Abstract
Aims
Bystander cardiopulmonary resuscitation and defibrillation can double survival to hospital discharge in out-of-hospital cardiac arrest. Mobile phone applications, such as GoodSAM, alerting nearby volunteer first-responders about out-of-hospital cardiac arrest could potentially improve bystander cardiopulmonary resuscitation and defibrillation, leading to better patient outcomes. The aim of this study was to determine GoodSAM’s effect on survival to hospital discharge following out-of-hospital cardiac arrest.
Methods and results
We collected data from the Out-of-Hospital Cardiac Arrest Outcomes Registry (University of Warwick, UK) submitted by the London Ambulance Service (1 April 2016 to 31 March 2017) and East Midlands Ambulance Service (1 January 2018 to 17 June 2018) and matched out-of-hospital cardiac arrests to GoodSAM alerts. We constructed logistic regression models to determine if there was an association between a GoodSAM first-responder accepting an alert and survival to hospital discharge, adjusting for location type, presenting rhythm, age, gender, ambulance service response time, cardiac arrest witnessed status, and bystander actions. Survival to hospital discharge was 9.6% (393/4196) in London and 7.2% (72/1001) in East Midlands. A GoodSAM first-responder accepted an alert for out-of-hospital cardiac arrest in 1.3% (53/4196) cases in London and 5.4% (51/1001) cases in East Midlands. When a responder accepted an alert, the adjusted odds ratio for survival to hospital discharge was 3.15 (95% CI: 1.19–8.36, P = 0.021) in London and 3.19 (95% CI: 1.17–8.73, P = 0.024) in East Midlands.
Conclusion
Alert acceptance was associated with improved survival in both ambulance services. Alert acceptance rates were low, and challenges remain to maximize the potential benefit of GoodSAM.
Keywords: Out-of-hospital cardiac arrest, CPR, Bystander defibrillation, Volunteer first-responders
Graphical abstract
Introduction
Around 1 in 10 people survive to hospital discharge or 1 month after an out-of-hospital cardiac arrest (OHCA).1–4 Bystander cardiopulmonary resuscitation (CPR) and defibrillation using an automated external defibrillator (AED) can at least double survival.5,6
Providing early access to trained bystanders capable and willing to provide these interventions could impact upon survival. There are a number of mobile-phone applications (‘apps’) worldwide that aim to activate volunteer first-responders near to an OHCA, who may then be able to provide assistance such as CPR and/or AED use.7
GoodSAM is one such system. It operates internationally and is now fully integrated into ambulance services in the UK, Australia, and New Zealand. During an emergency call, if an Emergency Medical Service (EMS) call-handler suspects OHCA and allocates a relevant code from the Computer-Assisted Dispatch (CAD) system, the system automatically activates GoodSAM. GoodSAM automatically sends an alert via the app to trained responders within a specified radius of the OHCA incident, asking if they are able to respond. The GoodSAM response is supplementary to the statutory EMS response.8
The aim of this study was to determine the effect of GoodSAM on OHCA survival to hospital discharge in two UK ambulance services.
Methods
System description
London Ambulance Service (LAS) covers an area of ∼620 square miles and serves ∼8.9 million people. It responds to over 1.1 million emergency incidents each year.9 East Midlands Ambulance Service (EMAS) covers an area of ∼6425 square miles and serves ∼4.8 million people. It responds to over 730 000 emergency incidents each year.10
GoodSAM is a mobile phone app that users can download on multiple platforms (Android, iOS, Windows). Registration and subsequent use of the app requires at least a valid CPR qualification, or evidence of a professional healthcare qualification (e.g. doctor, nurse, paramedic, Emergency Medical Technician). GoodSAM verifies this information, by upload of certificate or professional registration number. Partner organizations such as ambulance and other emergency services can also verify users who work for their own organization. GoodSAM has a Code of Conduct for its users, a detailed Data Protection Policy, and is registered with the Information Commissioner’s Office (ref: ZA094052). Data are encrypted using a 256-bit Advanced Encryption Standard (AES-256) cipher. There is some limited data sharing at the time that a responder accepts an alert (e.g. local ambulance services will have access to information about responders’ locations at that time), but no data are shared with other third parties.8
Both LAS and EMAS use a version of the Medical Priority Dispatch System. During a 999 call, GoodSAM is automatically activated if a call-handler allocates a dispatch code on the CAD system that indicates a possible OHCA—each ambulance service determines this in advance (see Table 1). These codes include both adult and paediatric cases but exclude traumatic cases where cardiac arrest is confirmed or suspected. If the call-handler subsequently changes the dispatch code during the 999 call to a code not shown in Table 1, the GoodSAM responder automatically receives an updated notification that the alert has been cancelled.
Table 1.
Criteria for GoodSAM activation following potential OHCA (at the time of study)
| London Ambulance Service | East Midlands Ambulance Service |
|---|---|
| Cardiac arrest—not breathing at all | Breathing problems—ineffective breathing |
| Respiratory arrest—breathing uncertain (agonal) | Burns/explosion—unconscious or cardiac arrest |
| Respiratory arrest/ineffective breathing | Cardiac/respiratory arrest |
| Complete obstruction/ineffective breathing | Chest pain—not alert/breathing problems |
| Fitting and not breathing | Choking—complete obstruction |
| Fitting and not breathing—fitting history | Convulsions—not breathing |
| Unconscious or fainting—ineffective breathing | Drowning—unconscious or cardiac arrest |
| Unconscious, agonal/ineffective breathing | Electrocution—not breathing |
| Unconscious or cardiac arrest | Fall—unconscious or cardiac arrest |
| Heart problems—not alert/just resuscitated | |
| Unconscious or cardiac arrest | |
| Unconscious fainting—ineffective breathing | |
| Unconscious fainting—agonal breathing | |
| Unknown problem—life status questionable | |
| Call from 111—possible cardiac arrest | |
| Call from 111—unconscious and ineffective breathing | |
| Call from 111–8 min response required |
GoodSAM tracks real-time position of a responder’s phone and can send an alert (as an audible siren) to responder(s) within a specified radius of the potential OHCA. In London, at the time of the study, GoodSAM alerted up to three responders within a 300 m radius of an incident and in East Midlands, GoodSAM alerted up to five responders within 800 m. This process is fully automated and not visible to the 999 call-handler. If the allocated code changes (to one not indicating cardiac arrest), GoodSAM issues a ‘stand-down’ or cancel notification to the phone.
Responders accept or reject an alert via a button press in-app. If one or more responders rejects or does not respond to the alert within 15 s, GoodSAM sends a further alert to the next nearest responder (up to the maximum number) if they are within the specified range. The alert will remain active on the phone for up to 15 min.
Once a GoodSAM responder accepts an alert, the app provides a route to the patient. The screen also displays a brief descriptor, based on the criteria in Table 1 (e.g. ‘cardiac arrest—not breathing at all’). If there is a nearby AED, this is displayed via the app too. There is additional functionality in the app for a responder to indicate when they have reached the scene, and if they are carrying an AED. As these are optional, we could not use this to reliably indicate who arrived on scene (with or without an AED).
