The Wolf Is Crying in the Operating Room

ARTEMIS: An alarm threshold and policy mining system for the intensive care unit

BACKGROUND

Alarm fatigue is a major technology-induced hazard for patients and staff in intensive care units. Too many - mostly unnecessary - alarms cause desensitisation and lack of response in medical staff. Unsuitable alarm policies are one reason for alarm fatigue. But changing alarm policies is a delicate issue since it concerns patient safety.

OBJECTIVE

We present ARTEMIS, a novel, computer-aided clinical decision support system for policy makers that can help to considerably improve alarm policies using data from hospital information systems.

METHODS

Policy makers can use different policy components from ARTEMIS' internal library to assemble tailor-made alarm policies for their intensive care units. Alternatively, policy makers can provide even more highly customised policy components as Python functions using data the hospital information systems. This can even include machine learning models - for example for setting alarm thresholds. Finally, policy makers can evaluate their system of policies and compare the resulting alarm loads.

RESULTS

ARTEMIS reports and compares numbers of alarms caused by different alarm policies for an easily adaptable target population. ARTEMIS can compare policies side-by-side and provides grid comparisons and heat maps for parameter optimisation. For example, we found that the utility of alarm delays varies based on target population. Furthermore, policy makers can introduce virtual parameters that are not in the original data by providing a formula to compute them. Virtual parameters help measuring and alarming on the right metric, even if the patient monitors do not directly measure this metric.

CONCLUSION

ARTEMIS does not release the policy maker from assessing the policy from a medical standpoint. But as a knowledge discovery and clinical decision support system, it provides a strong quantitative foundation for medical decisions. At comparatively low cost of implementation, ARTEMIS can have a substantial impact on patients and staff alike - with organisational, economic, and clinical benefits for the implementing hospital.

Management of Intraoperative Cardiac Arrest

Perioperative arrests are both uncommon and heterogeneous and have not been described or studied to the same extent as cardiac arrest in the community. These crises are usually witnessed, frequently anticipated, and involve a rescuer physician with knowledge of the patient's comorbidities and coexisting anesthetic or surgically related pathophysiology ultimately leading to better outcomes. This article reviews the most probable causes of intraoperative arrest and their management.

Tiered escalation response systems in practice: A post hoc analysis examining the workload implications

Objective

Many rapid response systems now have multiple tiers of escalation in addition to the traditional single tier of a medical emergency team. Given that the benefit to patient outcomes of this change is unclear, we sought to investigate the workload implications of a multitiered system, including the impact of trigger modification.

Design

The study design incorporated a post hoc analysis using a matched case-control dataset.

Setting

The study setting was an acute, adult tertiary referral hospital.

Participants

Cases that had an adverse event (cardiac arrest or unanticipated intensive care unit admission) or a rapid response team (RRT) call participated in the study. Controls were matched by age, gender, ward and time of year, and no adverse event or RRT call. Participants were admitted between May 2014 and April 2015.

Main outcome measures

The main outcome measure were the number of reviews, triggers, and modifications across three tiers of escalation; a nurse review, a multidisciplinary review (MDT-admitting medical team review), and an RRT call.

Results

There were 321 cases and 321 controls. Overall, there were 1948 nurse triggers, of which 1431 (73.5%) were in cases and 517 (26.5%) in controls, 798 MDT triggers (660 [82.7%] in cases and 138 [17.3%] in controls), and 379 RRT triggers (351 [92.6%] in cases and 28 [7.4%] in controls). Per patient per 24 h, there were 3.03 nurse, 1.24 MDT, and 0.59 RRT triggers. Accounting for modifications, this reduced to 2.17, 0.88, and 0.42, respectively. The proportion of triggers that were modified, so as not to trigger a review, was similar across all the tiers, being 28.6% of nurse, 29.6% of MDT, and 28.2% of RRT triggers. Per patient per 24 h, there were 0.61 nurse reviews, 0.52 MDT reviews, and 0.08 RRT reviews.

Conclusions

Lower-tier triggers were more prevalent, and modifications were common. Modifications significantly mitigated the escalation workload across all tiers of a multitiered system.

Clinical Alarms in a Gynaecological Surgical Unit: A Retrospective Data Analysis

Alarm fatigue refers to the desensitisation of medical staff to patient monitor clinical alarms, which may lead to slower response time or total ignorance of alarms and thereby affects patient safety. The reasons behind alarm fatigue are complex; the main contributing factors include the high number of alarms and the poor positive predictive value of alarms. The study was performed in the Surgery and Anaesthesia Unit of the Women's Hospital, Helsinki, by collecting data from patient monitoring device clinical alarms and patient characteristics from surgical operations. We descriptively analysed the data and statistically analysed the differences in alarm types between weekdays and weekends, using chi-squared, for a total of eight monitors with 562 patients. The most common operational procedure was caesarean section, of which 149 were performed (15.7%). Statistically significant differences existed in alarm types and procedures between weekdays and weekends. The number of alarms produced was 11.7 per patient. In total, 4698 (71.5%) alarms were technical and 1873 (28.5%) were physiological. The most common physiological alarm type was low pulse oximetry, with a total of 437 (23.3%). Of all the alarms, the number of alarms either acknowledged or silenced was 1234 (18.8%). A notable phenomenon in the study unit was alarm fatigue. Greater customisation of patient monitors for different settings is needed to reduce the number of alarms that do not have clinical significance.

Computational approaches to alleviate alarm fatigue in intensive care medicine: A systematic literature review

Patient monitoring technology has been used to guide therapy and alert staff when a vital sign leaves a predefined range in the intensive care unit (ICU) for decades. However, large amounts of technically false or clinically irrelevant alarms provoke alarm fatigue in staff leading to desensitisation towards critical alarms. With this systematic review, we are following the Preferred Reporting Items for Systematic Reviews (PRISMA) checklist in order to summarise scientific efforts that aimed to develop IT systems to reduce alarm fatigue in ICUs. 69 peer-reviewed publications were included. The majority of publications targeted the avoidance of technically false alarms, while the remainder focused on prediction of patient deterioration or alarm presentation. The investigated alarm types were mostly associated with heart rate or arrhythmia, followed by arterial blood pressure, oxygen saturation, and respiratory rate. Most publications focused on the development of software solutions, some on wearables, smartphones, or headmounted displays for delivering alarms to staff. The most commonly used statistical models were tree-based. In conclusion, we found strong evidence that alarm fatigue can be alleviated by IT-based solutions. However, future efforts should focus more on the avoidance of clinically non-actionable alarms which could be accelerated by improving the data availability.

