Abstract
Introduction
Patients with cardiovascular diseases (CVD) are at significant risk of developing critical events. Early warning scores (EWS) are recommended for early recognition of deteriorating patients, yet their performance has been poorly studied in cardiac care settings. Standardisation and integrated National Early Warning Score 2 (NEWS2) in electronic health records (EHRs) are recommended yet have not been evaluated in specialist settings.
Objective
To investigate the performance of digital NEWS2 in predicting critical events: death, intensive care unit (ICU) admission, cardiac arrest and medical emergencies.
Methods
Retrospective cohort analysis.
Study cohort
Individuals admitted with CVD diagnoses in 2020; including patients with COVID-19 due to conducting the study during the COVID-19 pandemic.
Measures
We tested the ability of NEWS2 in predicting the three critical outcomes from admission and within 24 hours before the event. NEWS2 was supplemented with age and cardiac rhythm and investigated. We used logistic regression analysis with the area under the receiver operating characteristic curve (AUC) to measure discrimination.
Results
In 6143 patients admitted under cardiac specialties, NEWS2 showed moderate to low predictive accuracy of traditionally examined outcomes: death, ICU admission, cardiac arrest and medical emergency (AUC: 0.63, 0.56, 0.70 and 0.63, respectively). Supplemented NEWS2 with age showed no improvement while age and cardiac rhythm improved discrimination (AUC: 0.75, 0.84, 0.95 and 0.94, respectively). Improved performance was found of NEWS2 with age for COVID-19 cases (AUC: 0.96, 0.70, 0.87 and 0.88, respectively).
Conclusion
The performance of NEWS2 in patients with CVD is suboptimal, and fair for patients with CVD with COVID-19 to predict deterioration. Adjustment with variables that strongly correlate with critical cardiovascular outcomes, that is, cardiac rhythm, can improve the model. There is a need to define critical endpoints, engagement with clinical experts in development and further validation and implementation studies of EHR-integrated EWS in cardiac specialist settings.
Keywords: COVID-19, adult cardiology, adult intensive & critical care, risk management, health informatics
STRENGTHS AND LIMITATIONS OF THIS STUDY.
Our study is the first to examine the performance of universal early warning score (EWS) (National Early Warning Score 2, NEWS2) in patients with cardiovascular diseases in a cardiac specialist hospital.
We were able to extract data from electronic health record systems where NEWS2 is integrated and automated, reflecting the accuracy of captured parameters and enabled us to integrate other data sources for critical outcomes and COVID-19 cases.
The study followed a retrospective data collection from three data sources where there was less control of missingness of NEWS2 recordings, heart rhythm at several points in time and other parameters that could be examined like FiO2 level.
We conducted external validation of NEWS2; with our hospital data and internally; with the same data set used for supplementing the model with other parameters—external validation studies are needed for generalisability.
The endpoints examined were favoured by researchers in validation studies but may not be the ideal or precise points to measure triggered EWS against.
Introduction
Disease severity classification of patients with cardiovascular diseases (CVD) is challenging for nurses and physicians. Individuals with CVD can present with various sorts of critical events due to the disease’s pathophysiology or the comorbidities associated.1 The aetiology of the disease and the specialised care provided may impose the standardised use of deterioration risk scores, such as the widely adopted National Early Warning Score 2 (NEWS2).2 Heterogeneity in the process of patient deterioration complicates detection and escalation.
