3 understanding and applying epidemiology hb lead-conf

Lead Litigation Conference 2013

November 14-15, 2013

Understanding & Applying Epidemiology Studies

Cris A. Williams, PhDENVIRON

The Speakers

Gary H. AbelsonHiscock & Barclay LLP

Nicholas SzokolyLaw Offices of Evan K. Thalenberg, P.A.

Karl Kieburtz, MD, MPHUniversity of Rochester Medical Center

Types of Studies

Observationalcase series – clinical observationecologic associations – existing datacross sectional (prevalence) – existing/new

datacase-control – new datacohort – retrospective – existing/new data prospective – new data

Interventionalnon-randomized trialrandomized, controlled trials

Evidence of Causal Relationship

Major: TemporalBiological PlausibilityConsistencyAlternative Explanations Explored

Other: Dose-response RelationshipStrength of AssociationCessation Effects

Gordis 1990

‘Learning’ Study Designs

Intended to answer the following questions:Does the agent get to where it may exert injury?

At a sufficient concentration?In a biologically active form?

Does it engage the target of interest?Does it influence downstream

biology/pharmacology?At what dose?

‘Confirming’ Study Designs

Studies that are designed to show that

an exposure/agent modifies a health state

RCT’s – Why Bother?

• Randomization• Blinding to treatment

Attempts to reduce bias and to improve the quality of evidence.

Moving from populations to

personsMedicine is always confronted with estimating what the benefits and

toxicities of interventions might be in a specific patient, given the

evidence available from populations, usually with little data

about ‘matched’ individuals

Attempts to address causality (relatedness) in individuals

Karch and Lasangna

Widely used in assessing adversityUncommon in thinking about benefit

Principles are the same

Outline Karl’s topic – Basics of Clinical Trials/How

Medicine Looks at Causation Introduction to Epidemiology Epi Study Designs and Strengths/Limitations Karl’s topic – How to Determine Causation in

an Individual Lead Studies Overview Lead Studies from the Plaintiffs’ Perspective Lead Studies from the Defense Perspective

Introduction to Epidemiology

Definition – study of occurrence of disease in populations

Subtypes according to exposure/agent◦ Biological – i.e., infectious disease (cholera)◦ Environmental

Natural environment (radon, asbestos, arsenic, lead)

Man-made or “anthropogenic”, incl. occupational (e.g., benzene but also asbestos, arsenic, lead)

Epidemiology Study Designs

Descriptive studies◦ For hypothesis generation/identification of risk

factors◦ Use data/information from readily-available

sources (census, disease registries, vital stats)◦ Cannot establish causal associations between

exposure and disease◦ Two subtypes: ecological; cross-sectional

Analytical studies◦ Hypothesis testing◦ Quantify relative risk of disease◦ Can establish causal associations between

exposure and disease◦ Two subtypes: case-control; cohort

Epidemiology Study Designs – Ecological Studies

Compare outcome frequencies between different groups during the same time period or in the same population at different time periods

Example – birth weight distributions in two different regions with different levels of arsenic in drinking water

No information on individuals’ exposure

No control of confounding (e.g., smoking)

Epidemiology Study Designs – Cross-Sectional Studies

Presence or absence of both exposure and outcome are assessed simultaneously; i.e., a “snapshot”

Example – study of infertility and psychological stress

Cannot distinguish whether exposure preceded outcome

Epidemiology Study Designs – Case-Control Studies

Subjects selected on the basis of whether they have (cases) or don’t have (controls) a specific outcome

Example – study of lung cancer cases and controls and residential radon exposure

Good for diseases with long latency periods and for rare diseases

Time- and cost-efficient Susceptible to bias

◦ Selection (e.g., hospital admits)◦ Recall (e.g., regarding exposure)

Epidemiology Study Designs – Cohort Studies

A group of individuals are defined on the basis of the presence or absence of exposure (e.g., worker studies)

Prospective – cohort identified and followed forward in time (exposure has not yet occurred)

Retrospective – cohort identified and traced backward in time (exposure had already occurred)

Longitudinal – cohort is the same individuals tracked forward or backward

Most studies relating lead exposure to IQ are longitudinal retrospective studies

Epidemiology Study Designs – Strengths and Limitations

Study Type Strengths Limitations

Ecological Quick, inexpensive, uses readily-available information.

