COMMENTARY: Is a Change from the National Early Warning System (NEWS) Warranted in Patients with Chronic Respiratory Conditions?

Early warning systems (EWS) were introduced in hospital systems in a number of countries including the UK (NEWS-1 and NEWS-2 UK) and Ireland (NEWS now INEWS V2) with the aim of identifying patients with acute physiological deterioration in a timely manner in order to prevent or reduce cardiopulmonary arrest, admission to the intensive care unit and premature death [1]. Both NEWS-1 and NEWS-2 UK and INEWS are based on the ViEWS VitalPac EWS [2]. However, INEWS uses precisely the same parameter ranges and scoring system as ViEWS while some minor modifications were made in NEWS-1 and NEWS-2 UK. These modifications included NEWS-2 UK introduced a second SpO₂ scoring system for use in patients with hypercapnic respiratory failure who had a prescribed oxygen saturation requirement of 88 − 92%. The existing SpO₂ scoring system (SpO₂ scale 1) used in both NEWS-1 and NEWS-2 UK applies to the majority of patients. The new dedicated SpO₂ scoring system (SpO₂ scale 2) in NEWS-2 UK is used for patients with confirmed hypercapnic respiratory failure with the NEWS-2 UK scoring system adjusted accordingly. Previous reports have voiced concern that the SpO₂ scoring system within the NEWS-1 UK and the Irish NEWS is not suitable for specific patient sub-populations, in particular those with chronic respiratory conditions including chronic obstructive pulmonary disorder (COPD) [3, 4]. Some of these patients have a lived baseline of lower oxygen saturation levels, usually between 88 to 92% SpO₂. An SpO₂ below 95% triggers an alert within NEWS. This can result in excessive triggering and false alarms in this high risk sub-population of chronic respiratory patients as well as the potential for the inappropriate use of high levels of supplemental oxygen in hypercapnic patients.[5] As part of the update to the Irish National Early Warning System (INEWS) National Clinical Guideline No.1 [see INEWS Guideline Version 2], the authors undertook a systematic review of the evidence comparing the NEWS-1 UK to modified EWSs (EWSs in which parameters are adjusted) in sub-populations of patients with chronic respiratory conditions including chronic hypoxia, chronic hypoxaemia, chronic physiological abnormalities, pulmonary fibrosis or COPD. The effectiveness of the modified EWSs in predicting mortality in this sub-population was compared to the NEWS-1 UK only.

Systematic literature search

Following a systematic search of the literature (until Feb 2018), four studies were identified; all included patients with chronic respiratory conditions and compared a modified EWS (NEWS-2 UK, CREWS (Chronic Respiratory Early Warning System), S-NEWS (Salford NEWS), and CROS (Capital Region of Denmark NEWS Override System)) to NEWS-1 UK [6–9]. Three of these studies were conducted in the UK [6, 7, 9] and one in Denmark [8] and all four were retrospective studies within an acute hospital setting. The sample size ranged from 196 patients [6], to 251,266 admissions (number of patients not reported, and of the total, 48,898 admissions were associated with patients at risk of type 2 respiratory failure (T2RF) and 1,394 were patients with ‘documented T2RF’) [9]. The modified EWSs had the same seven parameters as NEWS-1 UK and modifications were largely in the SpO₂ weighting and cut-offs as this has been associated with excessive triggering and increased workloads particularly in patients with chronic respiratory conditions (see Table 1).

Table 1. Characteristics and key findings of the studies which compared modified EWS to the NEWS in a sub-population of patients with chronic respiratory disease.