Systematic Review Registration:https://www.crd.york.ac.uk/prospero/display_record.php?ID=CRD42021233461, identifier: CRD42021233461.

The effect of educational interventions in managing nurses' alarm fatigue: An integrative review

BACKGROUND

Alarm fatigue is becoming more widely acknowledged as a serious safety concern in modern clinical practice. Nurses are not always proficient in the alarms' functions and capabilities, and they do not undertake training regularly. Educating nurses on alarms maintains their knowledge and abilities in complex clinical settings. Some education has been undertaken to improve clinical alarm response, but the evidence for evaluating the effectiveness of nurse education interventions is limited.

OBJECTIVE

To evaluate the effects of educational interventions for reducing alarm fatigue in nurses, including the reduction of excessive, false and non-actionable alarms, which are major factors causing alarm fatigue in nurses.

DATA SOURCES

PUBMED, EMBASE, CINAHL, SCOPUS and OVID databases were systematically searched from 2016 to 2021.

DESIGN

Integrative Review.

REVIEW METHODS

An integrative review of literature was performed using the PRISMA checklist. Critical appraisal was done using Joanna Briggs Institute level of evidence.

RESULTS

Thirteen studies met the inclusion criteria. The results of most studies showed that educational intervention was beneficial for reducing the total number of alarms and false alarms. Furthermore, nurses' perceptions and knowledge improved, but the reduction in nurses' alarm fatigue is uncertain. A positive effect in alarm management practices was identified after the educational intervention.

CONCLUSION

Educational intervention may be the way to manage nurses' alarm fatigue. The use of medical devices in hospitals is increasing exponentially, and for this reason, alarms are inevitable. The introduction of effective and continuous education and training programs for nurses concerning clinical alarm management as well as raising nurses' awareness of the occurrence of alarm fatigue is vital.

Machine learning in anesthesiology: Detecting adverse events in clinical practice

The credibility of threshold-based alarms in anesthesia monitors is low and most of the warnings they produce are not informative. This study aims to show that Machine Learning techniques have a potential to generate meaningful alarms during general anesthesia without putting constraints on the type of procedure. Two distinct approaches were tested - Complication Detection and Anomaly Detection. The former is a generic supervised learning problem and for this a simple feed-forward Neural Network performed best. For the latter, we used an Encoder-Decoder Long Short-Term Memory architecture that does not require a large manually-labeled dataset. We show this approach to be more flexible and in the spirit of Explainable Artificial Intelligence, offering greater potential for future improvement.

Machine learning methods for perioperative anesthetic management in cardiac surgery patients: a scoping review

Background

Machine learning (ML) is developing fast with promising prospects within medicine and already has several applications in perioperative care. We conducted a scoping review to examine the extent and potential limitations of ML implementation in perioperative anesthetic care, specifically in cardiac surgery patients.

Methods

We mapped the current literature by searching three databases: MEDLINE (Ovid), EMBASE (Ovid), and Cochrane Library. Articles were eligible if they reported on perioperative ML use in the field of cardiac surgery with relevance to anesthetic practices. Data on the applicability of ML and comparability to conventional statistical methods were extracted.

Results

Forty-six articles on ML relevant to the work of the anesthesiologist in cardiac surgery were identified. Three main categories emerged: (I) event and risk prediction, (II) hemodynamic monitoring, and (III) automation of echocardiography. Prediction models based on ML tend to behave similarly to conventional statistical methods. Using dynamic hemodynamic or ultrasound data in ML models, however, shifts the potential to promising results.

Conclusions

ML in cardiac surgery is increasingly used in perioperative anesthetic management. The majority is used for prediction purposes similar to conventional clinical scores. Remarkable ML model performances are achieved when using real-time dynamic parameters. However, beneficial clinical outcomes of ML integration have yet to be determined. Nonetheless, the first steps introducing ML in perioperative anesthetic care for cardiac surgery have been taken.

Integrating Physiological Data Artifacts Detection With Clinical Decision Support Systems: Observational Study

Background

Clinical decision support systems (CDSS) have the potential to lower the patient mortality and morbidity rates. However, signal artifacts present in physiological data affect the reliability and accuracy of the CDSS. Moreover, patient monitors and other medical devices generate false alarms while processing physiological data, further leading to alarm fatigue because of increased noise levels, staff disruption, and staff desensitization in busy critical care environments. This adversely affects the quality of care at the patient bedside. Hence, artifact detection (AD) algorithms play a crucial role in assessing the quality of physiological data and mitigating the impact of these artifacts.

Objective

The aim of this study is to evaluate a novel AD framework for integrating AD algorithms with CDSS. We designed the framework with features that support real-time implementation within critical care. In this study, we evaluated the framework and its features in a false alarm reduction study. We developed static framework component models, followed by dynamic framework compositions to formulate four CDSS. We evaluated these formulations using neonatal patient data and validated the six framework features: flexibility, reusability, signal quality indicator standardization, scalability, customizability, and real-time implementation support.

Methods

We developed four exemplar static AD components with standardized requirements and provisions interfaces that facilitate the interoperability of framework components. These AD components were mixed and matched into four different AD compositions to mitigate the artifacts’ effects. We developed a novel static clinical event detection component that is integrated with each AD composition to formulate and evaluate a dynamic CDSS for peripheral oxygen saturation (SpO2) alarm generation. This study collected data from 11 patients with diverse pathologies in the neonatal intensive care unit. Collected data streams and corresponding alarms include pulse rate and SpO2 measured from a pulse oximeter (Masimo SET SmartPod) integrated with an Infinity Delta monitor and the heart rate derived from electrocardiography leads attached to a second Infinity Delta monitor.