No exposure information, doesn’t control for confounding.

Cross-sectional Quick, inexpensive, can provide valuable information on health status of a population.

Cannot determine whether exposure preceded or resulted from disease.

Case-control Good for diseases with long latency periods and for rare disease, time- and cost-efficient, can examine multiple etiologic factors for a single disease.

Good for rare diseases, temporal relationship between exposure and disease difficult to establish, results may be biased (selection and recall).

Cohort Good for rare exposures, can assess multiple effects of a single exposure and temporal relationship between exposure and disease, minimizes bias, allows for determination of disease incidence rates.

Not good for rare diseases, expensive and time-consuming (prospective), requires availability of adequate records (retrospective), susceptible to loss to follow up.

Source: Nielsen and Jensen (2005).

Lead Studies - Overview Dozens of studies relating blood lead and

neurobehavior/cognitive development in humans and animals

Dating from the late 1970’s

Key studies/reviews◦ Needleman and Gatsonis (1990) – Low-level Lead Exposure and

the IQ of Children. A Meta-analysis of Modern Studies (JAMA 263:673-678)

◦ Canfield et al. (2003a) – Intellectual Impairment in Children with Blood Lead Concentrations Below 10 µg per Deciliter (N Engl J Med 348:1517-1526)

◦ Lanphear et al. (2005) – Low-level Environmental Lead Exposure and Children’s Intellectual Function: An International Pooled Analysis (Environ Health Perspect 113:894–899)

Lead Studies – Overview (cont.)

Key reviews/commentary◦ Kaufman (2001a) – Do Low Levels of Lead Produce IQ Loss in Children?: A

Careful Examination of the Literature? (Archives of Clinical Neuropsychology 16:303-341)

◦ Needleman and Bellinger (2001) – Studies of Lead Exposure and the Developing Central Nervous System: A Reply to Kaufman (Archives of Clinical Neuropsychology 16:359-374)

◦ Kaufman (2001b) – How Dangerous are Low (Not Moderate or High) Doses of Lead for Children' s Intellectual Development? (Archives of Clinical

Neuropsychology 16:403-431)

Early studies – established the effect

Later studies – establishing the effect at increasingly lower PbB

“Plaintiff” studies and “Defense” studies

Lead Studies from the Plaintiffs’ Perspective

The hypothesis that Pb damages children's brains at low doses is widely accepted

There is no safe level of PbB/a linear inverse relationship exists between PbB and intelligence test scores (IQ) – i.e., no threshold

Correlations between PbB and IQ are socially relevant

Correlations between PbB and IQ remain when confounders were considered

Lead Studies from the Plaintiffs’ Perspective (cont.)

The hypothesis that Pb affects IQ at low doses is widely accepted (EPA 2013)

12 major prospective studies, 1992-2011 3 cross-sectional studies, 1987-2011 4 “meta” analyses/pooled studies, 1990-

2005

Lead Studies from the Plaintiffs’ Perspective (cont.)

There is no safe level of blood lead/a linear inverse relationship exists between PbB and intelligence test scores (IQ) – i.e., no threshold

Boston cohort – PbB = 1-9.3 μg/dL (Schwartz 1994) Rochester cohort – PbB = 0.5-8.4 μg/dL (Canfield et al.

2003b) Mexico City cohort – PbB = 0.8-4.9 μg/dL (Téllez-Rojo

2008) North Carolina cohort – PbB = 2 μg/dL (Miranda et al.

2009)**EOG (4th grade) as a surrogate for IQ.

Lead Studies from the Plaintiffs’ Perspective (cont.)

Correlations between PbB and IQ are socially relevant (Needleman and Bellinger 2001)

shift of IQ scores occurred across entire distribution of IQ scores

shift in median scores of 6 points associated with 4-fold increase in IQ scores <80

5% of population failed to achieve superior IQ scores >125

Exposures in early childhood associated with 7x increase in high school failure and a 6x increase in reading disabilities

Lead Studies from the Plaintiffs’ Perspective (cont.)

Correlations between PbB and IQ remain when confounders were considered (Needleman and Bellinger 2001)

Prospective studies controlled for at least some measure of maternal IQ

Most studies controlled for at least global measures of SES; in some cases controlled for home environment more specifically (e.g., via HOME, FACES, etc.)

