Geospatial Analysis for Targeting Out-of-Hospital Cardiac Arrest Intervention

      Background

      Bystander cardiopulmonary resuscitation (BCPR) provides an opportunity for decreasing cardiac mortality. Rates of out-of-hospital cardiac arrest (OHCA) in which resuscitation was performed vary within cities and across demographics.

      Purpose

      To identify contiguous geographic census tracts with high OHCA, low BCPR rates and high-risk demographics to effectively target culturally appropriate community-based intervention planning.

      Methods

      In 2012, a cohort of 11,389 emergency medical services (EMS) OHCA cases from Houston TX (2004–2011) was linked to census tracts. Multivariable logistic regression analyses were used to identify demographics of contiguous geographic census tracts with the highest OHCA rates. Within these tracts, BCPR rates were evaluated. The combination of information was used to develop a plan to better target interventions.

      Results

      Contiguous census tracts of high OHCA rates were identified; the average rate per 100,000 within versus outside the identified tracts is 106.0 (SD 23.7) to 55.8 (SD 19.7). Tracts with a low BCPR rate (37.7%) relative to a high OHCA rate were identified. In a separate analysis, individuals at highest relative risk of OHCA were found to be African Americans, to have low income or education levels, and to be older individuals. For every 1% increase in African Americans in a census tract, there is an increase of 2.7% in the relative risk of the census tract belonging to a high-OHCA-rate region (95% CI=2.0%, 3.5%).

      Conclusions

      Geospatial analysis can provide important information on the contiguous areas of high OHCA rates and low BCPR rates with the aim of more effectively targeting interventions and ultimately decreasing cardiac deaths.

      Background

      Approximately 300,000 people in the U.S. experience an out-of-hospital cardiac arrest (OHCA) each year, and more than 90% of those who experience an OHCA die.
      • McNally B.
      • Robb R.
      • Mehta M.
      • et al.
      Out-of-hospital cardiac arrest surveillance—Cardiac Arrest Registry to Enhance Survival (CARES), U.S., October 1, 2005–December 31, 2010. National Center for Chronic Disease Prevention and Health Promotion.
      Although there has been a decline in sudden cardiac arrest and overall heart disease mortality, currently the improvements are more likely attributable to prevention of heart disease as opposed to improvements in resuscitation outcomes.
      • Rea T.D.
      • Page R.L.
      Community approaches to improve after out-of-hospital sudden cardiac arrest.
      There is a 10% decrease in chance of survival for every minute a person experiencing an OHCA is left unattended, making early intervention critical to improving resuscitation outcomes. Reducing emergency medical services (EMS) response time would improve early intervention, but the earliest intervention is likely from a bystander. Bystander cardiopulmonary resuscitation (BCPR) is effective; one life is saved for every 26–36 people who receive it.
      • Valenzuela T.D.
      • Roe D.J.
      • Cretin S.
      • Spatite D.W.
      • Larsen M.P.
      Estimating effectiveness of cardiac arrest interventions: a logistic regression survival model.
      • Sasson C.
      • Rogers M.A.
      • Dahl J.
      • Kellermann A.L.
      Predictors of survival from out-of-hospital cardiac arrest: a systematic review and meta-analysis.
      • Cummins R.O.
      • Omato J.P.
