1 BACKGROUND

The coronavirus disease 2019 (COVID-19) outbreak is currently a worldwide pandemic and is placing tremendous pressure on global health care systems. As of July 26, 2021, more than 190 million people had been infected, and this had resulted in over 4000,000 deaths.1 A Chinese study with 155 refractory COVID-19 patients has shown that 23.2% of patients require mechanical ventilation.2 Moreover, an Italian study with 3420 COVID-19 patients has shown that 16% require intensive care.3 Facing a high infection rate, rapid progression, a complexity of treatments, and a shortage of medical resources, COVID-19 has brought many new challenges to physicians and nurses.4, 5

Studies have shown that patients who are infected by COVID-19 and have severely abnormal vital signs are more likely to deteriorate and suffer from serious events, such as death and unplanned intensive care unit (ICU) transfers.2, 6 However, due to the COVID-19 outbreak, medical staff and vital resources, such as ventilators, are scarce, which produce delays in treating the sickest patients.4, 5, 7 Nurses also carry a heavy burden on caring critical COVID-19 patients, such as the heavy nursing workload, difficulty in taking a decision on triage patients with COVID-19 and allocation of medical resources.8, 9 A triage criteria are needed to assign newly admitted patients to various risk levels of potential deterioration. The high-risk patients must then be transferred to sites where adequate medical resources and critical care can be obtained to ensure their timely treatment. In addition, a tool is also needed to help nurses make scientific and objective clinical decisions during the COVID-19 outbreak.

The National Early Warning Score 2 (NEWS2) is a risk stratification tool published by the Royal College of Physicians for bedside monitoring of patients' deterioration levels.10 The NEWS2 has performed well in predicting mortality and ICU transfers with severe outcomes, like acute respiratory failure.10-12 Through a standardized score, the NEWS2 can distinguish high-risk patients with a potential for deterioration, which may only appear hours later.13 It can alert nurses to pay more attention to such patients with high deterioration risk.10 It also provides a useful reference in helping physicians and nurses make clinical decisions in triage more rapidly and accurately.12, 13

Our aim was to assess the predictive value of NEWS2 for newly admitted patients with COVID-19 and investigate its performance in distinguishing patients with serious events. This may offer nurses and the wider health care team the opportunity to intervene at an earlier stage in the patient's illness, potentially averting serious events. We also aimed to explore its triaging ability for newly admitted patients with COVID-19.

2 METHODS 2.1 Design and setting

This retrospective single-centre observational case–control study was performed at the Xiangya ward (50 beds) at the West Campus of the Union Hospital of Tongji Medical College of Huazhong University of Science and Technology in Wuhan, Hubei province. The Union Hospital of Tongji Medical College is a tertiary general hospital with 1200 beds, where 810 beds were allocated for COVID-19 patients. This hospital has used the Diagnosis and Treatment Plan of Corona Virus Disease 2019 to confirm and classify patients into mild, ordinary, severe, and critical cases.14 The methods and preliminary data from this study have been previously published.15

Patients with fever went to the fever clinics and received the first triage. The mild cases were triaged and transferred to Fangcang shelter hospitals to decrease the pressure on traditional hospitals.16 Fangcang shelter hospitals were large temporary hospitals built by converting public venues (eg, stadiums) for patients with mild to ordinary COVID-19 symptoms, where these patients are isolated, and provided medical care and rapid transfer to the traditional hospitals should their conditions worsen.16 The ordinary, severe, and critical cases were admitted into isolation wards of traditional hospitals. The Xiangya ward is one of the isolation wards set in the traditional hospital. For a better assessment of patients with severe COVID-19, this ward used the NEWS2 to detect patients who were deteriorating. This ward was managed by the third batch of the national medical team, composed of 30 doctors and 100 nurses, from the Xiangya Hospital of Central South University, assisting Wuhan.

2.2 Study population

Eligible patients were 18 years of age and older and admitted to the Xiangya isolation ward from 29 January to March 23, 2020. Patients with more than three missing vital signs were excluded. When one to three vital signs were missing, they were counted as zero in NEWS2 calculations. Patients who died from other severe acute diseases, based on clinical judgment, were also excluded from the analysis. (Figure 1).

