Modified Early Warning Score Changes Prior to Cardiac Arrest in General Wards

Prediction of lung metastases in thyroid cancer using machine learning based on SEER database

PURPOSE

Lung metastasis (LM) is one of the most frequent distant metastases of thyroid cancer (TC). This study aimed to develop a machine learning algorithm model to predict lung metastasis of thyroid cancer for providing relative information in clinical decision-making.

METHODS

Data comprising of demographic and clinicopathological characteristics of patients with thyroid cancer were extracted from the National Institutes of Health (NIH)'s Surveillance, Epidemiology, and End Results (SEER) database between 2010 and 2015, which is employed to develop six machine learning algorithm models support vector machine (SVM), logistic regression (LR), eXtreme gradient boosting (XGBoost), decision tree (DT), random forest (RF), and k-nearest neighbor (KNN). Compared and evaluated models by the following indicators: accuracy, precision, recall rate, F1-score, the area under the ROC curve (AUC) value and Brier score, and interpreted the association between clinicopathological characteristics and target variables based on the best model.

RESULTS

Nine thousand nine hundred and fifty patients were selected, which including 212 patients (2.1%) with lung metastasis, and 9738 patients without lung metastasis (97.9%). Multivariate logistic regression showed that age, T stage, N stage, and histological type were independent factors in TC with LM. Evaluation indicators of the best model- RF were as following: accuracy (0.99), recall rate (0.88), precision (0.61), F1-score (0.72), AUC value (0.99), and the Brier score (0.016).

CONCLUSION

RF learning model performed better and can be applied to forecast lung metastasis of thyroid cancer, and offer valuable and significant reference for clinicians' decision-making in advance.

Introspection confidence predicts EEG decoding of self-generated thoughts and meta-awareness

The neurophysiological bases of mind wandering (MW)-an experiential state wherein attention is disengaged from the external environment in favour of internal thoughts-and state meta-awareness are poorly understood. In parallel, the relationship between introspection confidence in experiential state judgements and neural representations remains unclear. Here, we recorded EEG while participants completed a listening task within which they made experiential state judgements and rated their confidence. Alpha power was reliably greater during MW episodes, with unaware MW further associated with greater delta and theta power. Multivariate pattern classification analysis revealed that MW and meta-awareness can be decoded from the distribution of power in these three frequency bands. Critically, we show that individual decoding accuracies positively correlate with introspection confidence. Our results reaffirm the role of alpha oscillations in MW, implicate lower frequencies in meta-awareness, and are consistent with the proposal that introspection confidence indexes neurophysiological discriminability of representational states.

Accuracy of Machine Learning Models to Predict In-hospital Cardiac Arrest: A Systematic Review

PURPOSE/AIMS

Despite advances in healthcare, the incidence of in-hospital cardiac arrest (IHCA) has continued to rise for the past decade. Identifying those patients at risk has proven challenging. Our objective was to conduct a systematic review of the literature to compare the IHCA predictive performance of machine learning (ML) models with the Modified Early Warning Score (MEWS).

DESIGN

The systematic review was conducted following the Preferred Reporting Items of Systematic Review and Meta-Analysis guidelines and registered on PROSPERO CRD42020182357.

METHOD

Data extraction was completed using the Critical Appraisal and Data Extraction for Systematic Reviews of Prediction Modeling Studies checklist. The risk of bias and applicability were evaluated using the Prediction model Risk of Bias Assessment Tool.

RESULTS

Nine articles were included in this review that developed or validated IHCA prediction models and compared them with the MEWS. The studies by Jang et al and Kim et al showed that their ML models outperformed MEWS to predict IHCA with good to excellent predictive performance.

CONCLUSIONS

The ML models presented in this systematic review demonstrate a novel approach to predicting IHCA. All included studies suggest that ML models had similar or better predictive performance compared with MEWS. However, there is substantial variability in performance measures and concerns for risk of bias.

Evaluation of the Initial General Ward Early Warning Score and ICU Admission, Hospital Length of Stay and Mortality

Introduction

Despite widespread implementation of the Early Warning Score (EWS) in hospitals, its effect on patient outcomes remains mostly unknown. We aimed to evaluate associations between the initial EWS and in-hospital mortality, intensive care unit (ICU) admission, and hospital length of stay (LOS).

Methods

We performed a retrospective cohort study of adult patients admitted to a general hospital ward between July 1, 2014–December 31, 2017. Data were obtained from electronic health records (EHR). The primary outcome was in-hospital mortality. Secondary outcomes were ICU admission and hospital LOS. We categorized patients into three risk groups (low, medium or high risk of clinical deterioration) based on EWS. Descriptive analyses were used.

