Mey Akashah, "Risk Assessment and Improved Decision-Making," Harvard School of Public Health, Harvard Medical School, and Harvard Extension School, April 5 2012.
Course: Human Health and Global Environmental Change
1. Risk Assessment and
Improved Decision-Making
Dr. Mey Akashah
Instructor, Harvard School of Public Health
Human Health and Global Environmental Change
April 5, 2012
Harvard School of Public Health (EH 278)
Harvard Medical School (HO703.0)
Harvard Extension School (ENVR E-165)
2. Regulations Based on Risk
Assessments
• Water discharge limits (EPA)
• Air emissions (EPA)
• Cleanup limits (EPA)
• Fish advisories or bans (EPA, DI)
• Product recalls
– Food/drugs/medical devices (FDA)
– Consumer products (CPSC)
• Workplace standards (OHSA)
3. Risk Assessment: A definition
―The characterization of the
potential adverse health
effects of human exposures
to environmental hazards.‖
―The Red Book‖
National Research Council. Risk Assessment in
the Federal Government: Managing the
Process, 1983
6. What is a hazard?
An agent capable of causing an adverse effect
• Combustible • Chemical
• Explosive • Non-ionizing radiation
• Toxic • Ionizing radiation
• Corrosive • Physical
• Biological • Noise
• Mechanical • Psychosocial
7. Hazard Identification:
The Process
Gather information on:
– Toxicity
– Concentration
– Potential routes of exposure
– Receptor organs and tissues
– Biochemical responses/Biotransformation
8. Hazard Identification:
The Data
• Historical information
• Chemical & related structures
• In vitro data
• In vivo data
– Animal studies
– Human studies
• Clinical trails (drugs)
• Occupational exposure
• Epidemiology
– Post-marketing surveillance
9. Hazard Identification
• Is the substance dangerous?
• What adverse effects are associated with the
toxicant? (e.g., neurological, immunological)
• Must evaluate quality of studies
– Choice of control groups, number of animals
– Severity of effect described
– Relevance of toxic mechanisms in animals to those
in humans
• Result: a scientific determination that a toxicant
can, at some exposure concentration, cause a
particular adverse health effect in humans
12. Concentration, Exposure,
& Dose
• Concentration:
– How much of a substance is present in
the medium
• Exposure:
– What we actually come into contact with
• Dose:
– The amount of the substance that enters
the body’s ―envelope‖
13. Exposure Versus Dose
Dose
• ―All things are poison and not
without poison; only the dose
makes a thing not a poison‖
OR
• ―The dose makes the poison‖
— Paracelsus
(1493-1541)
14. What do we mean by dose in
RAs?
• ―…the amount of a substance
administered at one time…‖
• Sometimes referred to as the
concentration of a substance in the
body ~ ―per mL blood‖ or ―amount per
kg body weight‖
• In environmental health, we often look
to ―cumulative exposure‖ as an
estimate of dose
16. Quantifying Exposures
• Indirect
– Involves collecting information about where, when, and
how people spend their time, and about the concentrations
of a contaminant in surrounding media.
– Ex. Ambient concentration
• Direct
– Assesses a person's exposure, using monitors attached to
the individual, or through the sampling of biological media
such as blood or urine.
– Ex. Direct personal exposure measurements (PEMs)
– Ex. Biological monitoring to capture dose
Note: In many, many cases, you need to do a combination
of these!
19. Imperfect…
• Exposureby an individual. provide a representation of the dose
received
assessments can
• However, whilean agent, many factors will determine the
concentration of
an individual may be exposed to a certain
biologically effective concentration for organs, tissues, and cells
• Individual and environmental factors effect dose!
