Research Report

The Scottish Ambulance Service New Clinical Response Model Evaluation Report

Alternative title NCRM report for SAS



Stoddart K, Cowie J, Robertson T, Bugge C, Donaldson J & Andreis F (2019) The Scottish Ambulance Service New Clinical Response Model Evaluation Report [NCRM report for SAS]. Scottish Ambulance Service. Stirling: University of Stirling.

Executive Summary The Scottish Ambulance Service (SAS) responds to around 1.8 million calls per year, including responses to 700,000 emergency and unscheduled incidents. Of these responses, over 500,000 are received through the 999 call service. SAS transfers around 90,000 patients between hospitals each year and responds to over 150,000 urgent requests for admission, transfer and discharge from GPs and hospitals (SAS, 2015). In 2017 SAS began to implement a new clinical response model (NCRM). The aims of the NCRM are to: - Save more lives by more accurately identifying patients with immediately life-threatening conditions, such as cardiac arrest; - Safely and more effectively send a matched resource first time to all patients based on their clinical need. The University of Stirling, commissioned to carry out an independent evaluation of the NCRM using data provided by SAS and NHS Information Services Division (ISD), considered the following questions: 1. Are patients with Immediately Life Threatening (ILT) conditions more quickly and accurately identified? 2. Are more lives saved as a consequence of the best available resources being dispatched to the patient? 3. Are improved clinical outcomes achieved if the matched resources are sent first time for patients with non-ILT conditions? Methods A quantitative analysis was conducted comparing SAS data on response to 999 calls from a pre-NCRM implementation time-period (January 2016) and a post-implementation time-period (January 2017 and January 2018). NHS ISD linked additional data from the Unscheduled Care Data Mart (UCDM) to the SAS data. UCDM contains emergency department data (ED) and data from the National Records of Scotland (NRS) for mortality data. Data were examined for the purple code (the highest risk category of call to the 999 service) and within the purple category, those patients in cardiac arrest. The same analyses were conducted for the remaining colour codes and a selection of clinical categories within these colour codes: breathing difficulties (red), stroke (amber) and falls (yellow). Key Findings Interpreting this data It should be noted that data is taken from only three (and in some cases two) time points and only from the month of January. While this does allow some relevant comparisons between the years, the findings cannot be generalised to the whole year and the whole time-period in question (January 2016 – January 2018). In addition, call volume was approximately 9% higher in 2018 compared to 2017 and 2016 (which were similar) with over 4,000 more calls in January 2018. Further analysis of the data using data from each month, as well as individual-level data (rather than it being aggregated), would allow much more robust and relevant evidence of change and the impact on the service and patients. 1. Are patients with Immediately Life Threatening (ILT) conditions more quickly and accurately identified? Patients with ILT conditions (purple calls) would appear to be more accurately identified post-NCRM with a noticeable increase in patients coded with ILT conditions by 2018. The time to respond to ILT conditions was slightly longer (but not statistically significant). Speed Resource allocation was used as an indicator of speed of identification. We found that resource allocation (and in turn response times) did not differ significantly between January 2016 (pre-NCRM) and January 2017 (post-NCRM introduction) for ILT (purple) calls. However, there was a longer time to allocate resources (i.e. identify) purple calls in 2018 compared to 2016 and this was statistically significant. For all other colour codes, 2017 and 2018 resource allocation were also significantly slower than 2016 (except amber 2017 calls) as expected with a priority-based system. Call handlers were provided with further training and development in the process of triage over the course of 2016 onwards, with the aim of more accurately allocating patients into the most appropriate category, and therefore it was to be expected that time to allocate resources and identification into the correct category would take longer. Accuracy Comparing 2016 (pre-NCRM) and 2017 (post-NCRM introduction) outcomes data, we found that sensitivity (correctly identifying a purple, ILT condition) was higher in 2017 compared to 2016, but specificity (correctly identifying a non-ILT condition) was lower in 2017. Overall accuracy (the likelihood of being correctly identified as either ILT or non-ILT) was not different between the two-time points. Similar results were also seen for the cardiac arrest cases within the purple calls. 