[ad_1]
Discussion
The majority of Trusts responding reported having an EPR and over 20 different providers were identified as operational in NHS Trusts in England. Three-quarters of digital Trusts responded that they had a DSA and most these use NEWS2 as part of their sepsis alerting system. This is the approach included in the NHS National Standard Contract.16
There is evidence that the provider of the EPR in use in the hospital is associated with the underlying sepsis algorithm in use in the Trust. For example, SIRS-based alerts are only found in Trusts where the provider is Cerner and qSOFA is part of the System C sepsis alert system.
Given NEWS2 is the nationally recommended system for identifying patients who need to be screened for sepsis it is not surprising that NEWS2 is the most commonly used system, usually with a threshold of 5 as the trigger for review and consideration of sepsis. In addition, many Trusts include a score of 3 in any single parameter, despite national guidance moving away from this approach as a trigger for significant escalation as it is a poor predictor of risk.17 This may be a legacy of the overlap between indicators in Red Flag Sepsis and NEWS2.
There was no evidence that Trusts’ digitisation of patient health records was associated with the introduction of more complex algorithms, either data-driven, machine learning-based algorithms or algorithms which were able to include pre-existing conditions or patient information.
Our review of the EPR systems in use in English Trusts are in line with those found by Warren et al.18 In their study of NHS Trusts (2017–2018), 23% of Trusts reported having no electronic system, suggesting an increase in adoption of electronic systems since 2018. Cerner was the most commonly reported provider (18%), then DXC (13%) and System C (11%). DXC were the providers of the Lorenzo EPR in 2017, but were bought out by Dedalus, an Italian-based provider, in April 2021,19 which was still a common provider in our survey (10%). We found a higher proportion with a mixed system than Warren et al, which may reflect a less precise definition of EPR in our study or changes over time.
In 2007, the NIHR reported that ‘T&T systems were in widespread use in NHS acute hospitals’; it is therefore no surprise that we have found that the majority of digital trusts are using digital T&T systems as key components of their DSAs.20 In addition, Trusts are expected to use NEWS2 as a screening system for deteriorating patients. Advantages of T&T systems include the ability to monitor large numbers of patients without ‘a large increase in workload’, and digital enhancement of these systems has clear advantages, for example, one study showed errors in pen-and-paper T&T systems, with errors in 29% calculated scores reviewed (n=84), half of these led to incorrect clinical action.12 Although NEWS2 is a relatively simple system, there is an obvious advantage in clinical data being aggregated automatically and removing the need for busy clinical staff determining and totalling the ‘points’ value of each observation.
In addition to NEWS2, qSOFA and SIRS were relatively commonly used, with adoption associated with the EPR provider. qSOFA was also used in addition to the expected NEWS2. It is possible to trigger a qSOFA score of 2 or more without aggregating to a score of 5 or more in NEWS2, but there is no evidence that using the two combined leads to improved specificity.21
SIRS-based systems were only in use in Trusts where Cerner was the EPR provider. SIRS is not now currently considered a useful way of defining sepsis, but may be useful in predicting poor outcomes for patients.9 22 SIRS-based algorithms do make use of more detailed information contained within patient records, for example, recent lactate and bilirubin levels which reflect organ dysfunction. In addition, the Cerner-based algorithm can be set different thresholds for patients with diabetes or undergoing dialysis, the only system which automatically considers wider information about the patient. However, most studies suggest that SIRS has limited utility in accurately identifying sepsis21 or predicting mortality in patients.23
NEWS2 and qSOFA were designed to be easily performed at the bedside and qSOFA was developed using a parsimonious model to achieve a ‘simple scoring system with the fewest number of variables associated with the greatest predictive ability’.24 Although this approach makes sense in low-resource settings without EPRs, it does not take advantage of the available granular patient data. This includes information from the current visit, previous contacts with the hospital and potentially information from recent primary care appointments.
In this paper, we have not examined the potential benefits or harms of DSAs; a systematic review25 did not find a reduction in mortality, in contrast to Honeyford et al.9 Studies have shown improvements in achievement of process measures. In different parts of the hospital, alerts’ potential to improve patient outcomes varies; in modern EDs, there is often continuous electronic monitoring. In highly resourced EDs, unwell patients are usually reviewed early by a senior clinician, hence there may be limited value of an alert system. Where staff are under increased pressure, alerts may be more important.25 This contrasting evidence emphasises the need for robust evidence to determine the most appropriate DSA.
A minority of Trusts were unwilling or unable to give detailed information about their EPR or the underlying algorithm used for their sepsis alert. This is an important aspect of the introduction of digital alerts in hospitals in England/UK. Currently, there is no clear approval process for digital alerts, and, hence, no necessity for hospitals to use ‘approved’ digital alerts. Two high-profile alerts have recently been identified at best as having no utility, at worst, causing patient harm.14 26
Initially, it was hypothesised that Trusts who responded that they had an EPR would be paperless or heavily paper reduced. However, responses to the question indicate some Trusts are combining electronic and paper systems; we were, therefore, unable to determine how many Trusts are paper reduced/less. We had similar challenges in determining the level of ‘digital’ in the sepsis alert. While some Trusts answered ‘yes’ to DSAs, examination of the details provided suggested that the alert relied on paper. The UK Sepsis Trusts describes sepsis screening as a two-part process, recommending that patients are ‘screened for sepsis’ if they have a NEWS2 score of 5 or more. It was difficult to determine whether the DSA described by respondents was the ‘prescreen’ to identify which patients needed screening for sepsis. Trusts which did not explicitly state that they used NEWS2 are highly likely to be using NEWS2 as part of their sepsis screening system, however, this may not be digital or not be considered part of the DSA. The combined paper and digital model requiring significant staff input to determine the requirement for review reduces some of the advantages of automation.
We opted to use an FOI request to increase response rate. People completing FOIs in NHS Trusts will not necessarily have the knowledge to answer the questions and err on the side of caution. Although some Trusts responded to the FOI with ‘not known’ or equivalent we are sure that there are staff in the Trust who know the algorithm being used. Finally, a minority of Trusts were unwilling or unable to give detailed information about their EPR or the underlying algorithm used for their sepsis alert. This is an important aspect of the introduction of digital alerts in hospitals in England. Currently, there is no necessity for hospitals to use approved digital alerts and no clear approval process. Two high-profile alerts have recently been identified as having no utility and at worst, causing patient harm.14 26
Wong et al14 have highlighted that in the USA ‘the ease of integration within the EPR and loose federal regulations’ means that hospitals adopt algorithms with ease, without a detailed knowledge of real-world performance. This is also the case in England, however, the Medical Health Regulatory Authority are now recommending that software as a medical device should undergo proper scrutiny, ‘commensurate with risk’. There is a need for a strong methodological library for evaluating digital tools, including determining risk. This is the focus of the UK NIHR DiAlS study that is evaluating electronic sepsis screening tools which are currently in use in England.
[ad_2]
Source link
