Objectives of the Public Health Review

Evaluation of Scientific Literature Relevant to the

Objectives of the Public Health Review

Objectives of the Public Health Review

In order to evaluate the analysis of health impacts in the draft SGEIS in a broader
environmental and public health context, DOH reviewed and evaluated relevant
emerging scientific literature investigating the environmental health and community
health dimensions of HVHF. This was not intended to be a comprehensive review of all
the published scientific literature on HVHF. Rather, the emerging literature was
surveyed, and studies with direct environmental health relevance were reviewed to
better understand the adequacy of the current science to inform questions regarding
public health impacts of HVHF.
Two major types of peer-reviewed scientific studies were the focus of the literature
review process – studies of impacts to environmental media and studies of health
outcomes. As is very often true in environmental health science, both types of studies
have limitations that make it difficult to draw firm conclusions about environmental
causation of disease from any one study or small group of studies. Strong conclusions
about disease causation in environmental health derive from a collective assessment of
the weight of evidence from a large body of research that often takes many years to
conduct.4
Studies of environmental impacts investigate the effects of HVHF activities on
environmental media such as air, water and soil. Contamination of environmental media

has the potential to contribute to human health impacts if people experience exposures
to those contaminants (for example, through breathing contaminated air or drinking
contaminated water) that are large enough to cause a biological effect. However,
studies of environmental impacts often do not attempt to directly demonstrate whether
contamination of environmental media has resulted in significant human exposure or
whether a health effect occurs as a result of an exposure. Other studies report on
observed human health outcomes potentially associated with HVHF activity (i.e.,
environmental epidemiology studies). Health outcome studies related to HVHF activity
focus on health effects reported among people living near HVHF drilling sites. Most
health outcome studies can only suggest a potential statistical relationship between a
source of environmental contamination and the observed health outcomes. These
studies are limited in their ability to demonstrate that an actual exposure to the source
has occurred or that exposure to an environmental source causes a health outcome.
Health outcome studies vary in the complexity of their design and how rapidly they can
be carried out. Some health outcome study designs that are relatively simple and quick
to conduct are often also limited in their ability to account for other unrelated factors
(usually referred to as bias and confounding) that might contribute to the observed
health effects. Longitudinal prospective cohort studies are among the strongest study
designs, but are very expensive and take years to conduct.

HVHF Health Outcome Studies

The public health science surrounding HVHF shale-gas development is currently limited
and studies are largely exploratory in nature. Peer-reviewed epidemiologic studies were
not found that employ robust study designs addressing possible associations between
HVHF activities and adverse health outcomes while providing adequate control for
confounding and bias. Scientific studies that contain relevant information investigating

human health outcomes potentially associated with HVHF activities are briefly
summarized below.
Birth Outcomes
An unpublished 2013 revision to a 2012 working paper by Hill reports results of a study
using data on 2,459 natural gas wells completed in Pennsylvania between 2006 and
2010, along with vital records for the years 2003 through 2010. The study compared
birth outcomes for infants born to mothers living within selected fixed distances from
spudded Marcellus Shale wells (the "existing well” infant group) with outcomes for
infants born to mothers living within the same distances from future wells (the “future
well” infant group). The outcomes considered were birth weight, gestation, five-minute
APGAR (Appearance, Pulse, Grimace, Activity, Respiration) score (a health indicator
assessed immediately following birth), small-for-gestational-age (yes/no), premature
(yes/no), congenital anomalies (yes/no) and infant death (yes/no). The investigator
reported that after specifying a fixed distance of 2.5 km from an existing or future well,
and after controlling for multiple risk factors (e.g., maternal age, race, education, WIC
status, marital status, insurance status and smoking), the “existing well” infant group
had statistically significantly lower averages for birth weight and 5-minute APGAR
score, as well as statistically significantly higher prevalence of low birth weight and
small-for-gestational age, compared with the “future wells” infant group. No statistically
significant differences were observed for prematurity, congenital anomalies or
infant death.
Hill’s conclusion that a “causal” relationship between natural gas development and birth
outcomes was established may overstate the findings of this single study. The statistical
approach used by the investigator, the differences-in-differences method, had in the
past been employed primarily by social scientists but is increasingly used in public
health studies. In the context of this study, this statistical approach assumed that, in the absence of drilling, average outcomes for the “existing wells” and “future wells” infant
groups would have followed parallel paths over time. Because differences may have
existed between the two study groups with regard to potential risk factors not
incorporated into the statistical analyses (e.g., prenatal care adequacy, maternal
lifestyles, pre-existing chronic diseases, perinatal complications) it is possible that this
"parallel paths" assumption may not have been appropriate. However, the author was
able to demonstrate that, at least with regard to measured characteristics, there were no
indications that this key assumption was not met.
A similar study by McKenzie et al. (2014) evaluated potential associations between
maternal residence near natural gas wells and birth outcomes in a retrospective cohort
study of 124,842 births between 1996 and 2009 in rural Colorado. Specifically, the
authors investigated associations between natural gas well density and prevalence of
congenital heart defects, neural tube defects, oral clefts, preterm birth, and term low
birth weight. The least exposed (reference) group had no natural gas wells within a 10-
mile radius. After adjustments for maternal and infant covariates, prevalence of
congenital heart defects was significantly positively associated with increased exposure
to natural gas development, with an increase of 30% (95% CI: 20% to 50%) for the
highest exposure tertile when compared with the reference group. Prevalence of neural
tube defects was significantly positively associated with exposure to natural gas
development for the highest tertile of exposure, with an increase of 100% (95% CI: 0 to
390%) for the most exposed group when compared with the reference group. Exposure
was associated with lower odds of preterm birth and lower odds of low birth weight (i.e.,
the high exposure groups were less likely to be preterm or low birth weight). No
association was found between exposure and oral clefts.

It is notable that these two birth-outcome studies used similar study designs and
observed associations between birth-outcome measures and maternal proximity to
HVHF well pads. However, there is a lack of coherence between the observed
associations in the two studies. Hill reported associations with low birth weight and
APGAR score, but no associations with congenital defects. Conversely, McKenzie et al.
reported associations between proximity to well pads and some congenital defects, but
the highest exposure group had lower odds of preterm birth or low birth weight than the
reference group. Taken together, the relationship between maternal proximity to HVHF
well pads during pregnancy and birth outcomes, if any, is unclear.
Both birth-outcome studies used proximity to a drilling site as an exposure surrogate,
rather than actual environmental contaminant measurements. This was a reasonable
approach for an initial exploratory investigation, as it would be difficult and expensive to
characterize indoor and outdoor exposures to all potentially relevant environmental
agents (e.g., noise, air pollutants, groundwater pollutants, nighttime lighting) at
numerous homes and workplaces. However, studies that employ vicinity as a surrogate
for exposure cannot identify specific risk factors associated with the observed adverse
outcomes or establish how, if at all, these risk factors were related to HVHF. For
example, these studies cannot exclude the possibility that another factor unrelated to
HVHF also varied by residence proximity to drill pads and contributed to the observed
pattern of birth outcomes. The lack of coherent associations between this exposure
surrogate and comparable outcomes may reflect weaknesses in the use of this
exposure surrogate. The authors noted that greater specificity in exposure estimates
would be required to further explore the reported associations.