The Temporal Pattern of Mortality Responses to Air Pollution: A Multicity Assessment of Mortality Displacement
Zanobetti, Antonella1; Schwartz, Joel1; Samoli, Evi2; Gryparis, Alexandros2; Touloumi, Giota2; Atkinson, Richard3; Le Tertre, Alain4; Bobros, Janos5; Celko, Martin6; Goren, Ayana7; Forsberg, Bertil8; Michelozzi, Paola9; Rabczenko, Daniel10; Aranguez Ruiz, Emiliano11 and; Katsouyanni, Klea2
From the 1Environmental Epidemiology Program, Harvard School of Public Health, Boston, MA;
2University of Athens Medical School, Athens, Greece;
3Department of Public Health Sciences, St George’s Hospital Medical School, London, United Kingdom;
4Environmental Health Unit, National Institute of Public Health Surveillance, Saint-Maurice, France;
5Municipal Institute of Public Health, Budapest, Hungary;
6Charles University Medical Faculty, Prague, Czech Republic;
7Department of Epidemiology, Tel Aviv University, Tel Aviv, Israel;
8Department of Public Health and Clinical Medicine, Umeå University, Umeå, Sweden;
9Agency for Public Health, Lazio Region, Rome, Italy;
10National Institute of Hygiene, Department of Medical, Statistics, Warsaw, Poland; and
11Municipal Department of Public Health, Madrid, Spain.
Address correspondence to: Antonella Zanobetti, Department of Environmental Health, Environmental Epidemiology Program, Harvard School of Public Health, 665 Huntington Avenue, Boston, MA 02115; azanob@sparc6a. harvard.edu
This research was part of the APHEA-2 project, which was funded by the European Union contract number ENV4-CT97-0534. Joel Schwartz was also supported by U.S. Environmental Protection Agency Grant R827353.
Submitted October 16, 2000; final version accepted August 21, 2001.
Although the association between particulate matter and mortality or morbidity is generally accepted, controversy remains about the importance of the association. If it is due solely to the deaths of frail individuals, which are brought forward by only a brief period of time, the public health implications of the association are fewer than if there is an increase in the number of deaths. Recently, other research has addressed the mortality displacement issue in single-city analysis. We analyzed this issue with a distributed lag model in a multicity hierarchic modeling approach, within the Air Pollution and Health: A European Approach (APHEA-2) study. We fit a Poisson regression model and a polynomial distributed lag model with up to 40 days of delay in each city. In the second stage we combined the city-specific results. We found that the overall effect of particulate matter less than 10 μM in aerodynamic diameter (PM10) per 10 μg/m3 for the fourth-degree distributed lag model is a 1.61% increase in daily deaths (95% CI = 1.02–2.20), whereas the mean of PM10 on the same day and the previous day is associated with only a 0.70% increase in deaths (95% CI = 0.43–0.97). This result is unchanged using an unconstrained distributed lag model. Our study confirms that the effects observed in daily time-series studies are not due primarily to short-term mortality displacement. The effect size estimate for airborne particles more than doubles when we consider longer-term effects, which has important implications for risk assessment.
Air pollution, especially airborne particles, has been consistently reported to be associated with daily deaths in reports from all over the world. 1–8 More recently, systematic multicity analyses have confirmed these findings. 9–12 Nevertheless, some have questioned the public health significance of these associations, arguing that if these deaths are occurring only in those who would have died in a few days anyway, the public health significance of exposure is small. Were that the case, the increase in deaths during and immediately after exposure would be counterbalanced by a deficit in daily deaths a few days later, when those deaths would have otherwise occurred. If such a pattern were true, the positive correlation seen between daily deaths and exposure shortly before the death would be counterbalanced by a negative correlation between exposure and daily deaths at some longer lag. An example of such a hypothetical pattern, called mortality displacement or harvesting effect, is seen in Figure 1. Were such a phenomenon to exist, it should be detected readily in studies of acute episodes, but those patterns have not been observed in air pollution episodes. 13
It is useful to examine the reason for such a phenomenon. Assume there is a pool of people at high risk of dying at any given time. An air pollution episode, by increasing the risk in that pool, would increase the death rate out of the pool and result in a smaller pool size. The finite size of the risk pool creates the possibility of a negative association with pollution at some lags. This rebound (ie, drop in the number of deaths, after an initial increase) presupposes that air pollution does not affect recruitment into the pool. Yet numerous epidemiologic studies have shown particulate air pollution to be associated with exacerbation of illness, including increased hospitalizations, 14 decreased heart rate variability, 15 etc, thus suggesting that increased recruitment is possible. Recently, Zelikoff et al 16 have shown that particle exposure exacerbates pneumonia in animals. Hence, air pollution may intensify some illnesses, increasing the size of the risk pool. Further, this may occur with a different lag than that between exposure and death out of the risk pool. Hence, the direction of the effect of an air pollution episode on the size of the risk pool, and the effect of the risk pool on the death rate over time, may be positive or negative.