Results

After applying inclusion and exclusion criteria, we included 53,180 patients for analysis. We found that the initial (low- vs high-risk) EWS was associated with an increased in-hospital mortality (1.5% vs 25.3%, P <0.001), an increased ICU admission rate (3.1% vs 17.6%, P <0.001), and an extended hospital LOS (4.0 days vs 8.0 days, P <0.001).

Conclusion

Our findings suggest that an initial high-risk EWS in patients admitted to a general hospital ward was associated with an increased risk of in-hospital mortality, ICU admission, and prolonged hospital LOS. Close monitoring and precise documentation of the EWS in the EHR may facilitate predicting poor outcomes in individual hospitalized patients and help to identify patients for whom timely and adequate management may improve outcomes.

An Early Predictive Scoring Model for In-Hospital Cardiac Arrest of Emergent Hemodialysis Patients

BACKGROUND

Patients undergoing hemodialysis are prone to cardiac arrests.

METHODS

This study aimed to develop a risk score to predict in-hospital cardiac arrest (IHCA) in emergency department (ED) patients undergoing emergency hemodialysis. Patients were included if they received urgent hemodialysis within 24 h after ED arrival. The primary outcome was IHCA within three days. Predictors included three domains: comorbidity, triage information (vital signs), and initial biochemical results. The final model was generated from data collected between 2015 and 2018 and validated using data from 2019.

RESULTS

A total of 257 patients, including 52 with IHCA, were analyzed. Statistical analysis selected significant variables with higher sensitivity cutoff, and scores were assigned based on relative beta coefficient ratio: K > 5.5 mmol/L (score 1), pH < 7.35 (score 1), oxygen saturation < 85% (score 1), and mean arterial pressure < 80 mmHg (score 2). The final scoring system had an area under the curve of 0.78 (p < 0.001) in the primary group and 0.75 (p = 0.023) in the validation group. The high-risk group (defined as sum scores ≥ 3) had an IHCA risk of 47.2% and 41.7%, while the low-risk group (sum scores < 3) had 18.3% and 7%, in the primary and validation databases, respectively.

CONCLUSIONS

This predictive score model for IHCA in emergent hemodialysis patients could help healthcare providers to take necessary precautions and allocate resources.

In-hospital Mortality and the Predictive Ability of the Modified Early Warning Score in Ghana: Single-Center, Retrospective Study

BACKGROUND

The modified early warning score (MEWS) is an objective measure of illness severity that promotes early recognition of clinical deterioration in critically ill patients. Its primary use is to facilitate faster intervention or increase the level of care. Despite its adoption in some African countries, MEWS is not standard of care in Ghana. In order to facilitate the use of such a tool, we assessed whether MEWS, or a combination of the more limited data that are routinely collected in current clinical practice, can be used predict to mortality among critically ill inpatients at the Korle-Bu Teaching Hospital in Accra, Ghana.

OBJECTIVE

The aim of this study was to identify the predictive ability of MEWS for medical inpatients at risk of mortality and its comparability to a measure combining routinely measured physiologic parameters (limited MEWS [LMEWS]).

METHODS

We conducted a retrospective study of medical inpatients, aged ≥13 years and admitted to the Korle-Bu Teaching Hospital from January 2017 to March 2019. Routine vital signs at 48 hours post admission were coded to obtain LMEWS values. The level of consciousness was imputed from medical records and combined with LMEWS to obtain the full MEWS value. A predictive model comparing mortality among patients with a significant MEWS value or LMEWS ≥4 versus a nonsignificant MEWS value or LMEWS <4 was designed using multiple logistic regression and internally validated for predictive accuracy, using the receiver operating characteristic (ROC) curve.

RESULTS

A total of 112 patients were included in the study. The adjusted odds of death comparing patients with a significant MEWS to patients with a nonsignificant MEWS was 6.33 (95% CI 1.96-20.48). Similarly, the adjusted odds of death comparing patients with a significant versus nonsignificant LMEWS value was 8.22 (95% CI 2.45-27.56). The ROC curve for each analysis had a C-statistic of 0.83 and 0.84, respectively.

CONCLUSIONS

LMEWS is a good predictor of mortality and comparable to MEWS. Adoption of LMEWS can be implemented now using currently available data to identify medical inpatients at risk of death in order to improve care.

