Optimisation of the deployment of automated external defibrillators in public places in England
Background Ambulance services treat over 32,000 patients sustaining an out-of-hospital cardiac arrest annually, receiving over 90,000 calls. The definitive treatment for out-of-hospital cardiac arrest is defibrillation. Prompt treatment with an automated external defibrillator can improve survival s...
Saved in:
| Main Authors: | , , , , , , , , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
NIHR Journals Library
2025-02-01
|
| Series: | Health and Social Care Delivery Research |
| Subjects: | |
| Online Access: | https://doi.org/10.3310/HTBT7685 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1850284763536949248 |
|---|---|
| author | Terry P Brown Lazaros Andronis Asmaa El-Banna Benjamin KH Leung Theodoros Arvanitis Charles Deakin Aloysius N Siriwardena John Long Gareth Clegg Steven Brooks Timothy CY Chan Steve Irving Louise Walker Craig Mortimer Sandra Igbodo Gavin D Perkins |
| author_facet | Terry P Brown Lazaros Andronis Asmaa El-Banna Benjamin KH Leung Theodoros Arvanitis Charles Deakin Aloysius N Siriwardena John Long Gareth Clegg Steven Brooks Timothy CY Chan Steve Irving Louise Walker Craig Mortimer Sandra Igbodo Gavin D Perkins |
| author_sort | Terry P Brown |
| collection | DOAJ |
| description | Background Ambulance services treat over 32,000 patients sustaining an out-of-hospital cardiac arrest annually, receiving over 90,000 calls. The definitive treatment for out-of-hospital cardiac arrest is defibrillation. Prompt treatment with an automated external defibrillator can improve survival significantly. However, their location in the community limits opportunity for their use. There is a requirement to identify the optimal location for an automated external defibrillator to improve out-of-hospital cardiac arrest coverage, to improve the chances of survival. Methods This was a secondary analysis of data collected by the Out-of-Hospital Cardiac Arrest Outcomes registry on historical out-of-hospital cardiac arrests, data held on the location of automated external defibrillators registered with ambulance services, and locations of points of interest. Walking distance was calculated between out-of-hospital cardiac arrests, registered automated external defibrillators and points of interest designated as potential sites for an automated external defibrillator. An out-of-hospital cardiac arrest was deemed to be covered if it occurred within 500 m of a registered automated external defibrillator or points of interest. For the optimisation analysis, mathematical models focused on the maximal covering location problem were adapted. A de novo decision-analytic model was developed for the cost-effectiveness analysis and used as a vehicle for assessing the costs and benefits (in terms of quality-adjusted life-years) of deployment strategies. A meeting of stakeholders was held to discuss and review the results of the study. Results Historical out-of-hospital cardiac arrests occurred in more deprived areas and automated external defibrillators were placed in more affluent areas. The median out-of-hospital cardiac arrest – automated external defibrillator distance was 638 m and 38.9% of out-of-hospital cardiac arrests occurred within 500 m of an automated external defibrillator. If an automated external defibrillator was placed in all points of interests, the proportion of out-of-hospital cardiac arrests covered varied greatly. The greatest coverage was achieved with cash machines. Coverage loss, assuming an automated external defibrillator was not available outside working hours, varied between points of interest and was greatest for schools. Dividing the country up into 1 km2 grids and placing an automated external defibrillator in the centre increased coverage significantly to 78.8%. The optimisation model showed that if automated external defibrillators were placed in each points-of-interest location out-of-hospital cardiac arrest coverage levels would improve above the current situation significantly, but it would not reach that of optimisation-based placement (based on grids). The coverage efficiency provided by the optimised grid points was unmatched by any points of interest in any region. An economic evaluation determined that all alternative placements were associated with higher quality-adjusted life-years and costs compared to current placement, resulting in incremental cost-effectiveness ratios over £30,000 per additional quality-adjusted life-year. The most appealing strategy was automated external defibrillator placement in halls and community centres, resulting in an additional 0.007 quality-adjusted life-year (non-parametric 95% confidence interval 0.004 to 0.011), an additional expected cost of £223 (non-parametric 95% confidence interval £148 to £330) and an incremental cost-effectiveness ratio of £32,418 per quality-adjusted life-year. The stakeholder meeting agreed that the current distribution of registered publicly accessible automated external defibrillators was suboptimal, and that there was a disparity in their location in respect of deprivation and other health inequalities. Conclusions We have developed a data-driven framework to support decisions about public-access automated external defibrillator locations, using optimisation and statistical models. Optimising automated external defibrillator locations can result in substantial improvement in coverage. Comparison between placement based on points of interest and current placement showed that the former improves coverage but is associated with higher costs and incremental cost-effectiveness ratio values over £30,000 per additional quality-adjusted life-year. Study registration This study is registered as researchregistry5121. Funding This award was funded by the National Institute for Health and Care Research (NIHR) Health and Social Care Delivery Research programme (NIHR award ref: NIHR127368) and is published in full in Health and Social Care Delivery Research; Vol. 13, No. 5. See the NIHR Funding and Awards website for further award information.
