVE

This study aims to examine the efficiency of EHR use during prerounding among pediatric residents, assess the effect of experience level on EHR use, and identify areas for improvement in EHR design and training.

METHODS

A mixed methods approach was used, involving a self-reported survey and video analysis of prerounding practices of the entire population of pediatric residents from a large teaching hospital in the South Atlantic Region. The residents were stratified by experience level by postgraduate year. Data were collected on the number of pages accessed, duration of prerounding, task completion rates, and effective use of data sources. Observational and qualitative data complemented the quantitative analysis. Our study followed the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) reporting guidelines, ensuring completeness and transparency of reporting.

RESULTS

Of the 30 pediatric residents, 20 were included in the analyses; of these, 16 (80%) missed at least 1 step during prerounding. Although more experienced residents on average omitted fewer steps, 4 (57%) of the 7 most experienced residents still omitted at least 1 step. On average, residents took 6.5 minutes to round each patient and accessed 21 pages within the EHR during prerounding; no statistically significant differences were observed between experience levels for prerounding times (P=.48) or number of pages accessed (P=.92). The use of aggregated data pages within the EHR system neither seem to improve prerounding times nor decrease the number of pages accessed.

CONCLUSIONS

The findings suggest that EHR design should be improved to better support user needs, and hospitals should adopt more effective training programs to familiarize residents with the system's capabilities. We recommend implementing prerounding checklists and providing ongoing EHR training programs for health care practitioners. Despite the generalizability of limitations of our study in terms of sample size and specialization, it offers valuable insights for future research to investigate the impact of EHR use on patient outcomes and satisfaction, as well as identify factors that contribute to efficient and effective EHR usage.

Effects of Unit Census and Patient Acuity Levels on Discussions During Patient Rounds

OBJECTIVES

PICU teams adapt the duration of patient rounding discussions to accommodate varying contextual factors, such as unit census and patient acuity. Although studies establish that shorter discussions can lead to the omission of critical patient information, little is known about how teams adapt their rounding discussions about essential patient topics (i.e., introduction/history, acute clinical status, care plans) in response to changing contexts. To fill this gap, we examined how census and patient acuity impact time spent discussing essential topics during individual patient encounters.

DESIGN

Observational study.

SETTING

PICU at a university-affiliated children's hospital, Toronto, ON, Canada.

SUBJECTS

Interprofessional morning rounding teams.

INTERVENTIONS

None.

MEASUREMENTS AND MAIN RESULTS

We observed 165 individual patient encounters during morning rounds over 10 weeks. Regardless of census or patient acuity, the duration of patient introductions/history did not change. When census was high versus low, acute clinical status discussions significantly decreased for both low acuity patients (00 min:50 s high census; 01 min:39 s low census; -49.5% change) and high acuity patients (01 min:10 s high census; 02 min:02 s low census; -42.6% change). Durations of care plan discussions significantly reduced as a function of census (01 min:19 s high census; 02 min:52 s low census; -54.7% change) for low but not high acuity patients.

CONCLUSIONS

Under high census and patient acuity levels, rounding teams disproportionately shorten time spent discussing essential patient topics. Of note, while teams preserved time to plan the care for acute patients, they cut care plan discussions of low acuity patients. This study provides needed detail regarding how rounding teams adapt their discussions of essential topics and establishes a foundation for consideration of varying contextual factors in the design of rounding guidelines. As ICUs are challenged with increasing census and patient acuity levels, it is critical that we turn our attention to these contextual aspects and understand how these adaptations impact clinical outcomes to address them.

Effects of multidisciplinary rounds and checklist in an Intensive Care Unit: a mixed methods study

OBJECTIVES

to analyze the implementation of multidisciplinary checklist-directed rounds before health indicators and multidisciplinary team perception of an Intensive Care Unit.

METHODS

a mixed methods study, with an explanatory sequential design, carried out at a hospital in southern Brazil, from September 2020 to August 2021. The integration of quantitative and qualitative data was combined by connection.

RESULTS

after the implementation of checklist-directed rounds, there was a significant reduction in hospital stay from ventilator-associated pneumonia, urinary tract infection and daily invasive device use. The investigated practice is essential for comprehensive care, harm reduction, effective work and critical patient safety.

CONCLUSIONS

the multidisciplinary rounds with checklist use reduced data on health indicators of critically ill patients and was considered a vital practice in the intensive care setting.

Chlorhexidine (di)gluconate locking device for central line infection prevention in intensive care unit patients: a multi-unit, pilot randomized controlled trial

PURPOSE

Intensive care unit (ICU) patients are at risk for central line-associated bloodstream infection (CLABSI) with significant attributable mortality and increased hospital length of stay, readmissions, and costs. Chlorhexidine (di)gluconate (CHG) is used as a disinfectant for central line insertion; however, the feasibility and efficacy of using CHG as a locking solution is unknown.

