Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

  • Research Article
  • Published:

Estimating effects of ambient PM2.5 exposure on health using PM2.5 component measurements and regression calibration

Journal of Exposure Science & Environmental Epidemiology volume 16, pages 30–38 (2006)Cite this article

An Erratum to this article was published on 14 March 2007

A Corrigendum to this article was published on 14 September 2006

Abstract

Most air pollution and health studies conducted in recent years have examined how a health outcome is related to pollution concentrations from a fixed outdoor monitor. The pollutant effect estimate in the health model used indicates how ambient pollution concentrations are associated with the health outcome, but not how actual exposure to ambient pollution is related to health. In this article, we propose a method of estimating personal exposures to ambient PM2.5 (particulate matter less than 2.5 μm in diameter) using sulfate, a component of PM2.5 that is derived primarily from ambient sources. We demonstrate how to use regression calibration in conjunction with these derived values to estimate the effects of personal ambient PM2.5 exposure on a continuous health outcome, forced expiratory volume in 1 s (FEV1), using repeated measures data. Through simulation, we show that a confidence interval (CI) for the calibrated estimator based on large sample theory methods has an appropriate coverage rate. In an application using data from our health study involving children with moderate to severe asthma, we found that a 10 μg/m3 increase in PM2.5 was associated with a 2.2% decrease in FEV1 at a 1-day lag of the pollutant (95% CI: 0.0–4.3% decrease). Regressing FEV1 directly on ambient PM2.5 concentrations from a fixed monitor yielded a much weaker estimate of 1.0% (95% CI: 0.0–2.0% decrease). Relatively small amounts of personal monitor data were needed to calibrate the estimate based on fixed outdoor concentrations.

This is a preview of subscription content, access via your institution

Access options

Buy this article

  • Purchase on Springer Link
  • Instant access to full article PDF
Buy now

Prices may be subject to local taxes which are calculated during checkout

Figure 1
Figure 2

Similar content being viewed by others

Abbreviations

CI:

confidence interval

ETS:

environmental tobacco smoke

FEV1:

forced expiratory volume in 1 second

F INF :

indoor concentration of a pollutant divided by the ambient level of that pollutant

MSE:

mean-squared error

PM2.5:

particulate matter less than 2.5 μm in diameter

F PEX :

personal pollutant exposure concentration divided by the ambient level of that pollutant

RTI:

Research Triangle Institute

REML:

restricted maximum likelihood

SD:

standard deviation

SE:

standard error.

References

  • Agresti A. Categorical Data Analysis, 2nd edn. John Wiley & Sons, New Jersey, 2002.

    Book  Google Scholar 

  • Carroll R.J., Ruppert D., and Stephanski L.A. Measurement Error in Nonlinear Models. Chapman and Hall, London, 1995.

    Book  Google Scholar 

  • Ebelt S.T., Petkau A.J., Vedal S., Fisher T.V., and Brauer M. Exposure of chronic obstructive pulmonary disease patients to particulate matter: relationships between personal and ambient concentrations. J Air Waste Manage Assoc 2000: 50: 1081–1094.

    Article  CAS  Google Scholar 

  • Fuller W.A. Measurement Error Models. John Wiley & Sons, New York, 1987.

    Book  Google Scholar 

  • Haukka J.K. Correction for covariate measurement error in generalized linear models — a bootstrap approach. Biometrics 1995: 51: 1127–1132.

    Article  CAS  Google Scholar 

  • Janssen N.A.H., Hoek G., Brunekreef B., Harssema H., Mensink I., and Zuidhof A. Personal sampling of particles in adults: relation among personal, indoor, and outdoor air concentrations. Am J Epidemiol 1998: 146: 537–547.

    Article  Google Scholar 

  • Janssen N.A.H., Hoek G., Harssema H., and Brunekreef B. Childhood exposure to PM10: relation between personal, classroom, and outdoor concentrations. Occup Environ Med 1997: 54: 888–894.

    Article  CAS  Google Scholar 

  • Janssen N.A.H., Hoek G., Harssema H., and Brunekreef B. Personal exposure to fine particles in children correlates closely with ambient fine particles. Arch Environ Health 1999: 54: 95–101.

    Article  CAS  Google Scholar 

  • Lawless P.A. Portable air sampling apparatus including non-intrusive activity monitor and methods of using same, US Patent 6,502,469 B2, awarded 1/7/2003, Research Triangle Institute, Research Triangle Park, NC 27709.

