Association between air pollution and ischemic cardiovascular emergency room visits

Environmental Research 92 (2003) 57–63

Association between air pollution and ischemic cardiovascular emergency room visits$ Chin An Lin,a,b,* Luiz Alberto Amador Pereira,a,c Gleice Margarete de Souza Conceic¸a˜o,a Humberto S. Kishi,a Rodolfo Milani Jr.,b Alfe´sio Luı´ s Ferreira Braga,a,c,d and Paulo Hilario Nascimento Saldivaa a

Laboratory of Experimental Air Pollution, Department of Pathology, University of Sa˜o Paulo Medical School, Av. Dr. Arnaldo 455, 1 Andar, Sala 1105, Cerqueira Cesar, CEP 01246-903, Sa˜o Paulo, Brazil b Department of Internal Medicine, University of Sa˜o Paulo Medical School, Sa˜o Paulo, Brazil c Department of Internal Medicine, Federal University of Sa˜o Paulo School of Medicine, Sa˜o Paulo, Brazil d Environmental Pediatrics Program, Department of Pediatrics, University of Santo Amaro Medical School, Sa˜o Paulo, Brazil Received 16 November 2001; received in revised form 16 October 2002; accepted 21 October 2002

Abstract This study observed the relationship between air pollutants and ischemic cardiac diseases such as angina and acute myocardial infarction in a representative cardiovascular center emergency room in Sa˜o Paulo, Brazil. Daily emergency room admissions to the Institute of the Heart of the University of Sa˜o Paulo, as well as data concerning daily air pollutant levels and meteorological variables, were collected from January 1994 to August 1995. Generalized additive Poisson regressions were fitted to the logarithm of the expected values of total emergency room visits due to angina or acute myocardial infarction, controlling for smooth functions of season and weather and indicators for days of the week. All investigated pollutants were positively associated with ischemic cardiovascular disease emergency room visits, and the time lags were relatively short, but only CO presented an effect that was statistically significant. An interquartile range increase in CO was associated with an increase of 6.4% (95% CI: 0.7–12.1) in daily angina or acute myocardial infarction emergency room visits. This result did not change when estimates were done using linear models and natural cubic splines. This study showed that air pollution has a role in cardiovascular morbidity in Sa˜o Paulo, reinforcing the necessity for air pollutant emission-controlling polices in urban areas. r 2003 Elsevier Science (USA). All rights reserved. Keywords: Angina; Myocardial infarction; Emergency room visits; Air pollution; Environmental epidemiology

1. Introduction Exposure to urban air pollution has been associated with adverse cardiovascular health effects among $ This manuscript was based on epidemiological research, with data from the registry of emergency room visits of The Institute of Heart. Consultation of these data did not involve the direct interview of patients enrolled in the study. We did not have access to the name, gender, address, or profession of any patient. No patient underwent any medical, pharmaceutical, or surgical intervention; we thereby declare that this article is in accordance with national and institutional guidelines for the protection of the welfare human subjects. *Corresponding author. Laboratory of Experimental Air Pollution, Department of Pathology, University of Sa˜o Paulo Medical School, Av. Dr. Arnaldo 455, 1 Andar, Sala 1105, Cerqueira Cesar, CEP 01246-903, Sa˜o Paulo, Brazil. E-mail address: [email protected] (C.A. Lin).

sensitive subjects in different urban areas with different air pollution concentrations (Committee of the Environmental and Occupational Health Assembly of the American Thoracic Society, 1996). Increased cardiovascular emergency room visits were associated with CO and black smoke in London (Atkinson et al., 1999). Schwartz reported similar results in two studies using total cardiovascular hospital admissions data from American cities. In the first study (Schwartz, 1997), carried out in Tucson, Arizona, increases in PM10 and CO were associated with increased hospital admissions. In a subsequent study, using data from eight US counties (Schwartz, 1999), both pollutants presented the most consistent and significant associations. Particles (black smoke) and CO, besides NO2, and SO2, were also associated with