Figure 1 is a flow chart illustrating the GoodSAM activation process and Figure 2 shows the appearance of (the sequentially appearing) mobile phone screens seen during an alert.
Figure 1.
How GoodSAM is activated. Once the emergency call-handler determines that there is a potential OHCA, this activation process is automatic.
Figure 2.
(A) Mobile-phone screen with app open. Blue dot is current responder’s position, other responders indicated by green person icon (one is shown carrying their own AED). A nearby public-access AED is also indicated. (B) An alert is received and (C) accepted.
In the UK, there is no legal obligation for a GoodSAM responder to accept an alert, and by not attending the statutory ambulance service response is not affected in any way. In England and Wales, the Social Action, Responsibility and Heroism Act (2015) may provide additional protections, but there is no case law. There has been no successful litigation against anyone who intervened to provide life-saving treatment to a person who had sustained an OHCA.11
By 2020, GoodSAM had more than 40 000 registered responders for OHCAs, and had the locations (and, where known, the hours when it could be accessed) for more than 50 000 AEDs.
Data collection
LAS and EMAS both submit data on all-cause OHCAs, when they either initiated or continued resuscitation, to the national Out-of-Hospital Cardiac Arrest Outcomes (OHCAO) registry at the University of Warwick. We collected and have presented data for LAS between 1 April 2016 and 31 March 2017 and for EMAS between 1 January 2018 and 17 June 2018. In London, GoodSAM was non-operational from 30 August 2016 to 19 September 2016 and from 26 December 2016 to 30 December 2016, so results that we present exclude OHCAs on those dates.
We collected the following data from the OHCAO registry: patient age (years); patient gender (male/female); date/time of 999 call connected to EMS; time EMS stopped on scene; OHCA location; OHCA witnessed by (EMS/bystander/unwitnessed); CPR performed by (EMS/bystander/unwitnessed); public-access AED used by public (yes/no—available for LAS only); initial cardiac arrest rhythm [ventricular fibrillation or ventricular tachycardia (VF/VT)/pulseless electrical activity (PEA)/asystole]; return of spontaneous circulation (ROSC) at hospital handover (yes/no); and survival to hospital discharge (yes/no). We calculated the ambulance response time as the difference between the time EMS stopped on scene and the time of 999 call connected to EMS.
GoodSAM provided the following information: number of 999 calls meeting criteria for GoodSAM alert; number of GoodSAM alerts sent; date/time of GoodSAM alert; number of GoodSAM alerts accepted, not seen, or rejected; and incident location.
We manually matched cases submitted to the OHCAO registry with GoodSAM alerts using date/time of the 999 call and OHCA location. LAS provided incident location as an address and postcode, and we reviewed these to determine whether these were residential or non-residential. EMAS provided location type as per Utstein definition12 (e.g. residential, public building, street, workplace), supplemented by postcode only, and we dichotomized these as residential or non-residential.
Both ambulance services provided a list of public-access AEDs known to them. OHCA location accuracy was such in London (LAS) that we were able to calculate the distance from each OHCA to its nearest public-access AED.
Data analysis and reporting
We presented and collected data according to the Utstein guidelines.12 The primary outcome was survival to hospital discharge. We have presented data for the following groups: GoodSAM alert accepted; GoodSAM alert not seen or rejected; and No GoodSAM alert sent.
Given the different system configurations, we constructed a separate multiple logistic regression model for LAS and EMAS data to determine if there was an association between GoodSAM alert acceptance group (accepted/not seen or rejected/no alert sent) and survival to hospital discharge. In addition to GoodSAM alert acceptance group, the models contained the following independent variables: patient age (years), patient gender (male/female), OHCA witnessed by (EMS/bystander/unwitnessed), CPR performed by (EMS/bystander, not performed), bystander AED use (yes/no, LAS only), location type (residential/non-residential), initial cardiac arrest rhythm (VF or VT/PEA/asystole); EMS response time (min: s); and distance to nearest AED (m, LAS only). We entered all variables into the model for their potential to impact the outcome in OHCA, with no statistical procedure to determine entry into the model. We considered that all covariates were clinically important variables for which we should adjust. We have presented unadjusted odds ratios (OR) and adjusted ORs with 95% confidence intervals (95% CI).
We used SPSS Statistics (version 24; IBM, New York, USA) for statistical analyses. We analysed differences in patient and process variables between the GoodSAM alert accepted, GoodSAM alert not seen/rejected, and No GoodSAM alert groups using χ2 tests for categorical variables and Kruskal–Wallis test for continuous variables.
For the logistic regression modelling, we used Shapiro–Wilk’s test to assess for normality of distribution for continuous variables and performed multicollinearity tests, determining that there was no substantial association between variables in the model. We also calculated pseudo-R2 values (Cox and Snell, and Nagelkerke) to indicate how much the predictor variables in the logistic regression model explain the outcome of interest, and the Hosmer–Lemeshow Goodness of Fit test to test the fit of the model and indicate how reliable the estimates provided for the outcome measure were.13
Research governance and ethical approvals
The OHCAO registry has approval from the Health Research Authority (reference number 13/SC/036) and Confidentiality Advisory Group (reference number ECC8-04(C)/2013). The registry granted approval for the use of non-identifiable data for the analyses described above.
The Biomedical and Scientific Research Ethics Committee at the University of Warwick (REGO 2018-2157) approved the study-specific protocols for collection and analysis of data on 6 March 2018 (LAS data) and 29 October 2018 (EMAS data). We held all data on a secure University server in encrypted folders.
Results
London
There were 11 894 emergency calls that fulfilled the criteria for a GoodSAM alert, and 4196/11 894 (35%) were confirmed OHCAs. Amongst these 4196 confirmed OHCAs, GoodSAM issued 354 alerts for the 282 (6.7%) cases when at least one responder was within a 300 m radius. A GoodSAM responder accepted an alert on 56/354 (16%) occasions for 53 OHCAs (1.3% of total OHCAs). More than one person received an alert on 56 occasions, and on three occasions two people accepted an alert.
Baseline characteristics for OHCA patients
The median age (n = 4164, 32 unknown) was 69.3 (IQR: 52.8–80.8) years and 2695/4196 (64%) were male. EMS witnessed 739/4196 (18%) OHCAs, bystanders 1985/4196 (47%), and 1472/4196 (35%) were unwitnessed. Bystanders performed CPR in 2211/4196 (53%) OHCAs, or 2209/3457 (64%) of the non-EMS-witnessed cases. They attached a public-access AED in 179/4196 (4.3%) OHCAs, or 176/3457 (5.1%) of the non-EMS-witnessed cases. OHCAs occurred in residential locations in 3447/4111 cases (84%, 85 unknown). Median EMS response time was 7:39 (IQR: 5:45–10:24) min. The median distance to an AED (n = 4111, 85 unknown) was 407 (IQR: 222–642) m. The initial cardiac rhythm (n = 4172, 24 unknown) was VF/VT in 916/4172 (22%), PEA in 1179/4172 (28%), and asystole in 2077/4172 (50%).