Persons administering the IQ tests in most of the studies were adequately trained in psychometrics

Lead Studies from the Defense Perspective

The hypothesis that Pb affects IQ at low doses is controversial

There is a level of PbB below which effects on IQ are insignificant – i.e., a threshold

Correlations between PbB and IQ are not socially relevant

Correlations between PbB and IQ largely disappear when confounders were considered

Lead Studies from the Defense Perspective (cont.)

The hypothesis that Pb consistently affects IQ at low doses is controversial

Inconsistency in IQ findings both between studies and within studies (Ernhart 1995)

Studies purported to be “low-lead” studies in humans (and animals) are actually moderate- to high-lead studies (Kaufman 2001b)

Gender-specific effects – some studies showed > effects in girls than boys; others demonstrated the opposite (Hebben 2001)

Although many studies report a significant association between lead and IQ, lead tends to account for a very small amount of variance in IQ (Bellinger and Dietrich 1994)

Lead Studies from the Defense Perspective (cont.)

There is a level of PbB below which effects on IQ are insignificant – i.e., a threshold

Assumption of PbB/IQ linearity based not on actual data but rather on dubious regression analyses (Kaufman 2001b)

EPA’s Integrated Science Assessment for Lead (EPA 2013)◦ Lack of a reference population (PbB in pre-industrial

population) limits ability to identify a threshold◦ “. . . the current evidence does not preclude the

possibility of a threshold for neurodevelopmental effects in children existing with lower blood levels than those currently examined.”

Lead Studies from the Defense Perspective (cont.)

Correlations between PbB and IQ are not socially relevant (Kaufman 2001b)

A few IQ points is well within a reasonable band of error around the observed score

The usual IQ loss attributed to low PbB is similar in magnitude to the 2- to 3-point mean gender difference (favoring males)

IQ score is meaningless without measurement of other aspects of intellectual functioning – e.g., creativity, social intelligence, practical intelligence, adaptive behavior, mechanical ability, etc.

Lead Studies from the Defense Perspective (cont.)

Correlations between PbB and IQ largely disappear when confounders were considered (Kaufman 2001a and b)

For the 26 PbB/IQ studies where confounders may have affected the results:◦ 12 used only a global assessment of SES◦ 17 used a “poor” or no measure of maternal IQ◦ 24 did not test father’s IQ◦ 24 did not control for persistent otitis media◦ 18 did not control for pregnancy risk factors

(e.g., maternal drug use/abuse, smoking)

Lead Studies from the Defense Perspective (cont.)

Difference in HOME Scores

Difference in Parental Years of Education

Difference in Maternal IQ Scores

7.5 15 22.5

Difference in Child IQ Scores

3.6 0.75 7.58 9.42 11.27

1.5 9.11 10.96 12.81

2.25 10.65 12.50 14.35

7.2 0.75 11.77 13.62 15.46

1.5 13.30 15.15 17.00

2.25 14.84 16.69 18.54

10.8 0.75 15.96 17.81 19.65

1.5 17.50 19.34 21.19

2.25 19.03 20.88 22.73

Source: Mink et al. (2004).

Lead Studies from the Defense Perspective (cont.)

An additional “threat” to the validity of the Pb/IQ studies are variables associated with intelligence that are either unknown or unmeasurable – the “Flynn effect” (Kaufman 2001a)◦ 3 point-per-decade gain in IQ beginning in the

1930’s and continuing to this day◦ Exposure to technology/mass media?◦ Parenting/increased awareness of importance

of providing cognitive stimulation in infancy?◦ Improved nutrition?

References Bellinger, D., and Dietrich, K. N. 1994. Low-level lead exposure and cognitive

function in children. Pediatric Annals 23:601-605. Canfield, R.L., Henderson, C.R. Jr., Cory-Slechta, D.A., Cox, C., Jusko, T.A., and

Lanphear, B.P. 2003a. Intellectual Impairment in children with blood lead concentrations below 10 µg per deciliter. N Engl J Med 348:1517-1526.

Canfield, R.L.., Kreher, D.A., Cornwell, C., and Henderson, C.R., Jr. 2003b. Low-level lead exposure, executive functioning, and learning in early childhood. Child Neuropsychol 9:35-53.