      • Thies W.H.
      • Pepe P.E.
      Improving survival from sudden cardiac arrest: “chain of survival” concept.
      Taken as a whole, these facts indicate that though continuing to focus on prevention and EMS response, BCPR provides an opportunity for decreasing cardiac mortality. Geospatial analysis can provide a way to optimize efforts by serving as a tool to help identify large-scale target areas for interventions to reach the population most in need.
      Although a few smaller individual city studies (n ranging from 537 to 4379) and a large multi-city study have found that OHCA rates vary between and within cities,
      • Lerner B.E.
      • Fairbanks R.J.
      • Shah M.N.
      Identification of out-of-hospital arrest clusters using a geographic information system.
      • Reinier K.
      • Thomas E.
      • Andrusiek D.L.
      • et al.
      Resuscitation Outcomes Consortium Investigators. Socioeconomic status and incidence of sudden cardiac arrest.
      • Sasson C.
      • Keirns C.C.
      • Smith D.
      • et al.
      Small area variation in out-of-hospital cardiac arrest: does the neighborhood matter?.
      • Sasson C.
      • Kierns C.C.
      • Smith D.M.
      • et al.
      Examining the contextual effects of neighborhood on out-of-hospital cardiac arrest and the provision of bystander cardiopulmonary resuscitation.
      • Sasson C.
      • Magid D.J.
      • Chan P.
      • et al.
      CARES Surveillance Group
      Association of neighborhood characteristics with bystander-initiated CPR.
      and that BCPR rates vary within a city and by demographics,
      • Lerner B.E.
      • Fairbanks R.J.
      • Shah M.N.
      Identification of out-of-hospital arrest clusters using a geographic information system.
      • Reinier K.
      • Thomas E.
      • Andrusiek D.L.
      • et al.
      Resuscitation Outcomes Consortium Investigators. Socioeconomic status and incidence of sudden cardiac arrest.
      • Sasson C.
      • Keirns C.C.
      • Smith D.
      • et al.
      Small area variation in out-of-hospital cardiac arrest: does the neighborhood matter?.
      • Sasson C.
      • Magid D.J.
      • Chan P.
      • et al.
      CARES Surveillance Group
      Association of neighborhood characteristics with bystander-initiated CPR.
      • Ong M.E.
      • Earnest A.
      • Shahidah N.
      • Ng W.M.
      • Foo C.
      • Nott D.J.
      Spatial variation and geographic-demographic determinants of out-of-hospital cardiac arrests in the city-state of Singapore.
      • Iwashyna T.J.
      • Christakis N.A.
      • Becker L.B.
      Neighborhoods matter: a population-based study of provision of cardiopulmonary resuscitation.
      • Galea S.
      • Blaney S.
      • Nandi A.
      • et al.
      Explaining racial disparities in incidence of and survival from out-of-hospital cardiac arrest.
      the current paper adds three important aspects to the literature. First, because of the number of cases (n=11,389) and size of the city, it adds to the body of evidence regarding the presence of spatial variation and clustering in rates of OHCA and BCPR. Second, it describes a focused way to use geospatial analysis specifically to target areas of high OHCA and low BCPR and how geospatial analysis can be used to identify the demographics in a way that can help target interventions for at-risk populations. Finally, this paper describes geospatial analysis as a practical means of identifying baseline measures of OHCA, BCPR, and potentially, trends in intervention effectiveness.