Study (Year) Country	EWS Compared	Parameters included	SpO2 weighting in modified EWS	Outcome, AUC (95% CI)
Eccles (2014) UK	NEWS-1 UK CREWS	RR, SpO2, FiO2, SBP, HR, AVPU, Temperature	88-92%	30-day mortality in chronic hypoxaemia patients NEWS-1 UK: AUC 0.88 (95% CI 0.79-0.96) CREWS: AUC 0.91 (95% CI 0.85-0.98)
Hodgson (2017) UK	NEWS-1 UK CREWS S-NEWS	Not specifically reported	Not specifically reported	Inpatient mortality in acute exasperation of COPD cohort: NEWS-1 UK: AUC 0.74 (95% CI 0.66-0.82) CREWS: AUC 0.72 (95% CI 0.63 to 0.80) S-NEWS: AUC 0.62 (95% CI 0.53 to 0.70). Non-COPD cohort: NEWS-1 UK=AUC 0.77 (95% CI 0.75 to 0.78)
Pedersen (2018) Denmark	NEWS-1 UK CROS CREWS S-NEWS	RR, SpO2, FiO2, SBP, HR, AVPU, Temperature	CROS: Option for doctors to apply limits to all parameters except temperature for individual patients CREWS and S-NEWS: 88-92%	48 h mortality in COPD patients: NEWS-1 UK: AUC 0.85 (95% CI 0.85-0.86) CROS: AUC 0.82 (95% CI 0.82-0.83) CREWS: AUC 0.85 (95% CI 0.84-0.85) S-NEWS: AUC 0.84 (95% CI 0.84-0.85)
Pimentel (2018) UK	NEWS-1 UK NEWS-2 UK CREWS	RR, SpO2, FiO2, SBP, HR, AVPU, Temperature	Below 88%	In-hospital death within 24-hours Patients with ‘documented’ type 2 respiratory failure T2RF: NEWS-1 UK: AUC 0.86 (95% CI 0.85-0.88) NEWS-2 UK: AUC 0.84 (95% CI 0.83-0.85) CREWS: AUC 0.86 (95% CI 0.85 − 0.88) Patients ‘at risk’ of type 2 respiratory failure: OUH Trust (4 hospitals) NEWS-1 UK: AUC 0.89 (95% CI 0.88-0.90) NEWS-2 UK: AUC 0.88 (95% CI 0.87-0.89) CREWS: AUC 0.90 (95% CI 0.89 − 0.90) PH Trust (1 large hospital) NEWS-1 UK: AUC 0.88 (95% CI 0.88-0.88) NEWS-2 UK: AUC 0.86 (95% CI 0.85-0.86) CREWS: AUC 0.88 (95% CI 0.88 − 0.88) Patients ‘not at risk’ of T2RF (control) OUH Trust (4 hospitals) NEWS-1 UK: AUC 0.92 (95% CI 0.92-0.93) NEWS-2 UK: AUC 0.90 (95% CI 0.90-0.91) CREWS: AUC 0.92 (95% CI 0.91 − 0.92) PH Trust (1 large hospital) NEWS-1 UK: AUC 0.91 (95% CI 0.91-0.91) NEWS-2 UK: AUC 0.89 (95% CI 0.89-0.89) CREWS: AUC 0.90 (95% CI 0.90 − 0.91)

Legend: AUC: Area under the curve; AVPU: Alert, voice, pain, unresponsive; CI: Confidence interval; COPD: Chronic obstructive pulmonary disorder; CREWS: Chronic respiratory EWS; CROS: Capital Region of Denmark NEWS Override System; EWS: Early Warning Score/System; FiO₂: Inspired oxygen; HR: Heart rate; NEWS: National Early Warning Score/System, OUH Trust: Oxford University Hospitals; PH Trust: Portsmouth Hospitals NHS Trust; RR: Respiratory rate; S-NEWS: Salford NEWS; SBP: Systolic blood pressure; SpO₂: Oxygen saturation; T2RF: Type 2 respiratory failure.

Key comparisons and findings

Across the four studies, there was variation in terms of included participants and settings ranging from 196 patients with chronic hypoxaemia in the respiratory wards of two UK hospitals [6]; 39,470 patients with an acute exacerbation of COPD in two UK acute medical units (AMU) [7]; patients with COPD from a single Danish hospital [8]; and patients from 251,266 admissions including with ‘documented or at risk of’ type II respiratory failure (defined as patients with COPD, bronchiectasis, cystic fibrosis, obesity hypoventilation syndrome or respiratory failure) from five UK hospitals in two National Health Service (NHS) trusts [9].

The number and type of EWSs compared also varied: one study compared two different EWSs (NEWS-1 UK versus CREWS) [6], two studies compared three different EWSs (NEWS-1 UK versus NEWS-2 UK versus CREWS) [9] and (NEWS-1 UK versus CREWS versus S-NEWS) [7], and one study compared four different EWSs (NEWS-1 UK versus CROS versus CREWS versus S-NEWS) [8].

In addition, the timing of follow up for mortality varied across the four studies: in-hospital mortality within 24 h [9], 48-hour mortality [8], 30-day mortality [6] and inpatient mortality [death before discharge from hospital][7].

