Model choice in time series studies of air pollution and mortality

Summary.  Multicity time series studies of particulate matter and mortality and morbidity have provided evidence that daily variation in air pollution levels is associated with daily variation in mortality counts. These findings served as key epidemiological evidence for the recent review of the US national ambient air quality standards for particulate matter. As a result, methodological issues concerning time series analysis of the relationship between air pollution and health have attracted the attention of the scientific community and critics have raised concerns about the adequacy of current model formulations. Time series data on pollution and mortality are generally analysed by using log‐linear, Poisson regression models for overdispersed counts with the daily number of deaths as outcome, the (possibly lagged) daily level of pollution as a linear predictor and smooth functions of weather variables and calendar time used to adjust for time‐varying confounders. Investigators around the world have used different approaches to adjust for confounding, making it difficult to compare results across studies. To date, the statistical properties of these different approaches have not been comprehensively compared. To address these issues, we quantify and characterize model uncertainty and model choice in adjusting for seasonal and long‐term trends in time series models of air pollution and mortality. First, we conduct a simulation study to compare and describe the properties of statistical methods that are commonly used for confounding adjustment. We generate data under several confounding scenarios and systematically compare the performance of the various methods with respect to the mean‐squared error of the estimated air pollution coefficient. We find that the bias in the estimates generally decreases with more aggressive smoothing and that model selection methods which optimize prediction may not be suitable for obtaining an estimate with small bias. Second, we apply and compare the modelling approaches with the National Morbidity, Mortality, and Air Pollution Study database which comprises daily time series of several pollutants, weather variables and mortality counts covering the period 1987–2000 for the largest 100 cities in the USA. When applying these approaches to adjusting for seasonal and long‐term trends we find that the Study's estimates for the national average effect of PM10 at lag 1 on mortality vary over approximately a twofold range, with 95% posterior intervals always excluding zero risk.

[1]  H. Akaike,et al.  Information Theory and an Extension of the Maximum Likelihood Principle , 1973 .

[2]  R. Shibata Selection of the order of an autoregressive model by Akaike's information criterion , 1976 .

[3]  Larry D. Haugh,et al.  Identification of Dynamic Regression (Distributed Lag) Models Connecting Two Time Series , 1977 .

[4]  G. Schwarz Estimating the Dimension of a Model , 1978 .

[5]  M. Stone Comments on Model Selection Criteria of Akaike and Schwarz , 1979 .

[6]  J. Rice Convergence rates for partially splined models , 1986 .

[7]  P. Speckman Kernel smoothing in partial linear models , 1988 .

[8]  R. Tibshirani,et al.  Linear Smoothers and Additive Models , 1989 .

[9]  R. Tibshirani,et al.  Generalized Additive Models , 1991 .

[10]  D. Dockery,et al.  Particulate air pollution and daily mortality in Steubenville, Ohio. , 1992, American journal of epidemiology.

[11]  B. Silverman,et al.  Nonparametric Regression and Generalized Linear Models: A roughness penalty approach , 1993 .

[12]  Joel Schwartz,et al.  Nonparametric smoothing in the analysis of air pollution and respiratory illness , 1994 .

[13]  B. Silverman,et al.  Nonparametric regression and generalized linear models , 1994 .

[14]  D. J. Donnell,et al.  Analysis of Additive Dependencies and Concurvities Using Smallest Additive Principal Components , 1994 .

[15]  B. Silverman,et al.  Nonparametric Regression and Generalized Linear Models: A roughness penalty approach , 1993 .

[16]  J. Schwartz,et al.  Total suspended particulate matter and daily mortality in Cincinnati, Ohio. , 1994, Environmental health perspectives.

[17]  Joel Schwartz,et al.  REVIEW OF EPIDEMIOLOGICAL EVIDENCE OF HEALTH EFFECTS OF PARTICULATE AIR POLLUTION , 1995 .

[18]  U. Epa Air Quality Criteria for Particulate Matter , 1996 .