Recently, three papers have examined this issue indirectly, by estimating the association between air pollution and daily deaths in Philadelphia, 17 Boston, 18 and Chicago 19 after filtering out such rebounds. None of the studies found any evidence that the effect size for air pollution was reduced as a result of the mortality displacement, and indeed all three studies reported that the effect size approximately doubled. Schwartz 18 interpreted this as suggesting that, far from depleting the pool of critically ill people, air pollution increased the size of the pool over longer time scales by increasing the intensity of illness in general. None of these studies provided any direct estimate of what the time course of the rise and fall of mortality after exposure might be (eg, Figure 1).
One additional analysis has recently been published. 20 These authors assumed a model in which air pollution could only deplete the pool of susceptible individuals at high risk of dying and could not increase recruitment into that pool. This is equivalent to assuming that the correlation between air pollution and daily deaths must become negative after a lag of several days. That assumption is a testable hypothesis.
Another recent paper 21 applied a different approach that explicitly tests this hypothesis. Zanobetti et al21 estimated the association of air pollution at multiple lags simultaneously, providing a direct estimate of Figure 1. Because air pollution is generally correlated, putting a large number of lags of a pollutant into a model produces high levels of multicolinearity and unstable results. To counter this problem, these authors used a nonparametric smoothed distributed lag, looking out to 40 days after exposure, to estimate the effect of air pollution on daily deaths in Milan between 1980 and 1989. This constrained the estimated effects of air pollution to vary smoothly with the number of days of lag between exposure and death. This required special software that is not generally available. However, in a sensitivity analysis, they showed that essentially identical results could be obtained using a cubic polynomial distributed lag model, which can be implemented in any Poisson regression package. In both cases, the coefficients of air pollution at each lag are constrained to fit a smooth shape, in which the latter case is a polynomial. If the polynomial is flexible enough to fit the true pattern of the data reasonably well, little bias will be introduced.
We have adopted that approach for a systematic examination of the lag between air pollution and daily deaths in the Air Pollution and Health: A European Approach (APHEA-2) study. 22,23 This analysis focuses on particulate air pollution in a multicity hierarchic model.
Subjects and Methods
The APHEA-2 study is a comprehensive, multicenter study that examines the association between air pollution and daily deaths in 30 cities across Europe and associated regions (eg, Tel Aviv). Data collection included daily counts of all-cause mortality, excluding deaths from external causes (International Classification of Diseases, 9th revision, code >800). The years of study were 1990 through 1997, although mortality data in most cities were available only through 1995 or 1996. In some cases, air pollution data were available only for part of the period.
Because of resource and time constraints, it was decided a priori to limit the analysis of mortality displacement to ten cities. To maximize the power of the study, we chose the largest cities in the study, with the stipulation that only one city could be chosen in each country. The ten cities selected were Athens, Budapest, Lodz, London, Madrid, Paris, Prague, Rome, Stockholm, and Tel Aviv. Together, they comprise a population of about 28 million people, which is two-thirds of the population in the full study, and they represent northern Europe, central Europe, and the Mediterranean region. An earlier paper 23 examined the association of particulate air pollution in all available cities and addressed the issue of heterogeneity in response. That analysis did not examine the “harvesting” issue addressed in this paper.
Daily measurements of particulate air pollution were provided by each city participating in the APHEA-2 project. Particulate matter was measured as PM10 (particulate air matter less than 10 μM in aerodynamic diameter) in four cities, as PM13 (particulate air matter with aerodynamic diameter less than 13 μM) in Paris, and PM15 (particulate air matter with aerodynamic diameter less than 15 μM) in Rome. The Paris data were assumed to be equivalent to PM10 in this study. Rome data were converted to PM10 using a site-specific conversion factor based on colocated measurements. 24 In Athens, data were routinely collected only on black smoke. Because traffic is the dominant source of particles in Athens, there were some days of colocated PM10 and black smoke monitoring that allowed the establishment of a site-specific selective conversion. Also in Lodz only data for black smoke were available, whereas in Budapest the original data were measured as total suspended particulate. In these three cities, data were converted to PM10 as a function of both black smoke (total suspended particulate for Budapest) and season, again on the basis of regression modeling with limited PM10 data.
We conducted a weighted metaregression with a dummy variable equal to 1 for cities where the other particle measures were converted to PM10 on the basis of site-specific calibration. We found a somewhat higher coefficient in the converted cities (1.98% per 10 μg/m3 increase in PM10 compared with 1.48% in the cities that measured PM10), but the confidence interval for the incremental 0.5% effect was ±1.93%. These results indicate that the coefficients could in fact be 0. Further, three of the five cities where the conversion occurred were in southern Europe, where a previous hierarchic model of all 29 cities in APHEA-2 showed larger coefficients. We conclude that there is little reason to believe the effect estimates differ between the cities where the air pollutant measurement has been converted and the other cities. Hence, results were reported as the effect of PM10. Further details have been previously reported. 23
Generalized additive regression models 25 were fitted in each of the ten cities, controlling for seasonal patterns, long-term time trends for weather, influenza epidemics, holidays, and day of the week. The models were built following the APHEA-2 methodology. 23 Because of the substantial variability in seasonal patterns and weather between, for example, Stockholm and Tel Aviv, separate models were chosen in each city. All models controlled for temperature and humidity on the same day using nonparametric smooth function. 27 In addition, we examined whether nonparametric functions of weather variables on the previous day or up to 3 previous days or the average of a few days improved model fit (defined as lowering the Akaike information criterion 28 for the model). We similarly chose the number of degrees of freedom for each weather variable to minimize the Akaike information criterion. This approach has been used and discussed previously. 29,30
Seasonal patterns are controlled because there are unmeasured predictors of death, such as diet, which vary seasonally and have long-term trends over time. Because air pollution also shows seasonal variations and long-term trends, this creates a potential for confounding. Shorter-term fluctuations in diet are unlikely to be correlated with air pollution. Hence, the goal of our smooth function of time is to remove seasonal and long-term fluctuations.