Results

A total of 119 SpO2 alarms were evaluated. The lowest achievable SpO2 false alarm rate was 39%, with a sensitivity of 80%. This demonstrates the framework’s utility in identifying the best possible dynamic composition to serve the clinical need for false SpO2 alarm reduction and subsequent alarm fatigue, given the limitations of a small sample size.

Conclusions

The framework features, including reusability, signal quality indicator standardization, scalability, and customizability, allow the evaluation and comparison of novel CDSS formulations. The optimal solution for a CDSS can then be hard-coded and integrated within clinical workflows for real-time implementation. The flexibility to serve different clinical needs and standardized component interoperability of the framework supports the potential for a real-time clinical implementation of AD.

Ventilator Alarms in Intensive Care Units: Frequency, Duration, Priority, and Relationship to Ventilator Parameters

Ventilator alarms have long been presumed to contribute substantially to the overall alarm burden in the intensive care unit. In a prospective observational study, we determined that each ventilator triggered an alarm cascade of up to 8 separate notifications once every 6 minutes. In 1 intensive care unit with different ventilator manufacturers, the distribution of high-priority alarms was manufacturer dependent with 8.6% of alarms from 1 type and 89.8% of alarms from another type of ventilator. Alarm limits were not a function of patient-specific ventilator settings.

Comparing regression and neural network techniques for personalized predictive analytics to promote lung protective ventilation in Intensive Care Units

Mechanical ventilation is a lifesaving tool and provides organ support for patients with respiratory failure. However, injurious ventilation due to inappropriate delivery of high tidal volume can initiate or potentiate lung injury. This could lead to acute respiratory distress syndrome, longer duration of mechanical ventilation, ventilator associated conditions and finally increased mortality. In this study, we explore the viability and compare machine learning methods to generate personalized predictive alerts indicating violation of the safe tidal volume per ideal body weight (IBW) threshold that is accepted as the upper limit for lung protective ventilation (LPV), prior to application to patients. We process streams of patient respiratory data recorded per minute from ventilators in an intensive care unit and apply several state-of-the-art time series prediction methods to forecast the behavior of the tidal volume metric per patient, 1 hour ahead. Our results show that boosted regression delivers better predictive accuracy than other methods that we investigated and requires relatively short execution times. Long short-term memory neural networks can deliver similar levels of accuracy but only after much longer periods of data acquisition, further extended by several hours computing time to train the algorithm. Utilizing Artificial Intelligence, we have developed a personalized clinical decision support tool that can predict tidal volume behavior within 10% accuracy and compare alerts recorded from a real world system to highlight that our models would have predicted violations 1 hour ahead and can therefore conclude that the algorithms can provide clinical decision support.

Recurrent probabilistic neural network-based short-term prediction for acute hypotension and ventricular fibrillation

In this paper, we propose a novel method for predicting acute clinical deterioration triggered by hypotension, ventricular fibrillation, and an undiagnosed multiple disease condition using biological signals, such as heart rate, RR interval, and blood pressure. Efforts trying to predict such acute clinical deterioration events have received much attention from researchers lately, but most of them are targeted to a single symptom. The distinctive feature of the proposed method is that the occurrence of the event is manifested as a probability by applying a recurrent probabilistic neural network, which is embedded with a hidden Markov model and a Gaussian mixture model. Additionally, its machine learning scheme allows it to learn from the sample data and apply it to a wide range of symptoms. The performance of the proposed method was tested using a dataset provided by Physionet and the University of Tokyo Hospital. The results show that the proposed method has a prediction accuracy of 92.5% for patients with acute hypotension and can predict the occurrence of ventricular fibrillation 5 min before it occurs with an accuracy of 82.5%. In addition, a multiple disease condition can be predicted 7 min before they occur, with an accuracy of over 90%.

Situation Awareness-Oriented Patient Monitoring with Visual Patient Technology: A Qualitative Review of the Primary Research

Visual Patient technology is a situation awareness-oriented visualization technology that translates numerical and waveform patient monitoring data into a new user-centered visual language. Vital sign values are converted into colors, shapes, and rhythmic movements-a language humans can easily perceive and interpret-on a patient avatar model in real time. In this review, we summarize the current state of the research on the Visual Patient, including the technology, its history, and its scientific context. We also provide a summary of our primary research and a brief overview of research work on similar user-centered visualizations in medicine. In several computer-based studies under various experimental conditions, Visual Patient transferred more information per unit time, increased perceived diagnostic certainty, and lowered perceived workload. Eye tracking showed the technology worked because of the way it synthesizes and transforms vital sign information into new and logical forms corresponding to the real phenomena. The technology could be particularly useful for improving situation awareness in settings with high cognitive demand or when users must make quick decisions. This comprehensive review of Visual Patient research is the foundation for an evaluation of the technology in clinical applications, starting with a high-fidelity simulation study in early 2020.

Use of Notification and Communication Technology (Call Light Systems) in Nursing Homes: Observational Study

BACKGROUND

The call light system is one of the major communication technologies that link nursing home staff to the needs of residents. By providing residents the ability to request assistance, the system becomes an indispensable resource for patient-focused health care. However, little is known about how call light systems are being used in nursing homes and how the system contributes to safety and quality of care for seniors.

OBJECTIVE

This study aimed to understand the experiences of nursing home staff who use call light systems and to uncover usability issues and challenges associated with the implemented systems.

METHODS

A mix of 150 hours of hypothetico-deductive (unstructured) task analysis and 90 hours of standard procedure (structured) task analysis was conducted in 4 different nursing homes. The data collected included insights into the nursing home's work system and the process of locating and responding to call lights.

RESULTS

The data showed that the highest alarm rate is before and after mealtimes. The staff exceeded the administration's expectations of time to respond 50% of the time. In addition, the staff canceled 10.0% (20/201) of call lights and did not immediately assist residents because of high workload. Furthermore, the staff forgot to come back to assist residents over 3% of the time. Usability issues such as broken parts, lack of feedback, lack of prioritization, and low or no discriminability also contributed to the long response time. More than 8% of the time, residents notified the staff about call lights after they waited for a long time, and eventually, these residents were left unattended.

