Next Article in Journal
Dynamics of Chinese Diet Divergence from Chinese Food Pagoda and Its Association with Adiposity and Influential Factors: 2004–2011
Next Article in Special Issue
Response to Buonocore et al. Comments on Wendt Hess et al. “Assessing Agreement in Exposure Classification between Proximity-Based Metrics and Air Monitoring Data in Epidemiology Studies of Unconventional Resource Development.” Int. J. Environ. Res. Public Health 2019, 16, 3055
Previous Article in Journal
Effects of Teaching Games on Decision Making and Skill Execution: A Systematic Review and Meta-Analysis
Previous Article in Special Issue
Assessing Agreement in Exposure Classification between Proximity-Based Metrics and Air Monitoring Data in Epidemiology Studies of Unconventional Resource Development
 
 
Reply published on 14 January 2020, see Int. J. Environ. Res. Public Health 2020, 17(2), 512.
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Comment

Air Monitoring Stations Far Removed From Drilling Activities Do Not Represent Residential Exposures to Marcellus Shale Air Pollutants. Response to the Paper by Hess et al. on Proximity-Based Unconventional Natural Gas Exposure Metrics

1
Center for Climate, Health, and the Global Environment, Harvard T.H. Chan School of Public Health, Boston, MA 02115, USA
2
Environmental Health Sciences, Mailman School of Public Health, Columbia University, New York, NY 10027, USA
3
Department of Environmental Health, Harvard T.H. Chan School of Public Health, Boston, MA 02115, USA
4
Environmental and Occupational Health, Colorado School of Public Health, University of Colorado Denver, Denver, CO 80204, USA
*
Author to whom correspondence should be addressed.
Int. J. Environ. Res. Public Health 2020, 17(2), 504; https://doi.org/10.3390/ijerph17020504
Submission received: 25 October 2019 / Accepted: 7 January 2020 / Published: 13 January 2020
(This article belongs to the Special Issue Shale Gas and Fracking: Impacts on Health and the Environment)

Introduction

In their study “Assessing Agreement in Exposure Classification between Proximity-Based Metrics and Air Monitoring Data in Epidemiology Studies of Unconventional Resource Development” Hess et al. perform a validation of the well-activity (WA) proximity models used in epidemiologic research studies of unconventional natural gas development (UNGD) [1]. The study lacks sufficient evidence to support the authors’ conclusion that there is significant exposure misclassification in the epidemiology that finds health effects around unconventional oil and gas (O&G) wells. Major flaws in the Hess et al. study’s validation method include inappropriate selection of air pollution data, missing information on other air pollution sources and meteorology, inappropriate choice of statistical methods, and misrepresentation of the purpose of proximity metrics used in epidemiological studies. These substantial shortcomings negate the study’s ability to answer the question: does a WA proximity metric correlate with actual levels of air pollutants in a region with UNGD?

Distance Between Air Monitoring Sites and Wells: A Mismatch of Geographical Scale