Various smoothing parameters exist for producing residuals with no seasonality. To choose among them, we examined the partial autocorrelation function of the residuals. This is because, although each death is an independent event, seasonal patterns in the mortality data produce correlations between the number of deaths on one day and on the previous day. Eliminating short-term serial correlation is therefore a measure of how successful our seasonal control has been. On the other hand, the use of excessive degrees of freedom for seasonal control induces negative serial correlation in the residuals of the mortality series, 31 which can distort the association with air pollution. Therefore, we chose a smoothing parameter for time to reduce the residuals to white noise. Sometimes it was necessary to introduce autoregressive terms to accomplish this. 32 This approach has been used in a number of recent studies. 6,12,30
Distributed Lag Model
The goal of our analysis was to estimate the dependence of daily deaths (on day t) on PM10 on that day and up to the previous 40 days. If the pollution-related deaths are only being advanced by a few days to a few weeks, we will see this effect as a negative association between air pollution and deaths several days to several weeks subsequently. The net effect of air pollution, net of any such short-term rebound up to 40 days, is the sum of the effect estimates for all 41 days. In addition, plotting individual effect size estimates vs lag number gives us a direct estimate of what Figure 1 really looks like. This is an example of a distributed lag model, which has been described previously. 33,34
For Poisson regression, the unconstrained distributed lag model may be written as:
where Zt = pollution variable delayed over time, for j = 0 . . . q days.
Because this model produces unstable estimates for large q, it is common to constrain the coefficients to vary smoothly with lag number. 33 A polynomial distributed lag constrains the βj to follow a polynomial pattern in the lag number, that is:
where j is the number of lag of delay and k is the degree of the polynomial. Further details, including how to estimate the ηk in a Poisson model, have been published previously. 34 Too much constraint risks bias, producing a distorted shape, whereas too little constraint produces estimates that are too noisy to be informative. Although a cubic polynomial was sufficient to match the results of the smoothed distributed lag in Milan, 21 we have chosen a fourth-degree polynomial in this study, to ensure enough degrees of freedom to fit the pattern of response over time. Such a polynomial has enough degrees of freedom to model a curve such as that shown in Figure 1, or any other plausible shape. Therefore, we estimated in each city the five coefficients η0 . . . η4 for the fourth-degree polynomial that defines the shape of the distributed lag. As a sensitivity analysis, we used a cubic polynomial and an unconstrained distributed lag model. The unconstrained distributed lag model is too noisy to provide any information about the shape of the effect size vs lag, but it does give an unbiased estimate of the overall effect. A separate distributed lag model was fit for each of the ten cities.
The hierarchic model has two stages. In the first stage, the &eegrik values are estimated in each city i, as described in Eqs 1 and 2.
In the second stage, we combined the city-specific coefficients ηik, using the multivariate maximum likelihood method. 35
We assume that:
where &eegri is the vector of ηk in city i, Si is the estimated variance-covariance matrix in city i, and D is the random variance-covariance matrix component, reflecting heterogeneity in response among the cities.
After combining the coefficients &ik by city, the combined coefficients by lag (&bgrj) for the distributed lag model were obtained from Eq 2.
To see how the results compare with more traditional models, we fit the same model in each city using as our exposure index the mean PM10 concentration on the day of death and the previous day. 11,34,36,37 Note that this model is a highly constrained variant of our distributed lag model, with the constraints forcing β1 = β0, and β2 = β3 = . . . = β40 = 0. All analyses were done using the S-plus software (Mathsoft Inc, Seattle, WA).
Table 1 shows the ten cities, their populations, the study period in each location, and the mean and standard deviation of the number of daily deaths and environmental variables. Further details of the baseline models for each city have been published previously. 23
Table 2 shows, for each city, the estimated regression coefficients of PM10 (per 10 μg/m3 and its 95% confidence interval) for the traditional model (mean of the current and previous day), and the overall effect from the fourth-degree polynomial, the cubic, and unrestricted distributed lag models. The overall effect is the sum of the βj per 10 μg/m3. It also shows the combined effect estimates across all of the ten cities, based on a random-effect model to combine results across cities.
