Comparison of the Aotearoa New Zealand Early Warning Score and National Early Warning Score to predict adverse inpatient events in a vital sign dataset

We thank Dr Murali and Professor Inada-Kim for their editorial [1] that accompanied our article [2] and would like to respond to some of the issues they have raised. They asked why we chose to compare the New Zealand Early Warning Score (NZEWS) with the original UK National Early Warning Score (NEWS) and not its successor, NEWS2. There are several reasons for this. NZEWS was designed before the release of NEWS2 in 2017 so we took the aggregate scoring components from its predecessor, NEWS. As such, the almost 14 million vital signs available to us for analysis did not contain data that included the two significant NEWS2 modifications. First, we were unable to determine whether permissive hypoxaemia, introduced for patients with type-2 respiratory failure, was in use for a given set of vital signs. Second, the inclusion of new confusion or delirium is not available in NZEWS, which uses the ‘AVPU’ system to score altered level of consciousness. In the absence of these key data points from our dataset, the scoring systems (NZEWS vs. NEWS2) are not directly comparable. Both of these additions to NEWS were considered for inclusion in NZEWS. Rather than making these changes however, the NZEWS Expert Advisory Group chose to await evidence of improved outcomes with NEWS2. Several studies have since shown that the permissive hypoxaemia modification introduced in NEWS2 does not improve outcomes in patients with type-2 respiratory failure [3] and raised questions around the safety of this additional modification [4]. An additional study suggested that the inclusion of delirium scoring may cause a substantial increase in medium- and high-level alerts in emergency medical patients, with the potential to ‘overwhelm the capacity of clinical staff’ [5]. Authors of these studies recommended further testing of NEWS2 before widespread implementation. At the time of writing, and in the absence of compelling evidence otherwise, there is no intention at the moment to update NZEWS to align with the changes made in NEWS2. We would also like to correct a statement in the editorial stating ‘…much like NEWS2, NZEWS has not been formally evaluated in primary care, pre-hospital (ambulance) or a community setting’. A study of the NZEWS in over 40,000 pre-hospital adult patients showed it to have significant utility as a decision support tool for paramedics when discharging low-acuity patients at the scene. This study is reference 29 in our original paper. Finally, we would like to endorse the recommendations in the editorial that researchers into early warning scoring systems should both standardise outcome definitions and strive to obtain physiological data from before, during and after hospital admissions. Pooling outcome-linked anonymised datasets in different healthcare settings would improve the calibration of systems in current use and assist with the design of future physiologically-based deterioration prediction programmes.