CVD is the leading cause of death in the UK and worldwide, with an estimated healthcare cost of £9 billion annually.1 3 Short-term critical events, such as cardiac arrest and transfer to the intensive care unit (ICU), are common in patients with CVD.4 5 In addition, mortality and morbidity are major concerns in patients with CVD globally.1 Risk stratification tools for long-term outcomes have long been favoured in this disease subgroup. Models like the Global Registry of Acute Coronary Events (GRACE) and the CHADS2-VASc are validated long-term risk scores in CVDs.6 7 For early-stage risk prediction, risk stratification of critical deterioration was unified for all disease groups and settings using early warning scoring (EWS) systems.8 9
The recently implemented and developed NEWS2 has been recommended by the consensus of clinical experts. A minimum set of physiological parameters and standardised application across the National Health Service (NHS) to promote patient safety and unify clinicians’ practice were the aims of NEWS2.9–11 The predictability of NEWS and NEWS2 of critical events was found fair to be acceptable to the emergency department (ED)12 13 and medical and surgical settings. On the other hand, issues reported in specialised settings, like poor predictive performance in haematology settings13 and the need to supplement NEWS in emergency settings12 14 and for patients with COVID-19,9 indicate its inefficiency in particular contexts. In cardiac settings, studies on EWS performance in critical events are poor and insufficient to defend the application for escalating the acutely ill. A study in 2012 showed high predictive ability in cardiac patients using ViEWS (VitalpacTM-derived early warning score), a model with different parameters than NEWS.15 Another examined RACE (RApid Clinical Evaluation), a postoperative cardiac EWS tool for cardiac surgical ICU; a narrowly included subset of cardiac patients.16 Little is known about the predictive value of some NEWS2 components to be deemed reliable in this subgroup, that is, inclusion of temperature or missing heart rhythm (figure 1). Despite the need for reliable early deterioration detection, it is reported that specialised cardiac centres may overlook the value of developing and validating EWS. The complexity of models and difficulties in analysing electronic health record (EHR) data formed barriers to the validation of EWS.16 However, in the era of predictive modelling generated from EHRs, and EWS embedded in EHR, validating EWS in patient population with a high rate of critical events is necessary. In the cardiac setting examined, NEWS2 was integrated into patients’ EHRs, providing a quick and reliable method of calculation and scoring. In addition, automated monitoring was implemented to facilitate the routine monitoring recording into a ready-to-calculate and evaluate NEWS2. This digitalisation process has given a unique opportunity for digital NEWS2 validation in a poorly studied subgroup with high incidents of critical events (table 1).
Figure 1.
The National Early Warning Score 2 (NEWS2).CVPU, new Confusion, Voice, Pain, Unresponsive; SpO2, peripheral capillary oxygen saturation.
Table 1.
Terms and definitions
| Term | Definition |
| NEWS2 | The National Early Warning Score 2: a model designed to perform as a tool for standardised assessment of acute illness severity by assessing physiological parameters and producing a score that helps in stratifying the risk of developing a worsening event.29 |
| Supplementing a model | The process of combining variables with the original model, that were not part of it, to improve its discriminatory ability.18 30 |
| Discrimination: | The ability of a model to distinguish between the patients who will develop an outcome of interest and the ones who will not.31 |
| Automated monitoring | Integration of patients’ routine monitoring and EHRs by transmitting measurements directly from monitoring machine to patients’ charts and continuously calculating and updating NEWS2.32 33 |
EHR, electronic health record.
Objective
To investigate the performance of digital NEWS2 in predicting critical events, at admission and prior to deterioration, for cardiac patients in the COVID-19 context in a cardiac specialist setting.
Our specific aims are:
To explore the independent association of physiological parameters and NEWS2 at hospital admission and 24 hours prior critical events; with disease severity (ICU admission, cardiac arrest, medical emergency and death).
To examine the predictive ability of NEWS2 and the supplemented NEWS2 with potential determinants of disease severity on admission and 24 hours prior to the condition worsening.
To compare the predictive value of NEWS2 with supplemented NEWS2 models (table 1).
Methods
Study cohort
The study population was defined as adult patients admitted to St Bartholomew’s Hospital, a cardiac specialist and teaching hospital in London with heart and cancer centres, from January to December 2020, under cardiac specialty care; for more than 24 hours. Due to conducting the study during the time nature of the pandemic, we have identified patients with CVD with COVID-19 based on positive PCR test results on or during admission.
Measures
NEWS2 and physiological parameters
We included physiological parameters and NEWS2 scores routinely obtained at hospital admission and 24 hours prior to deterioration. Included parameters that form NEWS2 score (figure 1) were respiratory rate (breaths per minute), oxygen saturation (%), systolic blood pressure (SBP; mm Hg), heart rate (beats per minute), temperature (°C) and consciousness (measured by Glasgow Coma Scale total score). We also included diastolic blood pressure, which is not part of NEWS2. Measurement time was chosen as the most completed set of parameters; measurements done 48 hours and 7 days prior to events were not included due to missing and inconsistent data. Heart rhythm was included 24 hours prior to event due to measurement recording by cardiac resuscitation team (CRT).
Outcomes
The primary outcome was patients’ critical status following assessment on admission or 24 hours prior to a critical event. Outcomes were critical events categorised as in-hospital death, transfer to ICU, developing cardiac arrest or medical emergency. A medical emergency was defined as deterioration, excluding cardiac arrest, requiring a patient to be seen by a critical care outreach team due to vasovagal attack, breathing difficulty, bleeding, loss of consciousness, seizure, cardiac tamponade, chest pain or prearrest rhythm.