METHODS

Patients with a central venous access device (CVAD) in situ were randomized to standard care or a CHG lock solution (CHGLS) within 72 hours of ICU admission. The CHG solution was instilled in the lumen of venous catheters not actively infusing. CVAD blood cultures were taken at baseline and every 48 hours. The primary outcome was feasibility including recruitment rate, consent rate, protocol adherence, and staff uptake. Secondary outcomes included CVAD colonization, bacteraemia, and clinical endpoints.

RESULTS

Of 3,848 patients screened, 122 were eligible for the study and consent was obtained from 82.0% of the patients or substitute decision makers approached. Fifty participants were allocated to each group. Tracking logs indicated that the CHGLS was used per protocol 408 times. Most nurses felt comfortable using the CHGLS. The proportion of central line colonization was significantly higher in the standard care group with 40 (29%) versus 26 (18.7%) in the CHGLS group (P=0.009).

CONCLUSIONS

Using a device that delivers CHG into CVADs was feasible in the ICU. Findings from this trial will inform a full-scale randomized controlled trial and provide preliminary data on the effectiveness of CHGLS.

TRIAL REGISTRATION

ClinicalTrials.gov Identifier NCT03309137, registered on October 13, 2017.

Usability challenges with electronic health records (EHRs) during prerounding on pediatric inpatients

Objective

Prerounding is critical for a healthcare team to develop a shared understanding of the patient’s condition and to develop a care plan. However, the design of electronic health records (EHRs) often makes prerounding inefficient, ineffective, and time consuming. The goal of this study was to observe how residents use the EHR while prerounding to identify usability challenges associated with the design of EHRs.

Materials and Methods

Thirty residents were tasked to preround 2 pediatric patients using the think-aloud protocol. The data from the surveys, video recordings, and think-aloud comments were analyzed to identify usability issues related to EHR. The time it took for participants to complete the 6 required prerounding tasks were calculated and the pages most commonly accessed were noted.

Results

Participants spent on average 6.5 min prerounding each patient with the most time spent on checking lab results and reviewing notes. Twenty-eight distinct pages were visited by at least 2 participants, mostly due to a lack of interconnectivity between related data across pages. Usability issues with the most commonly used pages include: data overload, missing/hidden information, difficulty identifying trends, and having to conduct manual calculations.

Conclusions

We list usability issues and provide a set of recommendations to remedy these issues that include: reducing information access cost, creating a checklist, automate calculations, and standardizing notes and EHR training. Ideally, the outcome of this work will help improve EHR design to maximize the time clinicians spend interacting with and providing care to their patients.

Residents spend a significant amount of time using the electronic health record (EHR) system and these interactions can often be frustrating, inefficient, and time consuming. This study focuses on identifying challenges associated with EHR use during prerounding by residents. As part of this study, we observed 30 residents while they preround 2 pediatric care patients using the EHR system. The results of our study showed that the design of EHRs currently do not support the tasks the residents need to perform while prerounding. Patient data are spread across numerous pages within the EHR system and the information is often not displayed in the most user friendly manner. This adversely affects the effectiveness and efficiency of prerounding. This study provides a set of recommendations to support prerounding that include dashboard design guidelines, standardizing data entry, and automating certain tasks within the EHR.

Design, Implementation, Utilization, and Sustainability of a Fast Healthcare Interoperability Resources-Based Inpatient Rounding List

OBJECTIVE

We designed and implemented an application programming interface (API)-based electronic health record (EHR)-integrated rounding list and evaluated acceptability, clinician satisfaction, information accuracy, and efficiency related to the application.

METHODS

We developed and integrated an application, employing iterative design techniques with user feedback. EHR and application user action logs, as well as hospital safety reports, were evaluated. Rounding preparation characteristics were obtained through surveys before and after application integration. To evaluate usability, inpatient providers, including residents, fellows, and attendings were surveyed 2 weeks prior to and 6 months after enterprise-wide EHR application integration. Our primary outcome was provider time savings measured by user action logs; secondary outcomes include provider satisfaction.