A number of issues with the Hess et al. study stem from the large distance between most of the air monitoring stations and natural gas wells in Pennsylvania. Air pollutant measures from locations within a few kilometers of natural gas drilling sites would be most essential to proximity WA model validation for estimating exposures to air pollutants [2]. However, the closest distance between the air monitor and natural gas wells in the Hess et al. study range from 0.3 km to 56 km, the median distances range from 100 to 250 km, and the maximum distances exceed 400 km (Hess et al. Figure 4 and Table 2).
Hess et al. argue that the monitor locations are suitable surrogates for home residences, as would be the case in much of the O&G proximity epidemiology. Given that the 100–250 km median distances between wells and the monitors serving as surrogates, this assumption seems implausible. Additionally, most of the air monitoring sites used in the study do not represent the potential exposure to Marcellus Shale-associated air pollutants for the over one million Pennsylvanians living within 1.6 km of an active oil or gas (O&G) well [3].
The epidemiological studies that have used the proximity WA or similar models have all observed the greatest health implications in the densest areas of O&G well activity, generally at residences within 20 km of a well [2,4,5,6,7,8,9,10,11,12,13,14,15]. McMullin et al. (2018) recently evaluated air pollution measured from monitors at distances from O&G wells ranging from 0.1 km to 1.13 km to the closest O&G site and found a number of VOCs, notably benzene, ethylbenzene, and n-nonane [16]. In order to successfully assess the validity of the WA metrics used in the existing epidemiological studies, the study should be performed at a comparable geographical scale.
The distances used by Hess et al. do not match expected transport distances for many of the pollutants monitored. Benzene, SO2, NOx, and CO have fairly short atmospheric lifetimes and can only travel short distances, especially since these sources are close to the ground. For example, SO2 from point sources was recently found to have an atmospheric lifetime of 4–12 h [17]. This means that for the lowest median distance to an SO2 monitoring station, 119 km, the wind speed needed to be approximately 10–30 km/h along that vector for the monitor to detect it. Since wind speeds at 30m altitude are generally under this range, this renders SO2 from many of these wells undetectable to many of the monitors. In short, most of the air monitoring sites used in this validation study are incapable of detecting Marcellus Shale well-related air pollutants because they are located too far away. Furthermore, recent research by Schade and Roest using a regulatory monitor in Texas detected air pollution coming from the Eagle Ford shale [18]. The Schade and Roest study was done using methods that accounted for meteorology and other sources in the region, demonstrating the utility of these methods. If similar monitoring and analysis were performed in Pennsylvania, it may produce similar results.

Misrepresentation of the WA Metrics Used in Epidemiologic Studies

Prior literature has not claimed the WA metric only captures air pollution exposure. In fact, the metric was designed to capture complex, multifactorial exposures, including air, water, noise, and light pollution as well as community disruption, psychosocial stress, increased traffic, and other potential impacts related to O&G activity [19]. Most prior epidemiologic studies note this [2,4,5,6,7,8,9,10,11,12,13,14,15] and acknowledge the limitations of using a WA metric. Therefore, Hess et al. have misinterpreted the purpose of the WA metric. They have made an “apples to oranges” comparison—or in this case, they compared six air pollutants, with differing fate and transport properties, to a multifactorial exposures and stresses captured by the WA metrics that vary both spatially and temporally and act over short distances and time scales.

Lack of Incorporation of Meteorology

The Hess et al. study does not incorporate meteorology in its analyses, which helps explain the variance and fluctuations of UNGD contaminants that may be of significant public health concern [1]. It is fairly standard practice to incorporate at least upwind/downwind relationships between pollutant sources and monitors, and successful monitoring at this distance would likely require collection of fine-scale meteorological data, and ideally back-trajectory analysis. Other potentially important metrics include cloud cover, wind speed, wind direction, and time of day. These meteorological metrics affect intensity, frequency, and duration of air pollutants [20]. For example, in Brown et al.’s 2015 study, a box model for air pollution dispersion in an area with O&G activity showed that air pollution levels increased at night when there is less air mixing [20]. Direct measurements in areas of intensive O&G activity indicate that nighttime ambient benzene concentrations are double daytime concentrations [21].

Not Incorporating Other Sources of Monitored Air Pollution

The Hess et al. study also does not incorporate other air pollution sources that could be detected by these monitors. The pollutants which these air pollution monitors measure come from a number of sources common in the area, including vehicular traffic, combustion of coal and gas for electricity generation, and home heating, not to mention natural gas processing and existing conventional O&G wells. There is no mention of apportioning total air pollution to sources. If these other air pollution sources are not accounted and adjusted for, the study design used by Hess et al. should be unable to detect the signal of interest.