Plain language summary Ambulance services of the NHS treat over 32,000 people whose heart suddenly stops pumping effectively, a condition known as cardiac arrest. Despite ambulance services’ best efforts fewer than 1 in 10 survive. Electric shock treatment, known as defibrillation, is one of the most effective treatments, and if it is given within a few minutes of the heart stopping, over half the people treated survive. It is now possible for public to use an automatic machine (defibrillator) to safely give an electric shock to the heart before the emergency services arrive. For the public to make best use of these machines they need to be in the right places. In this study, we attempted to work out the best places to put defibrillators in communities, making them more accessible to use. We showed that defibrillators currently are sited disproportionately in more affluent areas of the country, and not used despite being within an accessible distance from where a cardiac arrest occurs. We assessed that if a defibrillator was installed at various points of interest the number of cardiac arrests that were covered increased significantly. We then used a computer to model the best locations for new defibrillators and calculate the optimal number needed. Placement based on this model showed that, for a smaller number of defibrillators, a similar improvement in coverage could be achieved. A health economic analysis that considered the cost of purchasing and installing defibrillators showed that installing additional defibrillators in specific points of interest improved coverage, but it was also more costly compared to current defibrillator placement. This research showed that significant improvement in cardiac arrest coverage could be achieved if defibrillators were placed intelligently in public settings. We also created a system that uses data to decide where to place public-access defibrillators in the community.
Scientific summary Background Annually, English ambulance services treat over 30,000 people who have sustained an out-of-hospital cardiac arrest (OHCA), about 25% of whom achieve a return of spontaneous circulation by the time of hospital handover and 8.5% survive to 30 days. The chain of survival shows the essential elements required in an emergency care system to improve outcomes from an OHCA. The first two links – early recognition and early cardiopulmonary resuscitation – can buy time for the OHCA patient but are not definitive treatments in themselves. The key and most effective treatment for an OHCA is defibrillation. Prompt treatment with an automated external defibrillator (AED), within 3–5 minutes of collapse, can lead to survival rates in excess of 50%. Public-access defibrillation (PAD) refers to the use of AEDs by members of the public. PAD programmes allow the community access to this life-saving intervention while waiting for ambulances to arrive. The importance of PAD is growing given the increasing demands on ambulance services that are making reaching OHCAs in a timely manner challenging. However, at present, only a small proportion of patients are treated by PAD (5%). A fundamental, structural barrier, which limits opportunity for the use of AEDs, is their location in the community. There has been no clear strategy in the UK on where AEDs should be placed; the choice of where to install them in public places has been driven mainly by local ad hoc initiatives. This approach is limited and there is a call for an evidence-based strategy, and a requirement to identify the optimal location for an AED to improve OHCA coverage, to improve the chances of survival. Objectives The primary objective of this study was to optimise the placement of public-access AEDs in England, using mathematical modelling techniques, to maximise the likelihood that an individual sustaining an OHCA will have access to PAD, improving their chances of survival. The secondary objective was to assess the cost-effectiveness of optimised public-access AED placement compared to current and alternative-placement strategies. Methods Ethics and regulatory approvals Following Health Research Authority guidelines, the study did not require formal NHS Research Ethics Committee approval. The study was approved by the University of Warwick’s Biomedical and Scientific Research Ethics Committee (BSREC 118/18-19). The project was co-sponsored by the University Hospitals Birmingham NHS Foundation Trust and the University of Warwick. The Out-of-Hospital Cardiac Arrest Outcomes (OHCAO) registry has approval from the Confidentiality Advisory Group to collect and process identifiable patient information where it is not practical to obtain consent (22CAG0072 and 22CAG0087). Ethics approval for the OHCAO registry was gained from the National Research Ethics Committee South Central (13/SC/0361). Design This was a secondary analysis of data that were collected by the OHCAO registry on historical OHCAs, data held by ambulance services on the location of AEDs registered with them, and locations of points of interest (POIs) available from Ordnance Survey. Also, data were obtained on the census neighbourhood characteristics of areas of England. The study was divided into four work packages (WPs): (1) exploration of the characteristics and coverage of current locations of AEDs relative to the location of historical OHCAs; (2) comparison of the OHCA coverage of various AED deployment strategies (POIs and grid-based) and an optimisation model with the