Flow diagram of patients with COVID-19 recruited to the study

2.3 Definition

According to the Diagnosis and Treatment Plan of Corona Virus Disease 2019 published by the National Health Commission of the People's Republic of China,14 the clinical classifications are categorized as follows,

(a) Mild cases include patients with mild clinical symptoms and no pneumonia manifestation found in imaging.

(b) Ordinary cases refer to patients with symptoms like fever and respiratory tract symptoms and pneumonia manifestation found in imaging.

(c) Severe cases are patients who meet any of the following: 1) respiratory distress, respiratory rate ≥ 30/min; 2) oxygen saturation is less than 93% at a rest state; or 3) arterial partial pressure of oxygen / oxygen concentration ≤ 300 mmHg.

(d) Critical cases are patients who meet any of the following: 1) respiratory failure occurs and mechanical ventilation is required; 2) shock occurs; or 3) complications with other organ failures that require monitoring and treatment in an ICU.14, 17

NEWS2 is a standardized clinical scoring system for improving the detection of exacerbation in patients with acute illnesses (Table S1).18 NEWS2 provides a comprehensive score based on six physiological parameters: respiratory rate, oxygen saturation, systolic blood pressure, pulse rate, body temperature, and level of consciousness. In addition, for patients requiring supplemental oxygen therapy, two points were added.10 The NEWS2 thresholds and triggers were recommended to divide the NEWS2 results into those with a low clinical risk (NEWS2 = 0-4), a low to medium clinical risk (NEWS2 = 3 in any individual parameter), a medium clinical risk (NEWS2 = 5-6), and a high clinical risk (NEWS2 ≥ 7).10

We defined serious events as a composite measure of death during hospitalisation, non-invasive ventilation treatment, or ICU treatment for any reason during the hospital stay.19

2.4 Data collection

Nurses working in the isolation wards oversaw the collection of data. A self-designed questionnaire was used to collect data on patients with COVID-19 first admitted to Xiangya ward after triage at fever clinic based on the Diagnosis and Treatment Plan of Corona Virus Disease 2019. Vital signs, blood oxygen saturation, and other parameters were collected within 30 min of admission. The concentration of oxygen therapy was recorded during the blood oxygen saturation measurement of the patient. Other data included demographic details, clinical characteristics, clinical classification of COVID-19, and serious events were collected at no specific time. The WenJuanXing web application (www.wjx.cn) was used to transfer the data to digital form.

2.5 Statistical analysis

For data analyses, IBM 26.0 (Chicago, USA) and MEDCALC statistical software were used. A P < 0.05 was regarded as being statistically significant. The normality analyses of the data were confirmed by the Kolmogorov–Smirnov and Shapiro–Wilk tests.20 Descriptive statistics were used to determine the patients' characteristics. Continuous variables were presented as the mean ± standard deviation (SD) and categorical variables as a number (%). The Mann–Whitney U-test and Student t-tests were used for means of continuous variables. Pearson's chi-square tests of independence were used for categorical variables to compare the characteristics of patient subgroups. Receiver operating characteristic (ROC) analyses were used to identify the ability of NEWS2 to predict serious events. The area under the ROC curve (AUROC), sensitivity and specificity with 95% confidence intervals (CIs) were also calculated. Cut-off values and the Youden index were used as recommended by the MEDCALC statistical software. Univariable logistic regression analyses were used to analyze the relationships between the NEWS2 results and serious events, adjusted based on demographic details and clinical characteristics. The odds ratio (OR) and 95% CIs were used as association measures. We undertook ROC analysis and univariable logistic regression analysis for every one-point increase on the NEWS2 (excepting NEWS2 score of 10 and 11).

2.6 Ethical considerations

The Medical Ethical Committee of the Xiangya Hospital approved the study (Protocol number: 202002004). Informed consent was waived as routine, care provided was not influenced, and no therapeutic intervention was introduced.