Inappropriate Use of Kappa Statistic

To determine how well the WA metrics match air pollution exposures, the authors employ a Kappa statistic [22,23]. The Kappa statistic is used to assess inter-rater agreement of qualitative categorization. It is common in clinical applications to assess diagnostic agreement between physicians, qualitative rating of pain severity, and similar applications. It is not designed to be applied to continuous data, like air pollution measurements and WA metrics. Since, as mentioned earlier, a number of sources can contribute to air pollution measured at a monitor, the use of the Kappa statistic to test whether WA categorization matches air pollution categorization is an inappropriate application.
Further, even if the use of the Kappa statistic was appropriate here, a paper entitled “Misinterpretation and Misuse of the Kappa Statistic” notes that “with continuous data grouped into ordinal categories, usually just for the convenience of the investigator, Kappa is so arbitrary it is virtually meaningless.” [24] Additionally, the Kappa statistic is highly sensitive to the distribution of the underlying data and can produce unreliable results, including giving low results when the underlying raters are actually concordant. Some recommend that the Kappa test be supplemented with a number of other tests. Alternative statistics, like running regressions, would be more suited to the authors’ intended application.

Implications of Oil and Gas Health Impacts Research

The results from epidemiological, exposure assessment, and other health impacts studies have ramifications for public health. These types of studies can form the evidentiary basis for a number of strategies intended to protect the health of populations around O&G development. These include pollution controls, permitting decisions, and setback rules designed to establish, and then require, minimum distances between O&G wells and sensitive receptors and vulnerable populations, like schools, daycares, hospitals, and residences. Therefore, to protect public health, it is critical that studies on health impacts around O&G development are robust.

Conclusions

Because of the unrepresentative air monitoring locations and inappropriate statistical methods, the Hess et. al. study does not improve our understanding of the residential exposures associated with O&G wells. For these same reasons, the Hess et al. study also does not provide information useful for decisions relevant to the health of communities nearby. Instead, the study only confirms well understood limitations of WA metrics at long distances that bear little relevance to the geographical scales in the existing epidemiology, or communities near O&G development. We recommend that future efforts and resources be directed at obtaining comprehensive, publicly available air pollutant emission rates from natural gas well sites suitable for modelling exposures to close-by residents, assessing exposure at relevant proximity to O&G wells, and validating and improving proximity WA models. Finally, we recommend that any future industry research in this area be done in collaboration with academic scientists and public health researchers.

Author Contributions

Conceptualization, J.J.B. and J.A.C.; Writing—Original Draft Preparation, J.J.B., J.A.C., R.C., and L.M.; Writing—Review & Editing: J.J.B., J.A.C., R.C., J.D.S., and L.M. All authors have read and agreed to the published version of the manuscript.

Funding

This research received no external funding.

Conflicts of Interest

The authors declare no conflict of interest.