current coverage; (3) determination of the cost-effectiveness of the strategies in WP2 compared to current placement (CP); and (4) development of a national consensus of the optimal location for public-access AEDs. Statistical analysis Descriptive statistics were used to analyse for any differences in neighbourhood characteristics where OHCAs occurred and registered AEDs were located. Using geographical information system software, we calculated the walking distance between OHCA locations and locations of registered AEDS and POIs designated as potential sites for AED (e.g. pubs, places of worship, schools, halls and community centres, and cash machines). Assumed coverage was calculated as the proportion of OHCAs within 500 m of a registered AED or POI, assuming the latter were accessible 24 hours a day, 7 days a week. Actual coverage was calculated based on when the OHCA occurred and the working hours of the AED locations. Using these definitions, we then calculated coverage loss as assumed coverage minus actual coverage divided by assumed coverage. Coverage efficiency was calculated as the number of OHCAs covered divided by the number of POIs. For the optimisation part of the study, mathematical models for maximal coverage location problem were adapted. These mathematical models were based on seminal literature in operational research, specifically within the area of facility location optimisation (maximal covering location problem). The model took in information for historical OHCAs, as well as locations of existing AEDs and candidate future AED locations. It then determined the optimal locations for future AED placement so that OHCA coverage was maximised and the number of AEDs required to maximise coverage. For the cost-effectiveness analysis a de novo decision-analytic model was developed and used as a vehicle for assessing the costs and benefits of deployment strategies at POIs compared to current AED placement. The model was developed in line with recommended ‘good practice’ guidelines and, where relevant and possible, in accordance with requirements for economic evaluation aiming to inform decision-making in the UK. A stakeholder meeting was convened towards the end of the study that brought together groups interested in AED deployment. Results of the study were presented and discussed. Then three groups were organised to discuss: (1) what strategies should be used to increase the distribution of AEDs in communities; (2) what guidance should be provided for where to place AEDs in communities; and (3) what further research would be required. Patient and public involvement The project was designed to ensure meaningful patient and public involvement (PPI) was embedded throughout the study. The PPI co-applicant was involved from the conception of the study and made valuable contribution to the development of the proposal, and reviewed and commented on work at various stages of the project. Presentations were made to the National Institute for Health and Care Research (NIHR) Clinical Research Ambassadors Group, based at the University Hospital Birmingham NHS Foundation Trust. There were two PPI representatives on the Steering Committee, and at the stakeholder meeting there were representatives of two charities. Results The study looked at the location of 147,278 historical OHCAs (2014–9) and 32,491 AEDs, and 14 potential POIs. Current coverage Historical OHCAs were observed to occur in more urban areas with a high population density. These areas had a larger proportion of workers in routine jobs or unemployed and people from non-white ethnic groups, and a great degree of deprivation. In contrast, AEDs were placed in more affluent areas that had lower proportions of people from non-white ethnic groups. About 43.4% of the areas (the lower-level super output area) were found to not have a single AED located within their boundary. The average OHCA–AED distance was 1014 m (median 638 m). The number of OHCAs occurring within 100 m of an AED was 7965 (5.4%) and within 500 m was 57,225 (38.9%). Alternative coverage: points of interest The numbers (percentages) of OHCAs that were covered by each POI, ignoring the currently registered AEDs, varied greatly, from 7.4% by state secondary schools to 55.1% by cash machines. Similarly, of those OHCAs not covered by a currently registered AED, the proportion then covered by one of the POIs ranged from 5.5% by state secondary schools to 46.0% by cash machines. The proportion of OHCAs that were covered by a registered AED or a POI also ranged significantly, from 42.2% by state secondary schools to 67.0% by cash machines. Coverage loss for cash machines, care homes and community halls was assumed to be zero because the AEDs at these locations were presumed to be accessible all the time. Loss was greatest for state schools, if we assume that all the OHCAs that occur outside of normal state school opening times there will be no access to an AED. Coverage efficiency also varied significantly and ranged, based on assumed coverage, from about 150% in pubs to just under 450% in post offices, chemists, dentists, general practitioner surgeries and supermarkets. Alternative coverage: 1 km2 grids and census output areas If the country was divided up into 1 km2 grids, coverage of OHCAs would be significantly greater if we were to deploy an AED at the centre of each grid: 116,016 (78.8%). If we were also to consider the location of existing registered AEDs, the coverage would be 133,901 (90.9%). Alternatively, if we were to place an AED at the centroid of every census output area the coverage would be 57,134 (38.8%), and 97,718 (66.3%) if we included the location of existing AEDs. Alternative coverage: optimisation model The optimisation model was observed to show that if AEDs were placed in each POI location, OHCA coverage levels would not reach that of optimisation-based placement. For example, if an AED were placed in every pub in the West Midlands (n = 3573), it would result in additional coverage of 17.8% above that provided by existing registered AEDs. However, the same amount of added OHCA coverage could be achieved with only 250 optimised grid points spaced 1 km apart. The coverage efficiency provided by the optimised grid points was unmatched by any POI in any region. Cost-effectiveness Compared to current AED placement, all of the alternative deployments assessed were associated with a greater expected total cost per OHCA for a small increase in quality-adjusted life-years (QALYs) and life-years gained (LYGs). The most cost-effective option was halls and community centres. Compared to CP, this deployment option resulted in higher costs [£223, 95% confidence interval (CI) generated from probabilistic sensitivity analysis (PSA) distribution: £148 to £330], a higher number of QALYs (0.007, 95% CI generated from PSA distribution: 0.004 to 0.011) and an incremental cost-effectiveness ratio (ICER) of £32,418 per QALY (£18,893 per LYG). This value is above the upper bound of the £20,000–30,000 per QALY range that is often seen as a maximum ICER considered for decision-making in health care. Stakeholder meeting The meeting agreed that the current distribution of registered publicly accessible AEDs was suboptimal and that there was a disparity in their location in respect of deprivation and other health inequalities. Conclusions Automated external defibrillators are potentially life-saving devices for people who sustain an OHCA. AEDs need to be placed intelligently in public settings so that they are likely to be used by bystanders. We have developed a data-driven framework to support public-access AED location decisions, using optimisation and statistical models. We applied the methodology to real data from England. Results have demonstrated that optimising AED locations can result in substantial improvement in coverage compared to the current approach to AED deployment. We developed a de novo decision-analytic model to determine the costs and benefits associated with AED placement strategies in each of the different POIs and compared these against current AED placement. Results of the economic analysis showed that all of the alternative placements considered were associated with ICERs above £30,000 per additional QALY. Study registration This study is registered as researchregistry5121. Funding This award was funded by the National Institute for Health and Care Research (NIHR) Health and Social Care Delivery Research programme (NIHR award ref: NIHR127368) and is published in full in Health and Social Care Delivery Research; Vol. 13, No. 5. See the NIHR Funding and Awards website for further award information. |
| format | Article |
| id | doaj-art-afee50e07e12490a8707e3faff3d0fea |
| institution | OA Journals |
| issn | 2755-0079 |
| language | English |
| publishDate | 2025-02-01 |
| publisher | NIHR Journals Library |
| record_format | Article |
| series | Health and Social Care Delivery Research |
| spelling | doaj-art-afee50e07e12490a8707e3faff3d0fea2025-08-20T01:47:28ZengNIHR Journals LibraryHealth and Social Care Delivery Research2755-00792025-02-01130510.3310/HTBT7685NIHR127368Optimisation of the deployment of automated external defibrillators in public places in EnglandTerry P Brown0Lazaros Andronis1Asmaa El-Banna2Benjamin KH Leung3Theodoros Arvanitis4Charles Deakin5Aloysius N Siriwardena6John Long7Gareth Clegg8Steven Brooks9Timothy CY Chan10Steve Irving11Louise Walker12Craig Mortimer13Sandra Igbodo14Gavin D Perkins15NIHR Applied Research Collaboration West Midlands, Warwick Clinical Trials Unit, Warwick Medical School, University of Warwick, Coventry, UKWarwick Clinical Trials Unit, Warwick Medical School, University of Warwick, Coventry, UKWarwick Clinical Trials Unit, Warwick Medical School, University of Warwick, Coventry, UKDepartment of Mechanical and Industrial Engineering, University of Toronto, Toronto, CanadaDigital Health Technology, University of Birmingham, Birmingham, UKSouthampton University Hospital, Southampton, UKCommunity and Health Research Unit, University of Lincoln, Lincoln, UKPatient and Public Involvement Representative, Warwick, UKUsher Institute, University of Edinburgh, Edinburgh, UKDepartment of Emergency Medicine, Queens University, Kingston, Ontario, CanadaDepartment of Mechanical and Industrial Engineering, University of Toronto, Toronto, CanadaAssociation of Ambulance Chief Executives, London, UKIsle of Wight NHS Trust, Isle of Wight, UKSouth-East Coast Ambulance Service NHS Foundation Trust, Coxheath, UKNorth-West Ambulance Service NHS Trust, Bolton, UKNIHR Applied Research Collaboration West Midlands, Warwick Clinical Trials Unit, Warwick Medical School, University of Warwick, Coventry, UKBackground Ambulance services treat over 32,000 patients sustaining an out-of-hospital cardiac arrest annually, receiving over 90,000 calls. The definitive treatment for out-of-hospital cardiac arrest is defibrillation. Prompt treatment with an automated external defibrillator