what biomonitoring can and cannot tell us about causality in human health and ecological risk...

This article was downloaded by: [York University Libraries]On: 12 August 2014, At: 11:42Publisher: Taylor & FrancisInforma Ltd Registered in England and Wales Registered Number: 1072954 Registeredoffice: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Human and Ecological Risk Assessment:An International JournalPublication details, including instructions for authors andsubscription information:http://www.tandfonline.com/loi/bher20

What Biomonitoring Can and Cannot TellUs about Causality in Human Health andEcological Risk AssessmentsRalph G. Stahl Jr. a , Timothy S. Bingman b , Annette Guiseppi-Elie c

& Robert A. Hoke da DuPont Company, Corporate Remediation Group , Wilmington, DE,USAb DuPont Company, Corporate Remediation Group , Sewickley, PA,USAc DuPont Company, Corporate Remediation Group , Anderson, SC,USAd DuPont Company, Haskell Global Centers for Health andEnvironmental Sciences , Newark, DE, USAPublished online: 01 Feb 2010.

To cite this article: Ralph G. Stahl Jr. , Timothy S. Bingman , Annette Guiseppi-Elie & Robert A. Hoke(2010) What Biomonitoring Can and Cannot Tell Us about Causality in Human Health and EcologicalRisk Assessments, Human and Ecological Risk Assessment: An International Journal, 16:1, 74-86

To link to this article: http://dx.doi.org/10.1080/10807030903459353

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the“Content”) contained in the publications on our platform. However, Taylor & Francis,our agents, and our licensors make no representations or warranties whatsoever as tothe accuracy, completeness, or suitability for any purpose of the Content. Any opinionsand views expressed in this publication are the opinions and views of the authors,and are not the views of or endorsed by Taylor & Francis. The accuracy of the Contentshould not be relied upon and should be independently verified with primary sourcesof information. Taylor and Francis shall not be liable for any losses, actions, claims,proceedings, demands, costs, expenses, damages, and other liabilities whatsoever orhowsoever caused arising directly or indirectly in connection with, in relation to or arisingout of the use of the Content.

This article may be used for research, teaching, and private study purposes. Anysubstantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,

http://www.tandfonline.com/loi/bher20

http://dx.doi.org/10.1080/10807030903459353

systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

Dow

nloa

ded

by [

Yor

k U

nive

rsity

Lib

rari

es]

at 1

1:42

12

Aug

ust 2

014

http://www.tandfonline.com/page/terms-and-conditions

http://www.tandfonline.com/page/terms-and-conditions

Human and Ecological Risk Assessment, 16: 74–86, 2010Copyright C© Taylor & Francis Group, LLCISSN: 1080-7039 print / 1549-7860 onlineDOI: 10.1080/10807030903459353

What Biomonitoring Can and Cannot Tell Us aboutCausality in Human Health and Ecological RiskAssessments

Ralph G. Stahl, Jr.,1 Timothy S. Bingman,2 Annette Guiseppi-Elie,3

and Robert A. Hoke4

1DuPont Company, Corporate Remediation Group, Wilmington, DE, USA;2DuPont Company, Corporate Remediation Group, Sewickley, PA, USA; 3DuPontCompany, Corporate Remediation Group, Anderson, SC, USA; 4DuPont Company,Haskell Global Centers for Health and Environmental Sciences, Newark, DE, USA

ABSTRACTBiomonitoring can provide exposure and effects information on various stressors

(chemical or biological) that can be useful for human health and ecological riskassessments. It has been applied over the years where harmful changes in humanhealth or the environment were observed and which may have warranted moredetailed investigation. Sometimes biomonitoring programs may have been usefulin determining the significance and/or cause of these harmful observations. Thesedata can help to infer, but not confirm, causality as exemplified in classical studiesconducted in humans and wildlife. However, in most cases we note that additionalwork was needed to provide the information necessary to support or refute causality.Today modern technology provides the ability to measure a wide variety of param-eters in environmental media, plants, animals, and humans. Finding a chemicalin an environmental medium or biological tissue may be helpful in understandingpotential exposure (and perhaps to begin estimating hazard) to humans and eco-logical receptors, but mere presence does not necessarily help to establish effectsor assign causality. In this article we evaluate the strengths and weaknesses, in a riskassessment context, of the use of biomonitoring data to support a determination ofcausality.

Key Words: biomonitoring, causality, risk assessment.