Data processing
Data were extracted for patients admitted from January to December 2020. Patients’ demographics, physiological parameters, NEWS2 score, death and transfer to ICU were extracted from EHR. Diagnosis and comorbidities were gathered using the International Classification of Diseases 10th Revision coded data. Cardiac arrest and medical emergency were extracted from the CRT database. Patients with positive COVID-19 cases were identified from the COVID-19 pathology data, a daily updated database submitted to NHS England. Data for critical events and COVID-19 cases were linked to extracted EHR data using SQL by clinical data analysts in the hospital. Data were pseudonymised, transferred to the principal investigator (PI; BA) via the NHS network and then moved to University College London Data Safe Haven (DSH) for data analysis. The DSH is a secure database system with restricted access to the PI and research team via safe gateway technology.
Statistical analysis
Analysis was done in the DSH using the R program. Data cleaning and preprocessing were done by BA. Dependent and independent values, categorical and missing values and splitting data into training and test data sets were all addressed prior to analysis.
Statistical significance was defined as p value <0.5 using two-tailed tests.
The missing data in NEWS2 scores were assessed based on the reason for missing data. The categorical variables are presented as percentages (count), and the continuous variables are presented as the mean±SD. The normality analysis of the data was assessed using box plots for the frequencies. The intergroup difference between categorical variables was evaluated using Pearson’s χ2 test. The difference between groups was compared using the Mann-Whitney U test for non-normally distributed data.
The correlations between NEWS2 and physiological parameters and outcomes were evaluated using the Pearson correlation coefficient. Correlation coefficient values have a range between −1.0 and 1.0, where −1.0 shows perfect negative correlation and 1.0 indicates perfect positive correlation. To supplement the model with parameters that could improve the prediction, we split data into training and testing data sets using the train/test method (70% for training and 30% for testing) using decision tree through classification tree method by CART package in R. Univariate and multivariate logistic regression analyses were conducted to assess the association between score and outcomes. The predictive value of NEWS2 and supplemented model for hospital death, transfer to ICU, cardiac arrest and medical emergency were evaluated using the receiver operating characteristic (ROC) analysis. ROC is a graphical plot that shows the diagnostic ability of binary classifiers by plotting sensitivity against specificity for cut-off values used to guide the discrimination. The value of the areas under the ROC curve (AUC) was measured. The cut-off points of the models were assessed using Youden’s index: sensitivity, specificity and positive and negative predictive values. The AUC values were interpreted using the reported criteria by Fischer et al: AUC >0.9, 0.7–0.9 and 0.5–0.6 indicate high, moderate and low predictive accuracy, respectively.17
Patient and public involvement
Patients were not involved in setting the research question, outcome measures nor in the design of the study. Patients were not involved in interpretation, writing up of the results. There is no plan for the results to be disseminated to the patient population affected.
Results
Baseline characteristics
The initial cohort comprised 16 978 admitted patients, forming 40 901 encounters and 68 867 admissions to the wards from various specialties in oncology, cardiology, medicine and surgery. Patients with a primary CVD diagnosis were 7313 (36%), 14 798 encounters and 24 792 ward admissions. Patients with missing NEWS2 or physiological parameter values were 21% of the total cohort and 16% of cardiac patients. Missing NEWS2 scores were unrelated to specific values in the data set and were more likely to result from failure to record or loss in transit. In this case, they were regarded as missing completely as random17 and were removed.18 Included cardiac patients were 6143 patients admitted under cardiology, cardiothoracic surgery, congenital heart diseases or cardiac surgery specialties (online supplemental appendix 1). Patients with COVID-19 were 248 (4%), 40% were cardiac patients.
bmjopen-2022-066131supp001.pdf (419.8KB, pdf)
The mean age of the cardiac population is 63.73+14.47, and 69% were males. The in-hospital mortality was 12% (743 patients), ICU admission was 15% (921 patients), and 117 cardiac arrests and 160 medical emergencies. The characteristics of the study population are tabulated in table 2. The difference between dead cases, patients admitted to ICU, who developed cardiac arrest, or medical emergencies, and those who did not develop critical outcomes according to the NEWS2 scoring category were statistically significant (p<0.001). The comparison is tabulated in table 3.
Table 2.