EPA. 2013. Integrated Science Assessment for Lead. EPA/600/R-10/075F. Office of Research and Development, National Center for Environmental Assessment, Research Triangle Park, NC.

Ernhart, C.B. 1995. Inconsistencies in the lead-effects literature exist and cannot be explained by "effect modification". Neurotoxicol Teratol. 17(3):227-233.

Hebben, H. 2001. Low lead levels and neuropsychological assessment: Let us not be mislead. Archives of Clinical Neuropsychology 16:353-357.

Kaufman, A.S. 2001a. Do low levels of lead produce IQ loss in children?: A careful examination of the literature? Archives of Clinical Neuropsychology 16:303-341.

Kaufman, A.S. 2001b. How dangerous are low (not moderate or high) doses of lead for children' s intellectual development? Archives of Clinical Neuropsychology 16:403-431.

References (cont.) Lanphear, B.P., Hornung, R., Khoury, J., Yolton, K., Baghurst, P., Bellinger, D.C.,

Canfield, R.L., Dietrich, K.N., Bornschein, R., Greene, T., Rothenberg, S.J., Needleman, H.L., Schnaas, L., Wasserman, G., Graziano, J., and Roberts, R. 2005. Low-level environmental lead exposure and children’s intellectual function: An international pooled analysis (Environ Health Perspect 113:894–899.

Mink, P.J., Goodman, M., Barraj, L.M., Imrey, H., Kelsh, M.A., and Yager. J. 2004. Evaluation of uncontrolled confounding in studies of environmental exposures and neurobehavioral testing in children. Epidemiology 15(4):385-393.

Miranda, M.L., Kim, D., Reiter, J., Overstreet Galeano, MA., and Maxson, P. 2009. Environmental contributors to the achievement gap. Neurotoxicology 30:1019-1024.

Needleman, H.L., and Bellinger, D. 2001. Studies of lead exposure and the developing central nervous system: A reply to Kaufman. Archives of Clinical Neuropsychology 16:359-374.

Needleman, H.L., and Gatsonis, C.A. 1990. Low-level lead exposure and the IQ of children. A meta-analysis of modern studies. JAMA 263(5):673-678.

Nielsen, J.B. and Jensen, T.K. 2005. Environmental Epidemiology. In: Essentials of Medical Geology – Impacts of the Natural Environment on Public Health. O. Selinus, Ed. Elsevier.

Schwartz, J. 1994. Low-level lead exposure and children's IQ: A meta-analysis and search for a threshold. Environ Res 65:42-55..

References (cont.) Téllez-Rojo, M.M., Bellinger, D.C., Arroyo-Quiroz, C., Lamadrid-Figueroa, H.,

Mercado-Garcia, A., Schnaas-Arrieta, L., Wright, RO., Hernandez-Avila, M., and Hu, H. 2006. Longitudinal associations between blood lead concentrations lower than 10 microg/dL and neurobehavioral development in environmentally exposed children in Mexico City. Pediatrics 118:e323-e330.

Gary H. Abelson, Esq.Hiscock & Barclay, LLP

2000 HSBC Plaza100 Chestnut Street

Rochester, New York 14604Tel: (585) 295-4412

Fax: (585) 295-8411Email: gabelson@hblaw.com

EPIDEMIOLOGIC STUDIES FOR LEADWHAT ARE THEY GOOD FOR?

• Can epidemiologic studies be used “across the board” to determine whether lead has affected a particular individual?

General Causation

v.

Specific Causation

• Population based studies

• They were not designed – nor could they be – to determine whether a specific child has been affected by lead exposure

Does A equal B?

1. Do all smokers get lung cancer?

2. Do all football players sustain brain injuries?

• Epidemiologic Studies may show “general causation”

• They cannot, by themselves, get to specific causation in an individual.

• Lead and IQ – is there a correlation?