RESULTS

The application was widely adopted by inpatient providers, with more than 69% of all inpatients queried by the application within 6 months of deployment. Application utilization was sustained throughout the study period with 79% (interquartile range [IQR]: 76, 82) of enterprise-wide unique patients accessed per weekday. EHR action logs showed application users spent -3.24 minutes per day (95% confidence interval [CI]: -6.8, 0.33), p = 0.07 within the EHR compared with nonusers. Median self-reported chart review time for attendings decreased from 30 minutes (IQR: 15, 60) to 20 minutes (IQR: 10, 45) after application integration (p = 0.04). Self-reported sign-out preparation time decreased by a median of 5 minutes (p < 0.01), and providers were better prepared for hand-offs (p = 0.02). There were no increased safety reports during the study period.

CONCLUSION

This study demonstrates successful integration of a rounding application within a commercial EHR using APIs. We demonstrate increasing both provider-reported satisfaction and time savings. Rounding lists provided more accurate and timely information for rounds. Application usage was sustained across multiple specialties at 42 months. Other application designers should consider data density, optimization of provider workflows, and using real-time data transfer using novel tools when designing an application.

Efeitos de rounds multidisciplinares e checklist em Unidade de Terapia Intensiva: estudo de método misto

RESUMO Objetivos: analisar a implementação de rounds multidisciplinares direcionados por checklist frente aos indicadores de saúde e a percepção da equipe multiprofissional de uma Unidade de Terapia Intensiva. Métodos: estudo de método misto, com desenho sequencial explanatório, realizado em um hospital do sul do Brasil, no período de setembro de 2020 a agosto de 2021. A integração dos dados quantitativos e qualitativos foi combinada por conexão. Resultados: após a implementação dos rounds direcionados por checklist, constatou-se redução significativa no tempo de internação por pneumonias associadas à ventilação mecânica, infecção do trato urinário e nos dias de uso de dispositivos invasivos. A prática investigada é essencial para o cuidado integral, a redução de danos, o trabalho eficaz e a segurança do paciente crítico. Conclusões: os rounds multidisciplinares com uso de checklist reduziram os dados dos indicadores de saúde de pacientes críticos e foi considerado como prática vital no cenário de cuidados intensivos.

Round multiprofissional com checklist: associação com a melhoria na segurança do paciente em terapia intensiva

RESUMO Objetivo Verificar a associação entre round multiprofissional com uso de checklist e práticas de segurança do paciente por profissionais de saúde de uma unidade de terapia intensiva. Método Estudo de método misto, delineado pela abordagem sequencial explanatória, realizado em um hospital do sul do Brasil. Os dados quantitativos foram analisados por meio de regressão de Poisson e os dados qualitativos, pela análise de conteúdo. Fez-se a análise integrada por meio da combinação explicada/conectada. Resultados No período pós-implementação dos rounds com uso sistemático de checklist houve melhora significativa da profilaxia de tromboembolia venosa, sedação leve, redução dos dias de uso de ventilação mecânica, cateter venoso central e de sonda vesical de demora. Conclusão O round multiprofissional com uso sistemático de checklist, associado com a melhoria nas práticas de segurança do paciente, foi considerado como uma estratégia que assegura melhores cuidados em terapia intensiva e favorece a satisfação no trabalho.

Multiprofessional round with checklist: association with the improvement in patient safety in intensive care

ABSTRACT Objective To verify the association between a multiprofessional round with the use of checklists and patient safety practices by health professionals in an intensive care unit. Method Mixed-method study, delineated by the sequential explanatory approach, conducted in a hospital in southern Brazil. Quantitative data were analyzed using Poisson regression, and qualitative data, using content analysis. The integrated analysis was performed through the explained/connected combination. Results In the post-implementation period of the rounds with systematic use of the checklist, there was a significant improvement in the prophylaxis of venous thromboembolism, light sedation, reduction in the days of use of mechanical ventilation, central venous catheter and indwelling urinary catheter. Conclusion The multiprofessional round with the systematic use of checklist, associated with the improvement in patient safety practices, was considered as a strategy that ensures better care in intensive care and favors job satisfaction.

The Next-Generation Electronic Health Record in the ICU: A Focus on User-Technology Interface to Optimize Patient Safety and Quality

As a knowledge-based field of medicine, critical care medicine has benefited from the use of the electronic health records (EHRs) in daily practice, as intensive care unit (ICU) patients generate thousands of pieces of clinical data each day.1 ICU teams must review, interpret, and take action on these data points when managing multiple patients in a time-constrained environment. The increasing number of available data facts to be processed by ICU clinicians for decision-making surpasses human cognitive capacity. ICU physicians described the current display and representation of patient data in the EHR as suboptimum. Performance dashboards are an information delivery system that display the most important information about performance objectives to ICU directors, allowing them to monitor and manage their ICU performance more effectively. The development of visualization dashboards that monitor ICU performance will still need to adhere to usability principles such as Jakob Nielsen's heuristics. The goal of improving EHR interfaces will directly enhance provider well-being, patient outcomes, and quality of care.