The opinions expressed in this document reflect those of the authors and are not necessarilythose of the DuPont Company.Address correspondence to Ralph G. Stahl, Jr., DuPont Company, Corporate RemediationGroup, Chestnut Run Plaza, Building 715/Rm. 232, Route 141 & Lancaster Pike, Wilmington,DE 19805, USA. E-mail: [email protected]

74

Dow

nloa

ded

by [

Yor

k U

nive

rsity

Lib

rari

es]

at 1

1:42

12

Aug

ust 2

014

Biomonitoring, Causality, Risk Assessment

INTRODUCTION

In a broad sense, biomonitoring can be defined as the process of collecting chem-ical or biological information from humans, plants, and animals (USEPA 2000). It isrecognized that this definition is broad, and may overlap and be confused with “mon-itoring.” Generally speaking, monitoring is the collection and analysis of chemicalinformation from samples of environmental media (air, water, soil), and not neces-sarily from sampling specific biological tissues (NAS 2000; Borja et al. 2008; Newmanet al. 2008). Rather than focus on the differences and similarities between these twodefinitions, we believe it more useful to focus on how the biomonitoring data areused and to explore which uses might be more appropriate than others. Questionsof interest in this regard include: What, if any, insights are gained on potential risksto humans or the environment?; can or should these data be used in the risk assess-ment process for humans or ecological receptors?; and can biomonitoring data helpestablish causality? In this article we take the view that advances in molecular andanalytical technology have provided the ability to measure a number of chemicalsand biomarkers in environmental media, plants, animals, and especially humans(Smolders et al. 2009), but the measurements often necessitate gathering supportivedata to help ascertain causality.

Biomonitoring is a valuable tool that has been used for a variety of purposesover many years. In a now classical study in the 1800s, Potts documented the linkbetween exposure to soot and the development of scrotal cancer in chimney sweeps(as described in Gallo 2008). Potts was able to infer a link between an occupation(and subsequent exposure to soot) and an increase in a particular disease. However,scientists at that time did not have many of the additional tools available todaythat would have helped in their evaluation of biomonitoring data. For one, the riskassessment process for human health, as we know it, did not exist at that time. Thus,Potts was not able to estimate risks to the chimney sweeps from soot exposure. Also,analytical instruments and methods were few or did not exist. Hence, Potts couldnot identify the components of soot or potential causative agents in this complexmixture. More importantly, toxicological methods were lacking (Gallo 2008), and itwas years later before appropriate methods and data were developed to the degreethat a causal link was demonstrated between dermal exposure to polycyclic aromatichydrocarbons in the soot and the development of skin (scrotal) cancer (Eaton andGilbert 2008).

From an ecological perspective, Rachel Carson’s Silent Spring (Carson 1962)alerted environmental scientists to the plight of raptors and other birds in the 1960s.Writing about the alarming reductions in bird populations, she set in motion thenational discussion about the possible links between the increased use of pesticidesand harm to birds. Carson inferred the link between the use of the organochlo-rine pesticides, including DDT, and bird population declines, but was not able tospecifically identify the underlying cause(s).

Indeed, risk assessment in the early 1960s was still not a well developed processeither for humans or ecological receptors. While analytical methods and instrumen-tation in the 1960s were improved over those in Potts’ time, the level of sophistica-tion in these areas did not approach that available today. Although methods werein place for limited scale toxicological investigations (see Gallo 2008; Newman et al.

Hum. Ecol. Risk Assess. Vol. 16, No. 1, 2010 75

Dow

nloa

ded

by [

Yor

k U

nive

rsity

Lib

rari

es]

at 1

1:42

12

Aug

ust 2

014

R. G. Stahl, Jr. et al.

2008) the ability to undertake mechanistic toxicological studies was limited, andhindered the understanding of exposure to DDT (and metabolites) and subsequenteffects (eggshell thinning) in birds. It was many years later that the mechanistictoxicological data were developed to support the effects of exposure to DDT and itsmetabolites, and to clarify causality (see Bowerman et al. 1995; Beyer et al. 1996).

Today, where well developed frameworks exist for evaluating risks to humansand ecological receptors (e.g., USEPA 1992), data from biomonitoring are oftenused as one of several underlying lines of evidence for exposure and/or effects.But, these data also offer unique challenges to risk assessors who are tasked withevaluating information from diverse investigations and using it appropriately in therisk assessment process (see also Langer and Sang 1997). While such data can proveuseful in their own right, the question remains as to the ability of the data to informquestions of causality with respect to adverse environmental and human healthoutcomes.

