Discussion

This review offers a broad overview of the 10 high-impact interventions identified by NHS England and their potential impact on two metrics. The evidence suggests that some initiatives, such as SDEC, acute frailty services, care transfer hubs with multidisciplinary teams and telestroke services may effectively reduce ED wait times or ED LOS. However, the impact of in-patient flow interventions on this outcome was generally inconclusive, with some studies reporting reductions and others showing no difference or even an increase. Regarding ambulance response times, there was limited evidence, with only a few interventions, such as integrating GPs within emergency medicine settings and prehospital telemedicine interventions, showing potential benefits. For ambulance offload delays, one review found that various interventions generally reduced offload delay times, but the evidence was limited. This review highlights the need for research to establish the effectiveness of these interventions in different settings and contexts.

The principal limitations affecting the evidence base concern both the categorisation of interventions and the headline outcomes used by NHS England. In terms of the interventions, initiatives such as SDEC, ARI hubs and the HAH component of ‘virtual wards’, had sufficiently robust definitions and ‘name recognition’ to facilitate targeted evaluation within the research literature, although there was still variation in the services delivered under these three initiatives. The remaining initiatives all exhibited a much higher degree of uncertainty in how key elements of the interventions might be defined. This, in turn, made it challenging to identify relevant interventions within the published literature, even allowing for the use of an extensive list of potential intervention synonyms (see table 2). Some initiatives included multiple, often indistinct interventions, that appear to hold little in common beyond their overall intent (eg, UCR or in-patient flow). A review based exclusively on the 10 ‘high-impact initiatives’ as itemised by NHS England would have returned very few studies. For example, NICE guidance on the effectiveness of ARI hubs failed to identify any specific published evidence at all.47 However, even an initiative like SDEC might encompass a variety of service differences, and this heterogeneity of interventions and intervention types, both within and across initiatives, also increases the uncertainty regarding some findings and limits their generalisability to different contexts.

The two NHS England ‘headline metrics’ were ED wait times and ambulance response times. In terms of the research identified for this review, ED wait times and potentially related metrics (such as ED LOS) were reported as primary or secondary outcomes for nine of the 10 initiatives (only community beds had no evidence). ED LOS, while different from ‘wait time’ or ‘time to be seen’, has been included here to capture a more extensive evidence base and because it is a known potential predictor of important outcomes, such as mortality.2 Ambulance response times and other time-related metrics for ambulances were not evaluated at all for 6 of the 10 initiatives. The remaining four initiatives measured outcomes such as mission or duty time for UCR, offload delay for some in-patient flow interventions, duty cycles and conveyance rates for single point of access and metrics such as ‘onset to door’ time (within ‘virtual wards’ for prehospital telemedicine). Given how the majority of these interventions work or are envisaged to work, the absence of evidence on these two outcomes is not surprising. ‘Ambulance response times’ might only apply to a small number of high-impact initiatives while ‘ED wait times’ fail to recognise the whole-system impact of alternative high-impact interventions. As a consequence of this review’s focus on specific outcome metrics, few of the thousands of reviews and primary research studies screened for the period of this review (2018–2024) were adjudged relevant to our particular review question. The evaluated interventions commonly focused on outcomes such as ED visits, admissions, ‘overcrowding’, ‘ambulance offload delay’ and ‘inpatient flow’. While ambulance response times and ED wait times constitute important ‘headline’ metrics for emergency care, they are not typically either the first or most informative choice when evaluating the impact of many urgent care initiatives. Therefore, it should be noted that this review did not aim to evaluate all of the potential effects or wider benefits of the initiatives. A review of other metrics, such as admission/utilisation and discharge rates, might produce quite different findings. This highlights the limitations of seeking to apply a limited number of headline metrics across all emergency care contexts.

While the diversity of ‘high-impact initiatives’ from the specific to the broad, from the clearly defined to the diffuse, and from the hospital based to the community focused, makes an even and consistent approach to evidence gathering and assessment challenging, this review made great efforts to identify available evidence to a level appropriate for each initiative and any related interventions. As such it offers a valuable compendium. Notwithstanding such limitations, clear patterns to findings can still be observed. For example, in terms of reduction in ED wait times, there is consistent primary study level evidence for a clearly defined initiative such as SDEC, or generally consistent review-level evidence for diverse in-patient flow initiatives, despite both approaches exhibiting high variability in terms of what service is offered, together with context-sensitivity by particular centre (online supplemental file 4). However, it is also worth noting that a SDEC unit might also simply involve the provision of more staff and resource elsewhere within the ED setting, such as an additional senior decision-maker,20 who might have had a similar impact if provided within the ED itself. The evidence base supporting the 10 high-impact initiatives might, therefore, be very limited for the outcomes of interest, but this review does offer a broad overview of the potential evidence for 10 high-impact interventions identified by NHS England.

If future research is to effectively inform policy in this field, then it needs to be high quality (in design, conduct and reporting), be conducted over adequate periods of time and target clearly defined interventions that satisfy the identifiable characteristics (and activating mechanisms) of known real-world interventions, such as the 10 initiatives. It needs to measure outcomes specified as important by policy-makers, professionals, patients and the public, as well as those that are directly relevant to each initiative and how it is intended to work (ie, which outcomes it targets directly). Alternatively, research is needed that demonstrates a clear link between outcomes that are frequently measured for interventions (such as ED visits) and those outcomes of particular interest to policy-makers, for example, shorter ED wait times. This confirms what the review team were told by both clinical experts and patient representatives: wider, more contextually sensitive communication around initiatives and their evaluation is required.

This scoping review has limitations. The work was originally designed as a systematic review of reviews, but a scoping review was performed. This was because the evidence proved so limited in some instances, and so extensive in others, that broader processes had to be applied, including searching for additional evidence (both primary research studies and the extensive complementary searches for studies based on outcomes), and the use of single reviewers for the review of some initiatives. Second, within this context, a decision was also taken to limit the number of bibliographic databases searched, as long as it included MEDLINE/PubMed because it is known to have excellent coverage of UEC.48 Finally, as a scoping review of many different study designs, quality assessment of the evidence base was not performed; this also needs to be considered when reflecting on the findings.