Incidence of in-hospital cardiac arrest at general wards before and after implementation of an early warning score

BACKGROUND

In order to reduce the incidence of in-hospital cardiac arrest (IHCA) at general wards, medical emergency teams (MET) were implemented in the Capital Region of Denmark in 2012 as the efferent part of a track and trigger system. The National Early Warning Score (NEWS) system became the afferent part. This study aims at investigating the incidence of IHCA at general wards before and after the implementation of the NEWS system.

MATERIAL AND METHODS

We included patients at least 18 years old with IHCA at general wards in our hospital in the periods of 2006 to 2011 (pre-EWS group) and 2013 to 2018 (post-EWS group). Data was obtained from a local database and the National In-Hospital Cardiac Arrest Registry (DANARREST). We calculated incidence rate ratios (IRR) for IHCA at general wards with 95% confidence interval (95% CI). Odds ratios (OR) for return of spontaneous circulation (ROSC) and 30-day survival were also calculated with 95% CI.

RESULTS

A total of 444 IHCA occurred before the implementation of NEWS at general wards while 494 IHCA happened afterwards. The incidence rate of IHCA at general wards was 1.13 IHCA per 1000 admissions in the pre-EWS group (2006-2011) and 1.11 IHCA per 1000 admissions in the post-EWS group (2013-2018). The IRR between the two groups was 0.98 (95% CI [0.86;1.11], p = 0.71). The implementation did not affect the chance of ROSC with a crude OR of 1.14 (95% CI [0.88;1.47], p = 0.32) nor did it change the 30-day survival with a crude OR 1.30 (95% CI [0.96;1.75], p = 0.09).

CONCLUSION

Implementation of the EWS system at our hospital did not decrease the incidence rate of in-hospital cardiac arrest at general wards.

Do patients suffering an out-of-hospital cardiac arrest present to the ambulance service with symptoms in the preceding 48 h?

AIMS

In-hospital cardiac arrests are often preceded by a period of physiological deterioration that has often gone unnoticed. We proposed that the same might be true for out-of-hospital cardiac arrests (OHCAs) where ambulance crews leave patients at home who then subsequently go on to suffer a cardiac arrest.

METHODS AND RESULTS

We identified all OHCA over a 12-month period that had been seen and assessed by an ambulance crew within the 48 h preceding their cardiac arrest. We retrospectively calculated the patient's NEWS2 score at the time of their initial assessment as a marker of their physiological status and need for hospital admission. Of 1960 OHCA patients, 184 (9.4%) had been assessed by ambulance crews within the preceding 48 h. Excluding those who had been taken to hospital (and then discharged), declined hospital conveyance or were on end-of-life care pathways, 79 (56% of total) were left at home through crew discretion. Thirty-four out of 79 (43%) patients not conveyed had either a NEWS score of 3 in a single parameter or a score of ≥5, which in hospital would mandate an urgent medical review. The most overlooked observation was respiratory rate.

CONCLUSIONS

In total, 1.7% of all OHCA had been assessed in the previous 48 h and inappropriately left at home by ambulance crews. This represents a missed opportunity to avert cardiac arrest. NEWS scoring has the potential to improve pre-hospital triage of these patients and avoid missing the deteriorating patient.

Technical considerations for evaluating clinical prediction indices: a case study for predicting code blue events with MEWS

Objective.There have been many efforts to develop tools predictive of health deterioration in hospitalized patients, but comprehensive evaluation of their predictive ability is often lacking to guide implementation in clinical practice. In this work, we propose new techniques and metrics for evaluating the performance of predictive alert algorithms and illustrate the advantage of capturing the timeliness and the clinical burden of alerts through the example of the modified early warning score (MEWS) applied to the prediction of in-hospital code blue events.Approach. Different implementations of MEWS were calculated from available physiological parameter measurements collected from the electronic health records of ICU adult patients. The performance of MEWS was evaluated using conventional and a set of non-conventional metrics and approaches that take into account the timeliness and practicality of alarms as well as the false alarm burden.Main results. MEWS calculated using the worst-case measurement (i.e. values scoring 3 points in the MEWS definition) over 2 h intervals significantly reduced the false alarm rate by over 50% (from 0.19/h to 0.08/h) while maintaining similar sensitivity levels as MEWS calculated from raw measurements (∼80%). By considering a prediction horizon of 12 h preceding a code blue event, a significant improvement in the specificity (∼60%), the precision (∼155%), and the work-up to detection ratio (∼50%) could be achieved, at the cost of a relatively marginal decrease in sensitivity (∼10%).Significance. Performance aspects pertaining to the timeliness and burden of alarms can aid in understanding the potential utility of a predictive alarm algorithm in clinical settings.