The utility of biomonitoring data in causal analysis depends on the nature of theeffect to which one is trying to ascribe causality. In a complex disease such as cancer,biomonitoring data from an individual may provide insufficient information fromwhich to determine causality. Little is likely to be known, without direct interviewsand home surveys, about occupational and lifestyle influences for example, that maybe important to understanding causality, but that would not necessarily be knownsolely through biomonitoring.

In this article we hope to illustrate that biomonitoring data, whether from humansor ecological receptors, can provide useful information for risk assessment purposesbut is limited, in most cases, for use in determining causality. We suggest that there isadditional investigation and research that is typically needed before biomonitoringdata can be useful in assigning causality.

BIOMONITORING IN HUMAN HEALTH RISK ASSESSMENTS

History of Human Biomonitoring

As we noted in the introduction, biomonitoring has been practiced in variousforms since the late 1800s. At that time, physicians began the practice of monitoringfor metabolites of aspirin in the urine of persons being treated for rheumatic fever.By the 1890s lead was being measured in the blood and urine of lead refineryworkers as a way to identify workers whose lead burdens were approaching toxiclevels (Sexton et al. 2004). A similar practice was undertaken in the felt hat industry,where workers were monitored for occupational mercury exposures. By the 1960s,limited, but useful, analytical methods had been developed for about 20 commonindustrial chemicals. Currently, there are methods available for measuring severalhundred chemicals and their metabolites in diverse human samples (e.g., blood,urine, hair).

While many of the early advances in our ability to identify and quantify chemicalsin human tissues began in industrial settings, the application of human biomoni-toring in a broader public health setting is a fairly recent phenomenon. One of thefirst major organized efforts to collect human biomonitoring data was the U.S. Envi-ronmental Protection Agency’s (USEPA’s) National Human Adipose Tissue Survey,

76 Hum. Ecol. Risk Assess. Vol. 16, No. 1, 2010

Dow

nloa

ded

by [

Yor

k U

nive

rsity

Lib

rari

es]

at 1

1:42

12

Aug

ust 2

014


Figure 1. Growth in CDC’s human biomonitoring program.

which began circa 1970 (USEPA 2008). This survey focused primarily on the iden-tification of persistent chlorinated materials in human fat tissue, particularly DDT,DDD, and PCBs. The largest current effort to use biomonitoring data to understandthe presence of chemicals in the bodies of the general populace is the Centers forDisease Control and Prevention’s survey of human exposures, which began in 1999,and continues to the present. Currently this effort involves the testing of approxi-mately 2,000 persons for the presence of a variety of environmental contaminants(CDC 2005). The number of chemicals that have been included in the variousiterations of this survey is depicted in Figure 1.

Types of Biomarkers

This discussion of the history of biomonitoring, while useful for understand-ing the early days of the discipline, is only a small part of the story. The type ofbiomonitoring that has existed historically has been limited to what is known asthe measurement of “biomarkers of exposure.” Biological markers, also known asbiomarkers, include a wide variety of chemical and biochemical measurements thatcan reflect the presence of chemicals or their metabolites in living systems, alter-ations in those systems that can be precursors to disease states, or measurements ofthe susceptibility of the organism to disease (Smolders et al. 2009). Understandingthe various types of biomarkers and their applications and limitations is importantto an understanding of what biomonitoring can and cannot tell us about causationin human risk assessment.

The development of a particular disease can be viewed as the culmination ofa series of biological and biochemical changes in the organism, which beginswith exposure to the toxic agent and ends with the disease state. This cascadeof events is depicted graphically in Figure 2. In this cascade, various biomarkerscan be used to characterize the different elements that lead from initial exposureto manifestation of disease. Each of these three types of biomarkers—biomarkers


Dow

nloa

ded

by [

Yor

k U

nive

rsity

Lib

rari

es]

at 1

1:42

12

Aug

ust 2

014


Figure 2. Variety of biomarkers. Extracted from NRC (2006).

of exposure, biomarkers of effect, and biomarkers of susceptibility—are discussednext.

Biomarkers of Exposure

In its report Human Biomonitoring for Environmental Chemicals, the National Re-search Council defines a biomarker of exposure as “the chemical or its metaboliteor the product of an interaction between a chemical and some target molecule orcell that is measured in a compartment in an organism” (NRC 2006).

Biomarkers of exposure have a variety of applications. In their simplest applica-tion, biomarkers of exposure can provide an indication of the specific chemicalsto which an organism or population has been exposed (Sexton et al. 2004). Anexample would be the presence of carboxyhemoglobin in a person’s blood as anindication of carbon monoxide exposure. Exposure biomarkers are also useful inestablishing the extent to which exposure has occurred, in that they can provideuseful measurements of the actual absorbed dose of a toxicant, as well as providingan indication of the concentration present in the target tissue (CIIT 2005).

