Public Access Defibrillators and Socioeconomic Factors on the Small-Scale Spatial Level in Berlin

A Cross-Sectional Analysis

Background: The use of a public access defibrillator (PAD) increases the probability of surviving an out-of-hospital cardiac arrest (OHCA). No strategies exist, however, for the optimal distribution of PADs in an urban area in order to meet existing needs and ensure equal access for all potential users. It thus seems likely that the accessibility of PADs on the spatial level varies widely as a function of living circumstances.

Methods: This cross-sectional study is based on registry data concerning PAD (2022, n = 776) and OHCA (2018–2020, n = 4051), along with data on socioeconomic factors on the spatial level in Berlin (12 districts and 137 subdistricts). Associations of socioeconomic factors with the number of PADs per 10 000 inhabitants and the PAD coverage rate of sites of previous OHCAs were investigated.

Results: The median number of PADs per 10 000 inhabitants ranged from 0.46 to 2.67 at the district level, and only five districts had a median PAD coverage rate of sites of previous OHCAs above 0%, after aggregation of the analyses at the subdistrict level. Subdistricts with a more favorable economic status and a greater income disparity had a higher PAD density. Socially disadvantaged subdistricts had no association with PAD density.

Conclusion: There are large deficits in the distribution of PADs at the small-scale spatial level in Berlin with respect to the goals of meeting existing needs and ensuring equal access for all potential users. The findings presented here will be of importance for the planning of future PAD programs so that the distributional efficiency and fairness of PAD in urban areas can be improved.

Out of hospital cardiac arrest (OHCA) constitutes a serious public health problem. In Germany in 2020, some 60 000 cases of OHCA occurred in which rescue services attempted resuscitation (1). To improve the probability of survival and prevent neurological sequelae and further long-term sequelae, cardiopulmonary resuscitation—including the early use of an automated external defibrillator—is urgently required.

A systematic review of 41 studies showed that 40% of OHCA patients would have survived up to the time of hospital discharge, had they been defibrillated before the arrival of the rescue team. Defibrillation by lay first aiders using defibrillators installed on site led to a much higher proportion of surviving patients than defibrillation provided by professional first aiders with mobile devices (53% versus 29%) (2). A multicenter study from the USA and Canada showed, furthermore, that OHCA patients with shockable rhythms often had a better functional result at hospital discharge if they had been defibrillated by lay first aiders than by rescue services (57% versus 33%) (3). The reason is that laypersons can be on site more quickly. Strategies for the deployment of lay first aiders who use at the site of the incident an automated external defibrillator are therefore recommended to improve the prognosis of OHCA (4).

Studies from abroad have shown that socioeconomic characteristics of the OHCA incident site are associated with the speediness and completeness of the resuscitation chain. In areas with a higher socioeconomic status, OHCA patients were resuscitated in more cases (5, 6, 7, 8) and transferred to hospital more quickly (9). In Philadelphia and Seoul, neighborhoods with a greater mean household income and a larger proportion of the population with a higher educational qualification had more public access defibrillators (PADs) per person, and PADs were more common at incident sites of earlier OHCA (10, 11). The preconditions for the association between socioeconomic factors and elements of the resuscitation chain can, however, differ between regions and cultures.

In Germany, the provision of PADs has so far barely been scientifically studied. The available empirical data are fragmentary (12). The installation of PADs in public and private spaces is mostly not done on a demand basis. Strategies for a demand-based and equal-opportunities access to PADs are mostly lacking. Concepts for provision of PADs are likely to differ by region. Accordingly, the availability of PADs at the spatial level can vary substantially by living conditions. We can assume in particular that previous incident sites of OHCA are not considered to a satisfactory degree and that socioeconomically less well off neighborhoods are disadvantaged. Our study aims to investigate whether and to what extent the provision of PADs in Berlin is associated with the socioeconomic status and income distribution of urban small-scale spaces.

Methods

The present study is a cross-sectional study at a small-scale spatial level in Berlin. The lowest analysis level are the district regions (n=137), which are allocated to 12 Berlin districts. Because of the small population size (n=89) and data protection considerations we did not consider the district region Forst Grunewald.