CONCLUSIONS

Nursing homes that are still using old call light systems risk the continuation of usability issues that can affect the performance of the staff and contribute to declining staff and resident outcomes. By incorporating feedback from nurses, nursing home management will better understand the influence that the perceptions and usability of technology have on the quality of health care for their residents. In this study, it has been observed that the call light system is perceived to be an important factor affecting the outcomes of the care process and satisfaction of both residents and staff as well as the staff's performance. It is important to recognize that communication and notification technology contributes to the challenges the staff faced during their work, making their working conditions more difficult and challenging.

Avatar-Based Patient Monitoring With Peripheral Vision: A Multicenter Comparative Eye-Tracking Study

BACKGROUND

Continuous patient monitoring has been described by the World Health Organization as extremely important and is widely used in anesthesia, intensive care medicine, and emergency medicine. However, current state-of-the-art number- and waveform-based monitoring does not ideally support human users in acquiring quick, confident interpretations with low cognitive effort, and there are additional problematic aspects such as alarm fatigue. We developed a visualization technology (Visual Patient), specifically designed to help caregivers gain situation awareness quickly, which presents vital sign information in the form of an animated avatar of the monitored patient. We suspected that because of the way it displays the information as large, colorful, moving graphic objects, caregivers might be able to perform patient monitoring using their peripheral vision, which may facilitate quicker detection of anomalies, independently of acoustic alarms.

OBJECTIVE

In this study, we tested the hypothesis that avatar-based monitoring, when observed with peripheral vision only, increases the number of perceptible changes in patient status as well as caregivers' perceived diagnostic confidence compared with a high-fidelity simulation of conventional monitoring, when observed with peripheral vision only.

METHODS

We conducted a multicenter comparative study with a within-participant design in which anesthesiologists with their peripheral field of vision looked at 2 patient-monitoring scenarios and tried to identify changes in patient status. To ensure the best possible experimental conditions, we used an eye tracker, which recorded the eye movements of the participants and confirmed that they only looked at the monitoring scenarios with their peripheral vision.

RESULTS

Overall, 30 participants evaluated 18 different patient status changes with each technology (avatar and conventional patient monitoring). With conventional patient monitoring, participants could only detect those 3 changes in patient status that are associated with a change in the auditory pulse tone display, that is, tachycardia (faster beeping), bradycardia (slower beeping), and desaturation (lower pitch of beeping). With the avatar, the median number of detected vital sign changes quadrupled from 3 to 12 (P<.001) in scenario 1, and more than doubled from 3 to 8 (P<.001) in scenario 2. Median perceived diagnostic confidence was confident for both scenarios with the avatar and unconfident in scenario 1 (P<.001), and very unconfident in scenario 2 (P=.024) with conventional monitoring.

CONCLUSIONS

This study introduces the concept of peripheral vision monitoring. The test performed showed clearly that an avatar-based display is superior to a standard numeric display for peripheral vision. Avatar-based monitoring could potentially make much more of the patient monitoring information available to caregivers for longer time periods per case. Our results indicate that the optimal information transmission would consist of a combination of auditory and avatar-based monitoring.

It's not you, it's the design - common problems with patient monitoring reported by anesthesiologists: a mixed qualitative and quantitative study

Background

Patient monitoring is critical for perioperative patient safety as anesthesiologists routinely make crucial therapeutic decisions from the information displayed on patient monitors. Previous research has shown that today’s patient monitoring has room for improvement in areas such as information overload and alarm fatigue. The rationale of this study was to learn more about the problems anesthesiologists face in patient monitoring and to derive improvement suggestions for next-generation patient monitors.

Methods

We conducted a two-center qualitative/quantitative study. Initially, we interviewed 120 anesthesiologists (physicians and nurses) about the topic: common problems with patient monitoring in your daily work. Through deductive and inductive coding, we identified major topics and sub themes from the interviews. In a second step, a field survey, a separate group of 25 anesthesiologists rated their agree- or disagreement with central statements created for all identified major topics.

Results

We identified the following six main topics: 1. “Alarms,” 2. “Artifacts,” 3. “Software,” 4. “Hardware,” 5. “Human Factors,” 6. “System Factors,” and 17 sub themes. The central statements rated for the major topics were: 1. “problems with alarm settings complicate patient monitoring.” (56% agreed) 2. “artifacts complicate the assessment of the situation.” (64% agreed) 3. “information overload makes it difficult to get an overview quickly.” (56% agreed) 4. “problems with cables complicate working with patient monitors.” (92% agreed) 5. “factors related to human performance lead to critical information not being perceived.” (88% agreed) 6. “Switching between monitors from different manufacturers is difficult.” (88% agreed). The ratings of all statements differed significantly from neutral (all p < 0.03).

Conclusion

This study provides an overview of the problems anesthesiologists face in patient monitoring. Some of the issues, to our knowledge, were not previously identified as common problems in patient monitoring, e.g., hardware problems (e.g., cable entanglement and worn connectors), human factor aspects (e.g., fatigue and distractions), and systemic factor aspects (e.g., insufficient standardization between manufacturers). An ideal monitor should transfer the relevant patient monitoring information as efficiently as possible, prevent false positive alarms, and use technologies designed to improve the problems in patient monitoring.

Electronic supplementary material

The online version of this article (10.1186/s12871-019-0757-z) contains supplementary material, which is available to authorized users.

