Risk Communication in EPA's Controlled Inhalation Exposure Studies and in Support
On March 28, 2017, the National Academy of Sciences, Engineering, and Medicine (NASEM) released a much-anticipated report on the Environmental Protection Agency's controlled human inhalation exposure studies. This essay reviews the ethical controversies that led to the genesis of the report, summarizes its key findings, and comments on its approach to informing human subjects about the risks of inhalation exposure studies. NASEM's report makes a valuable contribution to our understanding of the scientific and ethical issues involved in conducting human inhalation exposure studies. Its definition of "reasonably foreseeable risks" provides useful guidance to investigators, research participants, and institutional review board members.
On March 28, 2017, the national Academy of Sciences, Engineering, and Medicine (NASEM) released a much-anticipated report on the Environmental Protection Agency's (EPA) controlled human inhalation exposure studies. To understand the genesis of the document, a quick review of recent events is in order. [End Page 117]
A Short History of the EPA's Controlled Exposure Studies
Prior to 2006, the EPA adopted the Common Rule for intramural or extramural research funded by the agency.1 Although the EPA did not have a formal policy that applied to research sponsored by private companies (third-party research), it applied scientific and ethical standards to these studies on a case-by-case basis. In the 1990s, in support of requests to the agency to loosen its restrictions on pesticide residues on foods, private companies submitted data to the EPA from experiments that exposed human subjects to pesticides. The Environmental Working Group and other critics claimed these studies were unethical, because they exposed human subjects to excessive risks and would benefit industry but not society (Resnik 2012).
In 1998, the EPA decided it would no longer accept data from third-party human exposure studies for its regulatory decision-making and asked the National Research Council (NRC) to study the scientific and ethical issues involved in human pesticide experiments. Industry organizations objected to this decision and argued that the EPA had engaged in inappropriate rule-making, because it had formulated a policy without prior notice and public comment. A federal court agreed and ordered the agency to engage in appropriate rule-making concerning its human research regulations (Resnik 2012). In 2004, the NRC released its report, which concluded that the EPA could rely on third-party research for regulatory decision-making if it meets appropriate scientific and ethical standards. The report also recommended that the EPA adopt the Common Rule for third-party studies and establish a board to review such studies.
In the fall of 2004, as the EPA was beginning its rule-making, an ethical controversy emerged that had a dramatic impact on the outcome of this process. The Organic Consumers Association and other environmental groups alleged that a study known by the acronym CHEERS (Children's Environmental Exposure Research Study) unethically exposed children to pesticides and targeted poor, minority groups. CHEERS was an observational study sponsored by the EPA, the Centers for Disease Control and Prevention (CDC), and the Duvall County Health Department. The study was designed to measure children's in-home exposures to pesticides and other household chemicals. The investigators planned to recruit families of young children living in the Jacksonville, Florida, area who used high amounts of pesticides in their homes. Later, a control group of low pesticide-using families was added. The study involved five home visits over a two-year period and required parents to keep records of their use of pesticides and other chemicals and their children's activities. During the home visits, the investigators [End Page 118] would collect dust and soil samples from the home and blood and urine samples from the children.
Three institutional review boards (IRBs) approved the CHEERS study and classified it as minimal risk (Resnik 2012).2 However, critics claimed that CHEERS was an intentional exposure study, because parents would be exposing their children to pesticides and other chemicals. This criticism misrepresented the study design, because the investigators sought to collect data on children's exposures that were already occurring in the home and they did not ask the parents to start using or to increase their use of pesticides (or other chemicals) to participate in the study. Parents could also reduce or stop their use of pesticides or other chemicals and still remain in the study (Resnik 2012).
Despite the efforts of EPA to convince the public that CHEERS was not an intentional exposure experiment, criticism of the study intensified. The EPA suspended the study before any subjects were enrolled, pending further review by the agency. In April 2005, California Senator Barbara Boxer threatened to derail Stephen Johnson's candidacy for EPA administrator if the agency did not stop the study. The agency complied with this demand and cancelled the study shortly thereafter. Later that year, Congress approved an amendment to an appropriations bill, sponsored by Senator Boxer, that imposed a one-year moratorium on the EPA accepting any human exposure data from industry and that prohibited the agency from conducting human pesticide experiments involving children or pregnant or nursing women (Resnik 2007b).