Healthcare Process Modeling to Phenotype Clinician Behaviors for Exploiting the Signal Gain of Clinical Expertise (HPM-ExpertSignals): Development and evaluation of a conceptual framework

OBJECTIVE

There are signals of clinicians' expert and knowledge-driven behaviors within clinical information systems (CIS) that can be exploited to support clinical prediction. Describe development of the Healthcare Process Modeling Framework to Phenotype Clinician Behaviors for Exploiting the Signal Gain of Clinical Expertise (HPM-ExpertSignals).

MATERIALS AND METHODS

We employed an iterative framework development approach that combined data-driven modeling and simulation testing to define and refine a process for phenotyping clinician behaviors. Our framework was developed and evaluated based on the Communicating Narrative Concerns Entered by Registered Nurses (CONCERN) predictive model to detect and leverage signals of clinician expertise for prediction of patient trajectories.

RESULTS

Seven themes-identified during development and simulation testing of the CONCERN model-informed framework development. The HPM-ExpertSignals conceptual framework includes a 3-step modeling technique: (1) identify patterns of clinical behaviors from user interaction with CIS; (2) interpret patterns as proxies of an individual's decisions, knowledge, and expertise; and (3) use patterns in predictive models for associations with outcomes. The CONCERN model differentiated at risk patients earlier than other early warning scores, lending confidence to the HPM-ExpertSignals framework.

DISCUSSION

The HPM-ExpertSignals framework moves beyond transactional data analytics to model clinical knowledge, decision making, and CIS interactions, which can support predictive modeling with a focus on the rapid and frequent patient surveillance cycle.

CONCLUSIONS

We propose this framework as an approach to embed clinicians' knowledge-driven behaviors in predictions and inferences to facilitate capture of healthcare processes that are activated independently, and sometimes well before, physiological changes are apparent.

Human factors in escalating acute ward care: a qualitative evidence synthesis

Background

Identifying how human factors affect clinical staff recognition and managment of the deteriorating ward patient may inform process improvements. We systematically reviewed the literature to identify (1) how human factors affect ward care escalation (2) gaps in the current literature and (3) critique literature methodologies.

Methods

We undertook a Qualitative Evidence Synthesis of care escalation studies. We searched MEDLINE, EMBASE and CINHAL from inception to September 2019. We used the Critical Appraisal Skills Programme and the Grading of Recommendations Assessment-Development and Evaluation and Confidence in Evidence from Reviews of Qualitative Research tool to assess study quality.

Results

Our search identified 24 studies meeting the inclusion criteria. Confidence in findings was moderate (20 studies) to high (4 studies). In 16 studies, the ability to recognise changes in the patient’s condition (soft signals), including skin colour/temperature, respiratory pattern, blood loss, personality change, patient complaint and fatigue, improved the ability to escalate patients. Soft signals were detected through patient assessment (looking/listening/feeling) and not Early Warning Scores (eight studies). In contrast, 13 studies found a high workload and low staffing levels reduced staff’s ability to detect patient deterioration and escalate care. In eight studies quantifiable deterioration evidence (Early Warning Scores) facilitated escalation communication, particularly when referrer/referee were unfamiliar. Conversely, escalating concerning non-triggering patients was challenging but achieved by some clinical staff (three studies). Team decision making facilitated the clinical escalation (six studies).

Conclusions

Early Warning Scores have clinical benefits but can sometimes impede escalation in patients not meeting the threshold. Staff use other factors (soft signals) not captured in Early Warning Scores to escalate care. The literature supports strategies that improve the escalation process such as good patient assessment skills.

PROSPERO registration number

CRD42018104745.

The development and psychometric evaluation of the Clinicians' Attitudes towards Responding and Escalating care of Deteriorating patients scale

BACKGROUND

Validated measures of ward nurses' safety cultures in relation to escalations of care in deteriorating patients are lacking.

OBJECTIVES

This study aimed to develop and evaluate the psychometric properties of the Clinicians' Attitudes towards Responding and Escalating care of Deteriorating patients (CARED) scale for use among ward nurses.

METHODS

The study was conducted in two phases: scale development and psychometric evaluation. The scale items were developed based on a systematic literature review, informant interviews, and expert reviews (n = 15). The reliability and validity of the scale were examined by administering the scale to 617 registered nurses with retest evaluations (n = 60). The factor structure of the CARED scale was examined in a split-half analysis with exploratory and confirmatory factor analyses. The internal consistency, test-retest reliability, convergent validity, and known-group validity of the scale were also analysed.