Apart from Rome, the estimated effect of PM10 increased, and in many cities was more than doubled, when the lagged effects were considered, rather than reduced. These results are seen in all of the distributed lag models that we applied, including the unconstrained model.
The reason for this increase is clear from Figure 2, which shows the estimated effect at each lag, and its confidence interval from the fourth-degree polynomial. It shows that the effect of PM10 does decrease to close to 0 with a lag of 10 days, but remains positive, and rises again to a second smaller peak, before dying out to 0 by lag 40.
Figure 3 shows the combined effect for the cubic polynomial. The PM10 effect decreases with a minimum at 14 days of lag and then rises again. Although they differ in some detail, both figures show the same general pattern. The initial effect declines to 0 with a lag of 1–2 weeks and then shows a second peak.
To test whether the effect at longer lags made an important contribution to the overall effect, we computed the overall effect (and its standard error) for the first 10 days and for days 11–40 before the death. The effect estimate (×1000) was 0.922 ± 0.184 for the first 10 days of exposure, and 0.688 ± 0.261 for the deaths associated with PM10 11–40 days before. Hence, although the exposure in the first week (and indeed the first 2 days) before the event had a stronger impact, the exposure in the preceding month substantially increased the estimate of the overall effect.
Previous studies have addressed the mortality displacement issue in single-city analysis. Although these studies were both methodologically innovative and produced valuable information on the issue, the heterogeneity of response to air pollution that has been reported in single-city results 23 suggests that a multicity approach, in various locations and using a predefined sampling framework, would be quite valuable in furthering discussion of this issue. Such a study would be necessary to obtain reliable estimates of effect size by lag. Our study is the first report to obtain such stable estimates of effect size by lag in multiple locations.
Qualitatively, our study confirms the basic finding of the previous four studies that did not force harvesting to occur: we do not find that most of the effect of air pollution is short-term harvesting. These results have now been shown in five studies using three different methodologies and in 13 of 14 cities, suggesting that the finding is robust. These findings are also consistent with the results of the episode studies. 13 Quantitatively, our study also confirms the previous results by showing that the effect size estimate for airborne particles more than doubles when longer-term effects are taken into consideration.
Our study adds several things to the previous literature. One is the weight of ten cities, which were not selected haphazardly or according to having positive results. This gives considerable assurance that the results are not due to a chance selection of the study locations or selection bias. Second, our study provides insight into the shape of the longer-term response to particulate air pollution. In particular, it suggests that the adverse response to pollution persists up to a month or longer. Moreover, the smoothed distributed lag model of Zanobetti et al21 produced a very similar curve of effect over time in Milan. There was a prolonged response out to a month in that study as well, with the same dip after 1–2 weeks.
The curves shown in Figures 2 and 3 reflect two processes. One is the pattern of risk over time that occurs in an individual after exposure. This is presumably positive definite, as pollution cannot be expected to improve health. The second is the effect of pollution on the sensitive pool, which can be to expand or shrink that pool. One possible explanation for the observed results is that the effects of air pollution persist for over a month (ie, longer-term average exposures have cumulative effects), but that this is partially countered by a drop in the size of the frail pool in the week or two after exposure. A second possibility is that the direct effects of air pollution trail off by a week or so, but that enhanced recruitment into the frail pool results in a long tail of excess deaths triggered by other factors. This is an important issue that remains to be investigated. If there is a prolonged increase in individual risks, it should be possible to identify intermediary biomarkers that remain elevated for some time.
The two-fold increase in risk associated with longer time scales is consistent with the report of higher risk estimates in cohort studies 38,39 than in previous time-series studies, given that the cohort studies incorporate effects of longer-term exposure. Together with those studies, it suggests that risk assessment based on the short-term associations likely underestimate the number of early deaths that are advanced by a significant amount, and that estimates based on the cohort studies, or studies such as this one, would more accurately assess the public health impact. Nevertheless, it is important to note that the exposure on the day of death and the immediately preceding day have the greatest impact. This finding suggests that there are important short-term influences at work, which is consistent with recent reports of changes in electrocardiogram patterns within hours of exposure to airborne particles. 15
We note that there appears to be heterogeneity in the response to particles evident in Table 2. This heterogeneity in response has been noted in several studies recently. 11,37 Exploration of the cause of such heterogeneity is now a major priority. Demographic factors do not appear to be major predictors. 11,37 Chronic obstructive pulmonary disease has been noted as an effect modifier in one study. 40 The factors responsible for this heterogeneity in the APHEA-2 cities was the focus of an earlier paper 23 (which did not address harvesting), and the mean concentration of NO2 and the mean temperature appeared to explain most of the variability. Because this analysis is more limited, we have not attempted to repeat those analyses.