Influence of non-invasive blood pressure measurement intervals on the occurrence of intra-operative hypotension

The American Society of Anesthesiologists Standards for Basic Monitoring recommends blood pressure (BP) measurement every 5 min. Research has shown distractions or technical factors can cause prolonged measurement intervals exceeding 5 min. We investigated the relationship between prolonged non-invasive BP (NIBP) measurement interval and the incidence of hypotension, detected post-interval. Our secondary outcome was to determine independent predictors of these prolonged NIBP measurement intervals. Retrospective data were analyzed from 139,509 general anesthesia cases from our institution's Anesthesia Information Management System (AIMS). Absolute hypotension (AH) was defined a priori as a systolic BP < 80 mmHg and relative hypotension (RH) was defined as a 40% decrease in systolic BP from the preoperative baseline. Odds ratios (OR) with 95% confidence intervals and Pearson's Chi square Test reported the association of prolonged NIBP measurement intervals on hypotension detected post-NIBP measurement interval. Logistic regression models were developed to determine independent predictors of NIBP measurement intervals. The analysis revealed that NIBP measurement intervals greater than 6 and 10 min are associated with an approximately four times higher incidence of a patient transitioning into hypotension (AH/RH > 6 min OR 4.0 / 3.6; AH/RH > 10 min OR 4.3 / 3.9; p < 0.001). A key finding was that the "> 10-minute AH model" indicated that age 41-80, increased co-morbidity profile, obesity and turning (repositioning) of the operative room table were significant predictors of prolonged NIBP measurement intervals (p < 0.001). While we do not suggest NIBP measurement intervals cause hypotension, intervals greater than 6 and 10 min are associated with a fourfold increase in the propensity of an undetected transition into both RH or AH. These data support current monitoring guidelines.

Role of Large Clinical Datasets From Physiologic Monitors in Improving the Safety of Clinical Alarm Systems and Methodological Considerations: A Case From Philips Monitors

BACKGROUND

Large datasets of the audit log of modern physiologic monitoring devices have rarely been used for predictive modeling, capturing unsafe practices, or guiding initiatives on alarm systems safety.

OBJECTIVE

This paper (1) describes a large clinical dataset using the audit log of the physiologic monitors, (2) discusses benefits and challenges of using the audit log in identifying the most important alarm signals and improving the safety of clinical alarm systems, and (3) provides suggestions for presenting alarm data and improving the audit log of the physiologic monitors.

METHODS

At a 20-bed transplant cardiac intensive care unit, alarm data recorded via the audit log of bedside monitors were retrieved from the server of the central station monitor.

RESULTS

Benefits of the audit log are many. They include easily retrievable data at no cost, complete alarm records, easy capture of inconsistent and unsafe practices, and easy identification of bedside monitors missed from a unit change of alarm settings adjustments. Challenges in analyzing the audit log are related to the time-consuming processes of data cleaning and analysis, and limited storage and retrieval capabilities of the monitors.

CONCLUSIONS

The audit log is a function of current capabilities of the physiologic monitoring systems, monitor's configuration, and alarm management practices by clinicians. Despite current challenges in data retrieval and analysis, large digitalized clinical datasets hold great promise in performance, safety, and quality improvement. Vendors, clinicians, researchers, and professional organizations should work closely to identify the most useful format and type of clinical data to expand medical devices' log capacity.

Advanced Monitoring Is Associated with Fewer Alarm Events During Planned Moderate Procedure-Related Sedation: A 2-Part Pilot Trial

Supplemental Digital Content is available in the text.

Published ahead of print February 1, 2016

Diagnostic and interventional procedures are often facilitated by moderate procedure-related sedation. Many studies support the overall safety of this sedation; however, adverse cardiovascular and respiratory events are reported in up to 70% of these procedures, more frequently in very young, very old, or sicker patients. Monitoring with pulse oximetry may underreport hypoventilation during sedation, particularly if supplemental oxygen is provided. Capnometry may result in false alarms during sedation when patients mouth breathe or displace sampling devices. Advanced monitor use during sedation may allow event detection before complications develop. This 2-part pilot study used advanced monitors during planned moderate sedation to (1) determine incidences of desaturation, low respiratory rate, and deeper than intended sedation alarm events; and (2) determine whether advanced monitor use is associated with fewer alarm events.

A human factors perspective on medical device alarms: problems with operating alarming devices and responding to device alarms

Medical devices emit alarms when a problem with the device or with the patient needs to be addressed by healthcare personnel. At present, problems with device alarms are frequently discussed in the literature, the main message being that patient safety is compromised because device alarms are not as effective and safe as they should - and could - be. There is a general consensus that alarm-related hazards result, to a considerable degree, from the interactions of human users with the device. The present paper addresses key aspects of human perception and cognition that may relate to both operating alarming devices and responding to device alarms. Recent publications suggested solutions to alarm-related hazards associated with usage errors based on assumptions on the causal relations between, for example, alarm management and human perception, cognition, and responding. However, although there is face validity in many of these assumptions, future research should provide objective empirical evidence in order to deepen our understanding of the actual causal relationships, and hence improve and expand the possibilities for taking appropriate action.

Pattern discovery in critical alarms originating from neonates under intensive care

Patient monitoring generates a large number of alarms, the vast majority of which are false. Excessive non-actionable medical alarms lead to alarm fatigue, a well-recognized patient safety issue. While multiple approaches to reduce alarm fatigue have been explored, patterns in alarming and inter-alarm relationships, as they manifest in the clinical workspace, are largely a black-box and hamper research efforts towards reducing alarms. The aim of this study is to detect opportunities to safely reduce alarm pressure, by developing techniques to identify, capture and visualize patterns in alarms. Nearly 500 000 critical medical alarms were acquired from a neonatal intensive care unit over a 20 month period. Heuristic techniques were developed to extract the inter-alarm relationships. These included identifying the presence of alarm clusters, patterns of transition from one alarm category to another, temporal associations amongst alarms and determination of prevalent sequences in which alarms manifest. Desaturation, bradycardia and apnea constituted 86% of all alarms and demonstrated distinctive periodic increases in the number of alarms that were synchronized with nursing care and enteral feeding. By inhibiting further alarms of a category for a short duration of time (30 s/60 s), non-actionable physiological alarms could be reduced by 20%. The patterns of transition from one alarm category to another and the time duration between such transitions revealed the presence of close temporal associations and multiparametric derangement. Examination of the prevalent alarm sequences reveals that while many sequences comprised of multiple alarms, nearly 65% of the sequences were isolated instances of alarms and are potentially irreducible. Patterns in alarming, as they manifest in the clinical workspace were identified and visualized. This information can be exploited to investigate strategies for reducing alarms.

Locating Errors Through Networked Surveillance: A Multimethod Approach to Peer Assessment, Hazard Identification, and Prioritization of Patient Safety Efforts in Cardiac Surgery

OBJECTIVES

The objectives were to develop a scientifically sound and feasible peer-to-peer assessment model that allows health-care organizations to evaluate patient safety in cardiovascular operating rooms and to establish safety priorities for improvement.