Evaluating Electronic Health Record Limitations and Time Expenditure in a German Medical Center

OBJECTIVES

This study set out to obtain a general profile of physician time expenditure and electronic health record (EHR) limitations in a large university medical center in Germany. We also aim to illustrate the merit of a tool allowing for easier capture and prioritization of specific clinical needs at the point of care for which the current study will inform development in subsequent work.

METHODS

Nineteen physicians across six different departments participated in this study. Direct clinical observations were conducted with 13 out of 19 physicians for a total of 2,205 minutes, and semistructured interviews were conducted with all participants. During observations, time was measured for larger activity categories (searching information, reading information, documenting information, patient interaction, calling, and others). Semistructured interviews focused on perceived limitations, frustrations, and desired improvements regarding the EHR environment.

RESULTS

Of the observed time, 37.1% was spent interacting with the health records (9.0% searching, 7.7% reading, and 20.5% writing), 28.0% was spent interacting with patients corrected for EHR use (26.9% of time in a patient's presence), 6.8% was spent calling, and 28.1% was spent on other activities. Major themes of discontent were a spread of patient information, high and often repeated documentation burden, poor integration of (new) information into workflow, limits in information exchange, and the impact of such problems on patient interaction. Physicians stated limited means to address such issues at the point of care.

CONCLUSION

In the study hospital, over one-third of physicians' time was spent interacting with the EHR, environment, with many aspects of used systems far from optimal and no convenient way for physicians to address issues as they occur at the point of care. A tool facilitating easier identification and registration of issues, as they occur, may aid in generating a more complete overview of limitations in the EHR environment.

Why Is the Electronic Health Record So Challenging for Research and Clinical Care?

BACKGROUND

 The electronic health record (EHR) has become increasingly ubiquitous. At the same time, health professionals have been turning to this resource for access to data that is needed for the delivery of health care and for clinical research. There is little doubt that the EHR has made both of these functions easier than earlier days when we relied on paper-based clinical records. Coupled with modern database and data warehouse systems, high-speed networks, and the ability to share clinical data with others are large number of challenges that arguably limit the optimal use of the EHR OBJECTIVES:  Our goal was to provide an exhaustive reference for those who use the EHR in clinical and research contexts, but also for health information systems professionals as they design, implement, and maintain EHR systems.

METHODS

 This study includes a panel of 24 biomedical informatics researchers, information technology professionals, and clinicians, all of whom have extensive experience in design, implementation, and maintenance of EHR systems, or in using the EHR as clinicians or researchers. All members of the panel are affiliated with Penn Medicine at the University of Pennsylvania and have experience with a variety of different EHR platforms and systems and how they have evolved over time.

RESULTS

 Each of the authors has shared their knowledge and experience in using the EHR in a suite of 20 short essays, each representing a specific challenge and classified according to a functional hierarchy of interlocking facets such as usability and usefulness, data quality, standards, governance, data integration, clinical care, and clinical research.

CONCLUSION

 We provide here a set of perspectives on the challenges posed by the EHR to clinical and research users.

Methods and measures to quantify ICU patient heterogeneity

Patients in intensive care units are heterogeneous and the daily prediction of their days to discharge (DTD) a complex task that practitioners and computers are not always able to solve satisfactorily. In order to make more precise DTD predictors, it is necessary to have tools for the analysis of the heterogeneity of the patients. Unfortunately, the number of publications in this field is almost non-existent. In order to alleviate this lack of tools, we propose four methods and their corresponding measures to quantify the heterogeneity of intensive patients in the process of determining the DTD. These new methods and measures have been tested with patients admitted over four years to a tertiary hospital in Spain. The results deepen the understanding of the intensive patient and can serve as a basis for the construction of better DTD predictors.

Factors Associated With Parental Participation in Family-Centered Rounds

OBJECTIVES

Although families positively perceive family-centered rounds (FCR), factors associated with engagement have been examined in few studies. Our objective for this study was to test the hypothesis that inviting the parent to speak and nurse presence are associated with parent engagement during FCR.

METHODS

We conducted a cross-sectional study with English-speaking parents (N = 199) of inpatients on the pediatric hospital medicine service at an academic medical center. We used a standardized checklist to record outcomes of engagement (number of questions asked and participation occurrences), predictor variables (team invited parent to speak, nurse presence), and other encounter-related variables. Parents were surveyed to assess parent and child characteristics and experiences during FCR. We examined parent, child, and encounter characteristic associations with the above outcomes using bivariate analyses and (for those associated in bivariate analyses) Poisson regressions.