0013-9351/03/$ - see front matter r 2003 Elsevier Science (USA). All rights reserved. doi:10.1016/S0013-9351(02)00054-3

58

C.A. Lin et al. / Environmental Research 92 (2003) 57–63

increases in hospital admissions due to acute myocardial infarction in London (Poloniecki et al., 1997). In terms of mortality, increased cardiovascular deaths have been associated with CO, NO2, and particles in Los Angeles (Kinney and O¨zkaynak, 1991), TSP in Philadelphia (Schwartz and Dockery, 1992), and PM10 in Utah Valley (Pope et al., 1992). Therefore, both automotiveand industry-related pollutants and inhalable particles can be considered a serious risk to subjects with cardiovascular diseases. In Sa˜o Paulo, the largest city in South America, the main source of air pollution is the huge automotive fleet, with about 4,500,000 vehicles for almost 10,000,000 inhabitants (CETESB, 1995). In winter, thermal inversion episodes impair air pollution dispersion, and high toxicant concentrations are frequently observed. The problem reached alarming levels in the early 1990s, leading the State Government to adopt, from 1995 to 1998, restrictive measures to reduce by 20% the number of vehicles circulating in the downtown during the weekdays. In Sa˜o Paulo, most of the stoves use natural gas and cooking is one of the most important sources of indoor air pollution. However, Sa˜o Paulo is not a cold city and thus homes and offices generally use no form of heating. Deaths due to cardiovascular diseases represent 30% of the total annual deaths in the city (PROAIM, 2000). This study thus investigated the association between air pollutants and cardiovascular diseases using the daily number of emergency room visits in the Heart Institute of the University of Sa˜o Paulo Medical School over a 2-year period.

2. Data Information on daily emergency room (ER) visits due to angina (ICD–9th: 413) and acute myocardial infarction (ICD–9th: 410), for people between 45 and 80 years of age were obtained from the Institute of the Heart of the University of Sa˜o Paulo Medical School from January 1994 to December 1995. The Institute of the Heart is the reference center for cardiovascular disease care and research, receiving cases from all districts of the city. A permanent staff comprising at least three residents and two assistant professors of cardiology on duty 24 h per day assures reliable diagnosis. During the period of study, 50% of Sa˜o Paulo’s inhabitants were provided with care by the public health system (SEADE, 1996). Thirteen percent of the hospital admissions due to ischemic cardiovascular diseases among this group were initially attended at the emergency room service of the Institute of the Heart (Official Statistic Data on Public Health System–Datasus, 2001). Daily records of air pollutants were obtained from the Sa˜o Paulo State Sanitary Agency (CETESB, 1995,

1996). SO2 was measured in 13 stations, PM10 in 12 stations, NO2 and CO in 5 stations, and O3 in 4 stations. The daily means of the pollutants in each station were collected, averaged, and considered as indicative of the city-wide status. Information on daily minimum temperature and mean relative humidity were obtained from the Institute of Astronomy and Geophysics of the University of Sa˜o Paulo.