can improve survival significantly. However, their location in the community limits opportunity for their use. There is a requirement to identify the optimal location for an automated external defibrillator to improve out-of-hospital cardiac arrest coverage, to improve the chances of survival. Methods This was a secondary analysis of data collected by the Out-of-Hospital Cardiac Arrest Outcomes registry on historical out-of-hospital cardiac arrests, data held on the location of automated external defibrillators registered with ambulance services, and locations of points of interest. Walking distance was calculated between out-of-hospital cardiac arrests, registered automated external defibrillators and points of interest designated as potential sites for an automated external defibrillator. An out-of-hospital cardiac arrest was deemed to be covered if it occurred within 500 m of a registered automated external defibrillator or points of interest. For the optimisation analysis, mathematical models focused on the maximal covering location problem were adapted. A de novo decision-analytic model was developed for the cost-effectiveness analysis and used as a vehicle for assessing the costs and benefits (in terms of quality-adjusted life-years) of deployment strategies. A meeting of stakeholders was held to discuss and review the results of the study. Results Historical out-of-hospital cardiac arrests occurred in more deprived areas and automated external defibrillators were placed in more affluent areas. The median out-of-hospital cardiac arrest – automated external defibrillator distance was 638 m and 38.9% of out-of-hospital cardiac arrests occurred within 500 m of an automated external defibrillator. If an automated external defibrillator was placed in all points of interests, the proportion of out-of-hospital cardiac arrests covered varied greatly. The greatest coverage was achieved with cash machines. Coverage loss, assuming an automated external defibrillator was not available outside working hours, varied between points of interest and was greatest for schools. Dividing the country up into 1 km2 grids and placing an automated external defibrillator in the centre increased coverage significantly to 78.8%. The optimisation model showed that if automated external defibrillators were placed in each points-of-interest location out-of-hospital cardiac arrest coverage levels would improve above the current situation significantly, but it would not reach that of optimisation-based placement (based on grids). The coverage efficiency provided by the optimised grid points was unmatched by any points of interest in any region. An economic evaluation determined that all alternative placements were associated with higher quality-adjusted life-years and costs compared to current placement, resulting in incremental cost-effectiveness ratios over £30,000 per additional quality-adjusted life-year. The most appealing strategy was automated external defibrillator placement in halls and community centres, resulting in an additional 0.007 quality-adjusted life-year (non-parametric 95% confidence interval 0.004 to 0.011), an additional expected cost of £223 (non-parametric 95% confidence interval £148 to £330) and an incremental cost-effectiveness ratio of £32,418 per quality-adjusted life-year. The stakeholder meeting agreed that the current distribution of registered publicly accessible automated external defibrillators was suboptimal, and that there was a disparity in their location in respect of deprivation and other health inequalities. Conclusions We have developed a data-driven framework to support decisions about public-access automated external defibrillator locations, using optimisation and statistical models. Optimising automated external defibrillator locations can result in substantial improvement in coverage. Comparison between placement based on points of interest and current placement showed that the former improves coverage but is associated with higher costs and incremental cost-effectiveness ratio values over £30,000 per additional quality-adjusted life-year. Study registration This study is registered as researchregistry5121. Funding This award was funded by the National Institute for Health and Care Research (NIHR) Health and Social Care Delivery Research programme (NIHR award ref: NIHR127368) and is published in full in Health and Social Care Delivery Research; Vol. 13, No. 5. See the NIHR Funding and Awards website for further award information. Plain language summary Ambulance services of the NHS treat over 32,000 people whose heart suddenly stops pumping effectively, a condition known as cardiac arrest. Despite ambulance services’ best efforts fewer than 1 in 10 survive. Electric shock treatment, known as defibrillation, is one of the most effective treatments, and if it is given within a few minutes of the heart stopping, over half the people treated survive. It is now possible for public to use an automatic machine (defibrillator) to safely give an electric shock to the heart before the emergency services arrive. For the public to make best use of these machines they need to be in the right places. In this study, we attempted to work out the best places to put defibrillators in communities, making them more accessible to use. We showed that defibrillators currently are sited disproportionately in more affluent areas of the country, and not used despite being within an accessible distance from where a