Data sources

We obtained geo data of PADs from two sources: DefiNetz and Berlin Schockt. The installation of PADs is reported to these registries on a voluntary basis. When we accessed DefiNetz on 10 February 2022 we identified 661 PADs and when we accessed Berlin Schockt on 7 February 2022, we identified 322 PADs. After deduplication (n=201) and after excluding PADs in airports (n=4) and with incomplete geographical coordinates, we included 776 PADs in the analyses.

Geo data of OHCA were provided by the IGNIS i3web control center database system of the Berlin fire brigade. In the time period from 1 June 2018 through 10 February 2020, 4060 OHCA incidents confirmed by the rescue services were registered in the Berlin district regions. After excluding OHCA with unknown addresses (n=9), 4051 OHCA were included in the study.

The data of exposures of interests—that is, socioeconomic status and income distribution at the small-scale spatial level—come from two regional reports (Monitoring Soziale Stadtentwicklung Berlin 2017 [13], Regionaler Sozialbericht Berlin und Brandenburg 2017 [14]) and a response from the Berlin Senate Department for Finances to a query from 2017 (15). Social data from 2017 ensure a chronological sequence of exposures and study end points.

Geospatial data allocation

A spatial grid of 100×100 m was used to map Berlin’s area initially into small-scale spatial fields (longitude 0.00154 and latitude 0.0009, according to World Geodetic System 1984 and by using the software package QGIS Version 3.4 QGIS Development Team, Zurich, Switzerland). For the area of Berlin this resulted in a grid map consisting of 86 484 cells. Geographical data from earlier OHCA incidents from the Berlin fire brigade and installed PADs from DefiNetz and Berlin Schockt were integrated into this grid map.

Definition of risk grids

A grid cell is defined as an OHCA risk grid if at least one OHCA occurred in this grid cell between June 2018 and February 2020. If at least one PAD had been installed within a 100 meter radius of the OHCA risk grid in February 2022, this grid cell is defined as covered.

Primary and secondary study endpoints

The primary endpoint was the number of installed PADs per 10 000 population at the level of the district region. Population numbers were as of 31 December 2018 (16). The secondary study endpoint was the coverage share of covered OHCA risk grids in the district regions, defined as the share (%) of covered OHCA risk grids among all OHCA risk grids.

Socioeconomic factors in the district regions

Social status was captured by applying a multidimensional deprivation index based on information on unemployment, proportion of non-EU foreign citizens, and housing data (persons per living space) at the level of the district regions (13). Economic status was determined on the basis of the mean income tax per capita (15) and the income distribution on the basis of the Gini coefficient (14) at the district level and then allocated to the included district regions. The Gini coefficient is a measure of income inequality on a scale from 0 (expresses perfect equality) to 1 (expresses maximal inequality). All three socioeconomic variables were divided into quartiles on the basis of the distribution. The eBox contains a detailed description of the definitions of the variables.

eBox

Definitions of exposure variables

Study hypothesis

The study hypothesis was specified a priori: the higher the socioeconomic level, measured with the deprivation index, and the lower the income inequality, measured with the Gini coefficient, the more PADs per head of population are available and the greater the coverage rate of the OHCA risk grids (research proposal for the registration of a masters thesis on 11 March 2020 at the Berlin School of Public Health, Charité). Income taxes were included in the analysis subsequently as a socioeconomic factor.

Statistical analyses

In the descriptive analysis, characteristics of the districts were described, and the number of PADs per 10 000 residents as well as the coverage rate of the OHCA risk grids were visualized according to natural breaks (Jenks optimization method) and according to quartiles of socioeconomic status and income distribution. Classification according to Jenks minimizes the variance within categories and maximizes the variance between categories (17). The variables are reported as medians and interquartile ranges. Generalized linear regression models with Poisson link function (for number of PADs per 10 000 residents) and negative binomial link function (for coverage rate of OHCA risk grids) were developed on the basis of the data distribution, so as to study the association between these target variables and the explanatory variables (socioeconomic status, income distribution) on the basis of the district regions. A proxy variable for densely populated district regions is integrated into these models, defined on the basis of the locations of Intercity-Express train stations and shopping centers, so as to account for a difference in visits to the district regions. In a sensitivity analysis of the models for the number of PADs for every 10 000 residents, we furthermore adjusted for population density of the district regions. In the regression models we estimated adjusted rate ratios with 95% confidence intervals for quartiles of socioeconomic status and income distribution. We used the software package R Version 3.6.3. (R Foundation for Statistical Computing, Vienna, Austria) for our statistical analysis

