The Annals of Applied Statistics

A locally adaptive process-convolution model for estimating the health impact of air pollution

Duncan Lee

Full-text: Open access

Abstract

Most epidemiological air pollution studies focus on severe outcomes such as hospitalisations or deaths, but this underestimates the impact of air pollution by ignoring ill health treated in primary care. This paper quantifies the impact of air pollution on the rates of respiratory medication prescribed in primary care in Scotland, which is a proxy measure for the prevalence of less severe respiratory disease. A novel bivariate spatiotemporal process-convolution model is proposed, which: (i) has increased computational efficiency via a tapering function based on nearest neighbourhoods; and (ii) has locally adaptive weights that outperform traditional distance-decay kernels. The results show significant effects of particulate matter on respiratory prescription rates which are consistent with severe endpoint studies.

Article information

Source
Ann. Appl. Stat., Volume 12, Number 4 (2018), 2540-2558.

Dates
Received: October 2017
Revised: February 2018
First available in Project Euclid: 13 November 2018

Permanent link to this document
https://projecteuclid.org/euclid.aoas/1542078055

Digital Object Identifier
doi:10.1214/18-AOAS1167

Mathematical Reviews number (MathSciNet)
MR3875711

Keywords
Air pollution bivariate spatiotemporal modelling process-convolution models respiratory medication rates

Citation

Lee, Duncan. A locally adaptive process-convolution model for estimating the health impact of air pollution. Ann. Appl. Stat. 12 (2018), no. 4, 2540--2558. doi:10.1214/18-AOAS1167. https://projecteuclid.org/euclid.aoas/1542078055


Export citation

References

  • Banerjee, S., Gelfand, A. E., Finley, A. O. and Sang, H. (2008). Gaussian predictive process models for large spatial data sets. J. R. Stat. Soc. Ser. B. Stat. Methodol. 70 825–848.
  • Bhopal, R., Steiner, M., Cezard, G., Bansal, N., Fischbacher, C., Simpson, C., Douglas, A. and Sheikh, A. (2015). Risk of respiratory hospitalization and death, readmission and subsequent mortality: Scottish health and ethnicity linkage study. Eur. J. Public Health 25 769–774.
  • Blangiardo, M., Finazzi, F. and Cameletti, M. (2016). Two-stage Bayesian model to evaluate the effect of air pollution on chronic respiratory diseases using drug prescriptions. Spatial Spatio-temporal Epidemiol. 18 1–12.
  • Centre for Environmental Data Analysis (2016). HadISDH: Gridded global monthly land surface humidity data version 3.0.0.2016p.
  • Centre for Environmental Data Analysis (2017). Climatic Research Unit (CRU) Time-Series (TS) version 4.01 of high-resolution gridded data of month-by-month variation in climate (Jan. 1901–Dec. 2016).
  • Congdon, P. (2005). Bayesian Models for Categorical Data. Wiley, Chichester.
  • Datta, A., Banerjee, S., Finley, A. O. and Gelfand, A. E. (2016). Hierarchical nearest-neighbor Gaussian process models for large geostatistical datasets. J. Amer. Statist. Assoc. 111 800–812.
  • Dibben, C. and Clemens, T. (2015). Place of work and residential exposure to ambient air pollution and birth outcomes in Scotland, using geographically fine pollution climate mapping estimates. Environ. Res. 140 535–541.
  • Dominici, F., Samet, J. and Zeger, S. (2000). Combining evidence on air pollution and daily mortality from the 20 largest US cities: A hierarchical modelling strategy. J. Roy. Statist. Soc. Ser. A 163 263–302.
  • Furrer, R., Genton, M. G. and Nychka, D. (2006). Covariance tapering for interpolation of large spatial datasets. J. Comput. Graph. Statist. 15 502–523.
  • Gelman, A., Carlin, J. B., Stern, H. S. and Rubin, D. B. (2004). Bayesian Data Analysis, 2nd ed. Chapman & Hall/CRC, Boca Raton, FL.
  • Higdon, D. (1998). A process-convolution approach to modelling temperatures in the North Atlantic Ocean. Environ. Ecol. Stat. 5 173–190.
  • Huang, G., Lee, D. and Scott, E. (2018). Multivariate space-time modelling of multiple air pollutants and their health effects accounting for exposure uncertainty. Stat. Med.. 37. 1134–1148.
  • Lee, D. (2018). Supplement to “A locally adaptive process-convolution model for estimating the health impact of air pollution.” DOI:10.1214/18-AOAS1167SUPPA, DOI:10.1214/18-AOAS1167SUPPB.
  • Lee, D., Mukhopadhyay, S., Rushworth, A. and Sahu, S. (2017). A rigorous statistical framework for spatio-temporal pollution prediction and estimation of its long-term impact on health. Biostatistics 18 370–385.
  • Liechty, J. C., Liechty, M. W. and Müller, P. (2004). Bayesian correlation estimation. Biometrika 91 1–14.
  • Loader, A., Hector, D., Sykes, D., Rose, R., Telfer, S. and Gray, S. (2016). Air Pollution in Scotland 2015. Technical report, Ricardo Energy & Environment.
  • Lu, H., Reilly, C. S., Banerjee, S. and Carlin, B. P. (2007). Bayesian areal wombling via adjacency modeling. Environ. Ecol. Stat. 14 433–452.
  • Meyer, S. and Held, L. (2014). Power-law models for infectious disease spread. Ann. Appl. Stat. 8 1612–1639.
  • Riebler, A., Held, L. and Rue, H. (2012). Estimation and extrapolation of time trends in registry data—Borrowing strength from related populations. Ann. Appl. Stat. 6 304–333.
  • Royal College of Physicians (2016). Every breath we take: The lifelong impact of air pollution. Available at https://www.rcplondon.ac.uk/projects/outputs/every-breath-we-take-lifelong-impact-air-pollution.
  • Watanabe, S. (2010). Asymptotic equivalence of Bayes cross validation and widely applicable information criterion in singular learning theory. J. Mach. Learn. Res. 11 3571–3594.

Supplemental materials

  • Supplement A: Additional results and data analysis. Section 1 contains additional data summaries, while Section 2 presents predictive analysis for the pollution data. Section 3 presents exploratory analysis of the prescription data, while Section 4 presents the reproducibility materials. Section 5 provides theoretical results, while Section 6 presents sensitivity analyses.
  • Supplement B: Supplementary data and code. Code and data for applying the model proposed in Section 3 to the GP data.