cardiac arrest occurs. We assessed that if a defibrillator was installed at various points of interest the number of cardiac arrests that were covered increased significantly. We then used a computer to model the best locations for new defibrillators and calculate the optimal number needed. Placement based on this model showed that, for a smaller number of defibrillators, a similar improvement in coverage could be achieved. A health economic analysis that considered the cost of purchasing and installing defibrillators showed that installing additional defibrillators in specific points of interest improved coverage, but it was also more costly compared to current defibrillator placement. This research showed that significant improvement in cardiac arrest coverage could be achieved if defibrillators were placed intelligently in public settings. We also created a system that uses data to decide where to place public-access defibrillators in the community. Scientific summary Background Annually, English ambulance services treat over 30,000 people who have sustained an out-of-hospital cardiac arrest (OHCA), about 25% of whom achieve a return of spontaneous circulation by the time of hospital handover and 8.5% survive to 30 days. The chain of survival shows the essential elements required in an emergency care system to improve outcomes from an OHCA. The first two links – early recognition and early cardiopulmonary resuscitation – can buy time for the OHCA patient but are not definitive treatments in themselves. The key and most effective treatment for an OHCA is defibrillation. Prompt treatment with an automated external defibrillator (AED), within 3–5 minutes of collapse, can lead to survival rates in excess of 50%. Public-access defibrillation (PAD) refers to the use of AEDs by members of the public. PAD programmes allow the community access to this life-saving intervention while waiting for ambulances to arrive. The importance of PAD is growing given the increasing demands on ambulance services that are making reaching OHCAs in a timely manner challenging. However, at present, only a small proportion of patients are treated by PAD (5%). A fundamental, structural barrier, which limits opportunity for the use of AEDs, is their location in the community. There has been no clear strategy in the UK on where AEDs should be placed; the choice of where to install them in public places has been driven mainly by local ad hoc initiatives. This approach is limited and there is a call for an evidence-based strategy, and a requirement to identify the optimal location for an AED to improve OHCA coverage, to improve the chances of survival. Objectives The primary objective of this study was to optimise the placement of public-access AEDs in England, using mathematical modelling techniques, to maximise the likelihood that an individual sustaining an OHCA will have access to PAD, improving their chances of survival. The secondary objective was to assess the cost-effectiveness of optimised public-access AED placement compared to current and alternative-placement strategies. Methods Ethics and regulatory approvals Following Health Research Authority guidelines, the study did not require formal NHS Research Ethics Committee approval. The study was approved by the University of Warwick’s Biomedical and Scientific Research Ethics Committee (BSREC 118/18-19). The project was co-sponsored by the University Hospitals Birmingham NHS Foundation Trust and the University of Warwick. The Out-of-Hospital Cardiac Arrest Outcomes (OHCAO) registry has approval from the Confidentiality Advisory Group to collect and process identifiable patient information where it is not practical to obtain consent (22CAG0072 and 22CAG0087). Ethics approval for the OHCAO registry was gained from the National Research Ethics Committee South Central (13/SC/0361). Design This was a secondary analysis of data that were collected by the OHCAO registry on historical OHCAs, data held by ambulance services on the location of AEDs registered with them, and locations of points of interest (POIs) available from Ordnance Survey. Also, data were obtained on the census neighbourhood characteristics of areas of England. The study was divided into four work packages (WPs): (1) exploration of the characteristics and coverage of current locations of AEDs relative to the location of historical OHCAs; (2) comparison of the OHCA coverage of various AED deployment strategies (POIs and grid-based) and an optimisation model with the current coverage; (3) determination of the cost-effectiveness of the strategies in WP2 compared to current placement (CP); and (4) development of a national consensus of the optimal location for public-access AEDs. Statistical analysis Descriptive statistics were used to analyse for any differences in neighbourhood characteristics where OHCAs occurred and registered AEDs were located. Using geographical information system software, we calculated the walking distance between OHCA locations and locations of registered AEDS and POIs designated as potential sites for AED (e.g. pubs, places of worship, schools, halls and community centres, and cash machines). Assumed coverage was calculated as the proportion of OHCAs within 500 m of a registered AED or POI, assuming the latter were accessible 24 hours a day, 7 days a week. Actual coverage was calculated based on when the OHCA occurred and the working hours of the AED locations. Using these definitions, we then calculated