Results

Table 1 shows study data by district. The median number of PADs per 10 000 residents was between 0.46 and 2.67 in the districts, after aggregation of the analyses at the level of the district regions. In only five districts was the median coverage rate of the OHCA risk grids above 0%, relative to the district regions. Figures 1 and 2 show the district regions after natural breaks (Jenks method) of the PAD density and coverage rate. Altogether 3465 grids in the 137 district regions were classified as OHCA risk grids, and only 141 of these in 64 district regions were covered by PADs.

Figure 1

District regions by natural breaks (Jenks method) of the number of PADs per 10 000 population

Figure 2

District regions by natural breaks (Jenks method) in the coverage rate of OHCA risk grids

Table 1

Description of study data by district

Association with socioeconomic factors

Table 2 shows the number of PAD for every 10 000 residents and the coverage rate of the OHCA risk grids by quartiles of the socioeconomic factors of the district regions.

Table 2

Number of PADs per 10 000 population and coverage rate of OHCA risk grids by quartiles of socioeconomic factors of the district regions

With the Gini coefficient of the district regions, PAD density rises from 0.80/10 000 population in quartile 1 to 1.56/10 000 population in quartile 4. Regarding income tax and deprivation index, no positive or negative association with the PAD density was seen.

Slight differences in the coverage rates of the OHCA risk grids were seen for consideration by quartiles of the deprivation index, the tax amount per head, and the Gini coefficient.

Table 3 shows the results of the regression analyses of socioeconomic factors with the primary and secondary study endpoint. Associations between income tax per capita and the number of PADs/10 000 residents are inconsistent; the PAD density is higher in quartiles 3 and 4 and lower in quartile 2. An increase in the Gini coefficient, however, is associated with an increasing PAD density across the four quartiles. No association was seen between the deprivation index and PAD density at the level of the district regions. After additionally adjusting for population density, only slight changes were seen in the point estimates (eTable).

Table 3

Rate ratios and 95% confidence intervals for associations of socioeconomic factors with the study endpoints at the level of the district regions

eTable

Rate ratios and 95% confidence intervals for associations of socioeconomic factors with the study endpoints at the level of the district regions (sensitivity analysis)

For the secondary endpoint, the coverage rate of OHCA risk grids, the tax amount per head shows a positive association. No consistent trend was seen across the quartiles of the Gini coefficient and the deprivation index; a high Gini coefficient is associated with a higher coverage rate and a privileged deprivation index with a lower coverage rate.

Discussion

The present study was the first to investigate associations between the availability of PADs and socioeconomic characteristics at the spatial level in Germany. On the basis of data from Berlin, the study shows that district regions with an income tax amount above the median (quartile 3 and 4), but also with a great income inequality have a higher PAD density. Social status, measured on the basis of unemployment, proportion of foreigners, and housing situation of the district regions is, however, not associated with PAD density. The study furthermore shows a very low PAD coverage rate of earlier OHCA incident sites (time period June 2018 to February 2020); the observed associations with socioeconomic factors should therefore be interpreted with caution.

Installation of PADs

Our data indicate that provision of PADs in Berlin is very low per head of population compared with other large cities. Only 776 PADs are available in all of Berlin for 3.6 million residents (2.16 PADs/10 000 population). In cities or regions with PAD programs—such as Copenhagen (30.5 PADs/10 000 population) (18) and the northern Netherlands (6.61 PADs/10 000 population) (19), PAD density is much higher.