METHODS

The locating errors through networked surveillance study was conducted to identify hazards in cardiac surgical care. A multidisciplinary team, composed of organizational sociology, organizational psychology, applied social psychology, clinical medicine, human factors engineering, and health services researchers, conducted the study. We used a transdisciplinary approach, which integrated the theories, concepts, and methods from each discipline, to develop comprehensive research methods. Multiple data collection was involved: focused literature review of cardiac surgery-related adverse events, retrospective analysis of cardiovascular events from a national database in the United Kingdom, and prospective peer assessment at 5 sites, involving survey assessments, structured interviews, direct observations, and contextual inquiries. A nominal group methodology, where one single group acts to problem solve and make decisions was used to review the data and develop a list of the top priority hazards.

RESULTS

The top 6 priority hazard themes were as follows: safety culture, teamwork and communication, infection prevention, transitions of care, failure to adhere to practices or policies, and operating room layout and equipment.

CONCLUSIONS

We integrated the theories and methods of a diverse group of researchers to identify a broad range of hazards and good clinical practices within the cardiovascular surgical operating room. Our findings were the basis for a plan to prioritize improvements in cardiac surgical care. These study methods allowed for the comprehensive assessment of a high-risk clinical setting that may translate to other clinical settings.

Multi-Tiered Observation and Response Charts: Prevalence and Incidence of Triggers, Modifications and Calls, to Acutely Deteriorating Adult Patients

BACKGROUND

Observation charts are the primary tool for recording patient vital signs. They have a critical role in documenting triggers for a multi-tiered escalation response to the deteriorating patient. The objectives of this study were to ascertain the prevalence and incidence of triggers, trigger modifications and escalation response (Call) amongst general medical and surgical inpatients following the introduction of an observation and response chart (ORC).

METHODS

Prospective (prevalence), over two 24-hour periods, and retrospective (incidence), over entire hospital stay, observational study of documented patient observations intended to trigger one of three escalation responses, being a MER-Medical Emergency Response [highest tier], MDT-Multidisciplinary Team [admitting team], or Nurse-senior ward nurse [lowest tier] response amongst adult general medical and surgical patients.

RESULTS

PREVALENCE

416 patients, 321 (77.2%) being medical admissions, median age 76 years (IQR 62, 85) and 95 (22.8%) Not for Resuscitation (NFR). Overall, 193 (46.4%) patients had a Trigger, being 17 (4.1%) MER, 45 (10.8%) MDT and 178 (42.8%) Nurse triggers. 60 (14.4%) patients had a Call, and 72 (17.3%) a modified Trigger.

INCIDENCE

206 patients, of similar age, of whom 166 (80.5%) had a Trigger, 122 (59.2%) a Call, and 91 (44.2%) a modified Trigger. PREVALENCE and incidence of failure to Call was 33.2% and 68% of patients, respectively, particular for Nurse Triggers (26.7% and 62.1%, respectively). The number of Modifications, Calls, and failure to Call, correlated with the number of Triggers (0.912 [p<0.01], 0.631 [p<0.01], 0.988 [p<0.01]).

CONCLUSION

Within a multi-tiered response system for the detection and response to the deteriorating patient Triggers, their Modifications and failure to Call are common, particularly within the lower tiers of escalation. The number of Triggers and their Modifications may erode the structure, compliance, and potential efficacy of structured observation and response charts within a multi-tiered response system.

Reduction of clinically irrelevant alarms in patient monitoring by adaptive time delays

The problem of high rates of false alarms in patient monitoring in anesthesiology and intensive care medicine is well known but remains unsolved. False alarms desensitize the medical staff, leading to ignored true alarms and reduced quality of patient care. A database of intra-operative monitoring data was analyzed to find characteristic alarm patterns. The original data were re-evaluated to find relevant events and to rate the severity of these events. Based on this analysis an adaptive time delay was developed that individually delays the alarms depending on the grade of threshold deviation. The conventional threshold algorithm led to 4893 alarms. 3515 (71.84 %) of these alarms were annotated as clinically irrelevant. In total 81.0 % of all clinically irrelevant alarms were caused by only mild and/or brief threshold violations. We implemented the new algorithm for selected parameters. These parameters equipped with adaptive validation delays led to 1729 alarms. 931 (53.85 %) alarms were annotated as clinically irrelevant. 632 alarms indicated the 645 clinically relevant events. The positive predictive value of occurring alarms improved from 28.16 % (conventional algorithm) to 46.15 % (new algorithm). 13 events were missed. The false positive alarm reduction rate of the algorithm ranged from 33 to 86.75 %. The overall reduction was 73.51 %. The implementation of this algorithm may be able to suppress a large percentage of false alarms. The effect of this approach has not been demonstrated but shows promise for reducing alarm fatigue. Its safety needs to be proven in a prospective study.

Performance Evaluation of New-Generation Pulse Oximeters in the NICU: Observational Study

This crossover observational study compares the data characteristics and performance of new-generation Nellcor OXIMAX and Masimo SET SmartPod pulse oximeter technologies. The study was conducted independent of either original equipment manufacturer (OEM) across eleven preterm infants in a Neonatal Intensive Care Unit (NICU). The SmartPods were integrated with Dräger Infinity Delta monitors. The Delta monitor measured the heart rate (HR) using an independent electrocardiogram sensor, and the two SmartPods collected arterial oxygen saturation (SpO2) and pulse rate (PR). All patient data were non-Gaussian. Nellcor PR showed a higher correlation with the HR as compared to Masimo PR. The statistically significant difference found in their median values (1% for SpO2, 1 bpm for PR) was deemed clinically insignificant. SpO2 alarms generated by both SmartPods were observed and categorized for performance evaluation. Results for sensitivity, positive predictive value, accuracy and false alarm rates were Nellcor (80.3, 50, 44.5, 50%) and Masimo (72.2, 48.2, 40.6, 51.8%) respectively. These metrics were not statistically significantly different between the two pulse oximeters. Despite claims by OEMs, both pulse oximeters exhibited high false alarm rates, with no statistically or clinically significant difference in performance. These findings have a direct impact on alarm fatigue in the NICU. Performance evaluation studies can also impact medical device purchase decisions made by hospital administrators.