3 RESULTS

A total of 200 patients were ultimately included in this study. No patient was excluded based on the exclusion criteria.

3.1 Demographic data

The mean age of the patients was 62.1 years (median: 64, range 18-93), with 15 patients (7.5%) aged ≥80 years. A total of 104 patients (52%) were men. Among these 200 patients, there were 12 patients (6%) who had serious events. None died during the first 72 h after admission, 7 patients (3.5%) died after the first 72 h following admission, 7 patients (3.5%) experienced unplanned ICU admissions, and 11 patients (5.5%) experienced non-invasive ventilation. Patients' clinical characteristics by serious events are summarized in Table 1. Patients with serious events were significantly different in terms of gender and being affected by heart disease from patients with no serious events. Physiological parameters and NEWS2 scores at admission by serious events are shown in Table 1. Respiratory rates and inspired oxygen concentration were significantly different between patients with serious events and patients without serious events. Of the 12 patients who had serious events, one had a NEWS2 below 4 at admission, five patients' NEWS2 between 4 and 6, and six patients' NEWS2 ≥ 7.

TABLE 1. Clinical characteristics, physiological parameters and NEWS2 score of patients with COVID-19 Patients with no serious events (n = 188) Patients with serious events (n = 12) P Age, year, mean ± SD 61.95 ± 12.6 64.92 ± 13.7 0.42 Length of stay, day, mean ± SD 28.37 ± 15.6 34.83 ± 21.8 0.30 Gender, n(%) 0.03 Men 94 (50.0) 10 (83.3) women 94 (50.0) 2 (16.7) Co-morbidities, n(%) Total 79 (42.0) 7 (58.3) 0.27 Hypertension 28 (14.9) 4 (33.3) 0.20 Heart disease 21 (11.2) 5 (41.7) 0.01 Diabetes 21 (11.2) 3 (25.0) 0.33 Disease of respiratory system 11 (5.9) 2 (0.8) 0.39 Clinical classification, n(%) <0.01 Ordinary case 27 (14.4) 0 Severe case 157 (83.5) 6 (50.0) Critical case 4 (2.1) 6 (50.0) Vital signs at admission, mean ± SD Systolic blood pressure, mm Hg 132.9 ± 27.9 124.3 ± 18.4 0.15 Heart rate, beats/min 88.5 ± 16.0 93.8 ± 20.3 0.51 Respiratory rate, breaths/min 21.2 ± 3.0 27.5 ± 6.5 <0.01 Temperature, °C 36.8 ± 0.7 37.2 ± 0.8 0.14 Peripheral oxygen saturation, % 97.3 ± 4.4 96.0 ± 4.1 0.21 Supplemental oxygen, n (%) 165 (87.8) 12 (100.0) 0.41 Inspired oxygen concentration, % 39.3 ± 10.2 56.8 ± 16.7 <0.01 NEWS2 score, mean ± SD 3.5 ± 0.1 6.4 ± 0.7 <0.01 NEWS2 ≥ 4, n (%) 83 (44.1) 11 (91.7) <0.01 NEWS2 ≥ 7, n (%) 17 (9.0) 6 (50.0) <0.01 Abbreviations: COVID-19, coronavirus disease 2019; NEWS2, national early warning score 2; SD, standard deviation. 3.2 NEWS2 AUROC curve

The AUROC (95% CI) for a NEWS2 score of 3 were 0.83 (0.77-0.88), sensitivity was 0.92 and specificity was 0.56. The Youden index was 0.48 (Figure 2A).

A, ROC curve for NEWS2 of COVID-19 patients to predict the risk of serious events; B, Comparing the AUROC to predict the serious events for NEWS2 ≥ 3, NEWS2 ≥ 4, NEWS2 ≥ 5, NEWS2 ≥ 6 and NEWS2 ≥ 7, P < 0.05. Abbreviations: AUROC, area under the receiver operating characteristic curve; COVID-19, coronavirus disease 2019; NEWS2, national early warning score 2; ROC, receiver operating characteristic

3.3 Risk assessment

In comparing the predictive value of NEWS2 at different levels with serious events, the AUROC of NEWS2 ≥ 4 was the highest (0.74), with a higher sensitivity (91.7%) and a higher OR (16.4). The OR of NEWS2 ≥ 7 was the highest (18.2), where the AUROC was 0.71 and with higher specificity (91.0%; Tables 2, Figure 2B).