ROSC at hospital was 1219/4196 (29%), and 393/4111 (9.6%, 85 unknown) patients survived to hospital discharge. Table 2 summarizes these data by GoodSAM alert status.
Table 2.
Patient, process, and outcome data by GoodSAM response group (London)
| GoodSAM alert |
No GoodSAM alert (n = 3914) | ||
|---|---|---|---|
| Accepted (n = 53) | ‘Not seen’ or rejected (n = 229) | ||
| Age (median (IQR), years)* | 66.0 (50.0–77.1) | 69.4 (54.0–80.0) | 69.3 (52.7–80.9) |
| Unknown cases | 32 | ||
| Gender | |||
| Male | 64.2% (34/53) | 67.2% (154/229) | 64.1% (2507/3914) |
| Female | 35.8% (19/53) | 32.8% (75/229) | 35.9% (1407/3914) |
| OHCA witnessed by* | |||
| EMS | 3.8% (2/53) | 2.2% (5/229) | 18.6% (729/3914) |
| Bystander | 58.4% (31/53) | 60.7% (139/229) | 46.4% (1815/3914) |
| Unwitnessed | 37.8% (20/53) | 37.1% (85/229) | 35.0% (1367/3914) |
| Bystander CPR | |||
| All cases* | 67.9% (36/53) | 64.2% (147/229) | 51.8% (2028/3914) |
| Non-EMS-witnessed | 70.6% (36/51) | 65.6% (147/224) | 63.7% (2026/3182) |
| Bystander AED | |||
| All cases* | 9.4% (5/53) | 8.3% (19/229) | 4.0% (155/3913) |
| Non-EMS-witnessed* | 9.8% (5/51) | 8.5% (19/224) | 4.8% (152/3182) |
| Unknown cases | 1 | ||
| Location type* | |||
| Residential | 69.8% (37/53) | 72.0% (162/225) | 84.7% (3248/3833) |
| Non-residential | 30.2% (16/53) | 28.0% (63/225) | 15.3% (585/3833) |
| Unknown cases | 4 | 81 | |
| Initial rhythm* | |||
| VF/VT | 20.8% (11/53) | 29.3% (67/229) | 21.5% (838/3890) |
| PEA | 17.0% (9/53) | 21.0% (48/229) | 28.8% (1122/3890) |
| Asystole | 62.3% (33/53) | 49.8% (114/229) | 49.6% (1930/3890) |
| Unknown cases | 24 | ||
| EMS response time* | 06:21 | 06:41 | 07:45 |
| [median (IQR), min:s] | (04:40–08:15) | (05:08–08:46) | (05:48–10:30) |
| Distance from nearest AED* | 255 | 312 | 413 |
| [median (IQR), m] | (134–433) | (140–539) | (228–651) |
| Unknown cases | 5 | 82 | |
| ROSC at hospital | 39.6% (21/53) | 28.4% (65/229) | 28.9% (1133/3914) |
| Survival to hospital discharge | 17.6% (9/51) | 10.3% (23/223) | 9.4% (361/3837) |
| Unknown cases | 2 | 6 | 77 |
P < 0.05.
Differences between groups analysed using χ2 for categorical variables and Kruskal–Wallis for continuous variables.
East Midlands
There were 8768 emergency calls that fulfilled the criteria for a GoodSAM alert, and 1041/8768 (12%) were confirmed OHCAs. Amongst these 1041 confirmed OHCAs, GoodSAM issued 349 alerts for the 227 (22%) cases when at least one responder was within an 800 m radius. A GoodSAM responder accepted an alert on 54/349 (15%) occasions for 51 OHCAs (4.9% of total OHCAs). More than one person received an alert on 73 occasions, and on three occasions two people accepted an alert.
Baseline characteristics for OHCA patients
The median age (n = 1009, 32 unknown) was 72 (IQR: 59–83) years and 656/1041 (63%) were male. EMS witnessed 21/1041 (2%) OHCAs, bystanders 564/1041 (54%), and 456/1041 (44%) were unwitnessed. Bystanders performed CPR in 647/1041 (62%) OHCAs, or 647/1020 (63%) of the non-EMS-witnessed cases.OHCAs occurred in residential locations in 868/1033 cases (84%, 8 unknown). Median EMS response time was 9:59 (IQR: 6:16–16:02) min. The initial cardiac rhythm (n = 970, 71 unknown) was VF/VT in 172/970 (18%), PEA in 204/970 (21%), and asystole in 594/970 (61%).
ROSC at hospital was 1219/4196 (25%), and 72/1001 (7.2%, 40 unknown) patients survived to hospital discharge. Table 3 summarizes these data by GoodSAM alert status.
Table 3.
Patient, process, and outcome data by GoodSAM response group (East Midlands)
| GoodSAM alert |
No GoodSAM alert (n = 814) | ||
|---|---|---|---|
| Accepted (n = 51) | ‘Not seen’ or rejected (n = 176) | ||
| Age [median (IQR), years]* | 73.0 (67.8–79.3) | 70.0 (56.8–81.0) | 72.0 (59.0–82.0) |
| Unknown cases | 3 | 4 | 25 |
| Gender | |||
| Male | 76.5% (39/51) | 64.2% (113/176) | 61.9% (504/814) |
| Female | 23.5% (12/51) | 35.8% (63/176) | 38.1% (310/814) |
| OHCA witnessed by | |||
| EMS | 2.0% (1/51) | 0.6% (1/176) | 2.3% (19/814) |
| Bystander | 49.0% (25/51) | 51.1% (90/176) | 55.2% (449/814) |
| Unwitnessed | 49.0% (25/51) | 48.3% (85/176) | 42.5% (346/814) |
| Bystander CPR | |||
| All cases* | 58.9% (30/51) | 74.4% (131/176) | 59.7% (486/814) |
| Non-EMS-witnessed* | 60.0% (30/50) | 74.9% (131/175) | 61.1% (486/795) |
| Location type | |||
| Residential | 80.4% (41/51) | 86.4% (152/176) | 83.7% (675/806) |
| Non-residential | 19.6% (10/51) | 13.6% (24/176) | 16.3% (131/806) |
| Unknown cases | 8 | ||
| Initial rhythm | |||
| VF/VT | 26.5% (13/49) | 17.9% (30/168) | 17.1% (129/753) |
| PEA | 14.3% (7/49) | 16.1% (27/168) | 22.6% (170/753) |
| Asystole | 59.2% (29/49) | 66.1% (111/168) | 60.3% (454/753) |
| Unknown cases | 2 | 8 | 61 |
| EMS response time* | 07:59 | 07:29 | 10:46 |
| [median (IQR), min:s] | (05:23–12:57) | (05:26–11:36) | (06:46–17:00) |
| Unknown cases | 2 | 6 | |
| ROSC at hospital | 25.5% (13/51) | 23.3% (41/176) | 24.9% (203/814) |
| Survival to hospital discharge | 15.2% (7/46) | 5.3% (9/170) | 7.1% (56/785) |
| Unknown cases | 5 | 6 | 29 |
P < 0.05.