RESULTS

A high overall content validity index of 0.95 was obtained from the validations of 15 international experts from seven countries. A three-factor solution was identified from the final 22 items: 'beliefs about rapid response system', 'fears about escalating care', and 'perceived confidence in responding to deteriorating patients'. The internal consistency reliability of the scale was supported with a good Cronbach's alpha value of 0.86 and a Spearman-Brown split-half coefficient of 0.87. An excellent test-retest reliability was demonstrated, with an intraclass correlation coefficient of 0.92. The convergent validity of the scale was supported with an existing validated scale. The CARED scale also demonstrated abilities to discriminate differences among the sample characteristics.

CONCLUSIONS

The final 22-item CARED scale was tested to be a reliable and valid scale in the Singaporean setting. The scale may be used in other settings to review hospitals' rapid response systems and to identify strategies to support ward nurses in the process of escalating care in deteriorating ward patients.

Association between the frequency of admission for pneumonia and the incidence of in-hospital cardiac arrest: A population-based case–control study

Objective:

To examine the association between the frequency of admission for pneumonia and the incidence of in-hospital cardiac arrest.

Methods:

We enrolled 1739 patients with in-hospital cardiac arrest and 6956 randomly selected age- and sex-matched control patients using a longitudinal claims sub-dataset from 1996 to 2011 for 1 million people randomly selected from the population covered by the Taiwan National Health Insurance program. The odds ratio of in-hospital cardiac arrest associated with the number of hospital admissions for pneumonia was calculated.

Results:

During the 15-year study period, the in-hospital cardiac arrest group had a higher frequency (28.4% vs 8.1%, p < 0.0001) of admission for pneumonia compared to the control group. The comorbidities of heart failure, chronic pulmonary disease, diabetes, renal failure, liver disease, lymphoma, alcohol abuse, and drug abuse were higher in the in-hospital cardiac arrest group than in the control group. In addition, the risk of in-hospital cardiac arrest was 3.37 for the patients admitted for pneumonia, and the risk of in-hospital cardiac arrest increased for patients with multiple admissions for pneumonia (once, 3.03; two times, 3.44; and three times, 4.42). In cross-analysis, the more admissions for pneumonia and the greater the number of comorbidities, the higher the risk of in-hospital cardiac arrest (odds ratio = 21.37, 95% confidence interval = 13.6–33.9 for patients with more than three admissions for pneumonia and more than three comorbidities).

Conclusion:

Higher admission frequency for pneumonia was associated with a higher risk of in-hospital cardiac arrest. Awareness of this risk factor may help clinicians provide early prevention or detection for patients with potential in-hospital cardiac arrest risks immediately after admission.

Relationships between the National Early Warning Score 2, clinical worry and patient outcome at discharge: Retrospective observational study

AIMS AND OBJECTIVES

To examine the performance of the National Early Warning Score 2 and composite score for clinical worry in identifying patients at risk of clinical deterioration, and to determine relationships between National Early Warning Score 2, clinical worry score and patient outcome at discharge.

BACKGROUND

The efficacy of early warning systems depends on patient population and care settings. Based on a theoretical framework on factors affecting clinical deterioration and patient outcomes, studies exploring the relationship between early warning systems and patient outcomes at discharge are sparse.

DESIGN

Retrospective observational study.

METHODS

A random sample of 732 medical records were reviewed. The area under the receiver operating characteristic curve was calculated to evaluate predictive abilities regarding the events of unanticipated in-hospital mortality, unplanned intensive care unit/ higher dependency bed admission and cardiac arrest. Multiple logistic regression analyses were performed to determine relationships between National Early Warning Score 2, clinical worry score and patient outcome. Reporting followed the STROBE checklist.

RESULTS

National Early Warning Score 2 and clinical worry score significantly predicted the events within 24 hr of the assessment. After controlling for other patient, treatment and organisational characteristics, National Early Warning Score 2 was a significant factor associated with patient outcome, but clinical worry score was not. Specifically, patients at high risk based on National Early Warning Score 2 were less likely to have improved outcome.

CONCLUSIONS

National Early Warning Score 2 and clinical worry score performed well for predicting deteriorating condition of patients. National Early Warning Score 2 was significantly associated with patient outcome. It can be used for efficient patient management for safe, quality care.