TABLE 2. AUROC and odds ratio performance of NEWS2 at different values to predict serious events of newly admitted patients with COVID-19, adjusted for gender and heart diseases AUROC (95%CI) PAUROC Sensitivity (95%CI) Specificity (95%CI) OR (95%CI) POR NEWS2 ≥ 1 0.53 (0.46-0.60) 0.75 100.0 (73.5-100.0) 5.3 (2.6-9.6) - 0.99 NEWS2 ≥ 2 0.55 (0.48-0.62) 0.52 100.0 (73.5-100.0) 10.1 (6.2-15.3) - 0.99 NEWS2 ≥ 3 0.70 (0.63-0.76) <0.01 100.0 (73.5-100.0) 39.9 (32.8-47.3) - 0.99 NEWS2 ≥ 4 0.74 (0.67-0.80) <0.01 91.7 (61.5-99.8) 55.9 (48.4-63.1) 16.4 (1.9-140.9) 0.01 NEWS2 ≥ 5 0.72 (0.65-0.78) <0.01 75.0 (42.8-94.5) 69.2 (62.0-75.7) 6.9 (1.6-29.1) 0.01 NEWS2 ≥ 6 0.70 (0.63-0.76) 0.02 58.3 (27.7-84.8) 81.4 (75.1-86.7) 8.8 (2.1-37.2) <0.01 NEWS2 ≥ 7 0.71 (0.64-0.77) 0.03 50.0 (21.1-78.9) 91.0 (85.9-94.6) 18.2 (3.9-85.4) <0.01 NEWS2 ≥ 8 0.66 (0.59-0.72) 0.11 33.3 (9.9-65.1) 98.0 (94.6-99.4) 28.8 (5.2-159.1) <0.01 NEWS2 ≥ 9 0.54 (0.47-0.61) 0.67 8.3 (0.2-38.5) 99.5 (97.1-100.0) 40.5 (1.6-1018.8) 0.02 NEWS2 ≥ 12 0.54 (0.47-0.61) 0.32 8.3 (0.2-38.5) 100.0 (98.1-100.0) - - Abbreviations: AUROC, area under the receiver operating characteristic curve; COVID-19, coronavirus disease 2019; NEWS2, national early warning score 2; OR, odds ratio; PAUROC, P value of AUROC; POR, P value of OR. 4 DISCUSSION

A single-centre retrospective case–control study was used to investigate the ability of NEWS2 to predict and distinguish the risk of serious events in newly admitted patients with COVID-19. We found that NEWS2 had a good predictive value, wherein the higher the NEWS2 score, the higher the risk of deterioration. The results suggest that the NEWS2 could be used to triage newly admitted patients with COVID-19. Compared to a NEWS2 score of 0 to 3, there was a 16.4-fold increase in the risk of deterioration in a NEWS2 score of ≥4 points.

In the current pandemic, the number of newly admitted patients with COVID-19 has increased. The early detection of worsening conditions in patients with COVID-19 is important because their deterioration usually develops rapidly, and early treatment and critical care will offer the greatest opportunity to avoid an adverse outcome.4, 21 Therefore, there is an urgent need for a rational triage method to determine the early risk stratification and the level of care for hospitalized patients with COVID-19, to assist nurses and physicians in clinical decision-making. We chose NEWS2 because it is easy to apply immediately and has a high predictive ability for other critically ill patients.22-24 The main clinical characteristic of patients with COVID-19 is respiratory disease.25, 26 This changes the patient's respiratory rate and oxygen saturation.26 In addition, NEWS2 may be suitable for detecting silent hypoxemia because it uses oxygen saturation and supportive oxygen therapy as scoring parameters.4

In our study, the AUROC of NEWS2 was 0.83. An AUROC of >0.8 is a good predictive value,27 indicating that NEWS2 can successfully predict the deterioration of patients with COVID-19. The cut-off point of NEWS2 was 3, which was consistent with the higher OR value (16.4) for NEWS2 ≥ 4. Our results may support the use of NEWS2 to assess newly admitted patients with COVID-19.