[19]  Brian D. Ripley,et al.  Pattern Recognition and Neural Networks , 1996 .

[20]  D. Bates,et al.  Particulate air pollution. , 1996, Thorax.

[21]  Paul H. C. Eilers,et al.  Flexible smoothing with B-splines and penalties , 1996 .

[22]  Richard A. Davis,et al.  Introduction to time series and forecasting , 1998 .

[23]  S L Zeger,et al.  Air pollution and mortality in Philadelphia, 1974-1988. , 1997, American journal of epidemiology.

[24]  R. Shanmugam Introduction to Time Series and Forecasting , 1997 .

[25]  Paul H. C. Eilers,et al.  Direct generalized additive modeling with penalized likelihood , 1998 .

[26]  W. Navidi,et al.  Bidirectional case-crossover designs for exposures with time trends. , 1998, Biometrics.

[27]  S. Zeger,et al.  Does weather confound or modify the association of particulate air pollution with mortality? An analysis of the Philadelphia data, 1973-1980. , 1998, Environmental research.

[28]  R. Burnett,et al.  The effect of the urban ambient air pollution mix on daily mortality rates in 11 Canadian cities. , 1998, Canadian journal of public health = Revue canadienne de sante publique.

[29]  W. MacNee,et al.  Particulate Air Pollution: Injurious and Protective Mechanisms in the Lungs , 1999 .

[30]  S L Zeger,et al.  Harvesting-resistant estimates of air pollution effects on mortality. , 1999, Epidemiology.

[31]  A woman's own birth weight and gestational age predict her later risk of developing preeclampsia, a precursor of chronic disease. , 1999 .

[32]  D. Dockery,et al.  A case-crossover analysis of air pollution and mortality in Philadelphia. , 1999, Environmental health perspectives.

[33]  H. Kitchen,et al.  Comments on the Paper by , 1999 .

[34]  Scott L. Zeger,et al.  Harvesting-resistant estimates of air pollution effects on mortality. , 1999 .

[35]  M P Wand,et al.  Generalized additive distributed lag models: quantifying mortality displacement. , 2000, Biostatistics.

[36]  J Schwartz,et al.  The distributed lag between air pollution and daily deaths. , 2000, Epidemiology.

[37]  S L Zeger,et al.  Estimating particulate matter-mortality dose-response curves and threshold levels: an analysis of daily time-series for the 20 largest US cities. , 2000, American journal of epidemiology.

[38]  C. Morris,et al.  Inference for multivariate normal hierarchical models , 2000 .

[39]  F. Dominici,et al.  Combining evidence on air pollution and daily mortality from the 20 largest US cities: a hierarchical modelling strategy , 2000 .

[40]  Merlise A. Clyde,et al.  Model uncertainty and health effect studies for particulate matter , 2000 .

[41]  Philip J. Everson,et al.  Simulation From Wishart Distributions With Eigenvalue Constraints , 2000 .

[42]  S L Zeger,et al.  The National Morbidity, Mortality, and Air Pollution Study. Part I: Methods and methodologic issues. , 2000, Research report.

[43]  Joel Schwartz,et al.  Transitional Regression Models, with Application to Environmental Time Series , 2000 .

[44]  J Schwartz,et al.  Do respiratory epidemics confound the association between air pollution and daily deaths? , 2000, The European respiratory journal.

[45]  J. Schwartz,et al.  The National Morbidity, Mortality, and Air Pollution Study. Part II: Morbidity and mortality from air pollution in the United States. , 2000, Research report.

[46]  S. Wood Modelling and smoothing parameter estimation with multiple quadratic penalties , 2000 .

[47]  J. Schwartz,et al.  Harvesting and long term exposure effects in the relation between air pollution and mortality. , 2000, American journal of epidemiology.

[48]  J Schwartz,et al.  Is There Harvesting in the Association of Airborne Particles with Daily Deaths and Hospital Admissions? , 2001, Epidemiology.

[49]  R. Burnett,et al.  Associations between daily cause-specific mortality and concentrations of ground-level ozone in Montreal, Quebec. , 2001, American journal of epidemiology.