3. Methods Generalized additive Poisson regressions (Hastie and Tibshirani, 1990) were fitted to the logarithm of the expected values of total emergency room visits due to angina or acute myocardial infarction (MI) as the sum of the smooth and linear functions of the predictor variables. A smooth function is a nonparametric tool that allows one to control for potential nonlinear dependency of the variable of interest on covariates such as season and weather. We adopted loess, a moving regression smoother, as the nonparametric function in our analysis (Cleveland and Devlin, 1988). A smoothing parameter (span) must be chosen for each covariate. A smooth parameter for time was chosen to remove long-term trends from the data, removing seasonality and minimizing the autocorrelation of the residuals. When necessary, autoregressive terms were incorporated to eliminate autocorrelation from the residuals. Smoothing parameters were also defined for the minimum temperature of the day and relative humidity, with spans chosen to minimize Akaike’s information criteria (Akaike, 1973). For temperature and humidity we adopted 0.5 and for seasonality 0.25. Indicators were used for days of the week. To control for seasonal trends on air pollutant effects we performed analyses taking into account cool (winter) and warm (spring, summer, and fall) periods, separately. Extreme observations can influence regression estimates. To reduce the sensitivity to outliers in the dependent variable we used robust regression, a method that gives less weight to observations with large residuals. M-estimation (Venables and Ripley, 1997) was the method used for the estimates. Single-pollutant models, as well as copollutant and multipollutant models, using the daily concentration and up to 7-day moving averages of each pollutant were tested. Results were presented as regression coefficients standard errors for all pollutants, and the percentage increases in emergency room visits for interquartile range increases in air pollutant concentrations for those moving averages that presented bigger effects. The 95% confidence intervals were estimated assuming normality of the residuals.

C.A. Lin et al. / Environmental Research 92 (2003) 57–63

Also, we performed analyses using generalized linear models (GLM) (McCullagh and Nelder, 1989) and natural cubic splines (Green and Silverman, 1994) instead of Generalized Additive Models (GAM) models and loess to check for possible inconsistencies in GAM estimates (Dominici et al., 2002). For the pollutants that presented statistically significant associations with acute cardiovascular diseases, we reestimated the basic depicted models using categorical terms for each pollutant based on their quartiles to verify whether the association exhibited a dose response behavior. S-PlusTM version 4.5 (Statistical Sciences, 1997) was used to carry out the statistical analyses.

4. Results There were 2750 ER visits due to angina or MI during the 2 years of this study. Angina represented 66% of the cases. Table 1 presents the descriptive analysis of the variables in the study. During the study period, there were no measurements of NO2 levels. The PM10 mean level surpassed the annual mean standard (50 mg/m3) and the daily maximum surpassed the daily standard (150 mg/m3) 23 times. CO levels surpassed the 8-h primary standard (9 ppm) 54 times, while O3 levels surpassed the 1-h primary standard (235 mg/m3) once. SO2 levels surpassed neither the annual nor the daily standards. PM10, SO2, and CO reached their highest levels during winter, while O3, a secondary pollutant, was usually higher in summer.

59

Table 2 presents the Pearson correlation coefficients for the pollutants and the weather variables. Primary air pollutants presented high degrees of interdependence. High temperatures were positively associated with cardiovascular ER visits and presented an almost linear behavior. All investigated pollutants were positively associated with ischemic cardiovascular disease emergency room visits, and the time lags were relatively short (concurrent day for SO2, 2-day moving average for CO, and 3-day moving average for PM10), except for the O3 association, which presented the longest lag (5-day moving average). Table 3 presents regression coefficients and standard errors of both GAM and GLM estimates for all pollutants considered. There were no significant differences between both adopted approaches. In fact, the SO2 effect estimated by GLM was higher than that estimated by GAM.

Table 2 Pearson correlation coefficients between air pollutants and weather variables SO2 PM10 CO O3 Temperaturea Humidityc

PM10

0.59b 0.46b 0.02 0.22 0.40

0.59b 0.05 0.28 0.68

CO

O3

0.19b 0.20 0.32

0.34 0.61

Temperaturea

0.13

a

Minimum temperature. Po0:01: c Relative humidity. b

Table 1 Descriptive analysis for the variables in the study Variables ER visits Total Infarction Angina

Daily mean

SDa

5%

Nc

7 3 5

730 730 730

1.97 1.15 1.59

Weather Temperatured (1C) Humiditye(%)

15.32 80.01

3.31 9.75

10.20 61.83

4.95 12.50

20.00 92.91

730 730

Pollutantsf SO2 (mg/m3) (ppm) PM10 (mg/m3) CO (ppm) O3 (mg/m3) (ppm)