22. Dose
• ―All things are poison and not
without poison; only the dose
makes a thing not a poison‖
OR
• ―The dose makes the poison‖
— Paracelsus
(1493-1541)
23. Individual Dose-Response Function
(Dose-Effect)
Example - Aspirin in humans
Severity
death
hemorrhage
encephalophathy
acidosis
hyperventilation
nausea
0 100 200 300 400 500 600
therapeutic
Dose (mg/kg)
Harvard Center for Risk Analysis
24. Dose-Response Assessment
• Involves quantitative characterization
of chemical potency
• Must understand importance of:
– intensity of exposure
– concentration over time
– whether a chemical has a threshold
– shape of the dose response curve
– metabolism at different doses
– persistence over time
– Similarities/differences in behavior of
chemical in humans and animals
28. Acute V. Chronic Toxicity
• Acute toxicity
– Describes the adverse effects of a substance that
can result exposures in a short space of time
(days to weeks)
– Animal Testing – single, high dose
• Chronic Toxicity
– Describes the adverse health effects from
repeated exposures, often at lower levels, to a
substance over a longer time period (months or
years)
– Animal Testing – multiple, low doses
30. Dose-Response and Regulation
Noncarcinogens: Threshold
Exposure level that is likely to
be without an appreciable risk of
deleterious effects during a lifetime
Carcinogens: Nonthreshold
There is no exposure level that
does not pose a finite probability of
generating a carcinogenic response,
i.e., there is a risk associated with
any exposure level
33. Cancer Risk (CR)
• Cancer slope factor (CSF)
• CR = intake estimate* X CSF*
*mg/kg/day
34. Cancer Potency Estimates (CPE)
• Estimated upper bound on lifetime cancer
risk associated with the lifetime daily dose
of a chemical
• CPE developed by different regulatory
agencies for the same chemical can vary
substantially
– Level of risk sufficient to trigger
regulatory can vary considerably among
agencies and within an agency
40. THRESHOLD: level at which biological responses begin
(may not be observable)
NOAEL: no observed adverse effects level
LOAEL: lowest observed adverse effects level
? ?
45. Risk Characterization
• A synthesis and summary of information
about a hazard that addresses the needs and
interests of decision makers and of interested
and affected parties.
• Risk characterization is a prelude to decision
making and depends on an iterative, analytic-
deliberative process.
Report by the National Academy of Sciences' National Research
Council, 1996.
46. hazard
identification
exposure
assessment
dose-response
assessment
risk
management
and
communication
risk
characterization
risk assessment
47. Goal is rarely ―no risk‖
• Environmental risk assessments deal with
the incremental probability of some harm
occurring.
• For example, EPA attempts to control
exposure to toxicants to levels that will
pose incremental lifetime cancer risk
among the most exposed to roughly 1 to
100 additional cancers per 1 million
people
49. Cost-Benefit Analysis
• Systematic process for calculating and
comparing benefits and costs of an
intervention, project, change in policy
• Benefits and costs expressed in terms of
monetary equivalents
– These are adjusted for the ―time value of
money‖
51. Public Perceptions about Risk
• Preceding tables often used to put risk into
perspective for public
• Often, these numbers seem reasonable to those
familiar with the RA process, but confusing and
sometimes frightening for the public
• Factors affecting fear around risks:
– Lack of control over risk
– Lack of familiarity with risk (unknown risks scarier
than known risks
– Natural risks more acceptable (earthquakes Vs. plant
explosions, cars Vs. sharks)
52. Risk Characterization
• Where all the information comes together!
• Integrates hazard identification, dose-response
assessment, & exposure assessment
• Requires making scientific judgments:
– What do exposure data mean?
– Is dosage information available?
– How do exposures compare with doses used in
animal studies? Human data?
– Are doses higher than allowed?
– Should report all uncertainties & assumptions!!!
53. Important Types of Uncertainty
• Parameter uncertainty
– we know that we don’t know it.
• Measurement errors
• Sampling errors
• Use of ―generic‖ data when actual data exists
• Scenario uncertainty
– Uncertainty regarding missing or incomplete information
needed to fully define the exposure or dose (i.e., errors in
information)
• Model uncertainty
– we may not know that we don’t know it (i.e., we chose the
incorrect model)
• Uncertainty regarding gaps in scientific theory required to make
predictions on the basis of causal inferences (model structure –
linear vs. nonlinear)
54. Why is it important to characterize
uncertainty and variability?
• We are not concerned about uncertainty for
its own sake, but rather how it influences
our decisions.