The CHEERS controversy had a significant impact on the EPA's revisions of its human research regulations, which were adopted in 2006 and amended in 2013. EPA administrators and attorneys drafted special protections for children or pregnant or nursing women participating in EPA-funded studies to avoid the perception that investigators would be permitted to test pesticides or other dangerous substances on these vulnerable populations. The revised regulations went beyond the Congressional ban on pesticide testing on children or pregnant or nursing women and prohibited any EPA-funded intentional exposure studies involving members of these groups (EPA 2013; Resnik 2007a). The EPA can only fund observational studies involving children or pregnant or nursing women.
The regulations define intentional exposure research as: "a study of a substance in which the exposure to the substance experienced by a human subject participating in the study would not have occurred but for the human subject's participation in the study" (40 CFR 26.202a). Under this broad definition, the EPA cannot fund any exposure research involving children or pregnant or nursing women if the exposure would not have occurred without their participation in the research. [End Page 119] For example, the rule prohibits EPA-funded researchers from conducting studies in which children are intentionally exposed to substances they often encounter in their daily lives, such as sunscreens or insect repellants.3 EPA-funded researchers also cannot participate in clinical trials involving children or pregnant or nursing women (Resnik 2007a).
In addition to special protections for children and pregnant or nursing women participating in research, the revised EPA regulations include: ethical requirements for the participation of non-pregnant and non-nursing adults in third party intentional exposure pesticide studies; a prohibition of third party intentional exposure pesticide studies involving children or pregnant or nursing women; standards for relying on human subjects research for making regulatory decisions; a reaffirmation of the EPA's acceptance of the Common Rule for EPA-funded research on non-pregnant adults; and policies and procedures for EPA review of human subjects research, including the formation of a human studies board (EPA 2013).
The Nasem Report
As one can discern from the summary of recent events, EPA-related intentional exposure research, whether funded by the agency or third parties, has been embroiled in controversy since that late 1990s. It is against this background that the NASEM report was undertaken. The most recent dispute related to EPA human exposure research arose in 2012, when members of the public and Congress became concerned that the agency was not following legal, ethical, and scientific standards for conducting controlled experiments that exposed human subjects to airborne particles, diesel exhaust, and other forms of air pollution at its Human Studies Facility, a facility associated with the University of North Carolina at Chapel Hill (Bravender and Peterka 2015).
In the EPA's controlled human inhalation experiments (CHIEs), investigators observe non-pregnant, non-nursing adults breathing air in a chamber for up to several hours. The subjects breathe air that has first been purified and then had a defined amount of a pollutant added back in. Studies may also include a control group that breathes only purified air. The amount of air pollution subjects are exposed to is roughly equivalent to the amount they would be exposed to from breathing air for 24 hours in a city on a smoggy day (NASEM 2017). Participants may perform physical activities while in the breathing chamber, such as riding an exercise bike. Investigators perform a physical exam on participants and collect biological samples (such as blood or sputum) before and after the exposure; they also may ask subjects to perform a spirometry breathing test before and after the exposure and may perform a bronchoscopy on participants to collect cell samples [End Page 120] from their airways after the exposure. Investigators carefully monitor participants during the studies and during the recovery period from the bronchoscopy. Though most studies involve healthy volunteers, some include participants with mild to moderate asthma, so investigators can learn how pollutants affect them.
The purpose of the research is to obtain information about how air pollutants affect human physiology that cannot be obtained from observational studies on human populations or controlled, animal experiments. For example, investigators can learn how pollutants can affect respiratory and cardiovascular systems by triggering inflammation, oxidative stress, and other reactions. Data obtained from CHIEs may be useful in guiding EPA regulatory decisions concerning air quality standards (NASEM 2017). Subjects provide informed consent prior to participating in research and are financially compensated for their time and effort, usually receiving several hundred to a thousand dollars or more (EPA 2017). An IRB at the University of North Carolina reviews and oversees the studies.