General characteristics of patients
| Variables | Descriptive (n=6143) |
| Age (years) | 63.7±14.47 |
| Gender (%) | |
| Male | 4239 (69) |
| Female | 1904 (31) |
| Specialty (%) | |
| Cardiology | 3817 (62) |
| Cardiothoracic surgery | 2020 (33) |
| Congenital heart disease | 184 (3) |
| Cardiac surgery | 122 (2) |
| Patients with COVID-19 (%) | 248 (4) |
| Cardiology and cardiothoracic surgery (%) | 100 (40) |
| Other (oncology, haematology, respiratory and thoracic surgery) (%) | 148 (60) |
| NEWS2 numerical parameters | |
| Systolic blood pressure (mm Hg) | 128.3±23.2 |
| Mean arterial pressure (mm Hg) | 91.21±14.1 |
| Pulse rate (beats per minute) | 74.43±16.7 |
| Temperature (°C) | 36.5±0.6 |
| Oxygen saturation (%) | 96.7±2.23 |
| Diastolic blood pressure (mm Hg) | 72.7±11.9 |
| NEWS2 | 1.5±1.7 |
| Outcomes (%) | |
| In-hospital mortality | 743 (12) |
| ICU admission | 921 (15) |
| Cardiac arrest | 117 (2) |
| Medical emergency | 160 (3) |
ICU, intensive care unit; NEWS2, National Early Warning Score 2.
Table 3.
Comparison between categorical parameters of study population
| Characteristic | NEWS2 categories | P value | ||
| Low (0–4) n (%) | Moderate (5–6) n (%) | High (≥7) n (%) | ||
| Specialty | ||||
| Cardiology | 3431 (90) | 249 (7) | 128 (3) | <0.001 |
| Cardiothoracic surgery | 1931 (83) | 86 (4) | 10 (0.5) | <0.001 |
| Congenital heart disease | 164 (89) | 17 (9) | 3 (2) | <0.001 |
| Cardiac surgery | 113 (92) | 6 (5) | 3 (2) | <0.001 |
| Outcomes | ||||
| In-hospital mortality | 597 (80) | 62 (8) | 84 (12) | <0.001 |
| ICU admission | 850 (92) | 36 (4) | 35 (4) | 0.013 |
| Cardiac arrest | 85 (73) | 14 (12) | 18 (15) | <0.001 |
| Medical emergency | 140 (87) | 12 (8) | 8 (5) | <0.001 |
ICU, intensive care unit; NEWS2, National Early Warning Score 2.
The mean of NEWS2 in death cases was higher by a small difference than alive patients (difference=1.005, p<0.001). Between ICU admission and non-admitted cases, the mean was similar (difference=0.01, p<0.09). Between cardiac arrests and non-arrest cases, and medical emergency and stable cases, there was a small variation (difference=1.99, p<0.001 and difference=0.99, p<0.001, respectively) (figures 2–5).
Figure 2.
National Early Warning Score 2 (NEWS2) in survived and non-survived cases.
Figure 3.
National Early Warning Score 2 (NEWS2) in intensive care unit (ICU) admission and non-ICU cases.
Figure 4.
National Early Warning Score 2 (NEWS2) in cardiac arrest (CA) and non-arrest cases.
Figure 5.
National Early Warning Score 2 (NEWS2) in medical emergency (ME) and non-emergency cases.
Using the correlation matrix between parameters measured on admission and outcomes, we found a positive correlation between temperature with heart rate (0.32); respiration rate, heart rate and death with NEWS2 (0.41, 0.31, 0.30); and death with cardiac arrest (0.31). In the parameters 24 hours prior to critical events, there was a strong correlation of SpO2 with systolic pressure, CVPU (new Confusion, Voice, Pain, Unresponsive) with NEWS2 (0.42, 0.41, 0.42), and systolic pressure, SpO2 and death with age (0.30, 0.31, 0.34). Cardiac rhythm is strongly correlated with cardiac arrest and death (0.51, 0.90) (figure 6 and online supplemental appendix 2).
Figure 6.
Correlation matrix using Pearson’s correlation coefficient between parameters and National Early Warning Score 2 (NEWS2) on admission and outcomes. DBP, diastolic blood pressure; ICU, intensive care unit; SBP, systolic blood pressure; CVPU, new Confusion, Voice, Pain, Unresponsive; RR, Respiration rate; MAP, Mean arterial blood pressure.