RELEVANCE TO CLINICAL PRACTICE

National Early Warning Score 2 can be used for early assessment of not only clinical deterioration but also patient outcome and provide timely intervention, when coupled with clinical worry score.

Early warning score validation methodologies and performance metrics: a systematic review

BACKGROUND

Early warning scores (EWS) have been developed as clinical prognostication tools to identify acutely deteriorating patients. In the past few years, there has been a proliferation of studies that describe the development and validation of novel machine learning-based EWS. Systematic reviews of published studies which focus on evaluating performance of both well-established and novel EWS have shown conflicting conclusions. A possible reason is the heterogeneity in validation methods applied. In this review, we aim to examine the methodologies and metrics used in studies which perform EWS validation.

METHODS

A systematic review of all eligible studies from the MEDLINE database and other sources, was performed. Studies were eligible if they performed validation on at least one EWS and reported associations between EWS scores and inpatient mortality, intensive care unit (ICU) transfers, or cardiac arrest (CA) of adults. Two reviewers independently did a full-text review and performed data abstraction by using standardized data-worksheet based on the TRIPOD (Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis) checklist. Meta-analysis was not performed due to heterogeneity.

RESULTS

The key differences in validation methodologies identified were (1) validation dataset used, (2) outcomes of interest, (3) case definition, time of EWS use and aggregation methods, and (4) handling of missing values. In terms of case definition, among the 48 eligible studies, 34 used the patient episode case definition while 12 used the observation set case definition, and 2 did the validation using both case definitions. Of those that used the patient episode case definition, 18 studies validated the EWS at a single point of time, mostly using the first recorded observation. The review also found more than 10 different performance metrics reported among the studies.

CONCLUSIONS

Methodologies and performance metrics used in studies performing validation on EWS were heterogeneous hence making it difficult to interpret and compare EWS performance. Standardizing EWS validation methodology and reporting can potentially address this issue.

A qualitative exploration of escalation of care in the acute ward setting

BACKGROUND

"Failure to Rescue" includes failing to prevent avoidable patient deterioration and death. Despite its use, delays in care escalation still affect patient outcomes.

AIMS AND OBJECTIVE

The aim of this qualitative service evaluation was to map the barriers and facilitators to the escalation of care in the acute ward setting and identify those that are modifiable.

DESIGN

A total of 55 hours of qualitative observations were completed to capture care escalation events. These were conducted at two hospital sites in one National Health Service trust.

METHODS

Observations were iterative, with research team meetings being used to discuss the data and future methods. Field notes were analysed thematically by two researchers, extracting data on barriers and facilitators to escalation of care.

RESULTS

Clinical nursing staff challenged the sensitivity and specificity of Early Warning Scores, describing tool failings in certain clinical scenarios. Staff did not escalate based on the alerting Early Warning Scores alone but used other clinical factors, such as bleeding, which are not necessarily captured in the scoring systems. Staff frequently did not re-escalate low-level scores. Patient and non-patient factors identified as posing barriers to escalation were complex care needs, patient outlier status, and involvement of multiple care teams. Factors negatively affecting the chain of communication during escalation were team tension, staffing levels, and inadequate handover.

CONCLUSION

This service evaluation identified barriers and facilitators to the escalation of care in the acute ward setting. Unlike other studies, we found that re-escalation or tracking of deterioration was problematic. Patients identified as being at a higher risk of escalation failure included complex patients, outliers, and patients with multiple care teams.

RELEVANCE TO CLINICAL PRACTICE

This service evaluation demonstrates continuing health care communication barriers. Patient groups (complex patients and outliers) risk process failures during escalation. This can be applied in clinical practice by staff anticipating problems in these patients, documenting clear escalation pathways.

Performance of patient acuity rating by rapid response team nurses for predicting short-term prognosis

Background

Although scoring and machine learning methods have been developed to predict patient deterioration, bedside assessment by nurses should not be overlooked. This study aimed to evaluate the performance of subjective bedside assessment of the patient by the rapid response team (RRT) nurses in predicting short-term patient deterioration.

Methods

Patients noticed by RRT nurses based on the vital sign instability, abnormal laboratory results, and direct contact via phone between November 1, 2016, and December 12, 2017, were included. Five RRT nurses visited the patients according to their shifts and assessed the possibility of patient deterioration. Patient acuity rating (PAR), a scale of 1–7, was used as the tool of bedside assessment. Other scores, including the modified early warning score, VitalPAC early warning score, standardised early warning score, and cardiac arrest risk triage, were calculated afterwards. The performance of these scores in predicting mortality and/or intensive care unit admission within 1 day was compared by calculating the area under the receiver operating curve.