The House Committee on Investigations and Oversight asked the EPA's Office of Inspector General (OIG) to determine whether the EPA's CHIEs comply with applicable laws, regulations, policies, and procedures. The OIG report found that the EPA was complying with applicable scientific, legal, and ethical standards, but it also recommended that the agency ensure that consent documents consistently inform subjects about risks and address the long-term risks of developing cancer from breathing air in the exposure chambers (EPA 2014). In response to the OIG report, the EPA sought independent advice from NASEM, which appointed the Committee on Assessing Toxicologic Risks to Human Subjects Used in Controlled Exposure Studies of Environmental Pollutants to evaluate the studies. NASEM asked the committee to:
Assess the utility of controlled inhalation exposure studies to inform and reduce uncertainties in setting air pollution standards to protect public health, and assess whether continuation of such studies is warranted.
Assess health risks to test subjects who participated in recent studies of air pollutants at EPA's clinical research facility.
If the committee supports continued conduct of human exposure studies, provide further guidance on methods for estimating levels of risk in controlled human exposure studies.
Provide advice on a template for characterizing reasonably foreseeable risks, which could be used in obtaining informed consent from potential study participants.(NASEM 2017, ix)
The committee, which included distinguished researchers and scholars from the fields of toxicology, medicine, public health, epidemiology, pharmacology, environmental science, risk assessment, law, and ethics, held public meetings over a two-year period and heard testimony from experts and stakeholders representing divergent viewpoints (NASEM 2017). The committee examined eight human [End Page 121] inhalation exposure studies conducted by the EPA and reviewed documents related to scientific review, research design, IRB review, adverse event reporting and management, and informed consent.
To complete the first task, the committee conducted an extensive review of the scientific literature related to the effects of air pollution on health and to the utility of conducting human exposure experiments to fill in knowledge gaps and reduce uncertainties pertaining to causal mechanisms and physiologic responses. The committee concluded that the EPA's controlled inhalation studies provide useful information for air pollution policies. According to the committee, a CHIE is justified if it meets two conditions: "(1) only when a CHIE study is expected to provide additional knowledge that informs policy decisions and regulation of pollutants that cannot be obtained by other means and (2) when it is reasonably foreseeable that the risks for study participants will not exceed transient and reversible biomarker or physiologic responses" (NASEM 2017, 6). The committee observed that some of the "transient" and "reversible" risks of CHIE studies include cough, chest pain, decreased respiratory function, wheeze, mild irritation to eyes, nose, throat, and airways, and that these physiologic responses usually resolved within a few hours after exposure to the polluted air (66).
To complete the second task, the committee reviewed adverse event reports from 1,390 human exposures. Only one qualified as a serious adverse event. In this case, a subject developed atrial fibrillation shortly after exposure to particulate matter. The subject's heart returned to normal sinus rhythm two hours after exposure, but the subject stayed in the hospital overnight for observation. To protect participants from harm, the study had excluded individuals with known heart conditions, but this subject had an undiagnosed heart condition so was not excluded from the study. The report concludes the short-term risks associated with adverse events are "unlikely to be large enough to warrant concern but … it is never possible to conclude that there is no risk" (4). Concerning the long-term health effects related to breathing polluted air for a couple of hours, the report acknowledged that exposures may contribute to the cumulative risks of chronic conditions, such as heart disease and lung cancer, but concluded that the added risk of participating in an EPA inhalation exposure study is "vanishingly small" (4).
The Ethics of Risk Communication
One of the key issues addressed by the OIG and the NASEM reports is how to inform subjects about reasonably foreseeable risks. The Common Rule requires that consent discussions and documents inform subjects about "reasonably foreseeable risks," but it does not define this term (45 CFR 46.116a2).4 According to [End Page 122] the OIG report, failure to define this important term has led to inconsistency in risk communication in consent documents. The OIG report recommended that the EPA define "reasonably foreseeable risks" to provide guidance for investigators and oversight committees and ensure that risks are communicated consistently in consent documents and discussions (EPA 2014).
The NASEM report defined a risk as "reasonably foreseeable" if "there is some credible evidence that harm might occur" (NASEM 2017, 9), and it cited Resnik (2013b) to clarify that credible evidence should come from "empirical research, past experience, or scientific or mathematical principles" (NASEM 2017, 88, 94). The NASEM report also noted that "An overly detailed list of all possibilities can result in a less valid consent process, since it groups the anticipated or likely risks with those that are only distant possibilities" (88).