Regarding the discrimination, NEWS2 on admission showed moderate to low predictive accuracy with death, ICU admission, cardiac arrest and medical emergency (AUC: 0.63, 0.56, 0.69 and 0.63, respectively), while NEWS2 24 hours before event showed low predictive value (AUC: 0.58, 0.63, 0.54 and 0.56, respectively). In patients with COVID-19, NEWS2 on admission showed good to poor performance (AUC: 0.64, 0.51, 0.81 and 0.80). When NEWS2 was supplemented with age, there was insignificant change in the predictive performance for all patients (AUC: 0.63, 0.53, 0.73 and 0.64) However, there was significant improvement for patients with COVID-19 (AUC: 0.96, 0.69, 0.87 and 0.88). This was also true for the model of NEWS2 supplemented with heart rhythm for the cardiac patients (AUC: 0.75, 0.84, 0.95 and 0.94). The calculated optimum cut-off value for NEWS2 was >5 which showed sensitivity for NEWS2 model of 20% and specificity of 94%, while for the model supplemented with heart rhythm sensitivity was 30% and specificity was 85% (table 4 and online supplemental appendix 3).
Table 4.
NEWS2 as a predictor of critical events compared with supplemented NEWS2
| Model | Death AUC (95% CI) |
ICU admission AUC (95% CI) | Cardiac arrest AUC (95% CI) | Medical emergency AUC (95% CI) |
| Cardiac patients | ||||
| NEWS2 (admission) | 0.63 (0.58–0.67) | 0.56 (0.51–0.62) | 0.69 (0.65–0.74) | 0.63 (0.57–0.67) |
| NEWS2 (24 hours before outcome) | 0.58 (0.53–0.65) | 0.63 (0.58–69) | 0.54 (0.51–0.67) | 0.56 (0.52–0.60) |
| NEWS2+age | 0.63 (0.58–0.69) | 0.53 (0.50–0.59) | 0.73 (0.68–0.79) | 0.64 (0.59–0.68) |
| NEWS2+rhythm | 0.75 (0.67–0.80) | 0.84 (0.78–0.88) | 0.95 (0.89–0.98) | 0.94 (0.89–0.97) |
| Patients with COVID-19 | ||||
| NEWS2 (admission) | 0.64 (0.59–0.69) | 0.51 (0.50–0.56) | 0.81 (0.75–0.86) | 0.80 (0.74–0.86) |
| NEWS2+age | 0.96 (0.89–0.98) | 0.69 (0.63–0.76) | 0.87 (0.81–0.94) | 0.88 (0.82–0.94) |
AUC, area under the receiver operating characteristic curve; ICU, intensive care unit; NEWS2, National Early Warning Score 2.
Discussion
Our retrospective study is among the first to evaluate the prognostic power of the digital EWS (NEWS2) in a patient with CVD in a specialist cardiac setting. We were able to validate NEWS2 in patients with COVID-19 in the cardiac setting. The main findings of the study reveal that: (1) NEWS2 is inadequate on its own to predict deterioration in patients with CVD in the examined specialist cardiac setting; and (2) adjustment of the tool by supplementing with positively correlated parameters can strengthen the prognostic performance and therefore reduce the burden of critical events associated with CVD.
Our findings were consistent with a previous study in patients with chest pain19 while contrary to findings by a study that examined a subset of patients with CVD in a single hospital.15 MEWS (modified early warning score) showed low predictive accuracy in patients with chest pain in the ED.19 However, good discrimination was found of ViEWS in a subset of patients in Canada a decade ago and using a distinct EWS than the current NEWS2.15 Studies were limited in number, scope and population and varied in EWS models. Patients with ‘normal’ vital signs may be sicker than they look through traditional routine monitoring.20 When EWS was explicitly developed for postcardiac surgery patients, prognostic accuracy was excellent in predicting ICU mortality.16 It included a range of organ system-specific parameters that correlate with cardiac surgery outcomes, such as lactic acid, FiO2 and platelets. Relevant parameters were found using machine learning, including clinical signs and heart rate variability, to improve scoring systems for adverse cardiac events.19 The cardiac rhythm combined with NEWS2 in our study, and the heart rate variability, such as the average of the instantaneous heart rate or ratio of low frequency (LF) power to high frequency (HF) power (LF/HF) selected by Liu et al,19 may not be routinely measured or readily available for clinicians as SBP or temperature. Yet, their predictive value indicates the need for highly illness-correlated parameters to be present for clinicians through facilitating timely and thorough assessment.