coverage loss as assumed coverage minus actual coverage divided by assumed coverage. Coverage efficiency was calculated as the number of OHCAs covered divided by the number of POIs. For the optimisation part of the study, mathematical models for maximal coverage location problem were adapted. These mathematical models were based on seminal literature in operational research, specifically within the area of facility location optimisation (maximal covering location problem). The model took in information for historical OHCAs, as well as locations of existing AEDs and candidate future AED locations. It then determined the optimal locations for future AED placement so that OHCA coverage was maximised and the number of AEDs required to maximise coverage. For the cost-effectiveness analysis a de novo decision-analytic model was developed and used as a vehicle for assessing the costs and benefits of deployment strategies at POIs compared to current AED placement. The model was developed in line with recommended ‘good practice’ guidelines and, where relevant and possible, in accordance with requirements for economic evaluation aiming to inform decision-making in the UK. A stakeholder meeting was convened towards the end of the study that brought together groups interested in AED deployment. Results of the study were presented and discussed. Then three groups were organised to discuss: (1) what strategies should be used to increase the distribution of AEDs in communities; (2) what guidance should be provided for where to place AEDs in communities; and (3) what further research would be required. Patient and public involvement The project was designed to ensure meaningful patient and public involvement (PPI) was embedded throughout the study. The PPI co-applicant was involved from the conception of the study and made valuable contribution to the development of the proposal, and reviewed and commented on work at various stages of the project. Presentations were made to the National Institute for Health and Care Research (NIHR) Clinical Research Ambassadors Group, based at the University Hospital Birmingham NHS Foundation Trust. There were two PPI representatives on the Steering Committee, and at the stakeholder meeting there were representatives of two charities. Results The study looked at the location of 147,278 historical OHCAs (2014–9) and 32,491 AEDs, and 14 potential POIs. Current coverage Historical OHCAs were observed to occur in more urban areas with a high population density. These areas had a larger proportion of workers in routine jobs or unemployed and people from non-white ethnic groups, and a great degree of deprivation. In contrast, AEDs were placed in more affluent areas that had lower proportions of people from non-white ethnic groups. About 43.4% of the areas (the lower-level super output area) were found to not have a single AED located within their boundary. The average OHCA–AED distance was 1014 m (median 638 m). The number of OHCAs occurring within 100 m of an AED was 7965 (5.4%) and within 500 m was 57,225 (38.9%). Alternative coverage: points of interest The numbers (percentages) of OHCAs that were covered by each POI, ignoring the currently registered AEDs, varied greatly, from 7.4% by state secondary schools to 55.1% by cash machines. Similarly, of those OHCAs not covered by a currently registered AED, the proportion then covered by one of the POIs ranged from 5.5% by state secondary schools to 46.0% by cash machines. The proportion of OHCAs that were covered by a registered AED or a POI also ranged significantly, from 42.2% by state secondary schools to 67.0% by cash machines. Coverage loss for cash machines, care homes and community halls was assumed to be zero because the AEDs at these locations were presumed to be accessible all the time. Loss was greatest for state schools, if we assume that all the OHCAs that occur outside of normal state school opening times there will be no access to an AED. Coverage efficiency also varied significantly and ranged, based on assumed coverage, from about 150% in pubs to just under 450% in post offices, chemists, dentists, general practitioner surgeries and supermarkets. Alternative coverage: 1 km2 grids and census output areas If the country was divided up into 1 km2 grids, coverage of OHCAs would be significantly greater if we were to deploy an AED at the centre of each grid: 116,016 (78.8%). If we were also to consider the location of existing registered AEDs, the coverage would be 133,901 (90.9%). Alternatively, if we were to place an AED at the centroid of every census output area the coverage would be 57,134 (38.8%), and 97,718 (66.3%) if we included the location of existing AEDs. Alternative coverage: optimisation model The optimisation model was observed to show that if AEDs were placed in each POI location, OHCA coverage levels would not reach that of optimisation-based placement. For example, if an AED were placed in every pub in the West Midlands (n = 3573), it would result in additional coverage of 17.8% above that provided by existing registered AEDs. However, the same amount of added OHCA coverage could be achieved with only 250 optimised grid points spaced 1 km apart. The coverage efficiency provided by the optimised grid points was unmatched by any POI in any region. Cost-effectiveness Compared to current AED placement, all of the alternative deployments assessed were associated with a greater expected total cost per OHCA for a small increase