ORIGINAL CONTRIBUTIONOzone and Short-term Mortalityin 95 US Urban Comm unities, 1987-2000Michelle L. Bell, PhDAidan McDermott, PhDScott L. Zeger, PhDJonathan M. Samet, MDFrancesca Dominici, PhDEXPOSURE TO TROPOSPHERIC OZONEis widespread in the UnitedStates,1,2occurring also outsidesouthern California, where ozoneformation was first recognized.3Short-term exposure to ozone has been linkedto adverse health effects, including in-creased rates of hospital admissions andemergency department visits, exacerba-tion of chronic respiratory conditions (eg,asthma), and decreased lung func-tion.4-8Numerous time-series studieshave addressed the relationship be-tween ozone levels and mortality countson short-term intervals of 1 or a few days,including some studies involving mul-tiple locations; however, their findingshave been inconsistent.9-17Interpreta-tion of this evidence is constrained by thelimited range of locations included inthese reports, the variability of meth-ods used, and the imprecision of esti-mates from some of the studies. Thestudy of ozone and health is compli-cated by the complex, nonlinear chemi-cal formation of tropospheric ozone,which is temperature driven, with higherozone levels at higher temperatures.18In 1997, the US Environmental Pro-tection Agency (EPA) proposed revi-sions to the National Ambient Air Qual-ity Standard (NAAQS) for ozone,adding a daily maximum 8-hour stan-dard of 80 ppb (parts per billion by vol-ume) while phasing out the daily hourlymaximum standard of 120 ppb. Thesechanges were prompted by evidencefrom epidemiologic, controlled hu-man exposure, and toxicologic stud-ies that identified adverse health ef-fects at ozone concentrations below theexisting 1-hour NAAQS.19Because ofthe relevance of epidemiologic evi-dence to the NAAQS for ozone andother pollutants, updated and ex-panded time-series studies of ozone areinformative to the regulatory process.With the National Morbidity,Mortality, and Air Pollution Study(NMMAPS), we have developed na-tional approaches for multisite time-series analyses of particulate matter withan aerodynamic diameter less than 10µm (PM10) and mortality and hospital-ization data that provided evidence fordecision making.20-26As part of theNMMAPS, we developed 2-stage statis-tical models20,21,27for estimating the per-centage increase in mortality associ-Author Affiliations: School of Forestry and Environ-mental Studies, Yale University, New Haven, Conn (DrBell); Departments of Biostatistics (Drs McDermott,Zeger, and Dominici) and Epidemiology (Dr Samet),Johns Hopkins Bloomberg School of Public Health, Bal-timore, Md.Corresponding Author: Michelle L. Bell, PhD, YaleUniversity, School of Forestry and EnvironmentalStudies, 205 Prospect St, New Haven, CT 06511([email protected]).Context Ozone has been associated with various adverse health effects, includingincreased rates of hospital admissions and exacerbation of respiratory illnesses. Al-though numerous time-series studies have estimated associations between day-to-day variation in ozone levels and mortality counts, results have been inconclusive.Objective To investigate whether short-term (daily and weekly) exposure to ambi-ent ozone is associated with mortality in the United States.Design and Setting Using analytical methods and databases developed for the Na-tional Morbidity, Mortality, and Air Pollution Study, we estimated a national averagerelative rate of mortality associated with short-term exposure to ambient ozone for95 large US urban communities from 1987-2000. We used distributed-lag models forestimating community-specific relative rates of mortality adjusted for time-varying con-founders (particulate matter, weather, seasonality, and long-term trends) and hierar-chical models for combining relative rates across communities to estimate a nationalaverage relative rate, taking into account spatial heterogeneity.Main Outcome Measure Daily counts of total non–injury-related mortality andcardiovascular and respiratory mortality in 95 large US communities during a 14-yearperiod.Results A 10-ppb increase in the previous week’s ozone was associated with a 0.52%increase in daily mortality (95% posterior interval [PI], 0.27%-0.77%) and a 0.64%increase in cardiovascular and respiratory mortality (95% PI, 0.31%-0.98%). Effectestimates for aggregate ozone during the previous week were larger than for modelsconsidering only a single day’s exposure. Results were robust to adjustment for par-ticulate matter, weather, seasonality, and long-term trends.Conclusions These results indicate a statistically significant association between short-term changes in ozone and mortality on average for 95 large US urban communities,which include about 40% of the total US population. The findings indicate that thiswidespread pollutant adversely affects public health.JAMA. 2004;292:2372-2378 www.jama.com2372 JAMA, November 17, 2004—Vol 292, No. 19 (Reprinted) ©2004 American Medical Association. All rights reserved.ated with exposure to PM10or otherpollutants. In our 2-stage approach, atime-series analysis is first performedwithin each community, and in the sec-ond stage of analysis, the results are com-bined across communities to produce anational average estimate that ac-counts for the within-community sta-tistical uncertainty and the heterogene-ity of the effects across the country.20,21We use an updated NMMAPS data-base, including 95 large US urban areasfor 1987-2000, to perform a multisitetime-series study of ozone and mortal-ity. Because ozone concentrations aretypically available daily, at the first stageof our analysis we extend previous ap-proaches to develop constrained andunconstrained distributed-lag mod-els.26,28-34Distributed-lag models are ap-propriate for estimating relative ratesof mortality associated with exposureto pollution levels during several pre-vious days, thus allowing more flex-ibility for exploring the lag between ex-posure and death than single-lagmodels. At the second stage, we use hi-erarchical models35-38to combine therelative rate estimates obtained from thecommunity-specific distributed-lagmodels to produce a national averageestimate. With this 2-stage model, varia-tion across communities in the short-term effects of ozone can be exploredand an effect estimated for the nation.METHODSMortality, Weather,and Pollution DataThis analysis is based on daily cause-specific mortality counts for 1987-2000obtained from the National Center forHealth Statistics on