20.96 0.008 70.34 5.12 58.80 0.030

12.15 0.005 33.94 2.41 37.63 0.019

6.27 0.002 31.84 2.10 11.76 0.006

15.90 0.006 40.47 2.96 45.81 0.023

45.00 0.017 136.28 9.66 127.24 0.065

694

Standard deviation of the mean. Interquartile range. c Number of days of observation. d Minimum temperature for the day. e Relative humidity. f NO2 was not measured during the period of study. b

3 2 3

95%

3.81 1.25 2.53

a

1 0 0

IQRb

703 697 535

C.A. Lin et al. / Environmental Research 92 (2003) 57–63

60

Table 3 Regression coefficients and standard errors of the association between air pollutants and cardiovascular emergency room visits for different periods of timea Regression coefficients (standard errors)

SO2 PM10 CO O3

GAM GLM GAM GLM GAM GLM GAM GLM

Day

2-day mab

3-day ma

4-day ma

5-day ma

0.004 (0.002) 0.004c (0.002) 0.000 (0.001) 0.000 (0.001) 0.021 (0.008) 0.024 (0.009) 0.000 (0.001) 0.001 (0.001)

0.003 (0.002) 0.004 (0.003) 0.001 (0.001) 0.001 (0.001) 0.023c (0.009) 0.027c (0.011) 0.001 (0.001) 0.001 (0.001)

0.001 0.002 0.001 0.001 0.020 0.020 0.001 0.001

0.001 0.001 0.001 0.000 0.013 0.006 0.000 0.000

0.001 0.001 0.001 0.000 0.007 0.003 0.001 0.001

(0.003) (0.003) (0.001) (0.001) (0.011) (0.013) (0.001) (0.001)

(0.003) (0.004) (0.001) (0.001) (0.012) (0.015) (0.001) (0.001)

(0.003) (0.004) (0.001) (0.001) (0.012) (0.016) (0.001) (0.001)

6-day ma 0.001 10.003 0.000 0.000 0.005 0.008 0.001 0.001

(0.003) (0.004) (0.001) (0.001) (0.013) (0.017) (0.001) (0.002)

7-day ma 0.001 0.001 0.000 0.000 0.006 0.008 0.000 0.000

(0.003) (0.004) (0.001) (0.001) (0.014) (0.018) (0.001) (0.002)

a

Concurrent days and moving averages ranged from 2 to 7 days. Ma, moving averages. c Po0:05: b

Table 4 Regression coefficients and standard errors for SO2 (concurrent day), PM10 (3-day moving average), CO (2-day moving average), and O3 (5-day moving average) estimated in copollutant and multipollutant generalized additive models Regression coefficients and standard errors Copollutant

SO2 PM10 CO O3 a b

Multipollutant

SO2

PM10

CO

O3

— 0.001(0.001) 0.019 (0.010) 0.001 (0.001)

0.004 (0.002) — 0.022 (0.011)a 0.001 (0.001)

0.002 (0.002) 0.000 (0.001) — 0.001 (0.001)

0.004 (0.003) 0.002 (0.001) 0.037 (0.012)b —

0.002 0.000 0.033 0.001

(0.003) (0.001) (0.016)a (0.001)

Po0:05: Po0:01:

the inclusion of SO2 as a copollutant dropped the CO effect observed in the single pollutant model by 20%, in the multipollutant model, the CO effect increased by 40%. Air pollutant effects remained almost unchanged when analyses were stratified for cool and warm periods (Table 5). Only PM10 effects presented differences between the two periods, and its impact on cardiovascular diseases was higher during the winter. When indicators of quartiles of a 2-day moving average of CO were used a threshold was observed, with the relative risk for angina and MI ER visits increasing sharply when the CO levels were above 3.45 ppm (Fig. 2).