in quality-adjusted life-years (QALYs) and life-years gained (LYGs). The most cost-effective option was halls and community centres. Compared to CP, this deployment option resulted in higher costs [£223, 95% confidence interval (CI) generated from probabilistic sensitivity analysis (PSA) distribution: £148 to £330], a higher number of QALYs (0.007, 95% CI generated from PSA distribution: 0.004 to 0.011) and an incremental cost-effectiveness ratio (ICER) of £32,418 per QALY (£18,893 per LYG). This value is above the upper bound of the £20,000–30,000 per QALY range that is often seen as a maximum ICER considered for decision-making in health care. Stakeholder meeting The meeting agreed that the current distribution of registered publicly accessible AEDs was suboptimal and that there was a disparity in their location in respect of deprivation and other health inequalities. Conclusions Automated external defibrillators are potentially life-saving devices for people who sustain an OHCA. AEDs need to be placed intelligently in public settings so that they are likely to be used by bystanders. We have developed a data-driven framework to support public-access AED location decisions, using optimisation and statistical models. We applied the methodology to real data from England. Results have demonstrated that optimising AED locations can result in substantial improvement in coverage compared to the current approach to AED deployment. We developed a de novo decision-analytic model to determine the costs and benefits associated with AED placement strategies in each of the different POIs and compared these against current AED placement. Results of the economic analysis showed that all of the alternative placements considered were associated with ICERs above £30,000 per additional QALY. Study registration This study is registered as researchregistry5121. Funding This award was funded by the National Institute for Health and Care Research (NIHR) Health and Social Care Delivery Research programme (NIHR award ref: NIHR127368) and is published in full in Health and Social Care Delivery Research; Vol. 13, No. 5. See the NIHR Funding and Awards website for further award information.https://doi.org/10.3310/HTBT7685out-of-hospital cardiac arrestautomated external defibrillatorpublic access defibrillationoptimisationhealth inequalitiescost-effectiveness: observational study |
| spellingShingle | Terry P Brown Lazaros Andronis Asmaa El-Banna Benjamin KH Leung Theodoros Arvanitis Charles Deakin Aloysius N Siriwardena John Long Gareth Clegg Steven Brooks Timothy CY Chan Steve Irving Louise Walker Craig Mortimer Sandra Igbodo Gavin D Perkins Optimisation of the deployment of automated external defibrillators in public places in England Health and Social Care Delivery Research out-of-hospital cardiac arrest automated external defibrillator public access defibrillation optimisation health inequalities cost-effectiveness: observational study |
| title | Optimisation of the deployment of automated external defibrillators in public places in England |
| title_full | Optimisation of the deployment of automated external defibrillators in public places in England |
| title_fullStr | Optimisation of the deployment of automated external defibrillators in public places in England |
| title_full_unstemmed | Optimisation of the deployment of automated external defibrillators in public places in England |
| title_short | Optimisation of the deployment of automated external defibrillators in public places in England |
| title_sort | optimisation of the deployment of automated external defibrillators in public places in england |
| topic | out-of-hospital cardiac arrest automated external defibrillator public access defibrillation optimisation health inequalities cost-effectiveness: observational study |
| url | https://doi.org/10.3310/HTBT7685 |
| work_keys_str_mv | AT terrypbrown optimisationofthedeploymentofautomatedexternaldefibrillatorsinpublicplacesinengland AT lazarosandronis optimisationofthedeploymentofautomatedexternaldefibrillatorsinpublicplacesinengland AT asmaaelbanna optimisationofthedeploymentofautomatedexternaldefibrillatorsinpublicplacesinengland AT benjaminkhleung optimisationofthedeploymentofautomatedexternaldefibrillatorsinpublicplacesinengland AT theodorosarvanitis optimisationofthedeploymentofautomatedexternaldefibrillatorsinpublicplacesinengland AT charlesdeakin optimisationofthedeploymentofautomatedexternaldefibrillatorsinpublicplacesinengland AT aloysiusnsiriwardena optimisationofthedeploymentofautomatedexternaldefibrillatorsinpublicplacesinengland AT johnlong optimisationofthedeploymentofautomatedexternaldefibrillatorsinpublicplacesinengland AT garethclegg optimisationofthedeploymentofautomatedexternaldefibrillatorsinpublicplacesinengland AT stevenbrooks optimisationofthedeploymentofautomatedexternaldefibrillatorsinpublicplacesinengland AT timothycychan optimisationofthedeploymentofautomatedexternaldefibrillatorsinpublicplacesinengland AT steveirving optimisationofthedeploymentofautomatedexternaldefibrillatorsinpublicplacesinengland AT louisewalker optimisationofthedeploymentofautomatedexternaldefibrillatorsinpublicplacesinengland AT craigmortimer optimisationofthedeploymentofautomatedexternaldefibrillatorsinpublicplacesinengland AT sandraigbodo optimisationofthedeploymentofautomatedexternaldefibrillatorsinpublicplacesinengland AT gavindperkins optimisationofthedeploymentofautomatedexternaldefibrillatorsinpublicplacesinengland |