Differences between groups analysed using χ2 for categorical variables and Kruskal–Wallis for continuous variables.
Logistic regression models
London
Continuous variables were non-normally distributed (Shapiro–Wilk test all P < 0.001). There was little dependency between covariates in the model (variance inflation factor (VIF) no more than 1.22 for any variable).
The logistic regression model included data from 3971/4196 (95%) OHCAs. Cox and Snell R2 (0.185) and Nagelkerke R2 (0.395) suggested that 18.5–39.5% of the variation in survival to hospital discharge could be explained by this model. The P-value for the Hosmer–Lemeshow Goodness of Fit test was non-significant (0.24), suggesting overall goodness of fit.
The adjusted OR for survival to hospital discharge (compared with death) if a GoodSAM alert was accepted (compared with no alert sent) was 3.15 (95% CI: 1.19–8.36; P = 0.021). If the GoodSAM alert ‘not seen or rejected’ was taken as the reference, the adjusted OR for survival to hospital discharge (compared with death) in the GoodSAM alert accepted group was 3.06 (95% CI: 1.03–9.03; P = 0.04). We have presented unadjusted ORs and adjusted ORs for all variables in the model in Table 4 and Figure 3A.
Table 4.
Logistic regression model—impact on survival to hospital discharge (London)
| Unadjusted OR (95% CI) | AOR (95% CI) | |
|---|---|---|
| GoodSAM group | ||
| Accepted | 2.06 (0.99–4.27); P = 0.052 | 3.15 (1.19–8.36); P = 0.021 |
| Not seen/rejected | 1.11 (0.71–1.73); P = 0.66 | 1.03 (0.61–1.75); P = 0.908 |
| No alert | Reference | Reference |
| EMS response time | 1.00 (0.99–1.99); P = 0.011 | 0.99 (0.99–1.00); P = 0.005 |
| Age in years | 0.99 (0.98–0.99); P < 0.001 | 0.98 (0.97–0.98); P < 0.001 |
| Gender | ||
| Male | 1.59 (1.26–2.00); P < 0.001 | 0.97 (0.73–1.30); P = 0.844 |
| Female | Reference | Reference |
| OHCA witnessed status | ||
| EMS | 9.21 (6.41–13.2); P < 0.001 | 7.70 (4.76–12.5); P < 0.001 |
| Bystander | 4.12 (2.91–5.82); P < 0.001 | 1.84 (1.25–2.71); P = 0.002 |
| Unwitnessed | Reference | Reference |
| CPR performed by | ||
| EMS | 5.37 (3.90–7.41); P < 0.001 | Not calculateda |
| Bystander | 1.93 (1.42–2.62); P < 0.001 | 1.09 (0.77–1.56); P = 0.621 |
| Not performed | Reference | Reference |
| Bystander AED | ||
| Yes | 2.11 (1.41–3.16); P < 0.001 | 1.38 (0.82–2.33); P = 0.227 |
| No | Reference | Reference |
| Initial rhythm | ||
| VF/VT | 33.6 (22.8–49.4); P < 0.001 | 27.9 (18.4–42.3); P < 0.001 |
| PEA | 3.96 (2.55–6.14); P < 0.001 | 2.42 (1.51–3.90); P < 0.001 |
| Asystole | Reference | Reference |
| Location type | ||
| Non-residential | 2.88 (2.28–3.63); P < 0.001 | 1.66 (1.23–2.25); P = 0.001 |
| Residential | Reference | Reference |
| Distance from nearest AED | 1.00 (0.99–1.00); P = 0.13 | 1.00 (1.00–1.00); P = 0.35 |
EMS performed CPR for all of the cases that were EMS-witnessed, therefore not calculated by SPSS (redundancy). Bold values are statistically significant (P <0.05).
Figure 3.
Adjusted odds ratios (with 95% confidence intervals) for ‘Alert accepted’ and ‘alert not seen/rejected’ groups, with ‘No GoodSAM alert’ (dotted line) as reference in (A) London and (B) East Midlands.
East Midlands
Continuous variables were non-normally distributed (Shapiro–Wilk test all P < 0.001). There was little dependency between covariates in the model (VIF no more than 1.07 for any variable).
The logistic regression model included data from 907/1041 (87%) OHCAs. Cox and Snell R2 and Nagelkerke R2 values were 0.109 and 0.286, respectively. Hosmer–Lemeshow Goodness of Fit test suggested overall goodness of fit (P = 0.60).
The adjusted OR for survival to hospital discharge (compared with death) if a GoodSAM alert was accepted (compared with no alert sent) was 3.19 (95% CI: 1.17–8.73; P = 0.024). If the GoodSAM alert ‘not seen or rejected’ was taken as the reference, the adjusted OR for survival to hospital discharge (compared with death) in the GoodSAM alert accepted group was 4.84 (95% CI: 1.34–17.5; P = 0.016). We have presented unadjusted ORs and adjusted ORs for all variables in the model in Table 5 and Figure 3B.
Table 5.
Logistic regression model—impact on survival to hospital discharge (East Midlands)
| Unadjusted OR (95% CI) | AOR (95% CI) | |
|---|---|---|
| GoodSAM group | ||
| Accepted | 2.34 (1.00–5.46); P = 0.05 | 3.19 (1.17–8.73); P = 0.024 |
| Not seen/rejected | 0.73 (0.35–1.50); P = 0.39 | 0.66 (0.26–1.77); P = 0.378 |
| No alert | Reference | Reference |
| EMS response time | 1.00 (1.00–1.00); P = 0.97 | 1.00 (1.00–1.00); P = 0.818 |
| Age in years | 0.96 (0.95–0.97); P < 0.001 | 0.96 (0.94–0.97); P < 0.001 |
| Gender | ||
| Male | 1.76 (1.02–3.02); P = 0.041 | 1.29 (0.65–2.52); P = 0.467 |
| Female | Reference | Reference |
| OHCA witnessed status | ||
| EMS | 1.06 (0.14–8.28); P = 0.96 | 1.37 (0.15–12.3); P = 0.778 |
| Bystander | 2.06 (1.22–3.48); P = 0.007 | 2.13 (1.09–4.15); P = 0.028 |
| Unwitnessed | Reference | Reference |
| CPR performed by | ||
| EMS | 0.61 (0.08–4.71); P = 0.63 | Not calculated a |
| Bystander | 0.84 (0.52–1.38); P = 0.84 | 0.73 (0.39–1.41); P = 0.350 |
| Not performed | Reference | Reference |
| Initial rhythm | ||
| VF/VT | 12.1 (6.32–23.1); P < 0.001 | 10.7 (5.09–22.3); P < 0.001 |
| PEA | 2.85 (1.32–6.18); P = 0.008 | 3.94 (1.66–9.37); P = 0.002 |
| Asystole | Reference | Reference |
| Location type | ||
| Non-residential | 3.16 (1.84–5.42); P < 0.001 | 1.73 (0.87–3.44); P = 0.121 |
| Residential | Reference | Reference |
EMS performed CPR for all of the cases that were EMS-witnessed, therefore not calculated by SPSS (redundancy). Bold values are statistically significant (P <0.05).