References

  1. Wendt Hess, J.; Bachler, G.; Momin, F.; Sexton, K. Assessing Agreement in Exposure Classification between Proximity-Based Metrics and Air Monitoring Data in Epidemiology Studies of Unconventional Resource Development. Int. J. Environ. Res. Public Health 2019, 16, 3055. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  2. Allshouse, W.B.; Adgate, J.L.; Blair, B.D.; McKenzie, L.M. Spatiotemporal Industrial Activity Model for Estimating the Intensity of Oil and Gas Operations in Colorado. Environ. Sci. Technol. 2017, 51, 10243–10250. [Google Scholar] [CrossRef] [PubMed]
  3. Czolowski, E.D.; Santoro, R.L.; Srebotnjak, T.; Shonkoff, S.B.C. Toward Consistent Methodology to Quantify Populations in Proximity to Oil and Gas Development: A National Spatial Analysis and Review. Environ. Health Perspect. 2017, 125, 086004. [Google Scholar] [CrossRef] [PubMed]
  4. McKenzie, L.M.; Blair, B.; Hughes, J.; Allshouse, W.B.; Blake, N.J.; Helmig, D.; Milmoe, P.; Halliday, H.; Blake, D.R.; Adgate, J.L. Ambient Nonmethane Hydrocarbon Levels Along Colorado’s Northern Front Range: Acute and Chronic Health Risks. Environ. Sci. Technol. 2018, 52, 4514–4525. [Google Scholar] [CrossRef]
  5. McKenzie, L.M.; Guo, R.; Witter, R.Z.; Savitz, D.A.; Newman, L.S.; Adgate, J.L. Birth Outcomes and Maternal Residential Proximity to Natural Gas Development in Rural Colorado. Environ. Health Perspect. 2014, 122, 412–417. [Google Scholar] [CrossRef] [Green Version]
  6. McKenzie, L.M.; Allshouse, W.B.; Byers, T.E.; Bedrick, E.J.; Serdar, B.; Adgate, J.L. Childhood hematologic cancer and residential proximity to oil and gas development. PLoS ONE 2017, 12, e0170423. [Google Scholar] [CrossRef] [Green Version]
  7. Allshouse, W.B.; McKenzie, L.M.; Barton, K.; Brindley, S.; Adgate, J.L. Community noise and air pollution exposure during the development of a multi-well oil and gas pad. Environ. Sci. Technol. 2019, 53, 7126–7135. [Google Scholar] [CrossRef]
  8. McKenzie, L.M.; Allshouse, W.; Daniels, S. Congenital heart defects and intensity of oil and gas well site activities in early pregnancy. Environ. Int. 2019, 132, 104949. [Google Scholar] [CrossRef]
  9. McKenzie, L.M.; Allshouse, W.B.; Burke, T.; Blair, B.D.; Adgate, J.L. Population Size, Growth, and Environmental Justice Near Oil and Gas Wells in Colorado. Environ. Sci. Technol. 2016, 50, 11471–11480. [Google Scholar] [CrossRef]
  10. McKenzie, L.M.; Crooks, J.; Peel, J.L.; Blair, B.D.; Brindley, S.; Allshouse, W.B.; Malin, S.; Adgate, J.L. Relationships between indicators of cardiovascular disease and intensity of oil and natural gas activity in Northeastern Colorado. Environ. Res. 2019, 170, 56–64. [Google Scholar] [CrossRef]
  11. Whitworth, K.W.; Marshall, A.K.; Symanski, E. Maternal residential proximity to unconventional gas development and perinatal outcomes among a diverse urban population in Texas. PLoS ONE 2017, 12, e0180966. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  12. Rasmussen, S.G.; Ogburn, E.L.; McCormack, M.; Casey, J.A.; Bandeen-Roche, K.; Mercer, D.G.; Schwartz, B.S. Association Between Unconventional Natural Gas Development in the Marcellus Shale and Asthma Exacerbations. JAMA Intern. Med. 2016, 176, 1334–1343. [Google Scholar] [CrossRef] [PubMed]
  13. Casey, J.A.; Ogburn, E.L.; Rasmussen, S.G.; Irving, J.K.; Pollak, J.; Locke, P.A.; Schwartz, B.S. Predictors of Indoor Radon Concentrations in Pennsylvania, 1989–2013. Environ. Health Perspect. 2015, 123, 1130–1137. [Google Scholar] [CrossRef] [PubMed]
  14. Casey, J.A.; Goin, D.E.; Rudolph, K.E.; Schwartz, B.S.; Mercer, D.; Elser, H.; Eisen, E.A.; Morello-Frosch, R. Unconventional natural gas development and adverse birth outcomes in Pennsylvania: The potential mediating role of antenatal anxiety and depression. Environ. Res. 2019, 177, 108598. [Google Scholar] [CrossRef]
  15. Casey, J.A.; Savitz, D.A.; Rasmussen, S.G.; Ogburn, E.L.; Pollak, J.; Mercer, D.G.; Schwartz, B.S. Unconventional Natural Gas Development and Birth Outcomes in Pennsylvania, USA. Epidemiology 2015, 27, 163. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  16. McMullin, T.; Bamber, A.; Bon, D.; Vigil, D.; Van Dyke, M. Exposures and Health Risks from Volatile Organic Compounds in Communities Located near Oil and Gas Exploration and Production Activities in Colorado (U.S.A.). Int. J. Environ. Res. Public Health 2018, 15, 1500. [Google Scholar] [CrossRef] [Green Version]
  17. Fioletov, V.E.; McLinden, C.A.; Krotkov, N.; Li, C. Lifetimes and emissions of SO2 from point sources estimated from OMI. Geophys. Res. Lett. 2015, 42, 1969–1976. [Google Scholar] [CrossRef]
  18. Schade, G.W.; Roest, G. Analysis of non-methane hydrocarbon data from a monitoring station affected by oil and gas development in the Eagle Ford shale, Texas. Elem. Sci. Anth. 2016, 4, 000096. [Google Scholar] [CrossRef] [Green Version]
  19. Koehler, K.; Ellis, J.H.; Casey, J.A.; Manthos, D.; Bandeen-Roche, K.; Platt, R.; Schwartz, B.S. Exposure Assessment Using Secondary Data Sources in Unconventional Natural Gas Development and Health Studies. Environ. Sci. Technol. 2018, 52, 6061–6069. [Google Scholar] [CrossRef]
  20. Brown, D.R.; Lewis, C.; Weinberger, B.I. Human exposure to unconventional natural gas development: A public health demonstration of periodic high exposure to chemical mixtures in ambient air. J. Environ. Sci. Health Part A 2015, 50, 460–472. [Google Scholar] [CrossRef]
  21. Halliday, H.S.; Thompson, A.M.; Wisthaler, A.; Blake, D.R.; Hornbrook, R.S.; Mikoviny, T.; Müller, M.; Eichler, P.; Apel, E.C.; Hills, A.J. Atmospheric benzene observations from oil and gas production in the Denver-Julesburg Basin in July and August 2014. J. Geophys. Res. Atmos. 2016, 121, 11055–11074. [Google Scholar] [CrossRef] [Green Version]
  22. Rosner, B. Fundamentals of Biostatistics; Nelson Education: Scarborough, UK, 2015. [Google Scholar]
  23. McHugh, M.L. Interrater reliability: The kappa statistic. Biochem. Med. Zagreb 2012, 22, 276–282. [Google Scholar] [CrossRef] [PubMed]
  24. Maclure, M.; Willett, W.C. Misinterpretation and misuse of the kappa statistic. Am. J. Epidemiol. 1987, 126, 161–169. [Google Scholar] [CrossRef] [PubMed]