The report acknowledged that addressing risks of concern to participants is part of valid consent, even if those risks are not reasonably foreseeable. For example, the report reviews epidemiologic studies that show an association between long-term exposure to air pollution and heart disease and lung cancer, but it concludes that the risks of developing these illnesses as a result of breathing air in an exposure chamber for a few hours are so small that there is no credible evidence to view these risks as reasonably foreseeable. However, the NASEM report also notes that since some subjects may be concerned about these risks, investigators should include them in informed consent documents and discussions.
The report recommended that investigators should communicate risks to subjects using real-life comparisons likely to be meaningful to subjects, because quantitative risks estimates can be difficult to understand. For example, some informed consent documents for EPA CHIEs compare the risks of breathing air in the exposure chamber for several hours to the risk of breathing air in a typical U.S. city on a smoggy day for 24 hours. The report states that this type of comparison can be useful but cautions that it should be supported by evidence, since exposures may vary, depending on the city, time of year, and so on. Investigators should justify risk comparisons in the research protocol.
In thinking about NASEM's definition of "reasonably foreseeable risk," it is important to note that disputes about how to interpret this phrase predate the recent controversy over the EPA's CHIE research. On March 7, 2013, the Office of Human Research Protections (OHRP) sent a determination letter to the University of Alabama at Birmingham, claiming that a multisite study known as SUPPORT (Surfactant, Positive Pressure, and Oxygenation Randomized Trial), funded by the National Institutes of Health (NIH), did not inform parents about reasonably foreseeable risks their infants would face by participating in the study. The purpose of the study was to determine the optimal blood oxygenation for preterm infants receiving respiratory support in the intensive care unit (ICU). The standard of care at that time was to maintain infants on 85–95% blood oxygen saturation. When the study began, there was evidence that lower oxygen [End Page 123] levels were associated with brain damage and death, whereas higher levels were associated with increased risk of retinopathy (OHRP 2013a). The study randomly assigned infants who were receiving intensive care and were born at 24 to 28 weeks of gestation to respiratory support with blood oxygen saturation at 85–89% or 91–95% (Resnik 2013b). Since the infants would be receiving respiratory support irrespective of participation in the study, the reasonably foreseeable risks of the research included those associated with the study's procedures or methods. The consent documents contained information about the risks of monitoring oxygen saturation with pulse oximeters but did not mention risks associated with assignment to the two different treatment groups. OHRP maintained that infants assigned to the 85–89% oxygen saturation group could have increased risks of brain damage or death, whereas those assigned to the 91–95% group could have an increased risk of retinopathy. OHRP claimed that because the consent form did not discuss these risks, the study failed to comply with the Common Rule's requirement to disclose reasonably foreseeable risks (OHRP 2013a).
Defenders of the study objected to OHRP's findings of noncompliance, arguing that since infants in both treatment groups would receive blood oxygenation within the standard of care, assignment to the different groups did not result in increased risks (Hudson, Guttmacher, and Collins 2013; Magnus and Caplan 2013; Wilfond et al. 2013). Moreover, defenders argued that the empirical data did not support the conclusion that blood oxygenation as low as 85% would increase the risk of brain injury of death or that oxygenation as high as 95% would increase the risk of retinopathy (Hudson, Guttmacher, and Collins 2013; Magnus and Caplan 2013). Indeed, the purposes of study were to obtain empirical data to determine whether there would be increased risks of adverse outcomes at either end of the oxygenation range within the standard of care and to establish an optimum level of oxygenation that minimized risks.
OHRP responded to this argument by claiming that, for some of the infants in the study, assignment to one of the treatment groups could increase their risks, because they could receive lower or higher blood oxygenation than they would have received if they had not been in the study. For example, if an infant would have received blood oxygenation of 90% if he or she had not been in the study, assignment to the 85–89% group could increase risks of brain injury or death, whereas assignment to the 91–95% group could increase the risk of retinopathy (OHRP 2013b). Other scholars agreed with OHRP's claim that the parents should have been told about the risks of assignment to the different treatment groups (Macklin and Shepherd 2013; Macklin et al. 2013).