Discussion
Main findings
Accepting a GoodSAM alert was associated with improved survival to hospital discharge in adjusted analyses during the study periods in London (adjusted OR: 3.15; 95% CI: 1.19–8.36; P = 0.021) and East Midlands (adjusted OR: 3.19; 95% CI: 1.17–8.73; P = 0.024). Thirty-five per cent of 999 calls meeting the criteria for GoodSAM activation in London, and 12% in East Midlands, were for confirmed OHCAs. A GoodSAM responder was close enough to receive an alert for 6.7% confirmed OHCAs in London and 22% in East Midlands. A GoodSAM responder accepted alerts for 1.3% confirmed OHCAs in London and 4.9% in East Midlands.
Comparison with the literature
A 2019 Cochrane Library systematic review14 found one RCT15 evaluating volunteer first-responder systems. This study, of 676 non-traumatic OHCAs in patients aged ≥8 years, found no improvement in 30 day survival when patients received a supplementary response from volunteer first-responders within a 500 m radius, compared with a standard ambulance service response alone (OR: 1.34; 95% CI: 0.79–2.29). However, the study was not powered for this outcome and did not mention public-access AED use.
Two observational studies reported improved survival to hospital discharge when volunteer first-responders were activated16 or attended an OHCA patient.17 Both these studies attempted to account for confounders, by propensity score matching16 or by multiple logistic regression,17 but neither accounted for the impact of bystander AED use. A further before-and-after study from South Korea reported an adjusted OR of 1.84 (95% CI: 1.29–2.63) for survival to hospital discharge and 2.31 (95% CI: 1.44–3.70) for survival with favourable neurological outcome for an intervention bundle to increase OHCA survival. This included a volunteer first-responder system introduced in 2015, whose contribution to the reported outcomes is unclear.18
Additionally, a 2020 meta-analysis19 pooled data from the RCT15 and two of these observational studies17,18 to report an OR for survival of 1.51 (95% CI: 1.24–1.84) when a volunteer first-responder was activated via mobile phone (compared with a standard response group).
In a study published since these reviews, in Denmark (2017–18) volunteer first-responders arriving before EMS was associated with increased bystander CPR (OR: 1.8; 95% CI: 1.1–2.9) and bystander defibrillation rates (OR: 3.7; 95% CI: 2.0–6.8). However, there was no improvement in survival.20
There are many mobile-phone apps worldwide to alert members of the public to nearby OHCAs. A recent review reported how such systems differ markedly in their alerting radii (from 200 m up to 10 km) and the number of volunteers alerted (up to 30).7 Other systems have reported higher alert acceptance rates than in our study in the Netherlands (27%),21 USA (23%),22 Denmark (51%),20 and Sweden (65%).15
Other systems also report many more responders available per incident20,23 than in our study, and so the chance that any given alert will have at least one person accept is higher. GoodSAM should attempt to increase its responder numbers but, until this happens, overcoming behavioural barriers to an individual accepting an alert is important to efforts to optimize the GoodSAM system.
In this study, we did not have precise information about the number of responders and AEDs registered on the GoodSAM app in either region, and so cannot reliably comment on the effect of either responder- or AED-density on outcomes. A study from the Netherlands suggests that at least 10 volunteer first-responders and two public-access AEDs are needed in every km2 for an optimum response.24
Clinical implications
The imprecision, indirectness, and risk of bias mean that this study likely represents a very low certainty of evidence according to the Grading of Recommendations, Assessment, Development and Evaluations framework25: i.e. the true effect might be different to the effect estimated in this study. The improvement in survival to hospital discharge associated with a GoodSAM response may be due to increased and earlier provision of CPR and/or AED use, but it is unclear whether this approach is superior to other strategies to strengthen the community response to OHCA.
A number of alerts were ‘not seen’ by the first-responder. Since this project GoodSAM have updated their system so that the audible alert siren can sound even when the mobile phone is in silent mode—a feature not available at the time of the study, and one possible explanation for this.
Almost all GoodSAM responders reported confidence in their ability to provide CPR and use an AED if asked to respond, and they rarely reported anxiety about responding.26 In a similar volunteer first-responder system in the Netherlands, 13% of 189 first-responders questioned (in 2013–14) reported a ‘severe’ short-term impact on their psychological health, but by 4–6 weeks after the event, all first-responders reported either ‘no’ (81%) or ‘mild’ (19%) stress on the validated Impact of Event Scale.27 In Denmark, perceived stress was low in 102 people (surveyed in 2018) 3 weeks after taking part in a resuscitation effort, and very few reported symptoms indicating post-traumatic stress disorder.28
The ‘true positive’ rate for OHCA following activation was 35% in London and 12% in East Midlands. There is marked difference in these figures, which might be partly explained by differences in the criteria used for ‘potential OHCA’ (see Table 1). Furthermore, other studies have reported that 53% (Denmark)20 and 33% (Sweden)23 of alerts were for subsequently confirmed OHCA. Activation for a non-OHCA may reduce future motivation to respond,26 so there is a need to optimize activation criteria across different systems.
EMAS notified GoodSAM responders up to (a radius of) 800 m away, whilst LAS did so up to 300 m. The proportion of alerts for OHCA when a GoodSAM responder was in range was higher in East Midlands (22%) than in London (6.7%), but alert acceptance as a proportion of GoodSAM alerts sent for OHCA was similar in both ambulance services—16% in London; 15% in East Midlands. It is therefore unclear how far the response radius can be increased before it affects the likelihood of accepting an alert. There are marked differences in geography between local ambulance service areas, and responder density may vary as well, so this optimum radius is likely to vary across the UK.
Set-up and maintenance costs for using GoodSAM in the OHCA response are modest,8 so pursuing strategies to improve both volunteer first-responder and other interventions to improve the community response to OHCA is entirely reasonable. This should include widespread CPR/AED training, particularly of school-aged children.29,30
There is also ongoing work in the UK to accurately locate and optimize the placement of public-access AEDs,31 as has been done elsewhere.32–34 There is now a national public-access database (‘The Circuit’) whose information is available to all UK local ambulance services.35
Efforts to maximize the number of responders using the app will increase the number of potential OHCAs for which a responder is alerted and/or accepts the alert. We also need further data about the proportion of GoodSAM responders who arrive prior to EMS and both quantitative and qualitative evaluations of their actions. The decision to respond is complex, so we should focus on behavioural science methodologies to help design interventions to improve response rates.26
Strengths and limitations
We have reported survival to hospital discharge as the main outcome measure, as this is important to clinicians, researchers, patients, and their families.36 Data on favourable neurological outcome and functional performance post-OHCA are not currently available from the OHCAO registry. This limitation means that we cannot report longer-term and favourable neurological outcomes, nor about what investigations or procedures survivors underwent in-hospital (e.g. percutaneous coronary intervention (PCI)). There are factors that might affect survival—such as time from collapse to CPR interval, time from CPR to ROSC, which form part of the validated Cardiac Arrest Hospital Prognosis score37—that we were also not able to account for.