Effect of altering alarm settings: a randomized controlled study

Medical alarm signals are important for alerting clinicians to life-threatening conditions, but the high rate of false alarms can be problematic. Reduction in alarm signals may lead to increased staff responsiveness to alarms and create a quieter environment for patients. The effect of these changes on patient outcomes is uncertain.

METHODS

We conducted a pilot, prospective, randomized, controlled trial in the cardiac care unit (CCU) to test a study protocol and data collection instruments and to examine the differences in alarms between usual care and altered settings. Subjects were randomized daily to either standard or altered CCU alarm settings. Secondary outcomes included the number of clinically significant events (CSEs) detected, event-triggered interventions (ETIs), frequency of alarms per monitored bed, and patient complications.

RESULTS

Over the two-week study time frame, 22 unique patients were enrolled. There were 1,710 alarms over 163 hours of monitoring in the standard group and 1,165 alarms over 169 hours in the study group (P < 0.001). There were more CSEs detected (14 vs. 3) and ETIs (12 vs. 2) in the study group, but sample size was too small to determine efficacy. No cardiac arrests or adverse patient outcomes were observed in either group. All patients were discharged from the hospital. Study protocol and outcomes were feasible and lessons were learned.

CONCLUSION

This study demonstrated feasibility of a study protocol for conducting a randomized controlled trial to evaluate CSEs, ETIs, frequency of alarms, and adverse patient outcomes when altering default alarm settings. A longer study can be performed using a similar study design.

Nurses' Perceptions and Practices Toward Clinical Alarms in a Transplant Cardiac Intensive Care Unit: Exploring Key Issues Leading to Alarm Fatigue

Background

Intensive care units (ICUs) are complex work environments where false alarms occur more frequently than on non-critical care units. The Joint Commission National Patient Safety Goal .06.01.01 targeted improving the safety of clinical alarm systems and required health care facilities to establish alarm systems safety as a hospital priority by July 2014. An important initial step toward this requirement is identifying ICU nurses’ perceptions and common clinical practices toward clinical alarms, where little information is available.

Objective

Our aim was to determine perceptions and practices of transplant/cardiac ICU (TCICU) nurses toward clinical alarms and benchmark the results against the 2011 Healthcare Technology Foundation’s (HTF) Clinical Alarms Committee Survey.

Methods

A quality improvement project was conducted on a 20-bed TCICU with 39 full- and part-time nurses. Nurses were surveyed about their perceptions and attitudes toward and practices on clinical alarms using an adapted HTF clinical alarms survey. Results were compared to the 2011 HTF data. Correlations among variables were examined.

Results

All TCICU nurses provided usable responses (N=39, 100%). Almost all nurses (95%-98%) believed that false alarms are frequent, disrupt care, and reduce trust in alarm systems, causing nurses to inappropriately disable them. Unlike the 2011 HTF clinical alarms survey results, a significantly higher percentage of our TCICU nurses believed that existing devices are complex, questioned the ability and adequacy of the new monitoring systems to solve alarm management issues, pointed to the lack of prompt response to alarms, and indicated the lack of clinical policy on alarm management (P<.01). Major themes in the narrative data focused on nurses’ frustration related to the excessive number of alarms and poor usability of the cardiac monitors. A lack of standardized approaches exists in changing patients’ electrodes and individualizing parameters. Around 60% of nurses indicated they received insufficient training on bedside and central cardiac monitors. A correlation also showed the need for training on cardiac monitors, specifically for older nurses (P=.01).

Conclusions

False and non-actionable alarms continue to desensitize TCICU nurses, perhaps resulting in missing fatal alarms. Nurses’ attitudes and practices related to clinical alarms are key elements for designing contextually sensitive quality initiatives to fight alarm fatigue. Alarm management in ICUs is a multidimensional complex process involving usability of monitoring devices, and unit, clinicians, training, and policy-related factors. This indicates the need for a multi-method approach to decrease alarm fatigue and improve alarm systems safety.

Alarm system management: evidence-based guidance encouraging direct measurement of informativeness to improve alarm response

Although there are powerful incentives for creating alarm management programmes to reduce 'alarm fatigue', they do not provide guidance on how to reduce the likelihood that clinicians will disregard critical alarms. The literature cites numerous phenomena that contribute to alarm fatigue, although many of these, including total rate of alarms, are not supported in the literature as factors that directly impact alarm response. The contributor that is most frequently associated with alarm response is informativeness, which is defined as the proportion of total alarms that successfully conveys a specific event, and the extent to which it is a hazard. Informativeness is low across all healthcare applications, consistently ranging from 1% to 20%. Because of its likelihood and strong evidential support, informativeness should be evaluated before other contributors are considered. Methods for measuring informativeness and alarm response are discussed. Design directions for potential interventions, as well as design alternatives to traditional alarms, are also discussed. With the increased attention and investment in alarm system management that alarm interventions are currently receiving, initiatives that focus on informativeness and the other evidence-based measures identified will allow us to more effectively, efficiently and reliably redirect clinician attention, ultimately improving alarm response.

Human Factors Issues in the Intensive Care Unit

In the context of an aging population, more critically ill patients, and a change in intensive care unit (ICU) services stemming from advances in technology, prevalent medical errors and staff burnout in the ICU are not surprising. The ICU provides ample opportunity for human factors experts to apply their knowledge about the strengths and weaknesses of human capabilities to design more effective care delivery. Human factors experts design work processes, technology, and environmental factors to effectively and constructively channel the attention and behavior of those providing care; a few areas of focus can have marked impacts on care delivery and patient outcomes. In this review, we focus on these 3 areas and investigate the solutions and problems addressed by previous research.

Patient Injuries from Anesthesia Gas Delivery Equipment

Background:

Improvements in anesthesia gas delivery equipment and provider training may increase patient safety. The authors analyzed patient injuries related to gas delivery equipment claims from the American Society of Anesthesiologists Closed Claims Project database over the decades from 1970s to the 2000s.