      Methods

      Study Design and Setting

      All cases in which EMS personnel perform chest compressions are considered OHCA cases. In 2012, the OHCA data were taken from the Houston Fire Department EMS calls over the 8-year period from 2004 to 2011. The database consists of n=11,389 qualified cases of OHCA. Rice University and Baylor College of Medicine IRBs approved all data-collecting procedures for human subjects.

      Selection of Participants

      The study included all qualified OHCA patients aged ≥18 years presenting to EMS during the study period. Excluded were cases in which chest compressions were not initiated because the adults were considered dead on arrival as defined by decapitation; rigor mortis; dependent lividity; decomposition; incineration; or obvious mortal wounds; absence of any signs of life (pulse, respirations, or any spontaneous movement) on EMS arrival associated with a penetrating head injury (e.g., gunshot wound, stab); or penetrating extremity injury with obvious exsanguination, absence of any signs of life (pulse, respirations, or any spontaneous movement) for >5 minutes associated with a penetrating injury to the chest or abdomen and a >10-minute transport time to a trauma center, or absence of any signs of life (pulse, respirations, or any spontaneous movement) associated with blunt trauma and cases outside of the city service area.

      Measurement

      Emergency medical services data included the latitude and longitude of the event, the physical address, and whether the event was witnessed and BCPR was administered. In addition to recording the time and location of the event, other relevant information necessary for age, gender, and race stratification also was recorded electronically. This information was collected by EMS using Utstein guidelines.
      • Jacobs I.
      • Nadkarni V.
      ILCOR Task Force on Cardiac Arrest and Cardiopulmonary resuscitation Outcomes. Cardiac arrest and cardiopulmonary resuscitation outcome reports: update and simplification of the Utstein templates for resuscitation registries.
      If latitude and longitude were missing, the geocoded location was found using ArcGIS, version 10. Locations were complete for >98% of cases.
      Emergency medical services dispatch initiates an incidence record that is updated electronically with patient care information by responding EMS personnel. This record then is transferred electronically to a data warehouse where the information is compiled to create the patient care record. Responding EMS personnel also contact telemetry for medical control. Telemetry transcribes this information in a second separate database recording the incidence number, the interventions, intervention times, provision of BCPR, and other pertinent information. The data warehouse is used to cross-check and validate the number of OHCA events in the telemetry data.
      Bystander cardiopulmonary resuscitation (BCPR) is defined as anyone performing CPR other than an on-duty EMS member. Each fire department and EMS unit responding to a CPR case completes a record. Respondents usually include an engine company, ambulance, paramedic squad, and supervisor. In some instances, a Medic ambulance responds instead of an ambulance and a squad. Each of the three or four respondents is required to document whether BCPR was performed prior to firefighters taking over that duty. In addition, a report is made on each cardiac arrest at the base station documenting BCPR (or not). Therefore, BCPR is documented via five channels. The cases of cardiac arrest and information regarding provision of BCPR are recorded in the electronic database in unique fields titled “working assessment” and “BCPR,” respectively.
      Census tract information was obtained using the 2010 Census Summary File 1 and the 2006–2010 ACS 5-year estimates.

      FILES: Census 2010 Summary File 1, Houston, Texas. Prepared by Social Explorer. www.socialexplorer.com.

      FILES: 2006–2010 American Community Survey, Houston, Texas. Prepared by Social Explorer. www.socialexplorer.com.

      Two of the original 597 census tracts were omitted due to a population of fewer than 1000 residents. A third tract was omitted due to a large transient population workforce not captured in the data. Physical locations visited multiple times by EMS (e.g., senior citizen facilities) were retained but tracked through the analysis to guard against potential bias.

      Data Analysis

      Statistical analyses were conducted using SAS, version 9.3 and R Development Core Team, version 2.14.0. To identify spatial clusters of census tracts of high rates of incidence of OHCA, census tract annual rate of incidence was calculated, and adjoining census tracts with rates in the highest quartile in the city overall were identified as high-rate clusters. Calculation of rates by census tract assumes that OHCA occurred near the residence, consistent with research findings.
      • Lerner B.E.
      • Fairbanks R.J.
      • Shah M.N.
      Identification of out-of-hospital arrest clusters using a geographic information system.
      • Ong M.E.
      • Earnest A.
      • Shahidah N.
      • Ng W.M.
      • Foo C.
      • Nott D.J.
      Spatial variation and geographic-demographic determinants of out-of-hospital cardiac arrests in the city-state of Singapore.
      To account for instability of the observed data where census tract populations are small, an empirical Bayes method for stabilizing crude incidence rates (annual event average divided by tract population) was used. The empirical Bayes adjusted rates use the SD of the EMS assessed rates from nine cities as the prior estimate of variability.
      • Reinier K.
      • Thomas E.
      • Andrusiek D.L.
      • et al.
      Resuscitation Outcomes Consortium Investigators. Socioeconomic status and incidence of sudden cardiac arrest.
      • Sasson C.
      • Keirns C.C.
      • Smith D.
      • et al.
      Small area variation in out-of-hospital cardiac arrest: does the neighborhood matter?.
      • Devine O.J.
      • Luis T.A.
      • Halloran M.E.
      Empirical Bayes method for stabilizing incidence rates before mapping.
      The Houston OHCA incidence rate listed below is the total OHCA rate averaged over 8 years divided by the total population. Variations in the reported incidence of OHCA differ by location, in part because of definition and ascertainment of cardiac arrest data.
      • Rea T.D.
      • Eisenberg M.S.
      • Sinibaldi G.
      • White R.D.
      Stability of rates within census tracts was assessed by calculating the Spearman correlation of adjusted rates over the 8-year period.
      To distinguish demographic characteristics of the high-rate cluster of OHCA incidence from the city at large, census tract demographic data were used as independent variables in a logistic regression dichotomizing on cluster membership. The ratio of the BCPR rate to the adjusted OHCA rate was examined, to identify census tracts with both high-OHCA and low-BCPR rates.
      • Sasson C.
      • Keirns C.C.
      • Smith D.
      • et al.
      Small area variation in out-of-hospital cardiac arrest: does the neighborhood matter?.