Ambulance services submit data to the OHCAO registry when they start or continue resuscitation efforts, and the reliability of our findings partly depends on how accurate their determination of OHCA was. Additionally, EMAS did not submit data on bystander AED use and neither ambulance service provided data on whether or not public-access AEDs were actually available at the time of the OHCA, which is not always the case.33,38,39 We collected data from two different time periods (and durations) making direct comparisons between the two EMS services difficult.
Although there is evidence of a difference in the odds of surviving where an alert is accepted (compared with no alert sent or alert not seen/rejected), we made no formal power calculation to detect such a difference. We are alert to the chance that this is a spurious statistically significant finding. We separated out the ‘not seen/rejected’ and ‘no GoodSAM alert’ groups (even though the ultimate effect—a responder does not go to a patient—is the same) as we considered the possibility that factors not already included in our modelling resulting in volunteer first-responders being available within a certain radius might also impact upon clinical outcome.
There was no common case identifier for GoodSAM and OHCAO registry data, so we matched data manually using date/time and incident location. There is the potential for error, resulting in erroneously not reporting incidences when GoodSAM was activated for confirmed OHCA.
Recognizing and accounting for all factors potentially relevant to OHCA survival and choosing an appropriately non-biased comparison group is difficult. It is possible that patients who are more likely to survive, by nature of some factor that we have not been able to account for, are more likely to get a GoodSAM response in the first place. However, the survival effect that we have reported did appear independent of the factors leading to issuing an alert. There was no survival advantage in the ‘not seen or rejected’ group over the ‘no alert’ group, and the adjusted OR for survival after alert acceptance remained statistically significant when alerts ‘not seen or rejected’ were taken as the reference group.
An approach used elsewhere20 would be to take the alert accepted group and compare outcomes when responders did or did not arrive before EMS. Unfortunately, we could not determine whether or not a GoodSAM responder reached the patient after accepting an alert, nor what interventions they performed. Indeed, a GoodSAM responder could attend an alert without indicating alert acceptance (via a button press) on the app, leading to misclassification of the GoodSAM response in that circumstance. Thus, we have not been able to compare such outcomes here.
There were little missing outcome data. However, where data were missing, this was predominantly in the ‘No GoodSAM alert’ group [77/85 (91%) cases in London where survival to hospital discharge data was missing, and 29/40 (72.5%) cases in East Midlands]. This study suggests an independent association between a GoodSAM responder accepting an alert and improved survival to hospital discharge. The magnitude of the effect was similar in two different datasets from two different ambulance services, but the wide confidence intervals suggest imprecise results. Only 19–40% of the variability in the outcome measure was explained by the logistic regression model in London, and 11–29% in East Midlands. Additionally, many non-significant results in univariate analyses became significant in multivariate analysis. Multicollinearity tests suggest no association between covariates, but there are a few variables with small numbers. Our results may be spurious due to either chance or other confounding factors that we have not been able to account for fully in our modelling.
Our models may also be limited because survival to discharge has small samples within each of the GoodSAM response categories (e.g. London, n = 7, and East Midlands, n = 9, in the ‘accepted’ groups). Ideally, we would have used exact logistic models for the analyses, but such analysis was not possible with the software (SPSS) used to perform our statistical analyses.
Conclusion
We have reported a statistically significant improved survival to hospital discharge in the GoodSAM volunteer first-responder system, in two separate ambulance services, in analyses adjusted for factors that can influence cardiac arrest outcome. There is uncertainty in this finding. Alert acceptance rates were low, but there are substantial opportunities to increase the number of GoodSAM responders who accept an alert, and to optimize the interventions they provide thereafter.
Acknowledgements
For collating and providing access to data: Professor Mark Wilson, Deepti Bal, and Ali Ghorbangholi at GoodSAM; Scott Booth and Dr Chen Ji at the OHCAO registry, University of Warwick. For crucial insights into how GoodSAM was integrated with their systems: Christopher Hartley-Sharpe at LAS.
Data availability statement
The data underlying this article were provided by the Out-of-Hospital Cardiac Arrest Outcomes (OHCAO) registry at the University of Warwick, GoodSAM, London Ambulance Service and East Midlands Ambulance Service by permission. Data will be shared on reasonable request to the corresponding author with permission of the relevant party.
Funding
National Institute for Health Research (NIHR) Doctoral Research Fellowship (DRF-2017-10-095) to C.M.S. British Heart Foundation and Resuscitation Council UK to OHCAO Registry. This publication presents independent research funded by the National Institute for Health Research (NIHR).
Conflict of interest: C.M.S. was a National Institute for Health Research (NIHR) Doctoral Research Fellow (DRF-2017-10-095) between November 2017 and February 2021, including at the time of this study. The views expressed are those of the authors and not necessarily those of the NHS, the NIHR, or the Department of Health and Social Care. He also has volunteer roles at the Resuscitation Council UK and the International Liaison Committee on Resuscitation (ILCOR). G.D.P. is a NIHR Senior Investigator and editor for Resuscitation. He is supported by grant funding from British Heart Foundation and Resuscitation Council UK to support the OHCAO registry, paid to the University of Warwick. He is supported by grant funding from the NIHR Clinical Research Network, which supports Ambulance Services to contribute data to the OHCAO registry. He is supported by grant funding from the NIHR ARC West Midlands for an acute care programme, paid to the University of Warwick. He has volunteer roles at the Resuscitation Council UK and ILCOR. R.L., R.T.F., and R.S. have no conflicts of interest to declare.
References
- 1.OHCAO Registry. Out-of-Hospital Cardiac Arrest Outcomes Registry Epidemiology Report, 2019. https://warwick.ac.uk/fac/sci/med/research/ctu/trials/ohcao/publications/epidemiologyreports/ohcao_epidemiology_report_2019_-_england.pdf (7 October 2021).