The subset of admitted patients with COVID-19 showed improvement in NEWS2 when the model was adjusted with age. The finding is supported by a study that reported low to moderate discrimination of NEWS2 for the severity of COVID-19 disease.21 Adjusting the model with age alone did not improve prediction in the study by Carr et al, as shown in our findings.9 However, supplementing with age, routinely collected blood and physiological parameters enhanced discrimination in a multisite (UK and non-UK hospitals) study,9 which seems to further support our results yet may detect the need for additional criteria adjustment. Our study indicates that EWS are not a stand-alone rapid assessment tool as it accords with reported limitations in rapid risk scores for predicting cardiovascular complications.22 23 In clinical settings, and before implementing EWS, clinicians look for signs of abnormality related to body organs affected, disease pathophysiology or procedure side effect to critically assess the situation. Systems that existed to stratify the risk of long-term cardiac complications have been successfully validated and used for years, such as the thrombolysis in myocardial infarction score24 and GRACE.25 They included cardiac disease variables: heart rate variabilities and serum cardiac biomarkers, which may not be available routinely or at the first admission presentation for rapid assessment. It was also observed that combining nurses’ objective assessment with traditional EWS in the Dutch-Early-Nurse-Worry-Indicator-Score improved the prediction of ICU admission and mortality in surgical patients.26 Therefore, thorough tracking of short-term deterioration parameters to develop decisive intelligent scoring systems will potentially outperform EWS in various diseases and settings.
It is essential to consider possible issues in developing EWS for specialised subgroups. A potential complexity may be present when having a variety of multiple parameters measured at various times during admission to form a scoring system that is meant to be simplistic and standardised. In addition, the endpoints favoured by researchers in validation studies may not be the ideal points to measure triggered EWS against. In the clinical application of EWS, a high score triggers an action to prevent a critical event.27 In the event of clinically intervening at the right time, the examined adverse events may not occur. Prior to reaching or while preventing an adverse event, precise and proper deterioration endpoints may be more fitted than traditionally studied outcomes, such as death and ICU admission. At the current time of available EHR integration and data science techniques, it is possible and may be more valid to identify and define appropriate critical illness endpoints to examine EWS against.
We assessed the performance of digitally integrated EWS; the integration generates NEWS2 in patients’ charts from remotely captured parameters by automated monitoring. EHR integration and automation can improve the accuracy and alerting promptness.28 We were able to extract a good sample of patients with CVD and identify patients diagnosed with COVID-19 despite the missingness of some NEWS2 recordings. The issue in recording completion could be due to a lack of staff adherence to routine and timely monitoring, as each measurement would be automatically transmitted to patients’ charts.
Therefore, careful and selective modelling of algorithms from parameters that can be available routinely and reflect significant clinical meaning is needed. The validity of NEWS2 in specialist settings like cardiology indicates the need for either score enhancement or systemic supplementing and finer endpoints definition for better detection. Studying the clinical environment from a practical side of EWS will explain the adoption and implementation role in the success or failure of EWS. From various specialties, clinicians’ involvement in models’ development and validation is invaluable to produce a higher accuracy and finely clinical expertise-born warning scores.
Conclusion
The EWS (NEWS2) in patients with CVDs is suboptimal to predict deterioration early. Adjusting EWS with variables that strongly correlate with critical cardiovascular outcomes will improve the early scoring models. Thorough tracking of parameters in EHRs and data availability can support the generation of decisive, intelligent models for a readily feasible system in routine clinical work. There is a need for defining and revising critical endpoints and the involvement of clinicians in models’ development that reflect a significant meaning for deterioration detection. Further validation and implementation studies in cardiac specialist settings and other specialist subgroups are required to investigate methods needed to enhance the capacity of EWS where it is least investigated.
bmjopen-2022-066131supp002.pdf (1.2MB, pdf)
Supplementary Material
Footnotes
Twitter: @amibanerjee1
Contributors: AB, DM and TB conceived the study. BA carried out the data collection with guidance from DM. BA conducted the analysis and interpretation of findings with revisions from AB. BA wrote the manuscript. AB, DM, TB and RP contributed to the interpretation and revision of the manuscript. AB was guarantor for this work.
Funding: BA has received PhD funding from the Saudi Arabian Cultural Bureau to conduct the study (grant number: ELP003964).
Competing interests: None declared.
Patient and public involvement: Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Provenance and peer review: Not commissioned; externally peer reviewed.
Supplemental material: This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.
Data availability statement
All data relevant to the study are included in the article or uploaded as supplementary information.
Ethics statements
Patient consent for publication
Not applicable.