While OHRP reaffirmed its claim that the parents should have been informed about these risks, it acknowledged that the term "reasonably foreseeable risks" is subject to interpretation, and it held a public meeting with clinical investigators, parents, ethicists, and other stakeholders to discuss the issues related to the SUPPORT study. The agency issued some guidance for disclosing reasonably [End Page 124] foreseeable risks in research that evaluates standards of care but did not define this contested phrase:
If researchers design and conduct a study for the purpose of evaluating a particular risk, then that risk is significant enough that it should be disclosed to the prospective subjects who are actually exposed to it. Prospective subjects ought to be informed of a particular risk that they may be exposed to, if a research study is evaluating that risk and the subjects could avoid it if they were to decide not to participate.(OHRP 2014)
Much of the controversy concerning informed consent in the SUPPORT study resulted from different interpretations of reasonably foreseeable risks (Resnik 2013a, 2013b). While defenders and critics of the SUPPORT study agreed that a risk is an adverse outcome (or harm), they disagreed about what makes a risk "reasonably foreseeable." Defenders of the SUPPORT study argued that the risks of assignment to the different blood oxygenation groups were not reasonably foreseeable because there was no empirical evidence for these risks, whereas critics claimed the risks were reasonably foreseeable because they were theoretically possible outcomes that one could anticipate in advance (Resnik 2013a, 2013b).5 Because the disputants did not use a common definition of "reasonably foreseeable risks," they were talking past each other (Resnik 2013a).
Both ways of characterizing reasonably foreseeable risks can create problems with the consent process. Most legal jurisdictions in the United States and many other countries now follow a consent disclosure rule known as the "reasonable person standard": one should disclose the information that a reasonable person would want to know (Berg et al. 2001). It is often the case in research that we can anticipate possible adverse outcomes that are speculative, but that most participants would want to know about. For example, in testing a new drug or biologic in human subjects in a Phase I clinical trial, there may be risks that have not been detected in prior animal experiments that could still materialize in people, such as adverse immune reactions, genetic or chromosomal damage, or various types of toxicity. Indeed, one of the main goals of a Phase I trial is to identify some of these risks and establish a safe dose (Dresser 2009; Kimmelman 2009). It would be irresponsible for investigators not to disclose these possible harms to human subjects in Phase I studies just because they have not been observed in animal experiments: most subjects would want to know, for example, that they could develop liver toxicity as a result of taking an experimental medication, even if toxicity has not occurred in animals at the dose they receive. Thus, the "empirical evidence" interpretation of "reasonably foreseeable" sets a bar that it is arguably too high for valid consent. [End Page 125]
If we think of risks as reasonably foreseeable if they are theoretically possible, then we may undermine the consent process by disclosing too many risks and distracting participants from thinking about the most important ones (NASEM 2017). Consent documents for research participation are notoriously long, complex, and difficult to read and understand (Flory and Emanuel 2004; Menikoff 2006). Describing all the theoretically possible risks of a study in a consent document could increase length and complexity unnecessarily and decrease readability and understandability. For example, it is theoretically possible that a participant in a Phase I trial could be struck by lightning or stung by bees during the study, but there is no good reason to disclose these risks in a consent document. A reasonable person would not need to be informed about these risks to make a valid choice to participate in the trial.
NASEM's definition of a "reasonably foreseeable risk" as one based on "credible evidence" strikes a sound compromise between these two extreme interpretations. The credible evidence interpretation is more permissive than the empirical evidence interpretation, because it includes evidence from "empirical research, past experience, or scientific or mathematical principles" (NASEM 2017, 88, 94). Thus, disclosure of a risk of toxicity in a Phase I study that has not been observed in animal experiments would be consistent with the credible evidence interpretation, because the risk of toxicity could be based on general principles of mammalian physiology or metabolism. The credible evidence interpretation is also more restrictive than the theoretically possible interpretation, because there might be no credible evidence for a risk that is theoretically possible. Thus, there would be no need to disclose the risk of being struck by lightning while participating in a Phase I study, because there would be no credible evidence for this risk, even if it is theoretically possible.
