Delays in Emergency Care and Mortality during Major U.S. Marathons

Abstract

Background: Large marathons frequently involve widespread road closures and infrastructure disruptions, which may create delays in emergency care for nonparticipants with acute medical conditions who live in proximity to marathon routes.

Methods: We analyzed Medicare data on hospitalizations for acute myocardial infarction or cardiac arrest among Medicare beneficiaries (≥65 years of age) in 11 U.S. cities that were hosting major marathons during the period from 2002 through 2012 and compared 30-day mortality among the beneficiaries who were hospitalized on the date of a marathon, those who were hospitalized on the same day of the week as the day of the marathon in the 5 weeks before or the 5 weeks after the marathon, and those who were hospitalized on the same day as the marathon but in surrounding ZIP Code areas unaffected by the marathon. We also analyzed data from a national registry of ambulance transports and investigated whether ambulance transports occurring before noon in marathon-affected areas (when road closures are likely) had longer scene-to-hospital transport times than on nonmarathon dates. We also compared transport times on marathon dates with those on nonmarathon dates in these same areas during evenings (when roads were reopened) and in areas unaffected by the marathon.

Results: The daily frequency of hospitalizations was similar on marathon and nonmarathon dates (mean number of hospitalizations per city, 10.6 and 10.5, respectively; P=0.71); the characteristics of the beneficiaries hospitalized on marathon and nonmarathon dates were also similar. Unadjusted 30-day mortality in marathon-affected areas on marathon dates was 28.2% (323 deaths in 1145 hospitalizations) as compared with 24.9% (2757 deaths in 11,074 hospitalizations) on nonmarathon dates (absolute risk difference, 3.3 percentage points; 95% confidence interval, 0.7 to 6.0; P=0.01; relative risk difference, 13.3%). This pattern persisted after adjustment for covariates and in an analysis that included beneficiaries who had five or more chronic medical conditions (a group that is unlikely to be hospitalized because of marathon participation). No significant differences were found with respect to where patients were hospitalized or the treatments they received in the hospital. Ambulance scene-to-hospital transport times for pickups before noon were 4.4 minutes longer on marathon dates than on nonmarathon dates (relative difference, 32.1%; P=0.005). No delays were found in evenings or in marathon-unaffected areas.

Conclusions: Medicare beneficiaries who were admitted to marathon-affected hospitals with acute myocardial infarction or cardiac arrest on marathon dates had longer ambulance transport times before noon (4.4 minutes longer) and higher 30-day mortality than beneficiaries who were hospitalized on nonmarathon dates. (Funded by the National Institutes of Health.).

Figure 1. Unadjusted 30-Day Mortality among Medicare Beneficiaries Hospitalized for Acute Myocardial Infarction or Cardiac Arrest on Marathon Dates and Nonmarathon Dates in Marathon-Affected Hospitals and Control Hospitals

Panel A shows the unadjusted 30-day mortality according to hospitalizations on the marathon date (week 0) and in the weeks relative to the marathon date (i.e., the same day of the week as the day of the marathon in the 5 weeks before and the 5 weeks after the marathon). Hospitals located in ZIP Code areas through which the marathon route passed were defined as marathon-affected hospitals. Control hospitals were defined as all hospitals in the hospital referral regions that surrounded the hospital referral region in which a marathon-affected hospital was located. The dashed lines represent 95% confidence intervals, which account for clustering of admissions within hospitals. Panel B shows the unadjusted 30-day mortality among Medicare beneficiaries hospitalized on marathon dates and nonmarathon dates in marathon-affected hospitals and control hospitals. The I bars represent 95% confidence intervals.

Figure 2. Adjusted 30-Day Mortality among Medicare Beneficiaries Hospitalized for Acute Myocardial Infarction or Cardiac Arrest on Marathon Dates and Nonmarathon Dates in Marathon-Affected Hospitals and Control Hospitals

The analysis was adjusted for beneficiary sex, age, race, any of 10 chronic medical conditions, the median household income in the ZIP Code area, the interaction between city and marathon day, and fixed effects for hospital. Separate models were estimated for marathon-affected and control hospitals. The I bars represent 95% confidence intervals.

Figure 3. Ambulance Travel Times from Scene to Hospital on a Marathon Date and Nonmarathon Dates in Marathon-Affected Areas and Areas Not Affected by the Marathon

Morning was defined as the period from 3 a.m. through noon, when road closures attributable to marathons would be more likely. Evening was defined as the period from 7 p.m. through 11:59 p.m., when roads should be reopened. Transport times were computed for persons 70 years of age or older, which was chosen to reflect the mean age of Medicare beneficiaries with acute myocardial infarction and cardiac arrest. During the morning, the mean unadjusted difference in transport time between a marathon day and combined nonmarathon days in marathon-affected areas was 4.4 minutes (95% CI, 1.3 to 7.5) (P = 0.005 by a Student’s t-test). There was no significant difference in the evening (−1.1 minutes; 95% CI, −6.0 to −3.8; P = 0.65). Owing to a lack of geographic information in National Emergency Medical Services Information System data, the analysis could not account for clustering of admissions within hospitals or cities. The dashed lines represent 95% confidence intervals. Sensitivity analyses of morning and age cutoffs are provided in the Supplementary Appendix.