Fig. 1. Percentage increases and 95% confidence intervals in daily ischemic cardiovascular disease emergency room visits due to interquartile range increases in SO2 (concurrent day, 20.96 mg/m3), PM10 (3-day moving average, 70.34 mg/m3), CO (2-day moving average, 5.12 ppm), and O3 (5-day moving average, 58.80 mg/m3).

Fig. 1 shows the percentage increases in the outcome due to interquartile range increases in the pollutants. Only CO presented an effect that was statistically significant. CO, SO2, O3, and, to a lesser extent, PM10 resisted the inclusion of copollutants (Table 4). While

5. Discussion This study was designed to assess the effect of air pollutants on cardiovascular diseases in Sa˜o Paulo, Brazil. We found that air pollution was positively associated with increases in the daily number of total ischemic cardiovascular disease ER visits, represented by the sum of angina and myocardial infarction cases, and that this association presented, in general, a short time lag. Only CO presented a statistically significant effect on the outcome, and a threshold was observed for levels of CO above 3.45 ppm.

C.A. Lin et al. / Environmental Research 92 (2003) 57–63

61

Table 5 Regression coefficients and standard errors for SO2 (concurrent day), PM10 (3-day moving average), CO (2-day moving average), and O3 (5-day moving average) estimated during warm and cool seasons using generalized additive models Regression coefficients (standard errors) PM10 (3-day maa)

SO2 (day) Warm 0.005 (0.003) a b

Cool 0.005 (0.003)

Warm 0.001 (0.002)

CO (2-day ma) Cool 0.002 (0.001)

Warm 0.028 (0.018)

O3 (5-day ma) Cool b

0.027 (0.012)

Warm

Cool

0.003 (0.002)

0.002 (0.002)

Ma, moving average. Po0:05:

Fig. 2. Relative risk of occurring ischemic cardiovascular disease emergency room visits according to categories of CO.

We adopted multivariate models, controlling for the most commonly used confounders in the association between air pollution and cardiovascular morbidity: long time trend, temperature, humidity, and day of the week. Even so, the possibility of incomplete adjustment for confounders must always be considered. Despite the difference in air pollutant concentrations between warm and cool periods, the effects did not change when analyses were stratified by season and they are different from results presented by other studies (Morris and Naumova, 1998; Shumway et al., 1988). This may be attributed to meteorological, almost permanent traffic jams and housing characteristics. In previous studies carried out in Sa˜o Paulo, we have reported temperature effects with the same size and nearly linear shape observed in the present study. However, in studies using respiratory outcomes the effect was inverted, with low temperatures associated with the outcomes. Sa˜o Paulo cannot be considered ‘‘a cold’’ city in comparison to cities examined in studies of the effects of temperature on cardiovascular diseases in the United State and Europe. The adoption of the GLM models showed that the associations between air pollutants and cardiovascular diseases assessed by the GAM approach were not overestimated or artificial.

In a recent study, Linn et al. (2000) elucidated the effect of CO and other primary pollutants associated with increases in cardiopulmonary illness hospitalization. In the same study the author drew attention to the possibility of response misclassification of the effects of pollutants, involving problems with misdiagnoses leading to errors in specific disease counts. To minimize the error in the diagnosis of various diseases, we decided to use cases attended in the Institute of the Hearth of the University of Sa˜o Paulo School of Medicine, one of the most important cardiovascular medical centers in South America. The staff of at least two consultant physicians and more than four residents 24 h per day assures high quality in diagnosis. This is an important factor for studies that report air pollution effects and causespecific outcomes. Many authors have reported the effects of air pollutants on the cardiovascular system. CO is a wellrecognized cardiovascular toxicant, and its association with the exacerbation of both angina (Aronow, 1981) and myocardial infarctions (Marius-Nunes, 1990) has already reported. These effects were observed under moderate and high CO ambient levels, respectively. In the present study we also found a nonlinear relationship between CO exposure and acute cardiovascular effects, with a threshold after the second quartile of CO concentrations. The threshold observed for CO in this study is not an isolated occurrence. In 1998, Pereira et al. (1998) published a study in Environmental Health Perspectives on air pollution exposure and intrauterine mortality in Sa˜o Paulo, with a similar observed relationship with CO. In the same study an experimental procedure was carried out and measurements of cord blood carboxyhemoglobin were compared with the outdoor CO concentrations on the days of delivery. Variation in CO concentration similar to the threshold level observed in the present study increased the fetuses’ COHb by 40% compared with the first CO interquartile range. CO concentrations in Sa˜o Paulo used to be higher than those observed in the United States. When we adopted an increase of one interquartile range to estimate our results we obtained effects that seemed to be greater than the effects usually reported for US cities