      Results

      Figure 1 identifies the location of OHCA events for the 8-year period; 265 cases were eliminated because of missing information. Census tracts from the two major airports, and the business center in downtown Houston, were eliminated from the regression a priori because these OHCA patients were not considered to be representative of the census tract demographics. The adjusted overall OHCA rate (i.e., not by census tract) for the city of Houston was 59.7 per 100,000 people compared with the unadjusted rate listed previously as 48.0 per 100,000 people. Summary information regarding the study population OHCA and BCPR rates and other census tract demographics is provided in Table 1.
      Figure thumbnail gr1
      Figure 1OHCA events in Houston for 2004–2011
      EMS, emergency medical services; OHCA, out-of-hospital cardiac arrest
      Table 1Statistics of the census tracts in emergency medical services area in Houston TX, 2004–2011
      Characteristics of HoustonValue
      Number of OHCAs11,389
      Total population, n2,963,912
      Number of census tracts594
      Race, n (%)
      Caucasian1,551,790 (52)
      African-American648,496 (22)
      Other768,034 (26)
      Gender, n (%)
      Male1,482,595 (50)
      Female1,485,725 (50)
      OHCA rate per 100,00048.9 (59.7)
      Number in parentheses is the adjusted rate. BCPR, bystander cardiopulmonary resuscitation; OHCA, out-of-hospital cardiac arrest
      BCPR OHCA overall ratio39
      Characteristics across census tracts
      Less than high school education, average % (SD)25.7 (18.6)
      Average median age, years (SD)33.4 (5.8)
      Earn <$10,000/year, average % (SD)8.2 (7.0)
      Note: 2010 Census and Houston emergency medical services study data.
      a Number in parentheses is the adjusted rate. BCPR, bystander cardiopulmonary resuscitation; OHCA, out-of-hospital cardiac arrest
      Adjoining census tracts with OHCA empirical Bayesian adjusted incidence rates in the highest quartile, corresponding to 77 per 100,000 and above, were identified (Figure 2). These census tract rates remained stable over the 8-year period, with the exception of some perturbation in the 2005 tracts, likely attributable to Hurricane Katrina. Exclusion of 2005 data from the subsequent analyses does not affect the results.
      Figure thumbnail gr2
      Figure 2OHCA high-rate region for 2004–2011
      EMS, emergency medical services; OHCA, out-of-hospital cardiac arrest
      The demographics of the high-rate region compared with the rest of the city include a higher percentage of African Americans, an older population as represented by median age, a higher percentage of individuals without a high school degree, and a higher percentage earning <$10,000 annually (Figure 3). Similarly, for every 1-year increase in median age in a census tract, there is an increase of 11.3% in the relative risk that the tract is in the high-rate region (95% CI=7.3%, 15.5%). Likewise, for a 1% increase in the percentage of people in a tract without a high school education or tract earning <$10,000 per year, there is an increase in the relative risk of being in the high-rate region by 3.2% (95% CI=1.7%, 4.8%) and 4.4% (95% CI=2.2%, 6.7%), respectively.
      Figure thumbnail gr3
      Figure 3Demographics inside and outside of the high-rate region
      The average OHCA rate per 100,000 across census tracts within the high-rate region is 106.0 (SD 23.7%); outside the high-rate region the average OHCA rate is 55.8 (SD 19.7). The same census tract demographics associated with the high-rate region are found to be significant descriptors (p<0.0001 in all cases) of high OHCA rates overall by the multivariable regression model.
      The BCPR rates vary throughout the region with an overall rate of 39%. The mean rate over census tracts was 37.7%, with a large SD of 20.6%. Figure 4 is a graph of the BCPR rate compared with the OHCA rate. The size of the dot is proportional to the number of residents in the census tract. The tracts highlighted in black represent the greatest opportunity for community intervention (i.e., high OHCA and low BCPR rates). Tracts also were identified where EMS responded multiple times to senior care locations.
      Figure thumbnail gr4
      Figure 4Plot of census tract OHCA rate versus census tract BCPR rate
      Note: The size of the dot is proportional to the number of residents in the census tract. The tracts highlighted in black represent the greatest opportunity for community intervention (i.e., high OHCA and low BCPR rates). CPR, cardiopulmonary resuscitation; BCPR, bystander cardiopulmonary resuscitation; OHCA, out-of-hospital cardiac arrest