Results

A total of 1,426 patients were included in the study, of which 258 (18.1%) died or were admitted to the intensive care unit within 1 day. The area under the receiver operating curve of PAR was 0.87 (95% confidence interval [CI] 0.84–0.89), which was higher than those of modified early warning score (0.66, 95% CI 0.62–0.70), VitalPAC early warning score (0.69, 95% CI 0.66–0.73), standardised early warning score (0.67, 95% CI 0.63–0.70) and cardiac arrest risk triage (0.63, 95% CI 0.59–0.66) (P<0.001).

Conclusions

PAR assessed by RRT nurses can be a useful tool for assessing short-term patient prognosis in the RRT setting.

The value of vital sign trends in predicting and monitoring clinical deterioration: A systematic review

BACKGROUND

Vital signs, i.e. respiratory rate, oxygen saturation, pulse, blood pressure and temperature, are regarded as an essential part of monitoring hospitalized patients. Changes in vital signs prior to clinical deterioration are well documented and early detection of preventable outcomes is key to timely intervention. Despite their role in clinical practice, how to best monitor and interpret them is still unclear.

OBJECTIVE

To evaluate the ability of vital sign trends to predict clinical deterioration in patients hospitalized with acute illness.

DATA SOURCES

PubMed, Embase, Cochrane Library and CINAHL were searched in December 2017.

STUDY SELECTION

Studies examining intermittently monitored vital sign trends in acutely ill adult patients on hospital wards and in emergency departments. Outcomes representing clinical deterioration were of interest.

DATA EXTRACTION

Performed separately by two authors using a preformed extraction sheet.

RESULTS

Of 7,366 references screened, only two were eligible for inclusion. Both were retrospective cohort studies without controls. One examined the accuracy of different vital sign trend models using discrete-time survival analysis in 269,999 admissions. One included 44,531 medical admissions examining trend in Vitalpac Early Warning Score weighted vital signs. They stated that vital sign trends increased detection of clinical deterioration. Critical appraisal was performed using evaluation tools. The studies had moderate risk of bias, and a low certainty of evidence. Additionally, four studies examining trends in early warning scores, otherwise eligible for inclusion, were evaluated.

CONCLUSIONS

This review illustrates a lack of research in intermittently monitored vital sign trends. The included studies, although heterogeneous and imprecise, indicates an added value of trend analysis. This highlights the need for well-controlled trials to thoroughly assess the research question.

Learning from different perspectives: Robust cardiac arrest prediction via temporal transfer learning

Predicting and preventing cardiac arrest is one of the biggest challenges of contemporary cardiology, as a patients survival depends on the effectiveness of the emergency response teams. While black-box models have shown to have better predictive accuracies for cardiac risk stratification, early warning scoring systems are more prominent in the hospital setting due to their ease of implementation and interpretability. We propose a temporal transfer learning approach to utilize information from adjacent time points to yield an early cardiac arrest prediction model that is robust in predictive accuracies as well as maintains the interpretability of the model coefficients. Our model estimates the logistic regression coefficients simultaneously for various time points to share knowledge from different observation windows. This framework can overcome small sample size issues, and result in robust estimation of the model coefficients. We find that our model consistently outperforms a logistic regression model fit only on vital sign data from a single time slice for 763 intensive care unit patients. Moreover, we find that the estimated coefficients from our model captures temporal trends in the data.

Continuous Monitoring of Vital Signs Using Wearable Devices on the General Ward: Pilot Study

Background

Measurement of vital signs in hospitalized patients is necessary to assess the clinical situation of the patient. Early warning scores (EWS), such as the modified early warning score (MEWS), are generally calculated 3 times a day, but these may not capture early deterioration. A delay in diagnosing deterioration is associated with increased mortality. Continuous monitoring with wearable devices might detect clinical deterioration at an earlier stage, which allows clinicians to take corrective actions.

Objective

In this pilot study, the feasibility of continuous monitoring using the ViSi Mobile (VM; Sotera Wireless) and HealthPatch (HP; Vital Connect) was tested, and the experiences of patients and nurses were collected.

Methods

In this feasibility study, 20 patients at the internal medicine and surgical ward were monitored with VM and HP simultaneously for 2 to 3 days. Technical problems were analyzed. Vital sign measurements by nurses were taken as reference and compared with vital signs measured by both devices. Patient and nurse experiences were obtained by semistructured interviews.