According to our results, NEWS2 ≥ 4 had the highest AUROC (0.74) with the high sensitivity (91.7%) and high OR (16.4). This was similar to the study by Abbott, which reported a higher OR when the NEWS was >3, compared to NEWS >1, 2, 4, or 5.28 This suggests that patients with NEWS2 < 4 have a low probability of serious events. On the other hand, a NEWS2 score of 7 or more has a high AUROC (0.71), with a high specificity (91.0%) and a high OR (18.2). This indicates that serious events are likely to occur.

In our study, 83 (44.1%) patients with no serious events were NEWS2 ≥ 4. Six (50.0%) patients with serious events were not identified at a threshold of 7. The lower NEWS2 threshold has the lower specificity with a higher incidence of false positives, that is, more patients with no serious events may be captured and would give nurses a higher nursing workload.29 The higher NEWS2 threshold has the lower sensitivity with a higher incidence of false negatives, that is, more potential deterioration patients would be missed and not be treated correctly if simply relying on NEWS2 to assess patients' clinical condition and make a clinical decision.29 Therefore, selecting the appropriate threshold of NEWS2 for triage criteria could be important to balance the incidence from false positives and false negatives. Physicians and nurses should use the NEWS2 in conjunction with clinical experience, professional medical knowledge, patient's other clinical conditions and adequate NEWS2 pre-implementation education of nurses to make better clinical decisions.30, 31

Based on our findings, the risk stratification for NEWS2 of newly admitted patients with COVID-19 was suggested as follows: low risk (NEWS2 = 0-3), medium risk (NEWS2 = 4-6), and high risk (NEWS2 ≥ 7). We suggest that newly admitted patients with NEWS2 scores of 0 to 3 should be triaged into the general ward and get general observations and receive treatment from ward nurses and physicians. Ward nurses could decide whether to continue routine NEWS2 monitoring, or increase the frequency of monitoring and/or escalation of care regarding the patient's condition, clinical experience, and professional knowledge.10 Patients with NEWS2 scores of 4 to 6 should be triaged into the general ward and be monitored minimum hourly and cared by ward nurses and physicians with competencies in critical care, and comprehensively evaluated by intensive care specialists to discuss further treatments.10 Last, patients with NEWS2 ≥ 7 should be considered to triage into ICU or the higher level care unit and get continuous monitoring of vital signs and critical care, and comprehensively evaluated by intensive care specialists.10 We also recommend that nurses continuously use NEWS2 to monitor and evaluate the condition of patients with COVID-19 after admission. Our findings are similar to the thresholds recommended by the NEWS2 guideline, which also recommends that patients with NEWS2 ≥ 7 be managed more urgently.10 In this way, physicians and nurses could make clinical decisions in treatment and critical care more rapidly and accurately. Timely medical intervention of COVID-19 patients with medium and high risks may help reduce their potential deterioration and improve their clinical outcomes.31

Of the 12 patients who had worsened, only one had a NEWS2 below 4 at admission. That patient, aged 74, died during hospitalisation, and was treated with non-invasive ventilation and intensive care. Although older patients show more atypical or unapparent symptoms than younger patients,32 another study has also shown that the NEWS2 was not useful in predicting adverse outcomes for patients with COVID-19 over 80 years old. This suggests that the clinical risk score should be used with caution in this group of patients.33 Two studies, in which the mean ages were 64 years (range 31-93) and 67.9 years (range 30-95), respectively, found that NEWS2 has a significant predictive value for patients with COVID-19.19, 34 Therefore, we recommend the continuous use of NEWS2 to monitor patients' conditions and to help nurses promptly identify patients with potential deterioration early.