Ethics approval
This study is registered and approved by the Health Research Authority (HRA) and Health and Care Research Wales (HCRW). Approval was obtained from Stanmore Research Ethics Committee in London (REC reference: 20/PR/0286). A confirmation of capacity has been received from St Bartholomew’s Hospital. Data collected are essential for patient care and were sufficiently anonymised; therefore, it was not considered necessary to obtain consent from patients.
References
- 1.Mozaffarian D. Global scourge of cardiovascular disease. J Am Coll Cardiol 2017;70:26–8. 10.1016/j.jacc.2017.05.007 [DOI] [PubMed] [Google Scholar]
- 2.Inada-Kim M, Nsutebu E. News 2: an opportunity to standardise the management of deterioration and sepsis. BMJ 2018;360:k1260. 10.1136/bmj.k1260 [DOI] [PubMed] [Google Scholar]
- 3.BHF . n.d. UK factsheet. british heart foundation. 2019;(April):1–21.
- 4.Herlitz J, Bång A, Ekström L, et al. A comparison between patients suffering in-hospital and out-of-hospital cardiac arrest in terms of treatment and outcome. J Intern Med 2000;248:53–60. 10.1046/j.1365-2796.2000.00702.x [DOI] [PubMed] [Google Scholar]
- 5.Sandroni C, Nolan J, Cavallaro F, et al. In-Hospital cardiac arrest: incidence, prognosis and possible measures to improve survival. Intensive Care Med 2007;33:237–45. 10.1007/s00134-006-0326-z [DOI] [PubMed] [Google Scholar]
- 6.Aragam KG, Tamhane UU, Kline-Rogers E, et al. Does simplicity compromise accuracy in ACS risk prediction? A retrospective analysis of the TIMI and grace risk scores. PLoS One 2009;4:e7947. 10.1371/journal.pone.0007947 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Shariff N, Aleem A, Singh M, et al. AF and venous thromboembolism - pathophysiology, risk assessment and CHADS-vasc score. J Atr Fibrillation 2012;5:649. 10.4022/jafib.649 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Smith GB, Prytherch DR, Meredith P. The ability of the national early warning score (news) to discriminate patients at risk of early cardiac arrest, unanticipated intensive care unit admission, and death. Resuscitation 2013;84:465–70. 10.1016/j.resuscitation.2012.12.016 [DOI] [PubMed] [Google Scholar]
- 9.Carr E, Bendayan R, Bean D, et al. Supplementing the national early warning score (NEWS2) for anticipating early deterioration among patients with COVID-19 infection. MedRxiv [Internet] 2020. [Google Scholar]
- 10.Day T, Oxton J. The National early warning score in practice: a reflection. Br J Nurs 2014;23:1036–40. 10.12968/bjon.2014.23.19.1036 [DOI] [PubMed] [Google Scholar]
- 11.Royal College of Physicians of London . National early warning score (NEWS): standardising the assessment of acute-illness severity in the NHS. Royal College of Physician, 2012. [Google Scholar]
- 12.Dundar ZD, Kocak S, Girisgin AS. Lactate and NEWS-L are fair predictors of mortality in critically ill geriatric emergency department patients. Am J Emerg Med 2020;38:217–21. 10.1016/j.ajem.2019.02.006 [DOI] [PubMed] [Google Scholar]
- 13.Alhmoud B, Bonnici T, Patel R, et al. Performance of universal early warning scores in different patient subgroups and clinical settings: a systematic review. BMJ Open 2021;11:e045849. 10.1136/bmjopen-2020-045849 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Eckart A, Hauser SI, Kutz A, et al. Combination of the national early warning score (news) and inflammatory biomarkers for early risk stratification in emergency department patients: results of a multinational, observational study. BMJ Open 2019;9:e024636. 10.1136/bmjopen-2018-024636 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Kellett J, Kim A. Validation of an abbreviated vitalpacvalidation of an abbreviated vitalpac. Resuscitation 2012;83:297–302. 10.1016/j.resuscitation.2011.08.022 [DOI] [PubMed] [Google Scholar]
- 16.Badreldin AMA, Doerr F, Bender EM, et al. Rapid clinical evaluation: an early warning cardiac surgical scoring system for hand-held digital devices. Eur J Cardiothorac Surg 2013;44:992–7; 10.1093/ejcts/ezt232 [DOI] [PubMed] [Google Scholar]
- 17.Fischer JE, Bachmann LM, Jaeschke R. A readers’ guide to the interpretation of diagnostic test properties: clinical example of sepsis. Intensive Care Med 2003;29:1043–51. 10.1007/s00134-003-1761-8 [DOI] [PubMed] [Google Scholar]