Results

In total, 86 out of 120 MEWS measurements were used for the analysis. Vital sign measurements by VM and HP were generally consistent with nurse measurements. In 15% (N=13) and 27% (N=23) of the VM and HP cases respectively, clinically relevant differences in MEWS were found based on inconsistent respiratory rate registrations. Connection failure was recognized as a predominant VM artifact (70%). Over 50% of all HP artifacts had an unknown cause, were self-limiting, and never took longer than 1 hour. The majority of patients, relatives, and nurses were positive about VM and HP.

Conclusions

Both VM and HP are promising for continuously monitoring vital signs in hospitalized patients, if the frequency and duration of artifacts are reduced. The devices were well received and comfortable for most patients.

Correlation of the predictive ability of early warning metrics and mortality for cardiac arrest patients receiving in-hospital Advanced Cardiovascular Life Support

BACKGROUND

The Modified Early Warning Score (MEWS) helps identify patients experiencing a decline in physiological parameters that indicate risk for cardiac arrest (CA).

OBJECTIVES

To assess the association between MEWS values and patient survival following in-hospital CA.

METHODS

Retrospective cohort study of patients who experienced in-hospital CA. The relationship between CA survival and MEWS values as well as other risk factors such as age, gender and type of electrographic cardiac rhythms was analyzed using logistic regression.

RESULTS

Survival rate to hospital discharge was 21%. Strong predictors for survival were MEWS values at hospital admission (p < .002), younger age (p < .005), ventricular fibrillation (p < .0001), and ventricular tachycardia (p < .0001). Gender and MEWS 4 hours prior to CA were not significantly associated with survival.

CONCLUSIONS

Survival following CA was significantly associated with MEWS at hospital admission but not 4 hours prior to CA. The type of cardiac rhythm and age were also predictive of survival.

Can a Patient's In-Hospital Length of Stay and Mortality Be Explained by Early-Risk Assessments?

Objective

To assess whether a patient’s in-hospital length of stay (LOS) and mortality can be explained by early objective and/or physicians’ subjective-risk assessments.

Data Sources/Study Setting

Analysis of a detailed dataset of 1,021 patients admitted to a large U.S. hospital between January and September 2014.

Study Design

We empirically test the explanatory power of objective and subjective early-risk assessments using various linear and logistic regression models.

Principal Findings

The objective measures of early warning can only weakly explain LOS and mortality. When controlled for various vital signs and demographics, objective signs lose their explanatory power. LOS and death are more associated with physicians’ early subjective risk assessments than the objective measures.

Conclusions

Explaining LOS and mortality require variables beyond patients’ initial medical risk measures. LOS and in-hospital mortality are more associated with the way in which the human element of healthcare service (e.g., physicians) perceives and reacts to the risks.

A Protocolised Once a Day Modified Early Warning Score (MEWS) Measurement Is an Appropriate Screening Tool for Major Adverse Events in a General Hospital Population

Background

The Modified Early Warning Score (MEWS) was developed to timely recognise clinically deteriorating hospitalised patients. However, the ability of the MEWS in predicting serious adverse events (SAEs) in a general hospital population has not been examined prospectively. The aims were to (1) analyse protocol adherence to a MEWS protocol in a real-life setting and (2) to determine the predictive value of protocolised daily MEWS measurement on SAEs: death, cardiac arrests, ICU-admissions and readmissions.

Methods

All adult patients admitted to 6 hospital wards in October and November 2015 were included. MEWS were checked each morning by the research team. For each critical score (MEWS ≥ 3), the clinical staff was inquired about the actions performed. 30-day follow-up for SAEs was performed to compare between patients with and without a critical score.

Results

1053 patients with 3673 vital parameter measurements were included, 200 (19.0%) had a critical score. The protocol adherence was 89.0%. 18.2% of MEWS were calculated wrongly. Patients with critical scores had significant higher rates of unplanned ICU admissions [7.0% vs 1.3%, p < 0.001], in-hospital mortality [6.0% vs 0.8%, p < 0.001], 30-day readmission rates [18.6% vs 10.8%, p < 0.05], and a longer length of stay [15.65 (SD: 15.7 days) vs 6.09 (SD: 6.9), p < 0.001]. Specificity of MEWS related to composite adverse events was 83% with a negative predicting value of 98.1%.

Conclusions

Protocol adherence was high, even though one-third of the critical scores were calculated wrongly. Patients with a MEWS ≥ 3 experienced significantly more adverse events. The negative predictive value of early morning MEWS < 3 was 98.1%, indicating the reliability of this score as a screening tool.