The study of Myrstad et al on COVID-19 in Norway shows that NEWS2 ≥ 6 (AUROC = 0.822) have a higher predictive value than NEWS2 ≥ 5 (AUROC = 0.786).19 In our study, we found that, while NEWS2 ≥ 6 (AUROC = 0.70, OR = 8.8) had good predictive values, they were not as good as NEWS2 ≥ 5 (AUROC = 0.72, OR = 6.9) and NEWS2 ≥ 4 (AUROC = 0.74, OR = 16.4). This may be due to the conditions of the Myrstad patients, which were more severe (mortality rate = 20%); therefore, making their NEWS2 threshold higher than 4. The analysis of NEWS2 ≥ 7 is not included in Myrstad's results, so further studies may be needed to confirm these results.

In our study, we recorded the inspired oxygen concentration received by 177 patients, and the accumulated results of patients with serious events were significantly higher than those of patients with no serious events. The findings illustrate that the results of NEWS2 in our study might be affected by the situation of oxygen therapy. Because inspired oxygen can decrease the subscore of “peripheral oxygen saturation” within NEWS2 by increasing patients' blood oxygen saturation and increasing the NEWS2 score by adding 2 points for oxygen therapy. Pimentel et al suggested that the parameters of inspired oxygen concentration should be considered when using early warning scores for COVID-19 patients.26 However, NEWS2 did not stratify inspired oxygen concentration for patients. Therefore, Lim et al suggested that NEWS2 should be adjusted in terms of inspired oxygen concentration during the COVID-19 pandemic.35 Malycha et al had previously designed a “FiO2 threshold” that was combined with NEWS and found that the predictive value was higher than NEWS in predicting mortality and unplanned ICU admission.36 To the best of our knowledge, there were no studies that reported modifying inspired oxygen concentration in NEWS2 for COVID-19 patients. Therefore, further study may be conducted to evaluate the performance of NEWS2 in relation to inspired oxygen concentration for COVID-19 patients.

This study also has some limitations. First, it was a single-centre retrospective study with a small sample size. Further studies should be conducted in a larger prospective group to verify the application value of NEWS2 in the triage of newly admitted patients with COVID-19. In addition, the number of patients in terms of gender and those affected with heart disease was different between the case and control groups in the current study sample. Further studies are needed to analyze the relationships between gender and heart disease to serious events. Finally, there were no mild cases due to the admission condition controlled in our study, and this may affect the evaluation of NEWS2 predictive capacity.

5 CONCLUSION

The purpose of this study was to evaluate the ability of NEWS2 to triage newly admitted patients with COVID-19 in order to identify high-risk patients at risk of deterioration. This may allow early instigation of critical care and supportive therapies. Our study found that NEWS2 has a good predictive value for deterioration risk. Our results suggest that patients with NEWS2 scores of 0 to 3 (low-risk group) should receive general observations and treatment, patients with NEWS2 scores of 4 to 6 (medium-risk group) should be monitored minimum hourly, and patients with a NEWS2 ≥ 7 (high-risk group) should have continuous monitoring and critical care. Using NEWS2 to triage new patients with COVID-19 may help in the efficient allocation of medical resources, improving patient care and supporting clinical decision-making for nurses and other members of the multidisciplinary team.

ACKNOWLEDGEMENT

This study was supported by the grants from the Regional Collaboration Innovation Project of Xin Jiang Municipality (2017E0278).

CONFLICT OF INTEREST

The authors declare that they have no competing interests.

AUTHOR CONTRIBUTIONS

Lingli Peng: Conceptualisation, Methodology, Writing- Reviewing and Editing. Zhen Luo: Writing- Original draft preparation, Methodology, Software, Validation. Xiaobei Peng: Data curation, Software, Validation. Fangyi Zhou: Data curation. Lei Zhang: Data curation. Mengwei Guo: Software. All authors commented on previous versions of the manuscript and read and approved the final manuscript.

ETHICS APPROVAL STATEMENT

The Medical Ethical Committee of the Xiangya Hospital approved the study (Protocol number: 202002004). Informed consent was waived as routine, care was not influenced, and no therapeutic intervention was introduced.