There are, however, limitations to the use of exposure biomarkers. First, biomark-ers of exposure do not, of themselves, provide any insight into the susceptibility ofhumans, plants, or animals to a specific disease state (Forbes et al. 2006; Smolderset al. 2009). In isolation, biomarkers of exposure do not necessarily provide anyindication of the increased risk that the organism may experience as a result of theexposure. In humans such information can provide an indication of risk in situationswhere it is coupled with a physiologically based pharmacokinetic model that relatesthe observed biomarker to some health risk endpoint (NRC 2006), or in cases wherethe biomarker itself has been previously linked to the likelihood of developing adisease state (e.g ., blood lead levels related to neurotoxicity in children).

On the matter of causation, biomarkers of exposure can directly provide infor-mation only on the specific chemical to which an individual or population has beenexposed and, to a lesser extent, the degree to which that exposure has resulted inan internal dose of the specific chemical. Moreover, given the current state of thescience, biomarkers of exposure cannot, in isolation, be related causally to suscep-tibility, or to risk of developing disease. This forms the basis of one of the greatestmisuses of biomonitoring data that use biomarkers of exposure—namely, the inac-curate presumption by some that the detection of a chemical in the body is evidence


Dow

nloa

ded

by [

Yor

k U

nive

rsity

Lib

rari

es]

at 1

1:42

12

Aug

ust 2

014


of risk or of a precursor to disease. To establish this link would require more detailedtoxicological study, as well as an understanding of the biomechanistic basis for thepotential effect. Unfortunately, the ability to measure changes at the molecular level,as well as low levels of substances in human tissues, appears to continually outpaceour ability to understand their significance to human health.

Biomarkers of Effect

Biomarkers of effect are defined as “any change that is qualitatively or quan-titatively predictive of health impairment or potential impairment resulting fromexposure” (Decaprio 1997). As noted in Figure 2, these changes occur at multiplepoints along the exposure–disease continuum and can, in some instances, reflectmodifications in the organism that precede actual structural or functional impair-ment.

Effect biomarkers can include a wide array of biological and biochemical end-points, such as DNA or hemoglobin adducts, mRNA expressions, or the presenceof a specific transcribed protein (Paustenbach and Galbraith 2006). These markerscan be useful as predictors of ultimate toxicity and can be, in some instances, atleast qualitatively associated with increased health risk for certain conditions. How-ever, many biomarkers of effect tend to be less associated with exposures to specificchemical agents (Decaprio 1997). For example, a given biomarker of effect can bepredictive of a specific type of disease, but the underlying cause could have resultedfrom multiple chemical agents. Thus, from the standpoint of causation, the abilityto ascribe the observation of specific biomarkers of effect to a specific chemical incases of multiple chemical exposures that impact similar biomarkers is limited.

Biomarkers of Susceptibility

A biomarker of susceptibility is an indicator of the ability of an organism torespond to the challenge of exposure to a specific chemical substance. Suscepti-bility biomarkers range from direct genetic analysis that can link specific geneticpolymorphisms to increased incidence of disease in an exposed population, to mea-surement of specific enzyme levels responsible for potentiating or detoxifying aspecific chemical exposure (Paustenbach and Galbraith 2006). An example wouldbe those enzymes involved in the excision and repair of DNA adducts where somesegments of the population will have a highly competent form of the excision/repairenzyme, while others will not. Those without the highly competent form are morelikely to suffer increased persistence of DNA adducts. These biomarkers allow anassessment of an individual’s susceptibility to disease by modifying the progressionalong the continuum shown in Figure 2.

As the name implies, this type of biomarker is useful in helping determine thesusceptibility of an individual to manifest a toxic response or to develop a specificdisease state. However, it is of limited value in determining the specific chemical towhich an individual has been exposed, or the extent of that exposure.

Use of Biomonitoring to Inform Causation in Human Risk Assessment

As noted previously, the applicability of biomonitoring data in determining causa-tion in human risk assessment varies with the nature of the question one is attempting


Dow

nloa

ded

by [

Yor

k U

nive

rsity

Lib

rari

es]

at 1

1:42

12

Aug

ust 2

014


Figure 3. Biomarkers in human health assessments associated with causality.

to answer. We depict the relationship between the various types of biomarkers andthe aspects of causation that they are attempting to inform (a so-called causationmetric) in Figure 3.