1. Is pre-morbid IQ testing necessary?

a.In order to make an intelligent assessment of whether an individual has lost IQ points as a result of exposure to lead there must be a baseline.

b.Studies that claim to quantify IQ loss as a result of exposure to lead are speculative.

c.Not possible to “assign” a loss of IQ with any elevated lead level – only a comparison between pre-injury testing and post-injury assessment can yield a true measure of intellectual change.

d. Full scale IQ measure can be unreliable for determining a person’s actual level of cognitive capability.

e. Neuropsychological testing is used to test areas involving memory, attention , executive functions, verbal fluency.

f. How does an individual test in these areas? Have to look “behind the numbers”

g. If a child scores low, medium or high on any test – does not mean they would have scored otherwise had there been no lead exposure

• There is no specific/single neuropsychological test or result that allows for the conclusion that a specific child’s lead exposure had any impact.

2. Confounding VariablesSignificant and well-known risk factors for the development of cognitive difficulties mirror claims of lead exposure:

a.Low birth weight

b.Maternal pregnancy issues – drugs/alcohol in utero exposure

c.Prenatal smoking

d. Parental involvement

e. Structured environment/lack thereof

f. Abuse issues1) Mental2) Physical 3) Sexual

h. Potential foster care

i. Outside influences

j. Drug Use by individuals

k. Depression/anxiety

• Each individual different – cannot “control” for all variables

• Fallacy of argument that confounding factors are being counted twice

3. School Performance

a. Attendance – school and specific classes

b. Homework completion

c. Extra help

d. Behavior problemse. Discipline problemsf. Drug useg. Frequent changes in

school – disruptive learning

Conclusion

General population studies are just that, not capable of being used to determine that any individual has been affected by

alleged exposure to lead.

The earth is flat;and epidemiological studies

over the past half century cannot establish specific

causation.

Framingham heart study? ONE OF THE LONGEST EPIDEMIOLOGICAL

LONGITUDINAL STUDIES!

Quit smoking on the advice of your physician?Epidemiological research.

Epidemiological research.

Take Tamiflu in 2009?

Ignoring history. Lead is the most well studied toxin in history. International pooled analysis:

◦ Data, not average of averages.◦ Seven participating sites:

Boston, Mass. Cincinnati, Ohio Cleveland, Ohio Mexico City, Mexico Port Pirie, Australia Rochester, NY Yugoslovia

Approximately 1,300 children.

Confounding variables. International Pooled Analysis

◦ HOME Inventory ◦ Sex◦ Birth weight◦ Birth order◦ Maternal education◦ Maternal IQ◦ Maternal age◦ Marital status◦ Prenatal smoking status◦ Prenatal alcohol use

Q And I think you had said it once best, and tell if it sounds -- you said this, but just tell me

if you still agree with it, the researchers control for those factors so that the conclusion of their study is specific to what is the measurable effect that lead has on IQ loss?

A Correct. Q And that's in the context of the pooled analysis. And for

Needleman, it would be the researchers controlled for those factors so the conclusion of their study is specific to what is the measurable effect that lead has on behavior?

A Correct. Q And so it's incorrect to assume that in order to reach the opinions

you have in this case, that you need administer the HOME test to the plaintiffs to determine whether or not their lead exposure caused them IQ loss?

A That's correct. Q It's already been done. That's the point. A Well, and I'm only giving opinions about what IQ loss I would

attribute using this type of medical literature to this particular child. Q And I know it sounds pedantic. Just work with me here. But the

point that I'm making is that you don't have to go back in the forensic context and control for all these variables again. That's why the researchers did it in the first place?

MR. [Lawyer for the Defendant]: Objection. A That's correct. Yes. I agree with that.

School performance. If only Johnny with his 65 IQ, reading

disorder and ADHD wouldn’t have missed 30 days his 4th time in 9th grade, it would have all worked out okay.

Hammond et al., May 2007◦ Among the top risk factors for dropping out of

middle school and high school: Learning disabled Low academic achievement Retention/over-age for grade

Multiple retentions have additive effects, dramatically increasing the risk of dropping out.

Questions

Dr. Cris A. Williams Ph.D., Senior Science Advisor ENVIRON 850.668.3551 cwilliams@environcorp.com

Gary Abelson Hiscock & Barclay LLP 585.295.8411 gabelson@hblaw.com

Nicholas Szokoly Law Offices of Evan K. Thalenberg, P.A. 410.625.9200 nszokoly@ektlaw.com

Dr. Karl Kieburtz MD, MPH University of Rochester Medical Center 585.275.8762 Karl.Kieburtz@chet.rochester.edu

Speaker Contact Info