      Discussion

      Detection of elevated clusters of disease is needed for monitoring, etiology, or early warning.
      • Patil G.P.
      • Taillie C.
      Upper level set scan statistic for detecting arbitrarily shaped hotspots.
      The results of the analysis can be used as a generic tool in disease prevention and management
      • Waller L.
      Methods for detecting disease clustering in time or space.
      to ensure that educational efforts are targeted where they are needed rather than where it is convenient. The visual identification of high-rate clusters found in the study promoted discussions between the local health department and the community leaders and ultimately led to training and intervention.
      With 11,389 cases, this study represents the most comprehensive city-specific examination of the distribution of OHCA incidents and BCPR rates within a metropolitan region.
      • Lerner B.E.
      • Fairbanks R.J.
      • Shah M.N.
      Identification of out-of-hospital arrest clusters using a geographic information system.
      • Reinier K.
      • Thomas E.
      • Andrusiek D.L.
      • et al.
      Resuscitation Outcomes Consortium Investigators. Socioeconomic status and incidence of sudden cardiac arrest.
      • Sasson C.
      • Magid D.J.
      • Chan P.
      • et al.
      CARES Surveillance Group
      Association of neighborhood characteristics with bystander-initiated CPR.
      • Iwashyna T.J.
      • Christakis N.A.
      • Becker L.B.
      Neighborhoods matter: a population-based study of provision of cardiopulmonary resuscitation.
      • Nichol G.
      • Thomas E.
      • Callaway C.W.
      • et al.
      Resuscitation Outcomes Consortium Investigators. Regional variation in out-of-hospital cardiac arrest incidence and outcome.
      In this work, spatial variation in OHCA and BCPR rates was found. The high-rate census tracts were found to be contiguous, clustering in a high-rate region traversing the city, and this high-rate region was characterized by specific demographics.
      Although aspects of this approach have been explored separately (i.e., identification of clusters,
      • Lerner B.E.
      • Fairbanks R.J.
      • Shah M.N.
      Identification of out-of-hospital arrest clusters using a geographic information system.
      • Sasson C.
      • Keirns C.C.
      • Smith D.
      • et al.
      Small area variation in out-of-hospital cardiac arrest: does the neighborhood matter?.
      • Sasson C.
      • Magid D.J.
      • Chan P.
      • et al.
      CARES Surveillance Group
      Association of neighborhood characteristics with bystander-initiated CPR.
      high OHCA and low BCPR,
      • Lerner B.E.
      • Fairbanks R.J.
      • Shah M.N.
      Identification of out-of-hospital arrest clusters using a geographic information system.
      • Sasson C.
      • Keirns C.C.
      • Smith D.
      • et al.
      Small area variation in out-of-hospital cardiac arrest: does the neighborhood matter?.
      • Sasson C.
      • Kierns C.C.
      • Smith D.M.
      • et al.
      Examining the contextual effects of neighborhood on out-of-hospital cardiac arrest and the provision of bystander cardiopulmonary resuscitation.
      • Ong M.E.
      • Earnest A.
      • Shahidah N.
      • Ng W.M.
      • Foo C.
      • Nott D.J.
      Spatial variation and geographic-demographic determinants of out-of-hospital cardiac arrests in the city-state of Singapore.
      and demographics of high-risk individuals
      • Reinier K.
      • Thomas E.
      • Andrusiek D.L.
      • et al.
      Resuscitation Outcomes Consortium Investigators. Socioeconomic status and incidence of sudden cardiac arrest.
      • Sasson C.
      • Kierns C.C.
      • Smith D.M.
      • et al.
      Examining the contextual effects of neighborhood on out-of-hospital cardiac arrest and the provision of bystander cardiopulmonary resuscitation.
      • Sasson C.
      • Magid D.J.
      • Chan P.
      • et al.
      CARES Surveillance Group
      Association of neighborhood characteristics with bystander-initiated CPR.
      • Ong M.E.
      • Earnest A.
      • Shahidah N.
      • Ng W.M.
      • Foo C.
      • Nott D.J.
      Spatial variation and geographic-demographic determinants of out-of-hospital cardiac arrests in the city-state of Singapore.
      • Iwashyna T.J.
      • Christakis N.A.
      • Becker L.B.
      Neighborhoods matter: a population-based study of provision of cardiopulmonary resuscitation.
      • Galea S.
      • Blaney S.
      • Nandi A.
      • et al.
      Explaining racial disparities in incidence of and survival from out-of-hospital cardiac arrest.
      ), this work is the first to use all the tools systematically with the specific intent of implementing a BCPR training campaign based on the results. These tools serve as a guide to health-based community programs by identifying where efforts may prove most beneficial and focusing energies on neighborhoods and subpopulations at greatest risk, with the goal of reducing cardiac death. Houston was chosen as the city to be used in this research in order to target the neighborhoods and sensitive subpopulations at increased risk of OHCA and low provision of BCPR. City community outreach staff members have established relationships with community stakeholders, leaders, and neighborhood-based organizational leads, who trust them and typically support activities that are viewed as being for the “greater good.”
      The staff first presented slides depicting the high-rate OHCA and low-rate BCPR areas to these leaders to enhance willingness to advocate and initiate BCPR training. Then, together with the community leaders, the staff began public dissemination of the OHCA and BCPR statistics coupled with intensive BCPR training in these areas via churches and community centers. The American Heart Association Friend and Family CPR Anytime kits were used. Each person receiving a kit was asked to train 10–15 additional community members, whether friends, family, or fellow churchgoers. Within 2 weeks, follow-up (via phone and e-mail) was conducted to gather information regarding number and ZIP code of additional community members trained. Additional community members trained represents 20% of the total individuals trained in this initiative to date.