In Figure 3 is illustrated the importance of understanding the question that one isattempting to answer before deciding on whether to employ biomonitoring data toinform it. Ultimately, from a public health standpoint, the most important questionthat any investigator is hoping to answer is the question of whether exposure toa particular agent at a particular level will result in an increased risk of toxicity ordisease. As noted in Figure 3, all three types of biomarkers can, depending on the cir-cumstance, help to inform this question. Ultimately, a weight of evidence approachthat draws not only on data from biomonitoring, but also from the fields of pharma-cokinetic modeling, epidemiology and classical toxicology, and risk assessment willneed to be employed if this question is to be answered.

ECOLOGICAL RISK ASSESSMENT

A great deal of the previous discussion on the potential use of biomonitoringdata in human health assessment and determination of causality is directly applica-ble to ecological risk assessment. In most cases there is a strong parallel betweenthe value (or lack thereof) of biomonitoring data for risk assessment for humansand ecological receptors. Biomarkers of exposure, for example, provide a usefulillustration. Findings of elevated P450 levels in fish today is thought to be relativelyclear evidence of exposure to PAHs; however, other substances in the environmentcan also cause this response (McCoy et al. 2002) and elevated levels of these enzymesdo not necessarily suggest these fish are at risk of increased neoplasia. As a result,


Dow

nloa

ded

by [

Yor

k U

nive

rsity

Lib

rari

es]

at 1

1:42

12

Aug

ust 2

014


for ecological risk assessment we only discuss those points that reinforce the applica-bility of, or that illustrate potential distinctions between, human health assessmentand ecological risk assessment.

Biomonitoring has been in use for many years in the field of environmental andecological assessment (Rosenberg and Resh 1993; Muller and Lenz 2006; Newmanet al. 2008). Not surprisingly, many of the same issues and challenges noted forhuman health assessment also exist in ecological assessments and in establishingcausality for observed effects. Similar to human health assessment, tissue burdensof chemicals in fish and wildlife observed through biomonitoring do not necessarilyresult in observed effects or unacceptable levels of risk (Forbes et al. 2006). Clearly,while there has been exposure to these chemicals, without additional work verylittle can be concluded regarding potential hazard and resultant risk (see Beyeret al. 1996). Forbes et al. (2006) made a similar argument that biomonitoring andbiomarkers should not be used to draw conclusions prematurely about possibleeffects on populations and communities, or to individual organisms. Prior to drawingany conclusions, there should be careful evaluations of the exposure pathways aswell as assessing the potential for toxicity to occur at the observed tissue levels of thechemical. These are important steps in interpreting the utility and applicability ofbiomonitoring data. We illustrate these issues by reviewing USEPA (2000) guidance,which sets forth a useful framework for establishing particular stressors and potentialeffects to ecological receptors, as well as a suggested approach for assessing causality.The important steps in this approach provide a road map for highlighting thestrengths and weaknesses of using biomonitoring data to establish causality.

In Figure 4 is illustrated the approach detailed by the USEPA (2000) that canbe applied to the determination of stressors, their potential effects, where biomon-itoring data exist, and how those data might be useful in assessing causality. First,biomonitoring and other data are reviewed to help formulate a list of potentialcausative agents. Developing a list of plausible causative agents can be done on thebasis of professional judgment and experience, or, for example, through expertopinion surveys. Once the potential causative agents are listed, the approach guidesone through an analysis of whether or not each particular stressor or agent can orcannot be linked to the observed effects. The linking of an agent to the receptoris a fundamental aspect of exposure assessment (USEPA 1997, 1998). Evaluatingareas of uncertainty in the biomonitoring data is also a key step. For example, insome studies there may be little or no data available on where and when organismswere collected, little information on their “health” status when collected, or per-haps other confounding variables, such as whether appropriate QA/QC protocolswere followed in the analytical methods. These concerns need to be known andunderstood before the biomonitoring data can be evaluated properly.