While the report's definition of "reasonably foreseeable risk" provides useful guidance for investigators and IRBs, its application of this definition to disclosing the risks of developing a chronic health problem from participating in CHIE research is ill-founded. The report says that these risks are not reasonably foreseeable because there is no credible evidence for them. To substantiate this claim, the report reviews epidemiological studies of the long-term risks of developing cancer and other chronic illnesses from years of exposure to air pollutants, such particulate matter. The report notes that evidence supports the hypothesis that risks are cumulative and dose-dependent, and it acknowledges that breathing air in an exposure chamber contributes to one's overall risk. The exposure to particulate matter in a CHIE study, for example, is 1/925 the exposure that one would receive from 40 years of normal activities (NASEM 2017, 66). However, the report asserts that since the added risk is "vanishingly small," it is not reasonably foreseeable. But a "vanishingly small" risk is not equivalent to no risk, nor is it the same as a risk that is not reasonably foreseeable. For example, the risk of being killed by a shark while swimming in the ocean (1/3.7 million) is vanishingly small, but it is [End Page 126] still a reasonably foreseeable risk based on credible evidence (Florida Museum of Natural History 2017). Likewise, participating in a CHIE study may only increase one's risk of developing a chronic illness by less than 0.1%, but this added risk is still reasonably foreseeable.
Moreover, claiming that these risks are not reasonably foreseeable while also asserting they should nevertheless be disclosed because some participants may be concerned about them goes beyond what is required by the Common Rule and introduces disclosure issues that may be difficult to resolve. Participants may be concerned about many risks that are not reasonably foreseeable. How should investigators decide which of these risks to disclose? Disclosing too many may confuse participants and distract them from focusing on important potential harms that are reasonably foreseeable.
A better way of thinking about the risks of developing chronic diseases (such as cancer or cardiovascular problems) from breathing polluted air in a CHIE study would be to say that while these risks are reasonably foreseeable, they are not significant, since they only make a relatively small contribution to one's overall air pollution risk profile. Since most people are routinely exposed to various types of air pollution during their daily lives, the added exposure from participating in a CHIE study will not have an appreciable impact on risks associated with breathing polluted air. This way of conceiving of the added risks of CHIE studies also lends itself to real-life comparisons of risks. If you tell someone that the risk of breathing air in an exposure chamber is roughly equivalent to the risk of breathing air for 24 hours in a city on a smoggy day, this might help them understand that the amount of pollution they encounter in the exposure chamber will be much less than the amount of pollution they are likely to encounter during various activities of daily life, such as walking in a city, driving to work, cooking, and so on. The comparison will not only help that person put the CHIE exposure in perspective, but it might encourage him or her to take steps to reduce other air pollution exposures to counteract the impact of the study on his or her cumulative exposures.
In conclusion, the NASEM report makes a valuable contribution to our understanding of the scientific and ethical issues related to CHIEs and illuminates some of the controversies concerning the EPA's human exposure research and defining foreseeable risk within the context of informed consent. The report's definition of "reasonably foreseeable risks" provides useful guidance to investigators, research participants, and IRB members. Additional empirical research and philosophical reflection on risk assessment and risk communication in human exposure research may help to clarify and resolve the remaining issues. [End Page 127]
1. The "Common Rule" refers to the Department of Health and Human Service's human subjects regulations (HHS 2009), which have been adopted by 17 federal agencies and were revised by the Obama Administration on January 19, 2017.
2. The Common Rule defines minimal risk as "the probability and magnitude of harm or discomfort anticipated in the research are not greater in and of themselves than those ordinarily encountered in daily life or during the performance of routine physical or psychological examinations or tests" (45 CFR 46.102i).
3. These studies would probably be allowed under the Common Rule because they would be classified as minimal risk; see Resnik 2007a.
4. It is not clear why the drafters of the Common Rule chose to use the term "reasonably foreseeable risks," but one may speculate that they borrowed it from negligence law. Although negligence law includes examples of reasonably foreseeable risks, it offers little insight into how to define this term (NASEM 2017; Resnik 2013b).
5. A risk is theoretically possible if it is consistent with our current scientific knowledge. This category of possibility is narrower than logically possible, because it is constrained by scientific facts and well-grounded theories.