- 2. Hawkes C, Booth S, Ji C, Brace-McDonnell SJ, Whittington A, Mapstone J, Cooke MW, Deakin CD, Gale CP, Fothergill R, Nolan JP, Rees N, Soar J, Siriwardena AN, Brown TP, Perkins GD; OHCAO Orators. Epidemiology and outcomes from out-of-hospital cardiac arrests in England. Resuscitation 2017;110:133–140. [DOI] [PubMed] [Google Scholar]
- 3. Kiguchi T, Okubo M, Nishiyama C, Maconochie I, Ong MEH, Kern KB, Wyckoff MH, McNally B, Christensen EF, Tjelmeland I, Herlitz J, Perkins GD, Booth S, Finn J, Shahidah N, Shin SD, Bobrow BJ, Morrison LJ, Salo A, Baldi E, Burkart R, Lin C-H, Jouven X, Soar J, Nolan JP, Iwami T.. Out-of-hospital cardiac arrest across the world: first report from the International Liaison Committee on Resuscitation (ILCOR). Resuscitation 2020;152:39–49. [DOI] [PubMed] [Google Scholar]
- 4. Yan S, Gan Y, Jiang N, Wang R, Chen Y, Luo Z, Zong Q, Chen S, Lv C.. The global survival rate among adult out-of-hospital cardiac arrest patients who received cardiopulmonary resuscitation: a systematic review and meta-analysis. Crit Care 2020;24:61. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Olasveengen TM, Semeraro F, Ristagno G, Castren M, Handley A, Kuzovlev A, Monsieurs KG, Raffay V, Smyth M, Soar J, Svavarsdottir H, Perkins GD.. European Resuscitation Council Guidelines 2021: basic life support and automated external defibrillation. Resuscitation 2021;161:98–114. [DOI] [PubMed] [Google Scholar]
- 6. Olasveengen TM, Mancini ME, Perkins GD, Avis S, Brooks S, Castrén M, Chung SP, Considine J, Couper K, Escalante R, Hatanaka T, Hung KKC, Kudenchuk P, Lim SH, Nishiyama C, Ristagno G, Semeraro F, Smith CM, Smyth MA, Vaillancourt C, Nolan JP, Hazinski MF, Morley PT, Svavarsdóttir H, Raffay V, Kuzovlev A, Grasner J-T, Dee R, Smith M, Rajendran K.. Adult basic life support: International Consensus on Cardiopulmonary Resuscitation and Emergency Cardiovascular Care Science with Treatment Recommendations. Resuscitation 2020;156:A35–A79. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Valeriano A, Van Heer S, de Champlain F, Brooks SC.. Crowdsourcing to save lives: a scoping review of bystander alert technologies for out-of-hospital cardiac arrest. Resuscitation 2021;158:94–121. [DOI] [PubMed] [Google Scholar]
- 8. Smith CM, Wilson MH, Ghorbangholi A, Hartley-Sharpe C, Gwinnutt C, Dicker B, Perkins GD.. The use of trained volunteers in the response to out-of-hospital cardiac arrest – the GoodSAM experience. Resuscitation 2017;121:123–126. [DOI] [PubMed] [Google Scholar]
- 9.Care Quality Commission (23 May 2018). London Ambulance Service NHS Trust inspection report. https://www.cqc.org.uk/sites/default/files/new_reports/AAAH3466.pdf (7 October 2021).
- 10.Care Quality Commission (17 July 2019). East Midlands Ambulance Service NHS Trust inspection report. https://api.cqc.org.uk/public/v1/reports/c489573c-97d4-415c-aa41-df3b27eecfd5?20210116011441 (7 October 2021).
- 11.Resuscitation Council UK (April 2018). CPR, AEDs and the law. https://www.resus.org.uk/library/publications/publication-cpr-aeds-and-law (7 October 2021).
- 12. Perkins GD, Jacobs IG, Nadkarni VM, Berg RA, Bhanji F, Biarent D, Bossaert LL, Brett SJ, Chamberlain D, de Caen AR, Deakin CD, Finn JC, Gräsner J-T, Hazinski MF, Iwami T, Koster RW, Lim SH, Ma MH-M, McNally BF, Morley PT, Morrison LJ, Monsieurs KG, Montgomery W, Nichol G, Okada K, Ong MEH, Travers AH, Nolan JP; Utstein Collaborators. Cardiac arrest and cardiopulmonary resuscitation outcome reports: update of the Utstein Resuscitation Registry templates for out-of-hospital cardiac arrest: a statement for healthcare professionals from a Task Force of the International Liaison Committee on Resuscitation (American Heart Association, European Resuscitation Council, Australian and New Zealand Council on Resuscitation, Heart and Stroke Foundation of Canada, InterAmerican Heart Foundation, Resuscitation Council of Southern Africa, Resuscitation Council of Asia); and the American Heart Association Emergency Cardiovascular Care Committee and the Council on Cardiopulmonary, Critical Care, Perioperative and Resuscitation. Resuscitation 2015;96:328–340. [DOI] [PubMed] [Google Scholar]
- 13. Pallant J. SPSS Survival Manual. 3rd ed. Maidenhead: Open University Press; 2007. [Google Scholar]
- 14. Barry T, Doheny MC, Masterson S, Conroy N, Klimas J, Segurado R, Codd M, Bury G.. Community first responders for out-of-hospital cardiac arrest in adults and children. Cochrane Database Syst Rev 2019;7:CD012764. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Ringh M, Rosenqvist M, Hollenberg J, Jonsson M, Fredman D, Nordberg P, Järnbert-Pettersson H, Hasselqvist-Ax I, Riva G, Svensson L.. Mobile-phone dispatch of laypersons for CPR in out-of-hospital cardiac arrest. N Engl J Med 2015;372:2316–2325. [DOI] [PubMed] [Google Scholar]
- 16. Stroop R, Kerner T, Strickmann B, Hensel M.. Mobile phone-based alerting of CPR-trained volunteers simultaneously with the ambulance can reduce the resuscitation-free interval and improve outcome after out-of-hospital cardiac arrest: a German, population-based cohort study. Resuscitation 2020;147:57–64. [DOI] [PubMed] [Google Scholar]
- 17. Pijls RWM, Nelemans PJ, Rahel BM, Gorgels APM.. A text message alert system for trained volunteers improves out-of-hospital cardiac arrest survival. Resuscitation 2016;105:182–187. [DOI] [PubMed] [Google Scholar]
- 18. Lee SY, Shin SD, Lee YJ, Song KJ, Hong KJ, Ro YS, Lee EJ, Kong SY.. Text message alert system and resuscitation outcomes after out-of-hospital cardiac arrest: a before-and-after population-based study. Resuscitation 2019;138:198–207. [DOI] [PubMed] [Google Scholar]
- 19. Scquizzato T, Pallanch O, Belletti A, Frontera A, Cabrini L, Zangrillo A, Landoni G.. Enhancing citizens response to out-of-hospital cardiac arrest: a systematic review of mobile-phone systems to alert citizens as first responders. Resuscitation 2020;152:16–25. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Andelius L, Malta Hansen C, Lippert FK, Karlsson L, Torp-Pedersen C, Kjær Ersbøll A, Køber L, Collatz Christensen H, Blomberg SN, Gislason GH, Folke F.. Smartphone activation of citizen responders to facilitate defibrillation in out-of-hospital cardiac arrest. J Am Coll Cardiol 2020;76:43–53. [DOI] [PubMed] [Google Scholar]
- 21. Scholten AC, van Manen JG, van der Worp WE, Ijzerman MJ, Doggen CJM.. Early cardiopulmonary resuscitation and use of Automated External Defibrillators by laypersons in out-of-hospital cardiac arrest using an SMS alert service. Resuscitation 2011;82:1273–1278. [DOI] [PubMed] [Google Scholar]
- 22. Brooks SC, Simmons G, Worthington H, Bobrow BJ, Morrison LJ.. The PulsePoint Respond mobile device application to crowdsource basic life support for patients with out-of-hospital cardiac arrest: challenges for optimal implementation. Resuscitation 2016;98:20–26. [DOI] [PubMed] [Google Scholar]
- 23. Berglund E, Claesson A, Nordberg P, Djärv T, Lundgren P, Folke F, Forsberg S, Riva G, Ringh M.. A smartphone application for dispatch of lay responders to out-of-hospital cardiac arrests. Resuscitation 2018;126:160–165. [DOI] [PubMed] [Google Scholar]
- 24. Stieglis R, Zijlstra JA, Riedijk F, Smeekes M, van der Worp WE, Koster RW.. AED and text message responders density in residential areas for rapid response in out-of-hospital cardiac arrest. Resuscitation 2020;150:170–177. [DOI] [PubMed] [Google Scholar]
- 25. Guyatt GH, Oxman AD, Vist GE, Kunz R, Falck-Ytter Y, Alonso-Coello P, Schünemann HJ; GRADE Working Group. GRADE: an emerging consensus on rating quality of evidence and strength of recommendations. BMJ 2008;336:924–926. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Smith CM, Griffiths F, Fothergill RT, Vlaev I, Perkins GD.. Identifying and overcoming barriers to Automated External Defibrillator use by GoodSAM volunteer first-responders in out-of-hospital cardiac arrest using the Theoretical Domains Framework and Behaviour Change Wheel. A qualitative study. BMJ Open 2020;10:e034908. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Zijlstra JA, Beesems SG, De Haan RJ, Koster RW.. Psychological impact on dispatched local lay rescuers performing bystander cardiopulmonary resuscitation. Resuscitation 2015;92:115–121. [DOI] [PubMed] [Google Scholar]
- 28. Ries ES, Kragh AR, Dammeyer J, Folke F, Andelius L, Hansen M.. C. Association of Psychological Distress, Contextual Factors, and Individual Differences among citizen responders. J Am Heart Assoc 2021;10:e020378. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Resuscitation Council UK. CPR in schools, 2020. https://www.resus.org.uk/public-resource/cpr-schools (7 October 2021).