At this early stage of investigation the list of possible causative agents might bequite large and difficult to manage. If so, in some cases additional field or laboratorystudies might be needed to reduce this uncertainty and to streamline the numberof potential causative agents. While determining the mechanism of toxicity for asuspected agent or stressor is not discussed per se in the USEPA (2000) guidance, itmight in fact be necessary, and is one of the issues we have discussed earlier in thisarticle. Additional field or laboratory work is particularly relevant in the situationwhere an immediate risk management action may be under consideration, but the


Dow

nloa

ded

by [

Yor

k U

nive

rsity

Lib

rari

es]

at 1

1:42

12

Aug

ust 2

014


Figure 4. USEPA stressor identification framework. Extracted from USEPA (2000).

potential cause (stressor or agent) of the unacceptable risk has not been clearlyidentified. When faced with this type of situation, there may be compelling reasonsto more carefully determine the cause(s) of an observed effect so that limitedresources for risk management are applied most effectively to addressing the sourceof the problem. On the other hand, we recognize that there may be situations whereimmediate action is necessary, even in the absence of a known causative agent(see USEPA 1997). Such situations, where the potential for significant impacts toorganisms, populations, or communities arise, may require actions that otherwisecould be undertaken later when more facts have been collected.

Generally speaking, the fundamental steps in both the human and ecological riskassessment process are to assess information, estimate risk, assess the uncertainties,and focus management actions on those risks found to be unacceptable (USEPA1998). In such a situation where one is faced with a suite of possible causative agents,risk assessment principles suggest it is necessary to determine which ones warrantimmediate attention and which do not. This is not likely to be a wholly scientificprocess, but one that may also invoke policy mandates, or management actions whenregulatory issues or ethical considerations are involved (Stahl et al. 2001).


Dow

nloa

ded

by [

Yor

k U

nive

rsity

Lib

rari

es]

at 1

1:42

12

Aug

ust 2

014


Table 1. Hill’s criteria for determining causality.∗

1. Strength of association—relationship between independent and dependent variables.2. Consistency of findings—replication of results by different studies.3. Biological gradient—strength of the dose–response relationship.4. Temporal sequence—“cause” before effect.5. Biologic or theoretical plausibility—mechanism of action.6. Coherence with established knowledge—no competing hypotheses.7. Specificity of association—cause is tightly linked to an outcome.

∗Extracted from Eaton and Gilbert (2008).

To help in determining causality, the USEPA Stressor Identification approach(USEPA 2000) builds on Hill’s criteria (Table 1), which have been used by scientistsand physicians over a number of years (see Fox 1991; Gallo 2008). The USEPAguidance, and Hill’s criteria, provide an appropriate, reasoned approach to usingbiomonitoring data, which is concordant with approaches for using the same in-formation in an ecological risk assessment (USEPA 1992, 1997, 1998). The StressorIdentification Approach culminates with an assessment of the strength of the asso-ciation between observations, stressors, and possible agents. The guidance providesan important recommendation to those involved with the assessment—focus oneliminating possible causative agents with existing information so as to narrow thefield of choices sooner rather than later. Even so, additional field or laboratorydata might be needed to narrow the field of choices or these might be evaluatedthrough careful review of existing literature. In either case, it is important to takea careful, systematic approach to the evaluation so that time and resources are notwasted.

As noted earlier, the issue of establishing causality in ecological assessment hasalso been discussed in the context of “ecoepidemiology” (Fox 1991). The conceptsprovided in Fox’s paper, while considered dated by some, also show that the deter-mination of causality is approached stepwise, where each step is directed at usingexisting information or collecting new information that would allow one to judgethe strength of the association between a particular stressor and the observed ef-fect. Similar to the approach illustrated in Figure 4, the progression from one stepto the next and the data evaluation that occurs are key elements of attempting todetermine causality in ecological assessments. The approaches detailed in USEPA(2000), and Fox (1991), when applied, provide ample opportunity for ecologicalrisk assessors to evaluate the data obtained from biomonitoring programs. The im-portance of this systematic, reasoned approach is also noted in various guidelinesfor ecological risk assessment (USEPA 1992, 1997) and highlights the desire touse risk assessment as a fundamental tool for evaluating data and recommendingwhere risks may or may not be acceptable. In this regard, there are clearly substan-tial parallels between human health risk assessment and ecological risk assessmentwith respect to the use of biomonitoring data and in causality determinations. And,just as important, the risk assessment process provides a clear and reasoned ap-proach for evaluating a variety of data including those stemming from biomonitoringstudies.


Dow

nloa

ded

by [

Yor

k U

nive

rsity

Lib

rari

es]

at 1

1:42

12

Aug

ust 2

014


IMPLICATIONS FOR PUBLIC POLICY

While biomonitoring data are often collected for human health and environ-mental evaluations, their use in determining causality is less well evolved. This is thecrux of the issue we have endeavored to explore in this article. The resolution tothis issue, however, will not result from the discussion herein, but rather will dependon further dialog among scientists and decision-makers in the public health andenvironmental communities.