  • Lawless P.A., Rodes C.E., and Ensor D.S. Multiwavelength absorbance of filter deposits for determination of environmental tobacco smoke and black carbon. Atmos Environ 2004: 38: 3373–3383.

    Article  CAS  Google Scholar 

  • Liu L.-J.S., Box M., Kalman D., Kaufman J., Koenig J., Larson T., Lumley T., Sheppard L., and Wallace L. Exposure assessment of particulate matter for susceptible populations in Seattle, WA. Environ Health Perspect 2003: 111: 909–918.

    Article  Google Scholar 

  • Ozkaynak H., Xue J., Spengler J., Wallace L., Pellizzari E., and Jenkins P. Personal exposure to airborne particles and metals: results from the Particle TEAM Study in Riverside, CA. J Exposure Anal Environ Epidemiol 1996: 6: 57–78.

    CAS  Google Scholar 

  • Rabinovitch N., Zhang L., Murphy J., Vedal S., Dutton S.J., and Gelfand E.W. Health effects of wintertime ambient air pollution on children with moderate to severe asthma. J Allergy Clin Immunol 2004: 114: 1131–1137.

    Article  CAS  Google Scholar 

  • Richardson A.M., and Welsh A.H. Asymptotic properties of restricted maximum likelihood (REML) estimates for hierarchical mixed models. Aust J Stat 1994: 36: 31–43.

    Article  Google Scholar 

  • Rodes C.E., Lawless P.A., Evans G.F., Sheldon L.S., Williams R.W., Vette A.F., Creason J.P., and Walsh D. The relationships between personal PM exposures for elderly populations and indoor and outdoor concentrations for three retirement center scenarios. J Exposure Anal Environ Epidemiol 2001: 11: 103–115.

    Article  CAS  Google Scholar 

  • Rojas-Bracho L., Suh H.H., and Koutrakis P. Relationship among personal, indoor, and out-door particle concentrations for individuals with COPD. J Exposure Anal Environ Epidemiol 2000: 10: 294–306.

    Article  CAS  Google Scholar 

  • Rosner B., Willett W.C., and Spiegelman D. Correction of logistic regression relative risk estimates and confidence intervals for systematic within-person measurement error. Stat Med 1989: 8: 1051–1069.

    Article  CAS  Google Scholar 

  • Sarnat J.A., Long C.M., Koutrakis P., Coull B.A., Schwartz J., and Suh H.H. Using sulfur as a tracer of outdoor fine particulate matter. Environ Sci Technol 2002: 36: 5305–5314.

    Article  CAS  Google Scholar 

  • Williams R., Suggs J., Zweidinger R., Evans G., Creason J., Kwok R., Rodes C., Lawless P., and Sheldon L. The 1998 Baltimore Particulate Epidemiology — Exposure Study: Part 1. Comparison of ambient, residential outdoor, indoor and apartment particulate matter monitoring. J Exposure Anal Environ Epidemiol 2000: 10: 518–532.

    Article  CAS  Google Scholar 

  • Wilson W., Mage D., and Grant L. Estimating separately personal exposure to ambient and nonambient particulate matter for epidemiology and risk assessment: why and how. J Air Waste Manage Assoc 2000: 50: 1167–1183.

    Article  CAS  Google Scholar 

  • Wilson W., and Suh H. Fine particles and coarse particles: concentration relationships relevant to epidemiologic studies. J Air Waste Manage Assoc 1997: 47: 1238–1249.

    Article  CAS  Google Scholar 

  • Zeger S., Thomas D., Dominici F., Samet J.M., Schwartz J., Dockery D., and Cohen A. Exposure measurement error in time-series studies of air pollution: concepts and consequences. Environ Health Perspect 2000: 108: 419–426.

    Article  CAS  Google Scholar 

Download references

Acknowledgements

This work was funded in part by three grants: EPA R825702, Thrasher Research Fund 02816-8 and, Colorado Tobacco Research Program R2-001.