The APHEA-2 collaborative group consists of: K. Katsouyanni, G. Touloumi, E. Samoli, A. Gryparis, Y. Monopolis, E. Aga, and D. Panagiotakos (Greece, coordinating center); C. Spix, A. Zanobetti, and H. E. Wichmann (Germany); H. R. Anderson, R. Atkinson, and J. Ayres (U.K.); S. Medina, A. Le Tertre, P. Quenel, L. Pascale, and A. Boumghar (Paris); J. Sunyer, M. Saez, F. Ballester, S. Perez-Hoyos, J. M. Tenias, E. Alonso, K. Kambra, E. Aranguez, A. Gandarillas, I. Galan, J. M. Ordonez (Spain); M. A. Vigotti, G. Rossi, E. Cadum, G. Costa, L. Albano, D. Mirabelli, P. Natale, L. Bisanti, A. Bellini, M. Baccini, A. Biggeri, P. Michelozzi, V. Fano, A. Barca, and F. Forastiere (Italy); D. Zmirou and F. Balducci (Grenoble, France); J. Schouten and J. Vonk (The Netherlands); J. Pekkanen and P. Tittanen (Finland); L. Clancy and P. Goodman (Ireland); A. Goren and R. Braunstein (Israel); C. Schindler (Switzerland); B. Wojtyniak, D. Rabczenko, and K. Szafraniek (Poland); B. Kriz, M. Celko, and J. Danova (Prague); A. Paldy, J. Bobvos, A. Vamos, G. Nador, I. Vincze, P. Rudnai, and A. Pinter (Hungary); E. Niciu, V. Frunza, and V. Bunda, (Romania); M. Macarol-Hitti and P. Otorepec (Slovenia); Z. Dörtbudak and F. Erkan (Turkey); B. Forsberg and B. Segerstedt, (Sweden); F. Kotesovec and J. Skorkovski (Teplice, Czech Republic).
1. Schwartz J, Dockery DW. Increased mortality in Philadelphia associated with daily air pollution concentrations. Am Rev Respir Dis 1992; 145: 600–604.
2. Pope CA, Dockery DW, Schwartz J. Review of epidemiologic evidence of health effects of particulate air pollution. Inhal Toxicol 1995; 7: 1–18.
3. Schwartz J. Air pollution and daily mortality: a review and meta-analysis. Environ Res 1994; 64: 36–52.
4. Touloumi G, Pocock SJ, Katsouyanni K, Trichopoulos D. Short-term effects of air pollution on daily mortality in Athens: a time-series analysis. Int J Epidemiol 1994; 23: 957–967.
5. Saldiva PH, Pope CA, Schwartz J, et al
. Air pollution and mortality in elderly people: a time series study in Sao Paulo, Brazil. Arch Environ Health 1995; 50: 159–163.
6. Rossi G, Vigotti MA, Zanobetti A, Repetto F, Giannelle V, Schwartz J. Air pollution and cause specific mortality in Milan, Italy, 1980–1989. Arch Environ Health 1999; 54: 158–164.
7. Hoek G, Schwartz J, Groot B, Eilers P. Effects of ambient particulate matter and ozone on daily mortality in Rotterdam, The Netherlands. Arch Environ Health 1997; 52: 455–463.
8. Ostro B, Sanchez JM, Aranda C, Eskeland GS. Air pollution and mortality: results from a study of Santiago, Chile. J Expo Anal Environ Epidemiol 1996; 6: 97–114.
9. Katsouyanni K, Touloumi G, Spix C, et al.
Short-term effects of ambient sulfur dioxide and particulate mater on mortality in 12 European cities: results from time series data from the APHEA project. BMJ 1997; 314: 1658–1663.
10. Dominici F, Samet J, Zeger SL. Combining evidence on air pollution and daily mortality from the largest 20 U.S. cities: a hierarchical modeling strategy. R Stat Soc Ser A 2000; 163: 263–302.
11. Zanobetti A, Schwartz J, Dockery DW. Airborne particles are a risk factor for hospital admissions for heart and lung disease. Environ Health Perspect 2000; 108: 1071–1077.
12. Schwartz J. Assessing confounding, effect modification, and thresholds in the association between ambient particles and daily deaths. Environ Health Perspect 2000; 108: 563–568.
13. Anderson HR. Health effects of air pollution episodes. In: Holgate ST, Sament JM, Koren HS, Maynard RL, eds. Air Pollution and Health. London: Academic Press, 1999.
14. Peters A, Liu E, Verrier RL, et al.
Air pollution and incidence of cardiac arrhythmia. Epidemiology 2000; 11: 11–17.
15. Gold DR, Litonjua A, Schwartz J, Litonjua A, Schwartz J. Ambient pollution and heart rate variability. Circulation 2000; 101: 1267–1273.
16. Zelikoff JT, Nadziejko C, Fang T, Gordon C, Premdass C, Cohen MD. Short term, low-dose inhalation of ambient particulate matter exacerbates ongoing pneumococcal infections in Streptococcus pneumoniae
-infected rats. In: Phalen RF, Bell YM, eds. Proceedings of the Third Colloquium on Particulate Air Pollution and Human Health, Air Pollution Health Effects Laboratory. vol. 8. Irvine: University of California, 1999; 94–101.
17. Zeger SL, Dominici F, Samet J. Harvesting-resistant estimates of air pollution effects on mortality. Epidemiology 1999; 10: 171–175.