62

C.A. Lin et al. / Environmental Research 92 (2003) 57–63

(Linn et al., 2000). However, when we compared our results with those reported by Linn et al. (2000), it was clear that the effect per ppm in Sa˜o Paulo was smaller than the effect observed in Los Angeles. We also believe that in Sa˜o Paulo CO is a good proxy for traffic-related emissions. Although nonstatistically significant associations were seen between acute ischemic cardiac diseases and PM10, SO2, and O3, the associations were positive and the estimated effect sizes were almost the same as those for CO. In the last decade, researchers have devoted enormous attention to the study of particle-related cardiovascular effects. Some evidence has supported the biological plausibility of these effects. Urban particles are mainly derived from incomplete combustion processes. Soluble transition metals present in urban particles have been associated with cellular toxicity, leading to inflammatory processes in the lung and also presenting a systemic effect. Godleski et al. (2000) have proposed hypothetical mechanisms by which inhaled ambient particles might result in deleterious effects on heart, involving, from the lung, the stimulation of the central nervous system and/ or the beginning of an inflammatory process that can promote systemic effects. Both processes could lead to the damage of heart cells. In their study, particles were associated with changes in heart rate variability, alterations in electrocardiographic morphology, and increased sympathetic nervous system activity in exposed dogs. Moreover, inhalable particles have been associated with increases in plasma viscosity, one factor that can contribute to an increase in myocardial infarction (Peters et al., 1997) and an increased heart rate (Pope et al., 1999). Additionally, particles present an acute effect on myocardial infarction, with lags ranging from a few hours to 1 day after the exposure (Peters et al., 2001). SO2 has also been associated with cardiovascular effects. In Hong Kong (Wong et al., 1999) and London (Atkinson et al., 1999) it was associated with increases in total cardiovascular disease hospital admissions. On the other hand, reports of the cardiovascular effects of O3 have not been frequent, and despite its lack of statistical significance, the association observed in this study points to a low explored effect of O3 exposure. Gold and colleagues (Gold et al., 2000) showed ozone exposure, besides particles, to be associated with reduced heart rate variability, probably due to a decrease in vagal tone. Further analysis of O3 cardiovascular effects and their biological plausibility are necessary. Despite the natural tend to emphasize the effects that can be attributed to each one of the pollutants, one must keep in mind that urban air pollution is a very complex mixture of different elements and that most of them are not measured on a regular basis. We can generalize from