Author information

Authors and Affiliations

  1. Division of Biostatistics, National Jewish Medical and Research Center, Denver, Colorado, USA

    Matthew Strand

  2. Department of Environmental and Occupational Health Sciences, University of Washington School of Public Health and Community Medicine, Seattle, Washington, USA

    Sverre Vedal

  3. RTI International, Center for Aerosol Technology, Research Triangle Park, North Carolina, USA

    Charles Rodes

  4. Department of Civil, Environmental and Architectural Engineering, University of Colorado, Boulder, Colorado, USA

    Steven J Dutton

  5. Department of Pediatrics, National Jewish Medical and Research Center, Denver, Colorado, USA

    Erwin W Gelfand & Nathan Rabinovitch

Authors
  1. Matthew Strand
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Sverre Vedal
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Charles Rodes
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Steven J Dutton
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Erwin W Gelfand
    View author publications

    You can also search for this author in PubMed Google Scholar

  6. Nathan Rabinovitch
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Matthew Strand.

Appendix A

Appendix A

Estimating the Ratio of PM2.5 to Sulfate Exposure Factors (λ)

A general approach to estimating λ

(i) Perform a mixed model regression of total personal PM2.5 on ambient PM2.5, where fixed and subject-specific random terms for both the y-intercept and slope of ambient PM2.5 are included in the model. If nonambient and ambient PM2.5 exposures are approximately independent of each other, which our data suggested (also see Wilson and Suh, 1997), then the personal slope (fixed slope plus subject-specific random slope) indicates the average fraction of ambient PM2.5 that the subject is exposed to within the study period, while the y-intercept indicates average exposure to nonambient sources. (ii) Repeat the previous step, with personal sulfate and ambient sulfate. (iii) The estimate of λ for a given subject is then obtained by dividing the personal PM2.5 slope by the personal sulfate slope. (iv) The common λ can be obtained by averaging subject-specific estimates.

Notes on Calculations

Data for the 21 “low-ETS” children were used to estimate subject-specific λ values. (A “low-ETS” subject was defined as one who (i) claimed to not live in a home with a smoker, and (ii) infrequently had days with high measured personal ETS exposure within the study period.) The average estimated value across subjects (λ̂=(1/n)∑iλ̂i) was 0.78 (median=0.75, SD=0.285, min=0.44, max=1.59). The regression of total personal PM2.5 on ambient PM2.5 for the remaining 29 subjects produced counterintuitive results; the slope estimate for the group was near zero, with no strong outliers apparent in the data. But slope estimates were similar between ETS groups for the sulfate regressions (and confirmed when analyzing another strong ambient-sourced PM2.5 component, elemental carbon). This reaffirmed our belief that exposures to ambient PM2.5 were probably more similar between ETS groups and that more complex statistical methods were necessary to model accurately those not in the low-ETS group.

Details for Simulations

Based on observed fits, we assumed ωit′∼N(0,10) (i.e., ωit′ has a normal distribution with mean=0, variance=10), uit∼N(0,2), ωit∼N(0,8), ɛit∼N(0,0.14) and φi∼N(0, 0.014) for all i and t, and that the correlation between FEV1 responses within an individual on two consecutive days was 0.44. For the models below, values for the random terms were sampled independently from these distributions for all i and t, except for ɛit, which were autocorrelated within subjects (AR(1) model) with ρ=0.44. The variance of u was estimated by examining differences in estimated exposures between subjects within days since we did not have true ambient exposures for subjects. The variance for ωit was then determined using σω2=σω′2−σu2. Within each simulation, the following steps were taken.

The Health Model

(i) Ambient exposures were simulated as XitA=0.46Zt+φiZt+ωit. (ii) FEV1 outcomes were then simulated as Yit=2–0.00435XitA+ɛit. (iii) Some FEV1 outcomes were randomly set to missing values so that the pattern and amount of responses were similar to actual data. (Subjects missed having FEV1 values on certain days due to absence or because the recorded FEV1 values were flagged as invalid.) (iv) The generated FEV1 data were then fit as a simple linear function of actual ambient PM2.5 concentrations (with correlated errors) to obtain a simulated estimate of β1.

The Personal Exposure Model

(i) Ambient exposures were simulated (independently of those for the health model) using XitA=0.46Zt+φiZt+ωit for the same subjects and days that personal monitor information was available in the actual study. (ii) Exposures with measurement error were then created as XitA*=XitA+uit. (iii) The generated ambient PM2.5 exposures and actual ambient PM2.5 concentrations were then used to fit model (7), yielding a simulated estimate of θ1.

Rights and permissions

Reprints and permissions

About this article

Cite this article

Strand, M., Vedal, S., Rodes, C. et al. Estimating effects of ambient PM2.5 exposure on health using PM2.5 component measurements and regression calibration. J Expo Sci Environ Epidemiol 16, 30–38 (2006). https://doi.org/10.1038/sj.jea.7500434

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1038/sj.jea.7500434

Keywords

This article is cited by

Search

Advanced search

Quick links