18. Schwartz J. Harvesting and long-term exposure effects in the relationship between air pollution and mortality. Am J Epidemiology 2000; 151: 440–448.
19. Schwartz J. Is there harvesting in the association of airborne particles with daily deaths and hospital admissions? Epidemiology 2001; 12: 55–61.
20. Murray CJ, Nelson CR. State-space modeling of the relationship between air quality and mortality. J Air Waste Manag Assoc 2000; 50: 1075–1080.
21. Zanobetti A, Wand MP, Schwartz J, Ryan L. Generalized additive distributed lag models: quantifying mortality displacement. Biostatistics 2000; 13: 279–292.
22. Katsouyanni K, Schwartz J, Spix C, et al.
Short term effects of air pollution on health: a European approach using epidemiologic time-series data: the APHEA protocol. J Epidemiol Community Health 1996; 50 (suppl 1): S12–S18.
23. Katsouyanni K, Touloumi G, Samoli E, et al.
Confounding and effect modification in the short-term effects of ambient particles on total mortality: results from 29 European cities within the APHEA2 project. Epidemiology 2001; 12: 521–531.
24. Michelozzi P, Forastiere F, Fusco D, et al.
Air pollution and daily mortality in Rome, Italy. Occup Environ J 1998; 55: 605–610.
25. Hastie T, Tibshirani R. Generalized Additive Models. London: Chapman and Hall, 1990.
26. Deleted in proof.
27. Cleveland WS, Devlin SJ. Robust locally weighted regression and smoothing scatterplots. J Am Stat Assoc 1988; 74: 829–836.
28. Akaike H. Information theory and an extension of the maximum likelihood principal. In: Petrov BN, Csaki F, eds. Second International Symposium on Information Theory. Budapest: Akademiai Kiado, 1973.
29. Schwartz J, Spix C, Touloumi G, et al.
Methodological issues in studies of air pollution and daily counts of deaths or hospital admissions. J Epidemiol Community Health 1996; 50 (suppl 1): S3–S11.
30. Schwartz J. Air pollution and hospital admissions for heart disease in eight U.S. counties. Epidemiology 1999; 10: 17–22.
31. Diggle PJ. Time Series. New York: Oxford University Press, 1990.
32. Brumback BA, Ryan LM, Schwartz J, Neas LM, Stark PC, Burge HA. Transitional regression models with application to environmental time series. J Am Stat Assoc 2000; 95: 16–28.
33. Pope CA III, Schwartz J. Time series for the analysis of pulmonary health data. Am J Respir Crit Care Med 1996; 154: S229–S233.
34. Schwartz J. The distributed lag between air pollution and daily deaths. Epidemiology 2000; 11: 320–26.
35. Berkey CS, Hoaglin DC, Antczak-Bouckoms A, Mosteller F, Colditz GA. Meta-analysis of multiple outcomes by regression with random effects. Stat Med 1998; 17: 2537–2550.
36. Kelsall JE, Samet JM, Zeger SL, Xu J. Air pollution and mortality in Philadelphia: 1974–1988. Am J Epidemiol 1997; 146: 750–762.
37. Samet JM, Zegar SL, Dominici F, et al. The National Morbidity, Mortality, and Air Pollution Study. Part II. Morbidity and mortality from air pollution in the United States. Res Rep Health Effects Institute 2000; 94 (pt 2):5–79.
38. Dockery DW, Pope CA III, Xu X, et al. An association between air pollution and mortality in six U.S. cities. N Engl J Med 1993; 329: 1753–1759.
39. Pope CA III, Thun MJ, Namboodiri M, et al.
Particulate air pollution as a predictor of mortality in a prospective study of U.S. adults. Am J Respir Crit Care Med 1995; 151: 669–674.
40. Sunyer J, Schwartz J, Tobias A, Macfarlane D, Garcia J, Anto JM. Patients with chronic obstructive pulmonary disease are at increased risk of death associated with urban particle air pollution: a case-crossover analysis. Am J Epidemiol 2000; 151: 50–56.
This article has been cited 73 time(s).