There has been an increased trend for scientists to generate biomonitoring data,making them more readily available for consideration in public policy decisions (seeSmolders et al. 2009). Technological advances in analytical methods will only increaseour ability to detect and measure substances at ever smaller levels in the environmentand in biological tissues. The increase in these data and their availability to decision-makers will require that we address the broader implications for the use of thesedata in policy development and decision-making. At the present, while the useof biomonitoring data for public policy decisions (where the basis for the policydecision is as a result ascribing causality) may be reasonable in some instances, itsuse should be tempered by the current limited state of the science.

As an example, in the United States, the Centers for Disease Control and Pre-vention (CDC) often emphasizes the premise that the presence of a substance in abiomonitored medium (e.g ., human tissue) is not necessarily indicative of “harm.”Yet, the alternative (i .e., presence does equal harm) has become pervasive as amechanism for using the precautionary principle, including in a regulatory context.We caution that this (the inaccurate presumption by some that the detection of achemical in the human body is evidence of risk or of a precursor to disease) formsthe basis of one of the greatest misuses of biomonitoring data (see Smolders et al.2009). As we have attempted to illustrate herein, biomonitoring data can be usefulin a risk assessment context and to a limited degree in evaluating causality, but onlywhen appropriate supportive testing and research are undertaken. Supportive workincludes careful evaluation of the toxicological data, if they exist, that would supportor refute causality whether in humans or ecological receptors.

On the matter of ascribing causality, we stress that it is critical to understand thespecific question(s) to be addressed when using biomonitoring data. From a publichealth standpoint, one of the most important questions is whether exposure to aparticular agent at a particular level will result in an increased risk of toxicity ordisease (see also Eaton and Gilbert 2008). Even though this question arises moreoften in the public health arena than in an ecological context, it is nonetheless anissue in the environmental (ecological) arena. We know that stressors other thanchemicals can lead to reductions in populations of plants (Armenteras et al. 2006)and animals (Muller and Lenz 2006). We suggest a weight of evidence approachbe employed for decision-making that draws not only on data from biomonitoringbut other relevant fields (e.g ., pharmacokinetic modeling, epidemiology, ecology,classical risk assessment). Further, in order to make the best use of biomonitor-ing data for decisions that will become the basis for public policies, it is impor-tant to understand the strengths and weaknesses of using such data. We have at-tempted to highlight some of these areas of strength and weakness earlier in thisarticle.


Dow

nloa

ded

by [

Yor

k U

nive

rsity

Lib

rari

es]

at 1

1:42

12

Aug

ust 2

014


In terms of informing public policy decisions, we suggest that these decisionsfirst be guided by a robust evaluation of all the available scientific data that includesappropriate biomonitoring data when available. Further, we recommend that the useof biomonitoring information for determining causality be guided by the following:

� Scientists and policy-makers need to be engaged as dialogue proceeds, andneed to be clear on what questions they are attempting to answer, and whatis meant by “biomonitoring” (i.e., be clear whether the available data relate toexposure, effects, or susceptibility).

� A weight of evidence approach needs to be applied to disparate informationfrom biomarkers and other types of human or environmental data.

When the scientific results are presented and subjected to open, frank discussionsregarding their strengths and limitations, more informed public policy decisions canresult. This is true whether or not the results come from biomonitoring studies orother sources of information.

ACKNOWLEDGMENT

The authors appreciate the recommendations from three peer reviewers, whosecomments improved this article.

REFERENCES

Armenteras D, Rudas G, Rodriguez N, et al. 2006. Patterns and causes of deforestation in theColombian Amazon. Ecol Indicators 6:353–68

Beyer WN, Heinz GH, and Redmon-Norwood AW (eds). 1996. Environmental Contaminantsin Wildlife. Lewis Publishers, Boca Raton, FL, USA

Borja A, Bricker SB, Dauer DM, et al. 2008. Overview of integrative tools and methods inassessing ecological integrity in estuarine and coastal systems worldwide. Marine Poll Bull56:1519–37

Bowerman WW, Giesy JP, Best DA, et al. 1995. A review of factors affecting productivity ofbald eagles in the Great Lakes Region: Implications for recovery. Environ Health Perspect103(suppl 4):51–9

Carson R. 1962. Silent Spring. Fawcett Publications, Greenwich, CT, USACDC (Centers for Disease Control and Prevention). 2005. Telebriefing transcript of Dr.