Share and Cite

MDPI and ACS Style

Buonocore, J.J.; Casey, J.A.; Croy, R.; Spengler, J.D.; McKenzie, L. Air Monitoring Stations Far Removed From Drilling Activities Do Not Represent Residential Exposures to Marcellus Shale Air Pollutants. Response to the Paper by Hess et al. on Proximity-Based Unconventional Natural Gas Exposure Metrics. Int. J. Environ. Res. Public Health 2020, 17, 504. https://doi.org/10.3390/ijerph17020504

AMA Style

Buonocore JJ, Casey JA, Croy R, Spengler JD, McKenzie L. Air Monitoring Stations Far Removed From Drilling Activities Do Not Represent Residential Exposures to Marcellus Shale Air Pollutants. Response to the Paper by Hess et al. on Proximity-Based Unconventional Natural Gas Exposure Metrics. International Journal of Environmental Research and Public Health. 2020; 17(2):504. https://doi.org/10.3390/ijerph17020504

Chicago/Turabian Style

Buonocore, Jonathan J., Joan A. Casey, Rachel Croy, John D. Spengler, and Lisa McKenzie. 2020. "Air Monitoring Stations Far Removed From Drilling Activities Do Not Represent Residential Exposures to Marcellus Shale Air Pollutants. Response to the Paper by Hess et al. on Proximity-Based Unconventional Natural Gas Exposure Metrics" International Journal of Environmental Research and Public Health 17, no. 2: 504. https://doi.org/10.3390/ijerph17020504

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop