Lights Out: Impact of the August 2003 Power Outage on Mortality in New York, NY

Anderson, G. Brookea; Bell, Michelle L.b

doi: 10.1097/EDE.0b013e318245c61c

Background: Little is known about how power outages affect health. We investigated mortality effects of the largest US blackout to date, 14–15 August 2003 in New York, NY.

Methods: We estimated mortality risk in New York, NY, using a generalized linear model with data from 1987–2005. We incorporated possible confounders, including weather and long-term and seasonal mortality trends.

Results: During the blackout, mortality increased for accidental deaths (122% [95% confidence interval = 28%–287%]) and nonaccidental (ie, disease-related) deaths (25% [12%–41%]), resulting in approximately 90 excess deaths. Increased mortality was not from deaths being advanced by a few days; rather, mortality risk remained slightly elevated through August 2003.

Conclusions: To our knowledge, this is the first analysis of power outages and nonaccidental mortality. Understanding the impact of power outages on human health is relevant, given that increased energy demand and climate change are likely to put added strain on power grids.

Author Information

From the aDepartment of Biostatistics, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD; and bSchool of Forestry and Environmental Studies, Yale University, New Haven, CT.

Submitted 24 April 2011; accepted 25 October 2011; posted online 17 January 2012.

Supported by the National Institute for Environmental Health Sciences (R01-ES015028), the US Environmental Protection Agency through the Johns Hopkins Particulate Matter Research Center (RD-83241701) and Harvard Clean Air Center (EPA RD 83479801), and an US EPA STAR Graduate Fellowship (91689201-0). The authors reported no other financial interests related to this research.

Supplemental digital content is available through direct URL citations in the HTML and PDF versions of this article ( This content is not peer-reviewed or copy-edited; it is the sole responsibility of the author.

Correspondence Brooke Anderson, Room E3137, 615 North Wolfe St, Baltimore, MD 21205. E-mail:

Article Outline

Power blackouts are likely to increase, as growing energy use stresses aging power grids.1 Climate change could also increase power outages,2 and energy infrastructure may be targeted by national security threats.1 However, we know little about how power outages affect health. Studies have reported increases in accidental deaths and injuries, including carbon monoxide (CO) poisoning,3,4 food poisoning,5 and hypothermia3 during power outages, as well as increased respiratory hospitalizations.6 Calls to emergency services7 and poison control8 can increase, although this may reflect the inability to contact primary health providers.7 Although one article showed graphically that mortality increased during the New York 2003 blackout,6 to our knowledge, there has been no analysis of the effect of blackouts on total mortality, including nonaccidental (ie, disease-related) causes. Furthermore, estimated death tolls for disasters (eg, hurricanes) generally include only directly attributable deaths (eg, drowning),9,10 even though disasters often also cause power outages. If outages affect mortality from nontraumatic causes, current fatality estimates for natural disasters could be underestimated, particularly for developed countries where disaster-related deaths are considered relatively rare.

We investigated mortality in New York, NY, during the largest blackout in US history (14–15 August 2003).11 This was the first citywide blackout of New York, NY, since 19777 and constituted a natural experiment—power was available until 4:11 PM on August 14, and then immediately unavailable throughout the community.7 We compare population mortality on blackout days with days with power.

Back to Top | Article Outline



Mortality data were obtained from the National Center for Health Statistics12 through the National Morbidity, Mortality, and Air Pollution Study13 for the New York, NY, community for 1987–2005. Data include daily mortality counts by age (<65, 65–74, ≥75 years) and cause of death: accidental, cardiovascular, respiratory, and noncardiorespiratory (eTable1,

Temperature and dew-point temperature data are from the National Climatic Data Center.14 We used selected monitors operating over the entire study period (eTable 2 and eFigure 1, and averaged daily data across all reporting monitors to estimate community-wide weather. We adjusted dew-point temperature by temperature,15 as these variables were highly correlated (R2 = 0.94).

Because pollution can affect mortality,16 data for ozone (O3), nitrogen dioxide (NO2), sulfur dioxide (SO2), particulate matter with aerodynamic diameter ≤10 μm (PM10), and CO were obtained for the months of August in 2001–2005 from the US Environmental Protection Agency's Air Quality System.17 We included only those monitors reporting during 14–16 August 2003, excluding source-oriented monitors (eFigure 1 and eTable 3,

Back to Top | Article Outline

We modeled daily mortality rates using a Poisson distribution with overdispersion, based on methods used to study health effects of air pollution and heat waves.13,15 We modeled blackout days using an indicator variable, and controlled for potential time-varying confounders, including day of week, temperature, dew-point temperature, and long-term and seasonal mortality trends (details in the eAppendix, estimated the relative risk of mortality associated with the power outage, comparing mortality on blackout days with other days. We tested sensitivity of results using a model with temperature at lags ranging from same day to 4 days previous. In this model, we included temperature lags using a distributed lag model with strata: same day and average of lags 1–4.18

To estimate number of deaths related to the blackout, we calculated baseline mortality on blackout days (the expected mortality had the blackout not occurred) and then estimated excess deaths associated with the blackout (relative risk associated with the blackout multiplied by expected mortality). Estimates of uncertainty for excess mortality incorporated uncertainty in the blackout's association with mortality and in baseline mortality. We estimated effects by age group and cause of death through stratified analysis.

Separately, we considered changes in mortality rates through 31 August 2003 to investigate evidence of short-term mortality displacement.19 Following methods used to study mortality displacement after heat waves,20 we fit a spline to a 29-day period centered on 15 August 2003. We placed knots more closely near the blackout because effects likely change more quickly at times closer to the event.21

Because heat waves can affect mortality beyond single days of heat,15,22,23 we investigated whether the blackout coincided with a heat wave, defined as ≥2 consecutive days with temperatures ≥98th percentile of year-round temperatures in New York, NY.15,22

Previous work using airborne measurements found decreased pollution in the Northeast during the blackout, possibly from power plant shutdowns.24 We compared ground-level measurements of O3, NO2, SO2, PM10, and CO during the blackout with distributions of typical levels for the area and season (August 2001–2002 and 2004–2005).

Back to Top | Article Outline


During the blackout, total mortality rose 28%, resulting in approximately 90 excess deaths (Fig. 1, Table). Effect estimates were almost identical using a more complex model that included temperature effects at longer lags (eTable 4, Risk of accidental deaths increased most; however, excess deaths were primarily from nonaccidental causes. Although all age groups were affected, those 65–74 years were particularly susceptible (Table). We found no evidence of short-term mortality displacement; rather, mortality remained slightly elevated through August 2003.

Temperatures were hot but not extreme for a typical August day in New York (eFigure 2, Neither the 14th nor 15th of August was a heat wave day, making adjustment for added heat wave effects unnecessary.

Pollution monitors were interrupted during the blackout; therefore, we could not analyze pollution for the afternoon of 14 August through the morning of 15 August 2003. However, there are data for the morning of 14 August and immediately following power restoration (15 August). We compared pollutant concentrations for all available data on 14–16 August 2003 with distributions of typical levels (August 2001–2002, 2004–2005) (Fig. 2).

Pollution remained at normal August levels for O3 (Fig. 2A) and PM10 (Fig. 2E). SO2 concentrations were high (top 75th percentile of usual values) on the morning of 16 August 2003 (Fig. 2B). NO2 was higher than usual in the evening following power restoration (15 August 2003) (Fig. 2C). CO was close to normal at 1 of the 2 monitors but exceptionally high during the afternoon of 15 August 2003 at a monitor in Manhattan (Fig. 2D).

Back to Top | Article Outline


Our results indicate that power outages can immediately and severely harm human health. We found a much higher mortality than reported by the New York City Department of Health and Mental Hygiene, which counted 6 deaths directly attributable to the blackout (the majority from CO poisoning).25 Although an earlier study graphically depicted mortality rates during the blackout,6 there has been no analysis of the effects of a blackout on disease-related mortality risk or estimates of excess deaths associated with the August 2003 New York blackout. We found that most excess deaths during the blackout were from disease-related causes.

The conditions reported during the blackout help explain our mortality estimates. Some people were trapped in subways over an hour,26 and firefighters performed an estimated 800 elevator rescues.11 Residents of high-rise apartments were isolated27 and lacked potable water.25 Those who left higher floors had to descend many flights of stairs—one documented mortality was a heart attack after leaving a high-level office.28

The blackout also complicated the management of illness, which may help explain the range of causes of death. Most food sources and pharmacies were closed—a serious problem for patients with diabetes and anyone low on prescription medicines.28,29 Some power-operated home medical equipment (eg, ventilators, oxygen conservers) could not be used.30 Ambulances responded more slowly than usual,7 and because cellular phone service failed during part of the blackout,31 it was difficult to contact emergency services. Four of the 75 hospitals in New York, NY, lost power for part of the blackout,27 and hospital emergency rooms with power were overcrowded with people seeking electricity for medical equipment.27

Both heat and air pollution can affect mortality.16,23 Our estimate of excess mortality during the blackout cannot be explained solely by heat. We controlled for temperature in the model, and the blackout did not coincide with a heat wave. However, increased exposure to heat during the power outage (ie, lack of air conditioning, increased pedestrian traffic) could have aggravated the normal effects of heat. Because pollution monitors were interrupted, we were unable to fully analyze pollutant levels. However, we found elevated levels of some pollutants on the day following the blackout. Although this blackout may have decreased pollution from long-range sources such as power generators in the Midwest,24 local pollution sources were also affected, including traffic, public transportation, airlines, and generator use.26,27,31 Unusually high pollution levels were likely caused by the blackout, and therefore, they were probably pathways rather than confounders in the relationship between the blackout and mortality.

Among US cities, New York, NY, may be particularly vulnerable because of its many high-rise buildings and substantial dependence on public transportation. Other power outages have lasted longer or been accompanied by more extreme weather.3,4 Future research may determine whether health effects observed during this blackout are found during other blackouts and in other communities. Our findings indicate that health impacts of power outages have been underestimated, which has implications for emergency planning, as well as for assessment of the full impact of national disasters and other events related to blackouts.

Back to Top | Article Outline


1. Amin SM, Gellings CW. The North American power delivery system: balancing market restructuring and environmental economics with infrastructure security. Energy. 2006;31:967–999.
2. Miller NL, Hayhoe K, Jin J, Auffhammer M. Climate, extreme heat, and electricity demand in California. J Appl Meteorol. 2008;47:1834–1844.
3. Broder J, Mehrotra A, Tintinalli J. Injuries from the 2002 North Carolina ice storm, and strategies for prevention. Injury. 2005;36:21–26.
4. Daley WR, Smith A, Argandona E, Malilay J, McGeehin M. An outbreak of carbon monoxide poisoning after a major ice storm in Maine. J Emerg Med. 2000;18:87–93.
5. Marx MA, Rodriguez CV, Greenko J, et al.. Diarrheal illness detected through syndromic surveillance after a massive power outage: New York City, August 2003. Am J Public Health. 2006;95:547–553.
6. Lin S, Fletcher BA, Luo M, Chinery R, Hwang SA. Health impact in New York City during the Northeastern Blackout of 2003. Public Health Rep. 2011;126:384–393.
7. Freese J, Richmond NJ, Silverman RA, Braun J, Kaufman BJ, Clair J. Impact of a citywide blackout on an urban Emergency Medical Services system. Prehosp Disaster Med. 2006;21:372–378.
8. Klein KR, Herzog P, Smolinske S, White SR. Demand for poison center services ‘surged’ during the 2003 blackout. Clin Toxicol. 2007;45:248–254.
9. Centre for Research on the Epidemiology of Disasters. EM-DAT: The OFDA/CRED International Disaster Database. Available at: Accessed 19 April 2011.
10. National Climatic Data Center. National Climatic Data Center Storm Event Database. Available at: Accessed 19 April 2011.
11. Barron J. Lights go on after biggest blackout, but not without 2nd day of suffering. New York Times. 2003.
12. National Center for Health Statistics. Available at: Accessed 19 April 2011.
13. Samet JM, Dominici F, Curriero FC, Coursac I, Zeger SL. Fine particulate air pollution and mortality in 20 U.S. Cities, 1987–1994. N Engl J Med. 2000;343:1742–1749.
14. National Climatic Data Center. NCDC's GIS Daily Surface Data. Available at: Accessed 19 April 2011.
15. Anderson GB, Bell ML. Heatwaves in the United States: mortality risk during heatwaves and effect modification by heatwave characteristics in 43 US communities. Environ Health Perspect. 2011;119:210–218.
16. Bell ML, McDermott A, Zeger SL, Samet JM, Dominici F. Ozone and short-term mortality in 95 US communities, 1987–2000. JAMA. 2004;294:2372–2378.
17. US Environmental Protection Agency. US EPA's Air Quality System. Available at: Accessed 19 April 2011.
18. Gasparrini A, Armstrong B, Kenward MG. Distributed lag non-linear models. Stat Med. 2010;29:2224–2234.
19. Hajat S, Armstrong BG, Gouveia N, Wilkinson P. Mortality displacement of heat-related deaths: a comparison of Delhi, Sao Paulo, and London. Epidemiology. 2005;16:613–620.
20. Hertel S, Le Tertre A, Jockel KH, Hoffmann B. Quantification of the heat wave effect on cause-specific mortality in Essen, Germany. Eur J Epidemiol. 2009;24:407–414.
21. Pattenden S, Nikiforov B, Armstrong B. Mortality and temperature in Sofia and London. J Epidemiol Community Health. 2003;57:628–633.
22. Hajat S, Armstrong B, Baccini M, et al.. Impact of high temperatures on mortality: is there an added heat wave effect? Epidemiology. 2006;17:632–638.
23. Anderson GB, Bell ML. Weather-related mortality: how heat, cold, and heat waves affect mortality in the United States. Epidemiology. 2009;20:205–213.
24. Marufu LT, Taubman BF, Bloomer B, et al.. The 2003 North American electrical blackout: an accidental experiment in atmospheric chemistry. Geophys Res Lett. 2004;31:L13106.
25. Bureau of Vital Statistics, New York City Department of Health and Mental Hygiene. Summary of the Vital Statistics 2003. New York: New York City Department of Health and Mental Hygiene; 2004.
26. Kennedy R. Thousands stranded on foot by crippled trains, crawling buses and traffic gridlock. New York Times. 2003.
27. Beatty ME, Phelps S, Rohner C, Weisfuse I. Blackout of 2003: public health effects and emergency response. Public Health Rep. 2006;121:36–44.
28. Barstow D. In frustration, humor and greed, a powerless New York endures. New York Times. 2003.
29. Blackout creates influx of patients, generator woes. ED Manag. 2003;15:112–113.
30. Greenwald PW, Rutherford AF, Green RA, Giglio J. Emergency department visits for home medical device failure during the 2003 North America blackout. Acad Emerg Med. 2004;11:786–789.
31. Barron J. Power surge blacks out Northeast, hitting cities in 8 states and Canada; midday shutdowns disrupt millions. New York Times. 2003.

Supplemental Digital Content

Back to Top | Article Outline
© 2012 Lippincott Williams & Wilkins, Inc.