- 30. Semeraro F, Greif R, Böttiger BW, Burkart R, Cimpoesu D, Georgiou M, Yeung J, Lippert F, Lockey AS, Olasveengen TM, Ristagno G, Schlieber J, Schnaubelt S, Scapigliati A, Monsieurs KG.. European Resuscitation Council Guidelines 2021. Systems saving lives. Resuscitation 2021;161:80–97. [DOI] [PubMed] [Google Scholar]
- 31.National Institute for Health Research. Funding and awards. Optimisation of the deployment of automatic external defibrillators in public places in England, 2019. https://fundingawards.nihr.ac.uk/award/NIHR127368 (7 October 2021).
- 32. Chan TCY, Li H, Lebovic G, Tang SK, Chan JYT, Cheng HCK, Morrison LJ, Brooks SC.. Identifying locations for public access defibrillators using mathematical optimization. Circulation 2013;127:1801–1809. [DOI] [PubMed] [Google Scholar]
- 33. Hansen CM, Wissenberg M, Weeke P, Ruwald MH, Lamberts M, Lippert FK, Gislason GH, Nielsen SL, Køber L, Torp-Pedersen C, Folke F.. Automated external defibrillators inaccessible to more than half of nearby cardiac arrests in public locations during evening, nighttime, and weekends. Circulation 2013;128:2224–2231. [DOI] [PubMed] [Google Scholar]
- 34. Sun CLF, Demirtas D, Brooks SC, Morrison LJ, Chan TCY.. Overcoming spatial and temporal barriers to public access defibrillators via optimization. J Am Coll Cardiol 2016;68:836–845. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.British Heart Foundation. The Circuit – The National Defibrillator Network, 2021. https://www.thecircuit.uk (7 October 2021).
- 36. Haywood K, Whitehead L, Nadkarni VM, Achana F, Beesems S, Böttiger BW, Brooks A, Castrén M, Ong MEH, Hazinski MF, Koster RW, Lilja G, Long J, Monsieurs KG, Morley PT, Morrison L, Nichol G, Oriolo V, Saposnik G, Smyth M, Spearpoint K, Williams B, Perkins GD; COSCA Collaborators. COSCA (Core Outcome Set for Cardiac Arrest) in adults: an advisory statement from the International Liaison Committee on Resuscitation. Resuscitation 2018;127:147–163. [DOI] [PubMed] [Google Scholar]
- 37. Maupain C, Bougouin W, Lamhaut L, Deye N, Diehl J-L, Geri G, Perier M-C, Beganton F, Marijon E, Jouven X, Cariou A, Dumas F.. The CAHP (Cardiac Arrest Hospital Prognosis) score: a tool for risk stratification after out-of-hospital cardiac arrest. Eur Heart J 2016;37:3222–3228. [DOI] [PubMed] [Google Scholar]
- 38. Karlsson L, Malta Hansen C, Wissenberg M, Møller Hansen S, Lippert FK, Rajan S, Kragholm K, Møller SG, Bach Søndergaard K, Gislason GH, Torp-Pedersen C, Folke F.. Automated external defibrillator accessibility is crucial for bystander defibrillation and survival: a registry-based study. Resuscitation 2019;136:30–37. [DOI] [PubMed] [Google Scholar]
- 39. Deakin CD, Anfield S, Hodgetts GA.. Underutilisation of public access defibrillation is related to retrieval distance and time-dependent availability. Heart 2018;104:1339–1343. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The data underlying this article were provided by the Out-of-Hospital Cardiac Arrest Outcomes (OHCAO) registry at the University of Warwick, GoodSAM, London Ambulance Service and East Midlands Ambulance Service by permission. Data will be shared on reasonable request to the corresponding author with permission of the relevant party.
Funding
National Institute for Health Research (NIHR) Doctoral Research Fellowship (DRF-2017-10-095) to C.M.S. British Heart Foundation and Resuscitation Council UK to OHCAO Registry. This publication presents independent research funded by the National Institute for Health Research (NIHR).
Conflict of interest: C.M.S. was a National Institute for Health Research (NIHR) Doctoral Research Fellow (DRF-2017-10-095) between November 2017 and February 2021, including at the time of this study. The views expressed are those of the authors and not necessarily those of the NHS, the NIHR, or the Department of Health and Social Care. He also has volunteer roles at the Resuscitation Council UK and the International Liaison Committee on Resuscitation (ILCOR). G.D.P. is a NIHR Senior Investigator and editor for Resuscitation. He is supported by grant funding from British Heart Foundation and Resuscitation Council UK to support the OHCAO registry, paid to the University of Warwick. He is supported by grant funding from the NIHR Clinical Research Network, which supports Ambulance Services to contribute data to the OHCAO registry. He is supported by grant funding from the NIHR ARC West Midlands for an acute care programme, paid to the University of Warwick. He has volunteer roles at the Resuscitation Council UK and ILCOR. R.L., R.T.F., and R.S. have no conflicts of interest to declare.