Julie Gerberding—Third National Report on Human Exposures to Environmental Chem-icals. Department of Health and Human Services. Atlanta, GA, USA. Available athttp://www.cdc.gov/od/oc/ media/transcripts/t050721.htm

CIIT (Chemical Industry Institute of Toxicology). 2005. Physiologically-Based Pharmacoki-netic Modeling as a Tool to Interpret Human Biomonitoring Data. Centers for HealthResearch—CIIT Activities 25(4) October–December. Raleigh-Durham, NC, USA

Decaprio A. 1997. “Biomarkers” coming of age for environmental health and risk assessment.Environ Sci Technol 31:1837–48

Eaton DL and Gilbert SG. 2008. Chapter 2. Principles of toxicology. In: Klaassen CD (edit.),Casarett and Doull’s Toxicology: The Basic Science of Poisons, 7th ed, pp 11–43. McGrawHill, New York, NY, USA


Dow

nloa

ded

by [

Yor

k U

nive

rsity

Lib

rari

es]

at 1

1:42

12

Aug

ust 2

014


Forbes VE, Palmqvist A, and Bach L 2006. The use and misuse of biomarkers in ecotoxicology.Environ Toxicol Chem 25:272–80

Fox GA. 1991. Practical causal inference for ecoepidemiologists. J Toxicol Environ Health33:359–73

Gallo MA. 2008. Chapter 1. History and scope of toxicology. In: Klaassen CD (edit.), Casarettand Doull’s Toxicology: The Basic Science of Poisons, 7th ed, pp 3–10. McGraw Hill, NewYork, NY, USA

Langer J and Sang S. 1997. Endocrine disruption: Cause for concern, cause for action. CanChem News 49:16–8

McCoy DL, Jones JM, Anderson JW, et al. 2002. Distribution of cytochrome P4501A1-inducingchemicals in sediments of the Delaware River-Bay system, USA. Environ. Toxicol Chem21:1618–27

Muller F and Lenz R. 2006. Ecological indicators: Theoretical fundamentals of consistentapplications in environmental management. Ecol Indicators 6:1–5

NAS (National Academy of Sciences). 2000. Ecological Indicators for the Nation. NationalAcademy Press, Washington, DC, USA

Newman MC, Jorgensen SE, and Fath BD (eds). 2008. Ecotoxicology: The history and presentdirections. In: Jorgensen SE and Fath B (eds), vol 2, Encyclopedia of Ecology, pp 1195–201.Elsevier, London, UK

NRC (National Research Council). 2006. Human Biomonitoring for Environmental Chemi-cals. National Academy Press, Washington, DC, USA

Paustenbach D and Galbraith D. 2006. “Biomonitoring” is body burden relevant to publichealth? Regul Toxicol Pharmacol 44:249–61

Rosenberg DM and Resh VH (eds). 1993. Freshwater Biomonitoring and Benthic Macroin-vertebrates. Chapman and Hall, New York, NY, USA

Sexton K, Needham L, and Prikle J. 2004. Human biomonitoring of environmental chemicals:Measuring chemicals in human tissues is the “gold standard” for assessing exposure topollution. Am Sci 92:38–42

Smolders R., Schramm K-R, Stenium U, et al. 2009. A review on the practical application ofhuman biomonitoring in integrated environmental health impact assessment. J ToxicolEnviron Health, Part B 12:107–23

Stahl RG, Bachman RA, Barton AL, et al. (eds). 2001. Risk Management: Ecological Risk-BasedDecision Making. SETAC Press, Pensacola, FL, USA

USEPA (US Environmental Protection Agency). 1992. Framework for Ecological Risk Assess-ment. EPA/630/R-92/001. Risk Assessment Forum, Washington, DC, USA

USEPA. 1997. Ecological Risk Assessment Guidance for Superfund: Process for Designingand Conducting Ecological Risk Assessments. EPA 540-R-97-0C5. Office of Solid Wasteand Emergency Response, Washington, DC, USA

USEPA. 1998. Guidelines for Ecological Risk Assessment. EPA/630/R-95/002F. Office ofResearch and Development, Washington, DC, USA

USEPA. 2000. Stressor Identification Guidance Document. EPA-822-B-00-025. Office of Waterand Office of Research and Development, Washington, DC, USA

USEPA. 2008. National Human Adipose Tissue Survey (NHATS). National Center for Envi-ronmental Assessment. Cincinnati, OH, USA. Available at http://cfpub.epa.gov/ncea


Dow

nloa

ded

by [

Yor

k U

nive

rsity

Lib

rari

es]

at 1

1:42

12

Aug

ust 2

014

what biomonitoring can and cannot tell us about causality in human health and ecological risk...

Documents