RESULTS

Inviting the parent to speak was independently associated with the number of questions asked (incident rate ratio [IRR] 1.4; 95% confidence interval [CI] 1.1-1.7). Trusting the medical team was inversely associated with questions asked (IRR 0.8; 95% CI 0.6-0.97). Factors associated with total participation included invitation for the parent to speak (IRR 1.5; 95% CI 1.3-1.6), nurse presence (IRR 1.3; 95% CI 1.1-1.5), white race (IRR 1.2; 95% CI 1.1-1.4), clerkship student presentation (IRR 1.2; 95% CI 1.03-1.3), and parent inclusion in rounding arrangement (IRR 1.5; 95% CI 1.05-2).

CONCLUSIONS

Parents present during FCR are more engaged when invited to speak. Nurse presence was associated with total parent participation. Future studies to inform interventions to optimize engagement are warranted.

Artificial Intelligence and Surgical Decision-making

Importance

Surgeons make complex, high-stakes decisions under time constraints and uncertainty, with significant effect on patient outcomes. This review describes the weaknesses of traditional clinical decision-support systems and proposes that artificial intelligence should be used to augment surgical decision-making.

Observations

Surgical decision-making is dominated by hypothetical-deductive reasoning, individual judgment, and heuristics. These factors can lead to bias, error, and preventable harm. Traditional predictive analytics and clinical decision-support systems are intended to augment surgical decision-making, but their clinical utility is compromised by time-consuming manual data management and suboptimal accuracy. These challenges can be overcome by automated artificial intelligence models fed by livestreaming electronic health record data with mobile device outputs. This approach would require data standardization, advances in model interpretability, careful implementation and monitoring, attention to ethical challenges involving algorithm bias and accountability for errors, and preservation of bedside assessment and human intuition in the decision-making process.

Conclusions and Relevance

Integration of artificial intelligence with surgical decision-making has the potential to transform care by augmenting the decision to operate, informed consent process, identification and mitigation of modifiable risk factors, decisions regarding postoperative management, and shared decisions regarding resource use.

[Laboratory testing in intensive care medicine]

Despite the tremendous technological developments in medicine, careful history-taking and clinical examination remain the cornerstones of diagnostics. Numerous laboratory tests are ordered in intensive care and emergency medicine. The rate of overutilization of these tests during initial patient admission is almost 50%. Patient history may be frequently insufficient for conducting targeted laboratory testing, and concern about not overlooking a pathology also contributes to laboratory test overutilization. On the other hand, laboratory test profiles are frequently defined a priori to simplify the management process. However, these profiles are commonly based on symptoms rather than on a suspected diagnosis. Several laboratory variables are outside the normal range in critically ill patients. However, normal ranges are defined on the basis of data from healthy subjects, and these do not allow for a clear distinction between stress adaptation and clinically relevant changes that require correction. Pathophysiological changes due to the acute injury in critically ill patients and the reaction of the organism to the injury or even to the treatment itself can lead to changes in laboratory values. Untargeted laboratory tests contribute to iatrogenic anemia and increased costs. The results of such tests are either hardly noticed or, in the worst case, lead to further unnecessary diagnostic steps and unjustified therapeutic measures. Point-of-care laboratory tests, including blood gas analysis, blood count, serum electrolytes, and lactate, to assess the patient's homeostatic state and laboratory data for the relevant critical care scores are uniformly required. Beyond that, every laboratory test should be chosen wisely based on a concrete clinical question.

Using machine learning to selectively highlight patient information

BACKGROUND

Electronic medical record (EMR) systems need functionality that decreases cognitive overload by drawing the clinician's attention to the right data, at the right time. We developed a Learning EMR (LEMR) system that learns statistical models of clinician information-seeking behavior and applies those models to direct the display of data in future patients. We evaluated the performance of the system in identifying relevant patient data in intensive care unit (ICU) patient cases.

METHODS

To capture information-seeking behavior, we enlisted critical care medicine physicians who reviewed a set of patient cases and selected data items relevant to the task of presenting at morning rounds. Using patient EMR data as predictors, we built machine learning models to predict their relevancy. We prospectively evaluated the predictions of a set of high performing models.

RESULTS

On an independent evaluation data set, 25 models achieved precision of 0.52, 95% CI [0.49, 0.54] and recall of 0.77, 95% CI [0.75, 0.80] in identifying relevant patient data items. For data items missed by the system, the reviewers rated the effect of not seeing those data from no impact to minor impact on patient care in about 82% of the cases.

CONCLUSION

Data-driven approaches for adaptively displaying data in EMR systems, like the LEMR system, show promise in using information-seeking behavior of clinicians to identify and highlight relevant patient data.