Methods:

After the Institutional Review Board approval, the authors reviewed the Closed Claims Project database of 9,806 total claims. Inclusion criteria were general anesthesia for surgical or obstetric anesthesia care (n = 6,022). Anesthesia gas delivery equipment was defined as any device used to convey gas to or from (but not involving) the airway management device. Claims related to anesthesia gas delivery equipment were compared between time periods by chi-square test, Fisher exact test, and Mann–Whitney U test.

Results:

Anesthesia gas delivery claims decreased over the decades (P &lt; 0.001) to 1% of claims in the 2000s. Outcomes in claims from 1990 to 2011 (n = 40) were less severe, with a greater proportion of awareness (n = 9, 23%; P = 0.003) and pneumothorax (n = 7, 18%; P = 0.047). Severe injuries (death/permanent brain damage) occurred in supplemental oxygen supply events outside the operating room, breathing circuit events, or ventilator mishaps. The majority (85%) of claims involved provider error with (n = 7) or without (n = 27) equipment failure. Thirty-five percent of claims were judged as preventable by preanesthesia machine check.

Conclusions:

Gas delivery equipment claims in the Closed Claims Project database decreased in 1990–2011 compared with earlier decades. Provider error contributed to severe injury, especially with inadequate alarms, improvised oxygen delivery systems, and misdiagnosis or treatment of breathing circuit events.

User image mismatch in anaesthesia alarms: a cognitive systems analysis

In this study, principles of Cognitive Systems Engineering are used to better understand the human-machine interaction manifesting in the use of anaesthesia alarms. The hypothesis is that the design of the machine incorporates built-in assumptions of the user that are discrepant with the anaesthesiologist's self-assessment, creating 'user image mismatch'. Mismatch was interpreted by focusing on the 'user image' as described from the perspectives of both machine and user. The machine-embedded image was interpreted through document analysis. The user-described image was interpreted through user (anaesthesiologist) interviews. Finally, an analysis was conducted in which the machine-embedded and user-described images were contrasted to identify user image mismatch. It is concluded that analysing user image mismatch expands the focus of attention towards macro-elements in the interaction between man and machine. User image mismatch is interpreted to arise from complexity of algorithm design and incongruity between alarm design and tenets of anaesthesia practice.

PRACTITIONER SUMMARY

Cognitive system engineering principles are applied to enhance the understanding of the interaction between anaesthesiologist and alarm. The 'user image' is interpreted and contrasted from the perspectives of machine as well as the user. Apparent machine-user mismatch is explored pertaining to specific design features.

Learning medical alarms whilst performing other tasks

Two studies are reported which first observe, and then attempt to replicate, the cognitive demands of intensive care unit (ICU) activity whilst concurrently learning audible alarms. The first study, an observational study in an ICU ward, showed that the alarms are very frequent and co-occur with some activities more than others. The three most frequently observed activities observed in the ICU were drugs (calculation, preparation and administration), patient observation and talking. The cognitive demands of these activities were simulated in a second, laboratory-based experiment in which alarms were learned. The results showed that performance in the alarm task generally improved as participants were exposed to more repetitions of those alarms, but that performance decrements were observed in the secondary tasks, particularly when there were two or three of them. Some confusions between the alarms persisted to the end of the study despite prolonged exposure to the alarms, confusions which were likely caused by both acoustic and verbal labelling similarities.

PRACTITIONER SUMMARY

The cognitive demands of working in an ICU were observed and simulated whilst alarms were learned. Alarms should generally avoid sharing similar rhythmic (and other) characteristics. The simulation task described here could be used for testing alarm learning without requiring a clinical environment.

Intra-operative monitoring--many alarms with minor impact

Alarms are key components of peri-operative monitoring devices, but a high false-alarm rate may lead to desensitisation and neglect. The objective of this study was to quantify the number of alarms and assess the value of these alarms during moderate-risk surgery. For this purpose, we analysed documentation of anaesthesia workstations during 38 surgical procedures. Alarms were classified on technical validity and clinical relevance. The median (IQR [range]) alarm density per procedure was 20.8 (14.5-34.2 [3.7-85.6]) alarms.h⁻¹ (1 alarm every 2.9 min) and increased during induction and emergence of anaesthesia, with up to one alarm per 0.99 min during these periods (p < 0.001). Sixty-four per cent of all alarms were clinically irrelevant, whereas 5% of all alarms required immediate intervention. The positive predictive value of an alarm during induction and emergence was 20% (95% CI 16-24%) and 11% (95% CI 8-14%), respectively. This study shows that peri-operative alarms are frequently irrelevant, with a low predictive value for an emerging event requiring clinical intervention.

A novel sonification strategy for auditory display of heart rate and oxygen saturation changes in clinical settings

OBJECTIVE

The aim of this study was development of a sonification scheme to convey deviations in heart rate and oxygen saturation from a desired target level.

BACKGROUND

Maintaining physiologic parameters, such as oxygen saturation, within desired ranges, is challenging in many clinical situations. High rates of false positive alarms in clinical settings limit the utility of the alarms that trigger when thresholds are exceeded. Auditory displays that consider the semantic connotations of sounds and the processing limitations of human perception and cognition may improve monitoring.

METHOD

Across two experiments, clinical practitioners were tested on their ability to (a) discriminate pairs of sounds (two-note discrimination task), (b) infer and discern the intended physiological connotation of each acoustic attribute (name-the-variable task), and (c) categorize the amount of change in an implied physiological variable into three levels of change: none, small, and large (change-magnitude task).

RESULTS

Considerable variation in performance was observed across the set of practitioners, ranging from near-perfect performance on all tasks, even with no prior exposure to the stimuli, to failure to reach a target accuracy criterion of 87.5% after -80 min of training. On average, performance was well above chance on the name-the-variable and change-magnitude tasks during initial exposure and reached criterion within -20 min of training on each task.

CONCLUSION

The described sonification strategy may effectively communicate information about current heart rate and oxygen saturation status relative to desired target levels.

APPLICATION

The results can be applied to clinical monitoring settings in which a stream of discrete auditory informational items is indicated.