Furthermore, the installation of PADs in Berlin does not follow the sites with the greatest confirmed OHCA risk based on earlier incident sites. A closer look at our data shows that most of the PADs (634 PADs, 81.7% of all PADs) are not located within a 100 meter radius of earlier OHCA incident sites (time period June 2018 to February 2020). Furthermore, 51 grids (6.6% of all grids with PAD) had more than one PAD and 96 risk grids (2.8% of all risk grids) had no PAD at all, in spite of the fact that more than three OHCA incidents had occurred there. Redistributing the existing 776 PADs into risk grids—those where previous OHCA incidents occurred—could raise the coverage rate of the OHCA risk grids from 3.86% to 22% and therefore enable a much better efficiency in the distribution.

The reason for these deficits is likely to be the current installation strategy for PADs, because PADs are installed purely on the basis of private initiatives. Demand planning and specifications for private or public institutions of when PADs should be established, are lacking. The suggested redistribution is therefore purely theoretical.

Guidelines for the provision of PADs

PADs should be installed within 100 meters of an OHCA incident site, within 1–2 minutes’ walking distance of a first aider (20). The 2015 guidelines of the European Resuscitation Council (ERC) recommend positioning PADs on sites where at least one OHCA incident had taken place in the preceding five years (21). The five year period should be understood as a threshold for prioritizing installation sites in a scenario of limited resources. Our analytic approach to PAD installation sites is derived from this recommendation. With the two year period between OHCA registration (June 2018 to February 2020) and PAD registration we allowed a latency period for the installation of PADs. Because of this latency period and available OHCA data over 20 months, our study period is shorter than five years.

The current ERC guidelines from 2021 recommend a PAD density of two PADs per km² (22). The extent to which these recommendations are put into practice should be the subject of future studies. One PAD installation per km² should provide advantages regarding accessibility over installation per head of population, and equal access to a PAD per km² is likely to be of equal relevance.

Socioeconomic factors and provision of public access defibrillators

The result regarding income tax amounts is mostly consistent with results from earlier studies that investigated an association between economic power and PAD accessibility at the spatial level (10, 11). A higher amount of tax or greater economic power enables the expensive installation and regular maintenance of PADs (23, 24).

The result of our study vis-à-vis the deprivation index differs from earlier study results from abroad. Studies in New Zealand and Scotland found a low density of PADs in areas with a higher level of deprivation (25, 26). The deprivation indexes applied there, however, combined social and economic indicators, whereas our deprivation index is based on social indicators alone.

Earlier studies showed that greater income inequality at the spatial level is associated with higher mortality (27) and a poor prognosis for cardiovascular disorders (28, 29). Wherever social inequality is pronounced, it is not merely those at an individual disadvantage who suffer, but everybody. For this reason we formulated as our original hypothesis that a high Gini coefficient impairs the installation of PADs at the small-scale spatial level. Our results do, however, show a positive association between income inequality and PAD provision. A closer look at the spatial distribution shows that a low Gini coefficient is found in districts with a low level of tax; segregation by income in the districts thus might explain this reverse association.

Strengths and limitations

Our study provides initial insights from Germany regarding the provision of PADs by socioeconomic characteristics at the small-scale spatial level. The results provide a data foundation for initiatives to improve access to PADs in the urban space. The developed geospatial model provides an opportunity to study further aspects of the resuscitation chain, such as the use of first aid apps.

Further to the named strengths, our study is subject to the following limitations.

• Because of the voluntary registration to the PAD registries, the geo data may be incomplete.
• In contrast to earlier studies, our deprivation index is based only on social indicators.
• Population movement as a possible confounding factor is considered in our analysis only by means of a proxy variable.
• Our study period for analyzing earlier OHCA incident sites was less than five years. For this reason, the number of risk grids is an underestimate in the sense of the 2015 ERC guidelines.
• The generalizability to other regions in Germany may well be limited.