• We need to characterize uncertainty for risk
management decisions.
• It is generally not received well by the
public but it needs to be done.
55.
56. Fish Consumption and Mercury
Hg exposure in pregnant women may
increase the risk of subtle
neurotoxic effects in the fetus
57. Effect of Small Shifts in Population
Mean Intelligence Quotient (IQ)
58. Fish Advisories:
The Good and the Bad
• Good: Polyunsaturated fatty acids
• Bad: Hg (among other things)
• Science Policy Interpretation
– Science: complex
– Policy simplified for general consumption
• Interpretation of Public: Avoid fish
altogether
59. Risk Assessment Vs.
Risk Management
• Risk Assessment
– The characterization of the potential adverse health effects of human
exposures to environmental hazards
• Risk Management
– Agency-based decision-making process
– Uses the knowledge gained from risk assessment to evaluate
alternative regulatory actions and select among them
– Entails consideration of political, social, economic, and engineering
information with risk assessment information to develop, analyze, and
select regulatory guidelines
– Requires the use of value judgments on acceptability of risk and the
reasonableness of the costs of control
61. Classic Risk Assessment and
Environmental Change
Is the classic four-step risk assessment
framework sufficient to examine
environmental change, particularly climate
change?
62. Example: Climate Change
Three key considerations:
• System complexity
– Including cyclical risks
• Multiplicative uncertainty
– Each uncertain variable propagates uncertainty
• Sample size
– Limited data on climactic response to current
levels of atmospheric particulates
63. ―Non-Classic‖ approaches to
climate change using RA
• IFRC Vulnerability and Capacity Assessment tool
• UNDP Global Risk Assessment Programme
• UK Climate Change Risk Assessment methodology
• Australia-New Zealand Standard (assessments of
climate change risks)
• …
• Most include both risk assessment and risk
management in one framework
Some slides care of Dr. Susan Wason and the faculty of the Harvard School of Public Health Department of Environmental Health
Problems faced by reg. agencies:Prioritizing problems (may be very different) Fulfilling legislative responsibilitiesBalancing costs and benefits of beneficial productsJustifying regulatory decisionsdemonstration of significant risk clear, transparent, and logical method for setting standards“sound science” instead of professional judgment
Includes:Description of the potential adverse health effects based on an evaluation of results of epidemiologic, clinical,toxicological, and environmental research; Extrapolation from those results to predict the type and estimate the extent of health effects in humans under givenconditions of exposure; Summary judgments on the existence and overall magnitude of the public-health problemAlso includes characterization of the uncertainties inherent in the process of inferring risk.
Until, ultimately, you find out…
Xenobiotic – foreign chemicalExposureExposure pathwaysDose – passes “envelope” - Goes on to a number of categories – bio-effective dose, etc.
By 1:55 PM or 25 minWhat I mean by that is…We are often stuck out here (outside of the body with an Environmental Factor)For ex., antidepressants urinated into the water in Mississippi and drunk up in greater concentrations by Creole fisherman in Louisiana Often EFs new or w/o reliable epi studies to give info on human riskEpi studies are expensive and few and far between for EFsWe don’t have blood from folksThus must combine our limited toxicological information with info on the intensity, frequency, and duration of potentialhuman exposures to get an idea of internal dose and its effectsExposure assessments can provide an accurate representation of the dose received by an individual. Although an individual may be exposed to a certain concentration of an agent, many factors will determine the effective concentration of the agent that organs, tissues and cells are actually exposed to.
Rarely can you do only monitoring or only modeling
Personal Measurement (PM) as an example
PEM – Personal Exposure Monitors
Even biological monitoring is imperfectIndividual and environmental factors affect dose!Manioc root and susceptible population (your mother/sister)
People differDose mattersOccurrence & intensity of chemical-induced biologic effects are dose-relatedNot all chemical-induced biologic effects are harmfulThings Change
Famous for stating even water, in the right dose, is a poisonDrinking water excessively swelling/lysing of cells
Often think about dose-response in terms of pharmaceuticals
Can be easily shifted to EH using Parcelsus’ observations Simple when it is exposure and response like thisBut can change the shape of that lineDeleterious responses from startHormesis – low doses otherwise harmful chemicals beneficial (Vit A/aspirin)And what is something else missing from this graph?TIME!!!