      Limitations

      Some difficulties and considerations occur in this type of study. One of the first difficulties noted was the need to validate the EMS service area as it changed over the years. The size of the service area changes as the city annexes surrounding areas and expands EMS to that area. Records were not easily attainable for the early years. Another difficulty was recognizing and controlling for elements not on target with the point of the analysis identification of neighborhood clusters to target BCPR. Given this objective, a concern was bias owing to calls to an area that were not represented in the census: assisted living centers, airports, business centers, and the influx of temporary refugees from a nearby city devastated by a major hurricane.
      Comparison of OHCA rates among cities was difficult because of variation in EMS policies of when to attempt resuscitation. Information that would have been useful but was either unavailable or incomplete was whether the event happened at the victim’s residence (Wu et al.
      • Wu L.A.
      • Kottke T.E.
      • Brekke L.N.
      • et al.
      Opportunities to prevent sudden out-of-hospital death due to coronary heart disease in a community.
      found that only 50% of cases were at the residence) or in the same census tract as the residence and a record of cardiac-related pre-existing conditions.

      Conclusion

      Beginning in 2010, the study area began to provide dispatcher-assisted compression-only CPR instructions. Future work includes tracking the efficacy of these instructions and the targeted BCPR training, compared with the baseline established in this research and expanding the use of this tool for other interventions and health endpoints.

      Acknowledgements

      This work was supported by Houston Endowment and the City of Houston. The authors acknowledge Laura Campos and Jiao Li for their dedication and hard work on this study.
      No financial disclosures were reported by the authors of this paper.