Approaches to solution and conclusions

The strategic positioning and increase in the density/number of PADs could be attempted, for example, by installing these at points of intersection in near-distance public transport (bus and tram stops) and locations close to neighborhoods (squares, parks, residential developments) with geolocalization in first aider alarm systems. Such an approach should facilitate the installation of PADs in urban small-scale spaces with low densities of PADs as well as for frequent occurrences of OHCA incidents and while considering socioeconomic factors. In view of earlier experiences that regulations for fire extinguishers and sprinkler installations guarantee equal measures to protect against fires (30), a legal mandate to install PADs and the associated demand planning is likely to facilitate equal access to PADs. An evidence based strategy in consideration of our results is required in order to pursue the aims of greater efficiency in the distribution of and equal opportunities in the accessibility of PADs at the small-scale spatial level.

Acknowledgment

The authors thank DefiNetz and Berlin Schockt for providing the geo data from public access defibrillators.

Conflict of interest statement
The authors declare that no conflict of interests exist.

Manuscript received on 5 November 2021, revised version accepted on 30 March 2022.

Translated from the original German by Birte Twisselmann, PhD.

Corresponding author
Dokyeong Lee, MSc
Institut für Public Health, Charité – Universitätsmedizin Berlin
Charitéplatz 1, 10117 Berlin

Cite this as:
Lee D, Stiepak JK, Pommerenke C, Poloczek S, Grittner U, Prugger C: Public access defibrillators and socioeconomic factors on the small-scale spatial level in Berlin—a cross-sectional analysis. Dtsch Arztebl Int 2022; 119: 393–9. DOI: 10.3238/arztebl.m2022.0180

►Supplementary material

eTable, eBox:
www.aerzteblatt-international.de/m2022.0180

References

1. Fischer M, Wnent J, Gräsner JT, et al.: Jahresbericht des Deutschen Reanimationsregisters. Außerklinische Reanimation 2020. Anästh Intensivmed 2021; 62: V68–73.

2. Bækgaard JS, Viereck S, Møller TP, Ersbøll AK, Lippert F, Folke F: The effects of public access defibrillation on survival after out-of-hospital cardiac arrest—a systematic review of observational studies. Circulation 2017; 136: 954–65

3. Pollack RA, Brown SP, Rea T, et al.: Impact of bystander automated external defibrillator use on survival and functional outcomes in shockable observed public cardiac arrest. Circulation 2018; 137: 2104–13

4. Nolan JP, Maconochie I, Soar J, et al.: Executive summary: 2020 international consensus on cardiopulmonary resuscitation and emergency cardiovascular care science with treatment recommendation. Circulation 2020; 142: 2–27

5. Sasson C, Kerins CC, Smith DM, et al.: Examining the contextual effects of neighborhood on out-of-hospital cardiac arrest and the provision of bystander cardiopulmonary resuscitation. Resuscitation 2011; 82: 674–9

6. Rivera NT, Kumar SL, Bhandari RK, Kumar SD: Disparities in survival with bystander CPR following cardiopulmonary arrest based on neighbourhood characteristics. Emerg Med Int 2016; 2016: 6983750

7. Nassel AF, Root ED, Haukoos JS, et al.: Multiple cluster analysis for the identification of high-risk census tracts for out-of-hospital cardiac arrest (OHCA) in Denver, Colorado. Resuscitation 2014; 85: 1667–73

8. Root ED, Gonzales L, Persse DE, et al.: A tale of two cities: the role of neighborhood socioeconomic status in spatial clustering of bystander CPR in Austin and Houston. Resuscitation 2013; 84: 752–9

9. Hsia RY, Huang D, Mann NC, et al.: A US national study of the association between income and ambulance response time in cardiac arrest. JAMA Netw Open 2018; 1: e185202

10. Griffis HM, Band RA, Ruther M, et al.: Employment and residential characteristics in relation to automated external defibrillator location. Am Heart J 2016; 172: 185–91

11. Lee SY, Do YK, Shin SD, et al.: Community socioeconomic status and public access defibrillators: a multilevel analysis. Resuscitation 2017; 120: 1–7

12. Trappe HJ: Plötzlicher Herztod und automatisierte externe Defibrillatoren – Wo stehen wir 2012. Herz 2012; 37: 416–23

13. Senatsverwaltung für Stadtentwicklung und Wohnen Berlin (ed.): Monitoring Soziale Stadtentwicklung 2017. Berlin. www.stadtentwicklung.berlin.de/planen/basisdaten_stadtentwicklung/monitoring/download/2017/Monitoring_Soziale_Stadtentwicklung_2017-Bericht.pdf (last accessed on 7 April 2022).