Can we agree now that dose matters?What about this issue of time and that “Things Change”?
By 2:20 or 50 min – TAKE BREAKWhich is the cancer curve?
SF UF = studies!MF: Ranges between 0 and 1. 0 = complete certainty 1 = complete uncertaintyTherefore, as the estimate becomes more certain, the MF moves towards 0. Therefore, the denominator is smaller, and the RfD gets closer to the NOAEL. This makes sense – if you’re absolutely sure that your data is right, then that data will yield your reference dose.
Sensitivity analysisTake one parameter, change values to see if conclusions change. Provides screening-level determination of which parameters matter most.Characterize distributions of uncertain and variable parameters:Uniform Normal LognormalFormal uncertainty propagationCompare different results based on different models.Monte Carlo AnalysisRandom samples drawn for all parameters in a problemValues combined to yield estimate, then another random sample drawnAllows for distributions to be combined when no closed-form solution exists
Skip next 4 slides if after ___PM
By 2:52PM or 1hr, 22 min
Great example of EH at workSince it occurs on the population level, we can make huge changes through relatively small scale interventionsTake the Kuwait Fish StudyIf we institute an advisory (and people follow it), we can actually shift the mean IQ of the populationSmall decrements are reclaimed that the whole population shifts slightlyHUGE gains for the population, as a wholeLeads to greater economic growth and efficiency
Fish consumption and HgScience Policy interpretationScience says:Policy says:Interpretation is often to avoid fish altogether
System complexitySo many input and outcome variablesInputs include human CO2 emissions, non-human CO2 emissions, methane emissions, marine and terrestrial CO2 conversion capacity, etc.Outcomes include drought, flood, changed frequency of extreme weather events, rising sea levels, etc.Levels of variables are constantly changing (e.g. cutting down forests = less CO2 conversion capacity)Complex, multi-faceted relationships between variables and outcomesAlso dependent on location and level of analysis (e.g. floods from same rainfall are worse in places with clay soil; models may work on global level, but not at national or sub-national levels)Feedback loops/effects of adaptation – current models use projections, rather than forecasts, and assume a state of “if nothing changes, then in 50 years…”BUT, this doesn’t take into account 1) changes in current behaviors, etc., or 2) the impact of any adaptation strategiesCyclical risks – hazards, etc. are not linear, they are cyclicalClassic RA: burning coal more smog/particulates more carcinogens in lungs more cancerClimate change RA: ↑ atmospheric CO2 warmer ocean changed wind direction drought plants die less CO2 absorptive capacity ↑ atmospheric CO2…Multiplicative uncertaintySo many variables, many of which are very difficult to measure accuratelySo many outcomes, many of which are very difficult to measure accuratelySo many relationships between variables, difficult to measure accuratelySo many compartmentsMultiplicative uncertainty means that projections might be OK, butForecasts are hardModels will be inaccurateDecision-makers won’t act because of lack of certainty (link to risk management)Effects of interventions difficult to predict and measureDependence on location and level of analysis (state, region, globe, etc.)Sample size (esp. climate change)Though climate change has happened throughout history:Never had the tools to measure system variables like todayNever had some of the variables we have today (synthetic chemicals contributing to greenhouse effect/global warming)Never happened at the rate we’ve seen today (rate of CO2 emissions, rate of land clearance, etc.So, difficult to create “control group”, as only really a sample size of 1.Therefore, use current situation & limited comparative data as baseline
IFRC Vulnerability and Capacity Assessment toolCommunity-level risk assessmentUNDP Global Risk Assessment ProgrammeIdentifies and monitors climate change disaster riskUK Climate Change Risk Assessment methodologyMulti- sector gov. climate change risk assessmentAustralia-New Zealand Standard (assessments of climate change risks) Replicable framework to address climate change risks
Classic risk assessment may not be completely “correct” or applicable.BUT, is the framework still useful?