Introduction
As we have seen earlier, there have been innovations within ambulance services that have changed the way they respond to people who call 999, including advice over the telephone and treatment at home as well as hospital attendance. These changes have come about both to provide more appropriate clinical care for patients that is responsive to their needs, and to help improve efficiency and make best use of ambulance and hospital resources.
The response received by the patient will depend on the assessment by the ambulance service. However, similar conditions may potentially be assessed and treated differently and there is potential for inappropriate decisions to be made, which could waste ambulance service resources, lead to deaths or lead to future hospital admissions. The performance and quality indicators developed in workstream 3 have encompassed these elements by examining different processes and outcomes. Another important element is the impact that decisions have on costs to both the ambulance service and the wider health system. The creation of the linked data in workstream 2 and the development of indicators to identify potentially incorrect decisions provided a unique opportunity to assess the impact of different types of response on NHS costs. For this piece of work we have examined the economic effects of different types of response in two ways. First, by comparing the average costs of treatment associated with each response type and, second, by estimating the average costs associated with incorrect decisions to either leave patients who need to go to hospital or transport patients when this is unnecessary. This allows us to show current costs and the potential cost reductions associated with a change in the type of response.
Methods
We used the linked data to create a new data set with additional cost data added. For each incident we extracted the relevant data fields: age, sex, call type, urgency, response, admission and discharge details, treatment type, ICD-10 primary diagnosis code, primary operation code, episode start and end date. For inpatient hospital stays we allocated a Healthcare Resource Group code (based on diagnostic code) to assign inpatient costs and used these to calculate the cost per inpatient spell. For ED attendances we assigned a cost per attendance and for ambulance costs a cost per response type (‘hear and treat’, ‘see and treat’ and ‘see and convey’). All costs were assigned using National Reference Costs.59For individual patients, not all hospital attendances may be related to the ambulance incident, so we made some assumptions (e.g. excluding elective inpatient episodes and using only the first ED attendance within 3 days of the call) to account for this.
DOI: 10.3310/pgfar07030 PROGRAMME GRANTS FOR APPLIED RESEARCH 2019 VOL. 7 NO. 3
© Queen’s Printer and Controller of HMSO 2019. This work was produced by Turner et al. under the terms of a commissioning contract issued by the Secretary of State for Health and Social Care. This issue may be freely reproduced for the purposes of private research and study and extracts (or indeed, the full report) may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising. Applications for commercial reproduction should be addressed to: NIHR Journals Library, National Institute for Health Research, Evaluation, Trials and Studies Coordinating Centre, Alpha House, University of Southampton Science Park, Southampton SO16 7NS, UK.
We calculated mean costs per call for‘hear and treat’, ‘see and treat’ and ‘see and convey’ responses and compared categories. However, we know that there will be differences in the characteristics of calls within each type of response. To control for this we also created matched groups of calls for patients with the same characteristics. We applied exact matching on call code and condition, and Mahalanobis distance matching on age and sex.60Exact matching was used for the categorical variables to ensure that a‘hear and treat’ caller with a given call code and condition was matched to a ‘see and treat’ or ‘convey’ caller with the same call code and condition. Mahalanobis, matching is used to minimise the distance in terms of age and sex for the matched observations, for instance a caller is matched to another caller, within the same call code and condition category, that is closest in terms of age and sex. We created matched samples to compare costs for:
l ‘hear and treat’ with ‘see and treat’ l ‘hear and treat’ with ‘see and convey’
l ‘see and treat’ with ‘see and convey’.
Incorrect decisions were identified for calls where‘hear and treat’ or ‘see and treat’ patients attend ED or are admitted to hospital within 3 days, or patients are taken to hospital and discharged from ED using the same criteria set out in indicator 5 (workstream 3). Each incorrect decision was matched with an equivalent case that had the correct decision. We then compared the costs of an‘incorrect’ decision with a ‘correct decision’ for each response type to enable us to estimate the costs associated with less than ideal decision-making.
Results
A total of 182,566 cases were included in the analysis, and 10,151 (5.6%) calls received a‘hear and treat’ response, 51,223 (28.0%) calls received a‘see and treat’ response and 121,192 (66.4%) calls resulted in the patient being taken to hospital. Using the results of the matched case analysis, as these are more precise, we found that the total mean cost of a‘hear and treat’ call was £125, ‘see and treat’ was £415 and ‘see and convey’ was £1745. The main reasons for these differences are the differences in inpatient costs. For the calls that initially received a‘hear and treat’ response, the mean length of stay was 0.24 days, compared with 0.68 days in the‘see and treat’ group, and 4.46 days in the ‘see and convey’ group. The majority of patients have 0 days of inpatient stay with only 2.35% of‘hear and treat’ cases, 5.44% of the ‘see and treat’ cases and 46.73% of ‘see and convey’ cases having an inpatient stay.
For the analysis comparing costs of correct and incorrect decisions we found that the mean total cost of a correct conveyance decision is £3728.99 and the mean total cost of an incorrect non-conveyance decision is £4042.38; therefore, the additional cost to the emergency services of making an inappropriate decision of this type is £313.39. This difference is the result of higher call and inpatients costs (relating to, approximately, 1 day in hospital). The mean total cost of an appropriate non-conveyance decision is £109 and the mean total cost of an incorrect conveyance decision is £346, resulting from higher call and ED costs (relating to, approximately, the cost of one ambulance call out and one A&E attendance).
Summary
The cost analysis has allowed us, for the first time, to estimate the actual costs of different types of ambulance response using real data on all NHS contacts for each type of response. The matched analysis suggests that, for many conditions, alternative responses are possible with lower costs, but imperfect matching and the lack of patient outcome information means that these results should be treated with caution. However, this type of information is helpful to support planning and commissioning of services and where efficiencies might be achieved. Our analyses of two different types of incorrect decision are more robust and show that these decisions are associated with higher costs. The development of processes to identify the rates of incorrect decisions through the workstream 3 indicator and ability to now assign a cost to these decisions provides a way of estimating what costs could be saved if efforts are put in place to reduce the number of incorrect decisions. A more detailed description of the methods, results and limitations of the cost analysis can be accessed as