the results of toxicological studies conducted in controlled environments to the general population exposed to urban pollutants. However, the interactions between so many toxicants can maximize their individual effects. As briefly mentioned above, the main source of air pollutants in Sa˜o Paulo is motor vehicle exhaust (CETESB, 2000). The automotive fleet, including cars, buses, and trucks, generates 50% of the inhalable particles, almost 70% of the SO2, 95% of the NO2, and 97% of the CO. A significant reduction of industrial emissions of SO2 contributes to keeping their levels under the air quality standard of 80 mg/m3. All other pollutants have reached levels that surpass the specific standard. Because Sa˜o Paulo is a warm city and there is not the fully thermal isolation between outdoor and indoor air that occurs in many countries where the wide range of temperatures demand heating and air conditioning, which in both winter and summer mandates the isolation of outdoor and indoor environments. Indoor air pollution, such as that derived from cooking, cannot be considered as a confounder of the association between daily variations in outdoor air pollution and daily fluctuations of health outcomes. However, we must assume that it may be an effect modifier. In summary, this study showed a significant association between CO and ischemic cardiovascular ER visits for adults 45 years of age and older, and this association resisted the inclusion of the most common confounders to the stratification of the analyses by season and copollutant and multipollutant models. The outcome was also associated with PM10, SO2, and O3, despite the lack of statistical significance. Further analysis using different cardiovascular endpoints are necessary in order to estimate the impact of air pollutants on the burden of cardiovascular diseases in Sa˜o Paulo.

References Akaike, H., 1973. Information theory and an extension of the maximum likelihood principal. In: Petrov, B.N., Csaki, F. (Eds.), Second International Symposium on Information Theory, Akademiai Kiado, Budapest. Aronow, W.S., 1981. Aggravation of angina pectoris by two percent carboxyhemoglobin. Am. Heart 101, 154–157. Atkinson, R.W., Bremner, S.A., Anderson, H.R., Strachan, D.P., Bland, J.M., de Leon, A.P., 1999. Short-term associations between emergency hospital admissions for respiratory and cardiovascular disease and outdoor air pollution in London. Arch. Environ. Health 54 (6), 398–411. CETESB, 1995. Relato´rio de Qualidade do ar na Regia˜o Metropolitana de Sa˜o Paulo e Cubata˜o–1994. CETESB, Sa˜o Paulo (in Portuguese). CETESB, 1996. Relato´rio de Qualidade do ar na Regia˜o Metropolitana de Sa˜o Paulo e Cubata˜o-1995, CETESB, Sa˜o Paulo (in Portuguese).

C.A. Lin et al. / Environmental Research 92 (2003) 57–63 CETESB, 2000. Poluic-a˜oVeicular–MedidasdeControle. Sa˜o Paulo State Sanitary Agency. URL: http://www.cetesb.sp.gov.br/poluicaoveicular/medidas.htm (in Portuguese). Cleveland, W.S., Devlin, S.J., 1988. Robust locally weighted regression and smoothing scatterplots. J. Am. Stat. Assoc. 74, 829–836. Committee of the Environmental and Occupational Health Assembly of the American Thoracic Society, 1996. Health effects of outdoor air pollution. Am. J. Respir. Crit. Care Med. 153, 3–50. Dominici, F., McDermott, A., Zeger, S.L., Samet, J.M., 2002. On the use of generalized additive models in time-series studies of air pollution and health. Am. J. Epidemiol. 156 (3), 193–203. Godleski, J.J., Verrier, R.L., Koutrakis, P., et al., 2000. Mechanisms of morbidity and mortality from exposure to ambient air particles. Health Effects Inst. Res. Rep. 91, 5–83. Gold, D.R., Litonjua, A., Schwartz, J., Lovett, E., Larson, A., Nearing, B., Allen, G., Verrier, M., Cherry, R., Verrier, R., 2000. Ambient pollution and heart rate variability. Circulation 101 (11), 1267–1273. Green, P.J., Silverman, B.W., 1994. Nonparametric Regression and Generalized Linear Models: a roughness penalty approach. Chapman & Hall, London. Hastie, T., Tibshirani, R., 1990. Generalized Additive Models. Chapman & Hall, London. Kinney, P.L., O¨zkaynak, H., 1991. Associations of daily mortality and air pollution in Los Angeles County. Environ. Res. 54, 99–120. Linn, W.S., Szlachcic, Y., Gong Jr., H., Kinney, P.L., Berhane, K.T., 2000. Air pollution and daily hospital admissions in metropolitan Los Angeles. Environ. Health Perspect. 108, 427–434. Marius-Nunes, A.L., 1990. Myocardial infarction with normal coronary arteries after acute exposure to carbon monoxide. Chest 97 (2), 491–494. McCullagh, P., Nelder, J.A., 1989. Generalized Linear Models, 2nd Edition. Chapman & Hall, London. Morris, R.D., Naumova, E.N., 1998. Carbon monoxide and hospital admissions for congestive heart failure: evidence of an increased effect at low temperatures. Environ. Health Perspect. 106, 649–653. Official Statistic Data on Public Health System–Datasus; Brazil, http://www.datasus.gov.br. (in Portuguese). Pereira, L.A.A., Loomis, D., Conceic-a˜o, G.M.S., Braga, A.L.F., Arcas, R.M., Kishi, H., Singer, J.M., Bo¨hm, G.M., Saldiva, P.H.N., 1998. Association between air pollution and intrauterine mortality in Sa˜o Paulo, Brazil. Environ. Health Perspect. 106, 325–329.