Bmc Medical Research MethodologyReducing and meta-analysing estimates from distributed lag non-linear modelsBmc Medical Research Methodology
American Journal of EpidemiologyAre the acute effects of particulate matter on mortality in the national morbidity, mortality, and air pollution study the result of inadequate control for weather and season? A sensitivity analysis using flexible distributed lag modelsAmerican Journal of Epidemiology
Journal of Epidemiology and Community HealthLinking environmental effects to health impacts: a computer modelling approach for air pollutionJournal of Epidemiology and Community Health
Atmospheric EnvironmentAssessment and prediction of short term hospital admissions: the case of Athens, GreeceAtmospheric Environment
Journal of the Air & Waste Management AssociationRevisiting a Population-Dynamic Model of Air Pollution and Daily Mortality of the Elderly in PhiladelphiaJournal of the Air & Waste Management Association
European Respiratory JournalShort-term effects of ambient particles on mortality in the elderly: results from 28 cities in the APHEA2 projectEuropean Respiratory Journal
Bulletin of VolcanologyThe respiratory health hazards of volcanic ash: a review for volcanic risk mitigationBulletin of Volcanology
Bulletin of VolcanologyCharacterization of respirable volcanic ash from the Soufriere Hills volcano, Montserrat, with implications for human health hazardsBulletin of Volcanology
Clinical ScienceCardiovascular effects of air pollutionClinical Science
American Journal of EpidemiologyChronic Particulate Exposure, Mortality, and Coronary Heart Disease in the Nurses' Health StudyAmerican Journal of Epidemiology
Environmental Science & TechnologyA Probabilistic characterization of the relationship between fine particulate matter and mortality: Elicitation of European expertsEnvironmental Science & Technology
Environmental Health PerspectivesThe effect of dose and timing of dose on the association between airborne particles and survivalEnvironmental Health Perspectives
Filtration & Separation
Why we need effective air filtration
Filtration & Separation, 39(3):
EnvironmetricsAnalysis of health outcome time series data in epidemiological studiesEnvironmetrics
Journal of the Air & Waste Management Association
Health effects of fine particulate air pollution: Lines that connect
Journal of the Air & Waste Management Association, 56(6):
Inhalation ToxicologyMortality effects of longer term exposures to fine particulate air pollution: Review of recent epidemiological evidenceInhalation Toxicology
Journal of Exposure Science and Environmental EpidemiologyAssociation between PM2.5 and all-cause and specific-cause mortality in 27 US communitiesJournal of Exposure Science and Environmental Epidemiology
Journal of the Royal Statistical Society Series C-Applied StatisticsA Bayesian hierarchical distributed lag model for estimating the time course of risk of hospitalization associated with particulate matter air pollutionJournal of the Royal Statistical Society Series C-Applied Statistics
Inhalation ToxicologyExposure of mice to concentrated ambient particulate matter results in platelet and systemic cytokine activationInhalation Toxicology
Journal of Epidemiology and Community Health
Predicted health impacts of urban air quality management
Journal of Epidemiology and Community Health, 58(2):
American Journal of Respiratory and Critical Care MedicineReduction in fine particulate air pollution and mortality - Extended follow-up of the Harvard six cities studyAmerican Journal of Respiratory and Critical Care Medicine
Occupational and Environmental MedicineShort-term associations between fine and coarse particles and hospital admissions for cardiorespiratory diseases in six French citiesOccupational and Environmental Medicine
Environmental Health PerspectivesThe temporal pattern of respiratory and heart disease mortality in response to air pollutionEnvironmental Health Perspectives
Annual Review of Public HealthTime-series studies of particulate matterAnnual Review of Public Health
Science of the Total EnvironmentKey airborne pollutants - the impact on healthScience of the Total Environment
International Journal of BiometeorologyExcess deaths during the 2004 heatwave in Brisbane, AustraliaInternational Journal of Biometeorology
American Journal of Epidemiology
Airborne particulate matter and mortality: Timescale effects in four US cities
American Journal of Epidemiology, 157():
American Journal of EpidemiologyResponse to Dr. Smith: Timescale-dependent mortality effects of air pollutionAmerican Journal of Epidemiology
Journal of Epidemiology and Community HealthApheis: public health impact of PM10 in 19 European citiesJournal of Epidemiology and Community Health
Journal of Toxicology and Environmental Health-Part A-Current IssuesA new longitudinal design for identifying subgroups of the population who are susceptible to the short-term effects of ambient air pollutionJournal of Toxicology and Environmental Health-Part A-Current Issues
Environmental ResearchDetermining the threshold effect of ozone on daily mortality: an analysis of ozone and mortality in Seoul, Korea, 1995-1999Environmental Research
Inhalation ToxicologyMortality displacement and distributed lag modelsInhalation Toxicology
Environmental HealthHealth effects of ambient air pollution - recent research development and contemporary methodological challengesEnvironmental Health
Journal of the Royal Statistical Society Series A-Statistics in Society
Model choice in time series studies of air pollution and mortality - Comments on the paper by Peng, Dominici and Louis
Journal of the Royal Statistical Society Series A-Statistics in Society, 169():
Environmental Health PerspectivesAn evaluation of the environmental and health effects of vehicle exhaust catalysts in the United KingdomEnvironmental Health Perspectives
Analytical ChemistryApplication of infrared laser desorption vacuum-UV single-photon ionization mass spectrometry for analysis of organic compounds from particulate matter filter samplesAnalytical Chemistry