2. Are more lives saved as a consequence of the best available resources being dispatched to the patient? Survival for purple-coded patients is markedly lower with respect to all other causes (as one would expect) and reflects that purple-coded calls/conditions are a unique category (in terms of risk of death) and represent the majority of incidents where patients face an immediate threat to life (ILT). The risk of death across the other colour codes is small in comparison and therefore differences of survival seem to exist only for the purple-coded patients. The cardiac arrest rate within the purple coded is around 53%. Survival analysis for all patients within the purple code and specifically for those affected by cardiac arrest are considered next. There seems to be a considerable (~20%) increase in survival for all purple-coded patients comparing January 2016 to January 2017, which is constant over time from time 0 (confirmed dead when the ambulance arrives at the scene) to 30 days post-call. When comparing January 2016 to January 2018 for the same group, survival also increased (~10%). The number of lives saved, 30 days post-call, in patients with ILT conditions in January 2016 (pre-NCRM) was 32 (14.2% of purple calls), and in post-NCRM in January 2017 was 134 (28.6% of purple calls) and in January 2018 was 182 (26.6% of purple calls). Although the numbers of patients with ILT conditions has increased, the data from the specificity and sensitivity analysis (Table 14) shows that there is no difference in false positive rates between the years. This suggests that the acuity of these patients remains very high and that the increase in volume represents patients correctly identified with the highest requirement for immediate response. Therefore, the increase in survival probability with those with ILT conditions is not likely to be caused by artificial inflation caused by conservative allocation of patients with ILT conditions to the purple code but rather by appropriate allocation and intervention(s) to those patients at risk from death due to ILT conditions. In terms of the 2018 survival probability being lower than in 2017, it is possible that the higher call-load in 2018 has limited the impact previously seen in 2017. Continued monitoring of these data is needed to identify how mortality has been impacted by the NCRM over the longer-term. 3. Are improved clinical outcomes achieved if the matched resources are sent first time for patients with non-ILT conditions? Overall survival for all non-ILT codes (Red, Amber, Yellow and Green) was similar, as noted above (where purple calls carry much higher risk of death). For these codes there was also no clear difference in survival in 2017 versus 2016 or 2018 versus 2016. Breathing difficulty (a sub-set of the red calls) seems to have worsened between 2016 and 2017, with 451 patients having a decrease in survival from 3% to 6%, with the gap widening as time passes. However, by 2018, survival was at 2016 levels despite the number of incidents (n=2044) back to the levels seen in 2016 (n=2018). No differences between years seem to be present for stroke or falls. Data on further clinical outcomes were not available within this dataset to analyse in any further detail. Conclusions By January 2018 the number of incidents (n=52,871) had increased by 9% when compared to January 2016 (n=48,544), amounting to over 4000 more incidents in 2018 than seen in 2016 or 2017. During this time of high demand in 2017 and particularly 2018, the NCRM does accurately identify patients who have the greatest need for services from SAS. The NCRM’s identification and triage of patients into triage categories, although taking time for the call handler and dispatching system, can get the ambulance and its crew to patients with the greatest need and this has improved the survival of those with immediate life-threatening conditions. Those with lower acuity needs are responded to but in a longer time period as expected when using a priority-based system (but with no apparent impact on survival). These conclusions are reached in the context of analysing aggregated data over three fairly short time-periods and further research over a longer time frame, with longitudinal data on individual cases, would further improve the evidence base for the NCRM.

FundersScottish Ambulance Service
Publication date20/02/2019
Publication date online20/02/2019
Publisher URL…rcent-more-lives
Place of publicationStirling

People (3)


Professor Jayne Donaldson

Professor Jayne Donaldson

Dean of Faculty Health Sciences & Sport, FHSS Management and Support

Dr Tony Robertson

Dr Tony Robertson

Lecturer in Geographies of Public Health, Biological and Environmental Sciences

Dr Kathleen Stoddart

Dr Kathleen Stoddart

Senior Lecturer, Health Sciences Stirling

Projects (1)

SAS New Clinical Response Model