The discrimination of these modified EWSs (i.e. the area under the curve [AUC]) also varied (see Table 1). Eccles et al. [6], reported that the CREWS had better ability to predict 30-day mortality than the NEWS-1 UK in chronic hypoxaemia patients and found no difference between groups. In Hodgson et al. [7], the NEWS-1 UK had a slightly better ability to predict inpatient mortality than the CREWS and the S-NEWS in the COPD patient cohort, however the 95% confidence intervals overlap. In Pedersen et al. [8], the NEWS-1 UK, modified CREWS and the S-NEWS had similar predictive ability for 48 h mortality, and the NEWS-1 UK was slightly superior to the CROS EWS. In Pimentel et al. [9], NEWS-1 UK, NEWS-2 UK and CREWS had similar discriminatory ability in patients with documented type II respiratory failure, though in patients “at-risk” of type II respiratory failure NEWS-2 UK had marginally inferior discrimination.

The NEWS-1 UK is based on an EWS designed to maximise both sensitivity (the ability to detect patients at risk of dying) and specificity (the minimisation of false alarms) for unselected patients admitted to acute settings. Our systematic review found that modified EWSs were similar in their ability to predict mortality suggesting there is no clear advantage in introducing an alternative scoring system for this group of patients. However, it must be acknowledged that the certainty of the evidence was judged to be very low as the included studies were observational in design and had an increased risk of bias as well as variation in the populations and settings studied and in the mortality follow up times.

Implications for practice

Adopting different scoring systems within the one hospital for specific patients groups can lead to confusion among staff [10–12]. Given the move to electronic systems in a large number of hospitals however, confusion is less of a problem in these settings, but it must be acknowledged that many countries and hospital systems are still using paper-based records. Findings from qualitative research have shown that the inability to tailor alarm settings in EWSs and accommodate systems for patients whose vital sign measurements normally fall outside predetermined thresholds is a major barrier to effective use of the system amongst healthcare professionals [13, 14]. The findings from this literature review based on very low certainty evidence, supported by an expert group of experienced clinicians, resulted in a new approach to managing this cohort of patients while maintaining individualised care through the use of targeted SpO₂ ranges. The expert group concluded that there was insufficient evidence to modify the INEWS to assist in the identification of deterioration in this cohort of patients. The group determined that it was the response to escalation which needed to change rather than the parameters or score. An inappropriate response to a “callout” to a patient with hypoxaemia and hypercapnia in response to the NEWS-1 UK system could well lead to the death of the patient (and could lead authors of retrospective studies to predict that the NEWS-1 UK was excellent for predicting death in patients with exacerbated COPD). As a result, the expert group developed a Modified Escalation and Response Protocol for use by a senior clinical decision maker (Consultant or Registrar), using expertise and clinical judgement in situations where a patient’s ‘lived’ baseline observations fall outside of the normal INEWS parameter ranges. Thus the original escalation thresholds in INEWS (as per ViEWS) were retained and adjustments of either parameter or score were not permitted – a practice which had come about as a ‘workaround’ solution to the problem of over-triggering and alarm fatigue. The authors plan to undertake future work to evaluate the performance of this Modified Escalation and Response Protocol in the Irish setting.

One of the limitations of our review was the comparison of modified EWSs with the NEWS-1 UK at a time when NEWS-2 has now been implemented formally in many settings [9]. However, we would argue that the review findings are still valid as the review includes a comparison of all modified scores including the NEWS-2. A further limitation relates to our search date in 2018. The searches were conducted to inform a National Clinical Guideline and will be updated when that guideline is next updated. Since the search, an additional related paper by Echevarria et al.,[15] has been published which included patients with a validated diagnosis of COPD and found that NEWS-2 had better specificity than NEWS (NEWS-1 UK) for predicting adverse outcomes in patients with acute COPD presentations (even in the absence of confirmed hypercapnia) and NEWS-2 would generate a lower number of alerts than NEWS-1-UK with no deaths in 24 h amongst patients where alerts were not generated. A further limitation is the major weakness of all retrospective studies including those within this review, where the clinical response to the elevated EWS is ignored and only the EWS score and the final outcome is documented.

Going forward, there is a clear need for more robust prospective research on the effectiveness of modified EWSs which address clinical patient outcomes for example mortality and critical care admissions in the chronic respiratory patient cohort. Further contextualisation could be provided through qualitative research methods that inform how these scoring systems perform and are used in clinical settings and how they are balanced, as they must be, with clinical expertise and clinical judgement. In addition, we stress the need to examine the impact of the use of digital EWSs compared to paper-based EWSs. The phenomenon of ‘alarm fatigue’ and its impact on the perceived usefulness of EWSs also requires further study.

Disclosure of interests

The authors have none to declare

Additional information

Funding

This article was funded by the Irish Health Research Board Collaboration in Ireland for Clinical Effectiveness Reviews (Grant no: HRB-CICER-2016-1871).