63

Peters, A., Doring, A., Wichmann, H.E., Koenig, W., 1997. Increased plasma viscosity during an air pollution episode: a link to mortality? Lancet 349, 1582–1587. Peters, A., Dockery, D.W., Muller, J.E., Mittleman, M.A., 2001. Increased particulate air pollution and the triggering of myocardial infarction. Circulation 103, 2810–2815. Poloniecki, J.D., Atkinson, R.W., de Leon, A.P., Anderson, H.R., 1997. Daily time series for cardiovascular hospital admissions and previous day’s air pollution in London, UK. Occup. Environ. Med. 54 (8), 535–540. Pope III, C.A., Schwartz, J., Ransom, M.R., 1992. Daily mortality and PM10 pollution in Utah Valley. Arch. Environ. Health. 47, 211–217. Pope III, C.A., Dockery, D.W., Kanner, R.E., Villegas, G.M., Schwartz, J., 1999. Oxygen saturation, pulse rate, and particulate air pollution. Am. J. Respir. Crit. Care Med. 159, 365–372. PROAIM, 2000 (Mortality Information Improvement Program). O´bitos por grandes grupos de causa, [on line, in Portuguese]. Available from internet: http://www.prodam.sp.gov.br/sms/estatist/proaim/tabcidsexo.htm Schwartz, J., 1997. Air pollution and hospital admissions for cardiovascular disease in Tucson. Epidemiology 8 (4), 371–377. Schwartz, J., 1999. Air pollution and hospital admissions for heart disease in eight US counties. Epidemiology 10 (1), 17–22. Schwartz, J., Dockery, D.W., 1992. Increased mortality in Philadelphia associated with daily air pollution concentrations. A. Rev. Respir. Dis. 145, 600–604. SEADE, 1996 (State Data Analysis System Foundation). Pesquisa de Condic-o˜es de Vida na Regia˜o Metropolitana de Sa˜o Paulo (in Portuguese). Sa˜o Paulo State Annual Report (online). Accessed in 05-22-2000. URL: http://www.seade.gov.br/cgi-bin/titabpv98/ tab.htm?sau/MSP96/sau96011-COLTAB Shumway, R.H., Azari, A.S., Pawitan, Y., 1988. Modeling Mortality Fluctuations in Los Angeles as Functions of Pollution and Weather Effects. Environ. Res. 45, 224–241. Statistical Sciences, Inc. 1997. S-PLUS for Windows User Manual, VERSION 4.0. Data Analysis Products Division, MathSoft, Inc, Seattle, WA. Venables, W.N., Ripley, B.D., 1997. Modern Applied Statistics with S-PLUS, 2nd Edition. Springer-Verlag, New York, 247–266. Wong, T.W., Lau, T.S., Yu, T.S., et al., 1999. Air pollution and hospital admissions for respiratory and cardiovascular diseases in Hong Kong. Occup. Environ. Med. 56 (10), 679–683.