- 18.Katuwal S, Knadel M, Norgaard T, et al. Predicting the dry bulk density of soils across denmark: comparison of single-parameter, multi-parameter, and vis–NIR based models. Geoderma 2020;361:114080. 10.1016/j.geoderma.2019.114080 [DOI] [Google Scholar]
- 19.Liu N, Koh ZX, Goh J, et al. Prediction of adverse cardiac events in emergency department patients with chest pain using machine learning for variable selection. BMC Med Inform Decis Mak 2014;14:75. 10.1186/1472-6947-14-75 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Hong W, Earnest A, Sultana P, et al. How accurate are vital signs in predicting clinical outcomes in critically ill emergency department patients. Eur J Emerg Med 2013;20:27–32. 10.1097/MEJ.0b013e32834fdcf3 [DOI] [PubMed] [Google Scholar]
- 21.Kostakis I, Smith GB, Prytherch D, et al. The performance of the national early warning score and national early warning score 2 in hospitalised patients infected by the severe acute respiratory syndrome coronavirus 2 (sars-cov-2). Resuscitation 2021;159:150–7. 10.1016/j.resuscitation.2020.10.039 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Hollander JE, Robey JL, Chase MR, et al. Relationship between a clear-cut alternative noncardiac diagnosis and 30-day outcome in emergency department patients with chest pain. Acad Emerg Med 2007;14:210–5. 10.1197/j.aem.2006.09.053 [DOI] [PubMed] [Google Scholar]
- 23.Manini AF, Dannemann N, Brown DF, et al. Limitations of risk score models in patients with acute chest pain. Am J Emerg Med 2009;27:43–8. 10.1016/j.ajem.2008.01.022 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Antman EM, Cohen M, Bernink PJ, et al. The TIMI risk score for unstable angina/non-ST elevation MI: a method for prognostication and therapeutic decision making. JAMA 2000;284:835–42. 10.1001/jama.284.7.835 [DOI] [PubMed] [Google Scholar]
- 25.Eagle KA, Lim MJ, Dabbous OH, et al. A validated prediction model for all forms of acute coronary syndrome: estimating the risk of 6-month postdischarge death in an international registry. JAMA 2004;291:2727–33. 10.1001/jama.291.22.2727 [DOI] [PubMed] [Google Scholar]
- 26.Douw G, Huisman-de Waal G, van Zanten ARH, et al. Nurses’ “worry” as predictor of deteriorating surgical ward patients: a prospective cohort study of the dutch-early-nurse-worry-indicator-score. Int J Nurs Stud 2016;59:134–40. 10.1016/j.ijnurstu.2016.04.006 [DOI] [PubMed] [Google Scholar]
- 27.Pedrosa I, Browning T, Kwon JK, et al. Response to COVID-19: minimizing risks, addressing challenges and maintaining operations in a complex academic radiology department. J Comput Assist Tomogr 2020;44:479–84. 10.1097/RCT.0000000000001040 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.National Institute For Health And Care Excellence (NICE) . National early warning score systems that alert to deteriorating adult patients in hospital. Medtech Innovation Briefing 2020:1–18.:978-1–4731. [Google Scholar]
- 29.Royal College of Physicians of London . NHS england approves use of national early warning score (NEWS) 2 to improve detection of acutely ill patients. Royal College of Physician, 2017. [Google Scholar]
- 30.Schreuder HT, Czaplewski R, Bailey RG. Combining mapped and statistical data in forest ecological inventory and monitoring – supplementing an existing system. Environ Monit Assess 1999;56:269–91. 10.1023/A:1005984426987 Available: 10.1023/A:1005984426987 [DOI] [Google Scholar]
- 31.Altman DG, Royston P. What do we mean by validating a prognostic model? Statist Med 2000;19:453–73. [DOI] [PubMed] [Google Scholar]
- 32.Lockwood JM, Thomas J, Martin S, et al. AutoPEWS: automating pediatric early warning score calculation improves accuracy without sacrificing predictive ability. Pediatr Qual Saf 2020;5:e274. 10.1097/pq9.0000000000000274 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Khanna AK, Hoppe P, Saugel B. Automated continuous noninvasive ward monitoring: future directions and challenges. Crit Care 2019;23:194. 10.1186/s13054-019-2485-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
bmjopen-2022-066131supp001.pdf (419.8KB, pdf)
bmjopen-2022-066131supp002.pdf (1.2MB, pdf)
Data Availability Statement
All data relevant to the study are included in the article or uploaded as supplementary information.