Journal of Toxicology and Environmental Health-Part A-Current IssuesWhy do estimates of the acute and chronic effects of air pollution on mortality differ?Journal of Toxicology and Environmental Health-Part A-Current Issues
Journal of the Air & Waste Management Association
Analysis of primary fine particle National Ambient Air Quality Standard metrics
Journal of the Air & Waste Management Association, 56(2):
Environmental Health PerspectivesComparison of oxidative properties, light absorbance, and total and elemental mass concentration of ambient PM2.5 collected at 20 European sitesEnvironmental Health Perspectives
American Journal of Respiratory and Critical Care MedicineMortality displacement in the association of ozone with mortality - An analysis of 48 cities in the United StatesAmerican Journal of Respiratory and Critical Care Medicine
Environmental Health PerspectivesCause-specific mortality and the extended effects of particulate pollution and temperature exposureEnvironmental Health Perspectives
Cell Biology and Toxicology
Effects of ambient air particles on nitric oxide production in macrophage cell lines
Cell Biology and Toxicology, 20(4):
American Journal of EpidemiologyRecent exposure to particulate matter and C-reactive protein concentration in the Multi-Ethnic Study of AtherosclerosisAmerican Journal of Epidemiology
British Medical Journal
Admissions for myocardial infarction and World Cup football: database survey
British Medical Journal, 325():
British Medical BulletinAmbient air pollution and healthBritish Medical Bulletin
European Respiratory JournalShort-term effects of nitrogen dioxide on mortality: an analysis within the APHEA projectEuropean Respiratory Journal
Environmental Health PerspectivesParticulate Matter (PM) Research Centers (1999-2005) and the Role of Interdisciplinary Center-Based ResearchEnvironmental Health Perspectives
Environmental Health PerspectivesOn the relationship between time-series studies, dynamic population studies, and estimating loss of life due to short-term exposure to environmental risksEnvironmental Health Perspectives
Atmospheric EnvironmentAnalysis of particle-associated semi-volatile aromatic and aliphatic hydrocarbons in urban particulate matter on a daily basisAtmospheric Environment
Resource and Energy EconomicsDose response functions and the harvesting effectResource and Energy Economics
Occupational and Environmental MedicineInteracting effects of particulate pollution and cold temperature on cardiorespiratory mortality in ScotlandOccupational and Environmental Medicine
Environmental Health Perspectives
The US Environmental Protection Agency particulate matter health effects research centers program: A midcourse report of status, progress, and plans
Environmental Health Perspectives, 111(8):
Environmental Health PerspectivesOzone and daily mortality in Shanghai, ChinaEnvironmental Health Perspectives
International Journal of Mass SpectrometryAn on-line aerosol laser mass spectrometer with three, easily interchangeable laser based ionisation methods for characterisation of inorganic and aromatic compounds on particlesInternational Journal of Mass Spectrometry
International Statistical Review
Health effects of air pollution: A statistical review
International Statistical Review, 71(2):
Bulletin of VolcanologyMortality in England during the 1783-4 Laki Craters eruptionBulletin of Volcanology
BiometricsBayesian Distributed Lag Models: Estimating Effects of Particulate Matter Air Pollution on Daily MortalityBiometrics
Environmental Health Perspectives
The concentration-response relation between PM2.5 and daily deaths
Environmental Health Perspectives, 110():
Air pollution and health
European Respiratory JournalParticulate air pollution and hospital admissions for card iorespiratory diseases: are the elderly at greater risk?European Respiratory Journal
Environmental Health Perspectives
Is the association of airborne particles with daily deaths confounded by gaseous air pollutants? An approach to control by matching
Environmental Health Perspectives, 112(5):
Environmental ResearchDiurnal temperature range and daily mortality in Shanghai, ChinaEnvironmental Research
Environmental Science and Pollution ResearchApplication of GIS and modelling in health risk assessment for urban road mobilityEnvironmental Science and Pollution Research
Atmospheric EnvironmentShort-term effects of daily air pollution on mortalityAtmospheric Environment
Environment InternationalA framework to spatially cluster air pollution monitoring sites in US based on the PM2.5 compositionEnvironment International
Cadernos De Saude PublicaAir pollution and respiratory and cardiovascular diseases: a time series study in Cubatao, Sao Paulo State, BrazilCadernos De Saude Publica
Chemical Industry & Chemical Engineering QuarterlyShort-Term Health Effects of Particulate Air Pollution With Special Reference to the Needs of Southern European CountriesChemical Industry & Chemical Engineering Quarterly
Environment InternationalAssociations between short/medium-term variations in black smoke air pollution and mortality in the Glasgow conurbation, UKEnvironment International
Environmental Health PerspectivesAssociations between Fine and Coarse Particles and Mortality in Mediterranean Cities: Results from the MED-PARTICLES ProjectEnvironmental Health Perspectives
Atmospheric EnvironmentDifferential effects of fine and coarse particles on daily emergency cardiovascular hospitalizations in Hong KongAtmospheric Environment
air pollution; mortality; mortality displacement
© 2002 Lippincott Williams & Wilkins, Inc.
What does "Remember me" mean?
By checking this box, you'll stay logged in until you logout. You'll get easier access to your articles, collections,
media, and all your other content, even if you close your browser or shut down your
To protect your most sensitive data and activities (like changing your password),
we'll ask you to re-enter your password when you access these services.
What if I'm on a computer that I share with others?
If you're using a public computer or you share this computer with others, we recommend
that you uncheck the "Remember me" box.
Highlight selected keywords in the article text.
Data is temporarily unavailable. Please try again soon.
Readers Of this Article Also Read