14. Amt für Statistik Berlin-Brandenburg (ed.): Regionaler Sozialbericht Berlin und Brandenburg 2017. Potsdam 2017. https://digital.zlb.de/viewer/metadata/15601789_2017/1/LOG_0003/ (last accessed on 7 April 2022).

15. Perdoni S: Überraschende Zahlen: So viel Steuern zahlen die Berliner. Berliner Zeitung (21.08.2018). www.berliner-zeitung.de/berlin/ueberraschende-zahlen-so-viel-steuern-zahlen-die-berliner-31139632 (last accessed on 16 October 2019).

16. Senatsverwaltung für Stadtentwicklung und Wohnen Berlin (ed.): Monitoring Soziale Stadtentwicklung 2019. Berlin 2019. www.stadtentwicklung.berlin.de/planen/basisdaten_stadtentwicklung/monitoring/download/2019/MSS_Fortschreibung2019_Langfassung.pdf (last accessed on 7 April 2022).

17. Jenks GF, Caspall FC: Error on choroplethic maps: definition, measurement, reduction. Ann Assoc Am Geogr 1971; 61: 217–44

18. Karlsson L, Hansen CM, Wissenberg M, et al.: Automated external defibrillator accessibility is crucial for bystander defibrillation and survival: a registry-based study. Resuscitation 2019; 136: 30–7

19. Berdowski J, Blom MT, Bardai A, Tan HL, Tijssen JG, Koster RW: Impact of onsite or dispatched automated external defibrillator use on survival after out-of-hospital cardiac arrest. Circulation 2011; 124: 2225–32

20. Chan TCY, Li H, Lebovic G, et al.: Identifying locations for public access defibrillators using mathematical optimization. Circulation 2013; 127: 1801–9

21. Perkins GD, Handley AJ, Koster RW, et al.: European resuscitation council guidelines for resuscitation 2015 section 2 adult basic life support and automated external defibrillation. Resuscitation 2015; 95: 81–99

22. Semero F, Greif R, Böttiger BW, et al.: European Resuscitation Council Guidelines 2021: systems saving lives. Resuscitation 2021; 161: 80–97

23. Kuisma M, Castren M, Nurminen K: Public access defibrillation in Helsinki-costs and potential benefits from a community-based pilot study. Resuscitation 2003; 56: 149–52

24. Neumann PJ, Cohen JT, Weinstein MC: Updating cost-effectiveness–the curious resilience of the $ 50,000-per-QALY threshold. N Engl J Med 2014; 371: 796–7

25. Dicker B, Garrett N, Wong S, et al.: Relationship between socioeconomic factors, distribution of public access defibrillators and incidence of out-of-hospital cardiac arrest. Resuscitation 2019; 138: 53–8

26. Leung KHB, Brooks SC, Clegg GR, Chan TCY: Socioeconomically equitable public defibrillator placement using mathematical optimization. Resuscitation 2021; 166: 14–20

27. Kennedy BP, Kawachi I, Prothrow-Stith D: Income distribution and mortality: cross sectional ecological study of the Robin Hood index in the United States. BMJ 1996; 312: 1004–7

28. Lindenauer PK, Lagu T, Rothberg MB, et al.: Income inequality and 30 day outcomes after acute myocardial infarction, heart failure, and pneumonia: retrospective cohort study. BMJ 2013; 346: f521

29. Dewan P, Rørth R, Jhund PS, et al.: Income inequality and outcomes in heart failure: a global between-country analysis. JACC Heart Fail 2019; 7: 336–46

30. Mell HK, Sayre MR: Public access defibrillators and fire extinguishers: are comparisons reasonable? Prog Cardiovasc Dis 2008; 51: 204–12