      References

        • McNally B.
        • Robb R.
        • Mehta M.
        • et al.
        Out-of-hospital cardiac arrest surveillance—Cardiac Arrest Registry to Enhance Survival (CARES), U.S., October 1, 2005–December 31, 2010. National Center for Chronic Disease Prevention and Health Promotion.
        MMWR Surveill Summ. 2011; 60: 1-19
        • Rea T.D.
        • Page R.L.
        Community approaches to improve after out-of-hospital sudden cardiac arrest.
        Circulation. 2010; 121: 1134-1140
        • Valenzuela T.D.
        • Roe D.J.
        • Cretin S.
        • Spatite D.W.
        • Larsen M.P.
        Estimating effectiveness of cardiac arrest interventions: a logistic regression survival model.
        Circulation. 1997; 96: 3308-3313
        • Sasson C.
        • Rogers M.A.
        • Dahl J.
        • Kellermann A.L.
        Predictors of survival from out-of-hospital cardiac arrest: a systematic review and meta-analysis.
        Circ Cardiovasc Qual Outcomes. 2010; 3: 63-81
        • Cummins R.O.
        • Omato J.P.
        • Thies W.H.
        • Pepe P.E.
        Improving survival from sudden cardiac arrest: “chain of survival” concept.
        Circulation. 1991; 83: 1832-1847
        • Lerner B.E.
        • Fairbanks R.J.
        • Shah M.N.
        Identification of out-of-hospital arrest clusters using a geographic information system.
        Acad Emerg Med. 2005; 12: 81-84
        • Reinier K.
        • Thomas E.
        • Andrusiek D.L.
        • et al.
        Resuscitation Outcomes Consortium Investigators. Socioeconomic status and incidence of sudden cardiac arrest.
        Can Med Assoc J. 2011; 183: 1705-1712
        • Sasson C.
        • Keirns C.C.
        • Smith D.
        • et al.
        Small area variation in out-of-hospital cardiac arrest: does the neighborhood matter?.
        Ann Intern Med. 2010; 153: 19-22.
        • Sasson C.
        • Kierns C.C.
        • Smith D.M.
        • et al.
        Examining the contextual effects of neighborhood on out-of-hospital cardiac arrest and the provision of bystander cardiopulmonary resuscitation.
        Resuscitation. 2011; 82: 674-679
        • Sasson C.
        • Magid D.J.
        • Chan P.
        • et al.
        • CARES Surveillance Group
        Association of neighborhood characteristics with bystander-initiated CPR.
        N Engl J Med. 2012; 367: 1607-1615
        • Ong M.E.
        • Earnest A.
        • Shahidah N.
        • Ng W.M.
        • Foo C.
        • Nott D.J.
        Spatial variation and geographic-demographic determinants of out-of-hospital cardiac arrests in the city-state of Singapore.
        Ann Emerg Med. 2011; 58: 343-351
        • Iwashyna T.J.
        • Christakis N.A.
        • Becker L.B.
        Neighborhoods matter: a population-based study of provision of cardiopulmonary resuscitation.
        Ann Emerg Med. 1999; 34: 459-468
        • Galea S.
        • Blaney S.
        • Nandi A.
        • et al.
        Explaining racial disparities in incidence of and survival from out-of-hospital cardiac arrest.
        Am J Epidemiol. 2007; 166: 534-543
        • Jacobs I.
        • Nadkarni V.
        ILCOR Task Force on Cardiac Arrest and Cardiopulmonary resuscitation Outcomes. Cardiac arrest and cardiopulmonary resuscitation outcome reports: update and simplification of the Utstein templates for resuscitation registries.
        Resuscitation. 2004; 63: 233-249
      1. FILES: Census 2010 Summary File 1, Houston, Texas. Prepared by Social Explorer. www.socialexplorer.com.

      2. FILES: 2006–2010 American Community Survey, Houston, Texas. Prepared by Social Explorer. www.socialexplorer.com.

        • Devine O.J.
        • Luis T.A.
        • Halloran M.E.
        Empirical Bayes method for stabilizing incidence rates before mapping.
        Epidemiology. 1994; 5: 622-630
        • Rea T.D.
        • Eisenberg M.S.
        • Sinibaldi G.
        • White R.D.
        Incidence of EMS-treated out-of-hospital cardiac arrest in the U.S. Resuscitation. 2004; 63: 17-24
        • Patil G.P.
        • Taillie C.
        Upper level set scan statistic for detecting arbitrarily shaped hotspots.
        Environ Ecological Stat. 2004; 11: 183-197
        • Waller L.
        Methods for detecting disease clustering in time or space.
        Monitoring the Health of Populations: Statistical Principles and Methods for Public Health Surveillance. 2003; https://doi.org/10.1093/acprof:oso/9780195146493.003.0007
        • Nichol G.
        • Thomas E.
        • Callaway C.W.
        • et al.
        Resuscitation Outcomes Consortium Investigators. Regional variation in out-of-hospital cardiac arrest incidence and outcome.
        J Am Med Assoc. 2008; 300: 1423-1431
        • Wu L.A.
        • Kottke T.E.
        • Brekke L.N.
        • et al.
        Opportunities to prevent sudden out-of-hospital death due to coronary heart disease in a community.
        Resuscitation. 2003; 56: 55-58