Transcript Questions for Discussion

Chinese Benzene Cohort Study
• 74,828 exposed and 35,805 unexposed
workers
• Leukemia deaths: RR 2.3, 95% CI 1.1-5.0
• Malignant lymphoma deaths: RR 4.5,
95% CI 1.3-28.4
• No deaths from multiple myeloma in
either population
TABLE IV. Results of Studies Reporting on Benzene and Total Leukemia
First Author
(year)
Girard (1970)a
Ishimaru (1971)b
Linos (1980)
Thorpe (1974)
Rushton (1981)
Tsai (1983)
Decoufle (1983)
Bond (1986)
Austin (1986)
Wong (1987a)
Study
setting
Community
Community
Community
Industry
Industry
Industry
Industry
Industry
Industry
Industry
Yin (1989)
Hurley (1991)
Paxton (1994)
Greenland (1994)
Industry
Industry
Industry
Industry
Exposed
cases
30
24
4
8
18
0
3
4
15
6
1
25
5
14
NA
NA
Relative
risk
2.9
1.8
3.3
1.2
2.3
0
6.8
1.9
NA
1.4
0.7
5.7
0.4
3.6
0.9
1.4
95%CI
1.1-7.4
0.9-3.7
0.6-27.6
0.5-2.4
1.0-5.0
0.0-8.7
1.4-19.9
0.5-4.9
NA
0.5-3.0
0.0-3.7
4.0-8.0
0.1-1.0
2.0-6.0
0.3-3.1
0.6-3.2
Wong and Raabe, 2000
Meta-Analysis of Risk of
Multiple Myeloma in
Petroleum Workers
• 22 cohort studies of >250,000 workers
• 220.93 deaths expected; 205 observed
• SMR 0.93; 95% CI 0.81-1.07
Bergsagel et al; Blood, 1999
Nested Case Control Study of
Multiple Myeloma and AML
• >18,000 petroleum distribution workers
• Up to 5 controls for each of 11 cases with
multiple myeloma and 13 with AML
• Cumulative mean exposure (ppm-yrs
THC):
– MM 672; controls 800
– AML 773; controls 837
– Wong, Trent & Harris, 1999
Why is it harder to find an elevated
SMR for MM than for AML in
benzene-exposed cohorts?
• Benzene does not cause MM
• Benzene is a relatively less potent myelogen
than it is a leukemogen
• Benzene is as or more potent a myelogen than
it is a leukemogen, but the background
incidence of MM is less than AML
• The latency period for MM is longer than for
AML
MORTALITY OF DISTRIBUTION WORKERS EXPOSED TO GASOLINE
Cause of death (ICDA-8)B
All causes of death (1-999)
All malignant neoplasms (140-209)
Cancer of digestive organs and peritoneum (150-159)
Cancer of respiratory system (160-163)
Cancer of lung (162-163)
Cancer of kidney (189)
Leukemia and aleukemia (204-207)
Cancer of other lymphatic tissue (203-203, 208)
All lymphopoietic cancer (200-209)
Allergic, endocrine, metabolic, nutritional diseases (240-279)
Diabetes mellitus (250)
All diseases of circulatory systems (390-458)
All respiratory diseases (460-519)
Cirrhosis of liver (571)
Motor vehicle accidents (810-827)
Suicide (950-959)
OBSERVED
2066
520
151
173
165
12
27
18
55
30
20
1039
150
31
35
26
SMR
51.3**
66.4**
69.4**
65.8**
66.2**
65.4
89.1
91.9
75.4*
44.6*
35.0**
48.9**
56.5**
36.5**
44.9**
39.4**
Healthy Worker Effect: Reasons
Specific to Hematopoietic Cancer
• Acute Myelogenous Leukemia
– Down’s syndrome, Fanconi’s anemia, etc
– Status post cancer chemotherapy
• Non-Hodgkin’s Lymphoma
– HIV AIDs
– Immune suppression diseases or
therapies
Exposure Dose Relationships in
Determining Causality
Is this a useful approach when
there are only small numbers
and borderline statistical
significance?
Challenges to Epidemiological
Identification of Environmental Causes
of Cancer
•
•
•
•
•
Cancer as a final common pathway
Unknown susceptibility factors
Uncertain exposure levels
Powerless negative
Cluster fallacy
Different Levels of Certainty
Required for Decisions
• SCIENTIFIC
(p < .05)
• TORT
(p < .50)
• REGULATORY (p = whatever)
Is Relative Risk Greater Than Two
Required for Proof of Causation?
(Carruth and Goldstein, 2001)
Threshold: Required to support inference
of causation
12
Threshold: Determinative of ultimate issue
2
Not required for inference of causation
14
Sufficient for inference of causation (not
clear if required)
1
Total # of published decisions
29
Is Relative Risk Greater Than Two
Required for Admissibility?
(Carruth and Goldstein, 2001)
Required for admissibility
10
Not required for admissibility
11
Total # of published cases
21
Cases Before and After Daubert II
Addressing Whether Relative Risk Greater
Than Two is Required for Admissibility of
Causation Opinion
Before
Daubert II
Daubert II
and After
Required
Not Required
0
10
5
6
Not Considered
9
1
Shortcomings of Occupational Epidemiology for
Establishing an Odds Ratio for a Specific Risk
• Healthy worker effect
• Inadequate exposure data
• Dilution of high risk group
• Appropriateness of time period