[50]  J. Schwartz,et al.  Investigating regional differences in short-term effects of air pollution on daily mortality in the APHEA project: a sensitivity analysis for controlling long-term trends and seasonality. , 2001, Environmental health perspectives.

[51]  J Schwartz,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 , 2001, Epidemiology.

[52]  Investigating regional differences in short-term effects of air pollution on daily mortality in the APHEA project , 2001 .

[53]  F. Dominici,et al.  On the use of generalized additive models in time-series studies of air pollution and health. , 2002, American journal of epidemiology.

[54]  Chong Gu Smoothing Spline Anova Models , 2002 .

[55]  Antonella Zanobetti,et al.  The Temporal Pattern of Mortality Responses to Air Pollution: A Multicity Assessment of Mortality Displacement , 2002, Epidemiology.

[56]  Scott L. Zeger,et al.  Air Pollution and Mortality , 2002 .

[57]  Scott L Zeger,et al.  Temperature and mortality in 11 cities of the eastern United States. , 2002, American journal of epidemiology.

[58]  T STATEMEN,et al.  Revised Analyses of Time-Series Studies of Air Pollution and Health , 2003 .

[59]  Bin Yu,et al.  Minimum Description Length Model Selection Criteria for Generalized Linear Models , 2003 .

[60]  J. Ghosh,et al.  Approximations and consistency of Bayes factors as model dimension grows , 2003 .

[61]  J Schwartz,et al.  Investigating the dose-response relation between air pollution and total mortality in the APHEA-2 multicity project , 2003, Occupational and environmental medicine.

[62]  Richard T Burnett,et al.  A Review of Time-Series Studies Used to Evaluate the Short-Term Effects of Air Pollution on Human Health , 2003, Reviews on environmental health.

[63]  R. Burnett,et al.  The Effect of Concurvity in Generalized Additive Models Linking Mortality to Ambient Particulate Matter , 2003, Epidemiology.

[64]  T. Hastie,et al.  Improved Semiparametric Time Series Models of Air Pollution and Mortality , 2004 .

[65]  Joel Schwartz,et al.  Analysis of health outcome time series data in epidemiological studies , 2004 .

[66]  F. Dominici,et al.  Time-series studies of particulate matter. , 2004, Annual review of public health.

[67]  A. Biggeri,et al.  [Meta-analysis of the Italian studies on short-term effects of air pollution--MISA 1996-2002]. , 2004, Epidemiologia e prevenzione.

[68]  J. Schwartz Is the association of airborne particles with daily deaths confounded by gaseous air pollutants? An approach to control by matching. , 2003, Environmental health perspectives.

[69]  Michael J Daniels,et al.  The National Morbidity, Mortality, and Air Pollution Study. Part III: PM10 concentration-response curves and thresholds for the 20 largest US cities. , 2004, Research report.

[70]  B. Ripley,et al.  Semiparametric Regression: Preface , 2003 .

[71]  Michael J Daniels,et al.  Underestimation of Standard Errors in Multi-site Time Series Studies , 2004, Epidemiology.

[72]  S. Zeger,et al.  Are 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 models. , 2005, American journal of epidemiology.

[73]  Francesca Dominici,et al.  Revised Analyses of the National Morbidity, Mortality, and Air Pollution Study: Mortality Among Residents Of 90 Cities , 2005, Journal of toxicology and environmental health. Part A.

[74]  A. Biggeri,et al.  Meta-analysis of the Italian Studies of Short-term Effects of Air Pollution (MISA), 1990–1999 , 2005, International journal of occupational and environmental health.

[75]  F. Dominici,et al.  Seasonal analyses of air pollution and mortality in 100 US cities. , 2005, American journal of epidemiology.

[76]  S. Zeger,et al.  Are the Acute Effects of PM10 on Mortality in NMMAPS the Result of Inadequate Control for Weather and Season? A Sensitivity Analysis Using Flexible Distributed Lag Models , 2005 .