teachers' cancers la quinta middle school 1988-2005 sam milham, md
Post on 20-Jun-2015
257 Views
Preview:
TRANSCRIPT
L. Lloyd Morgan [bilovsky@aol.com] 1
Interphone Studies To DateInterphone Studies To Date
An Examination of Poor Study DesignAn Examination of Poor Study Design
Resulting in an UNDER-ESTIMATIONResulting in an UNDER-ESTIMATIONof the Risk of Brain Tumorsof the Risk of Brain Tumors
L. Lloyd MorganBEMS, San Diego, 12 June 2008
L. Lloyd Morgan [bilovsky@aol.com] 2
MethodologyMethodologyWhat If There Is No Risk of Brain Tumors?
ORs <1.0 would be ~equal ORs>1.0 Think coin tossing
• OR=1.0 are excluded
~5% of ORs would be significant ORs <1.0 would be ~equal ORs>1.0
Calculate ratio: OR<1.0/OR>1.0 13 Interphone brain tumor studies to date
• Exclude 2 overlapping studies and recent (Schlehofer) study
• Analysis restricted to 10 Interphone brain tumor studies
Calculate binomial p-values
L. Lloyd Morgan [bilovsky@aol.com] 3
MethodologyMethodologyCalculate Ratio by Categories by Studies
How to have statistically independent categories? Compare between studies, not within studies Categories
• Brain Tumors– All– Acoustic Neuroma– Glioma– Meningioma
• Years of use (Years)• Cumulative hours of use (Hours)• Cumulative number of calls (Call #)• “Regular” cellphone use (“Regular”)• Years of ipsilateral cellphone use (Years Ipsi)• Years of contralateral cellphone use (Yrs Contra)• Minutes of cellphone use per day (Min/Day)
L. Lloyd Morgan [bilovsky@aol.com] 4
ResultsResultsPercent Significant Findings By Category
Expectation: ~5%
0%
5%
10%
15%
20%
25%
All Years Hours Call # "Regular" Years Ipsi YearsContra
Min perDay
Categories
~ expected
Basedon 7
Findings0%
L. Lloyd Morgan [bilovsky@aol.com] 5
ResultsResultsRatio (OR<1.0/OR>1.0):
Exposures: >10 Year and <10 Year
0.0
1.0
2.0
3.0
4.0
5.0
>10 year <10 year
Ratio
Longest ExposureLowest Ratio
p=0.16
p=3.6 x 10-22
~ Expected ratio if no risk
L. Lloyd Morgan [bilovsky@aol.com] 6
ResultsResultsRatio by Category
0
1
2
3
4
5
6
7
8
All Call # Hours Years "Regular" YearsContra
Min perDay
YearsIpsi
Categories
Ratio
p=1.2x10-20
p=6.0x10-07
p=6.1x10-06
p=1.1x10-07
p=0.0011 p=0.0097p=0.16
p=0.097
HighestExposure
Lowest Ratio
Non-significant
Near-significant
L. Lloyd Morgan [bilovsky@aol.com] 7
ResultsResultsRatio by Brain Tumor Type
0
1
2
3
4
5
6
All AcousticNeuroma
Glioma Meningioma
Ratio
p=1.2x10-20
p=2.9x10-5
p=6.0x10-10
p=8.2x10-9
L. Lloyd Morgan [bilovsky@aol.com] 8
Interphone Protocol Design FlawsInterphone Protocol Design Flaws Flaw 1: Selection Bias
Participating controls use cellphones more than non-participating controls
• Weighted average control participation rate: 59%• Löon 2004: 20% control refused; 34% used, 59% did not use
Underestimates risk Flaw 2: Tumors outside the radiation plume are
unexposed Unexposed tumors treated as exposed Plume volume small relative to brain volume
• Well know since 1994 (4 previous papers) Underestimates risk
L. Lloyd Morgan [bilovsky@aol.com] 9
Flaw 2Flaw 2Tumors Outside Radiation Plume Are UnexposedTumors Outside Radiation Plume Are Unexposed
Radiation plume’s volume is small % of brain’s volume Ipsilateral: exposed Contralateral: unexposed
Absorbed radiation decreases rapidly with plume penetration depth• Half-way to the brain’s mid-line, >90% of energy is absorbed
Percentage of absorbed cellphone radiation Ipsilateral temporal lobe: 50-60% (wt. av.=53%)
• ~15% of brain’s volume “Ipsilateral” cerebellum: 12-25% (wt. av.=19%)
• ~5% of brain’s volume
62-85% of absorbed radiation is in ~20% of the brain’s volume Plume decreases rapid with depth (actual exposed
brain’s volume: <20%, perhaps 15%)
L. Lloyd Morgan [bilovsky@aol.com] 10
Flaw 2Flaw 2Absorbed Radiation Decreases Rapidly w DepthAbsorbed Radiation Decreases Rapidly w Depth
Relative Absorbed Radation and Penetration Depth in Temporal Lobe
0%
10%
20%
30%
40%
50%
60%
15-24 25-34 35-44 45-54 55-64 65-74 75-84
Depth (mm)
% Absorbed Radiation Relative to Max
Absorbed Radiation
900 MHz European Phones (worst case)
800-900 MHz Japanese Phones
Source: Cardis et al 2008
L. Lloyd Morgan [bilovsky@aol.com] 11
Interphone Protocol Design FlawsInterphone Protocol Design Flaws
Flaw 3: Latency time Known latency times
• Ionizing radiation & brain tumor: 20-40 years• Smoking & lung cancer: ~30 years• Asbestos & mesothelioma: 20-40+ years
Short latency times underestimates risk Flaw 4: Definition of “regular” user
“Regular” user: At least once a week for 6 months or more
• If definition of “regular” smoker were used, would a risk of lung cancer be found?
Definition of “regular” user underestimates risk
L. Lloyd Morgan [bilovsky@aol.com] 12
Flaws 3 and 4Flaws 3 and 4Latency Time and the Definition of “Regular Users”Latency Time and the Definition of “Regular Users”
UK Subscribers by Year
0
10
20
30
40
50
60
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Years from Eligibility Date(Latency Time)
Millions
0Wt. Ave.Eligibility
Date2002.5
123456789101112131415
<5 year latency85% User-Year
>5 year latency15% User-years
>10 year latency2% User-years
L. Lloyd Morgan [bilovsky@aol.com] 13
Interphone Protocol Design FlawsInterphone Protocol Design Flaws
Flaw 5: Young adults and children excluded Children and young adults at greater risk than
adults• Interphone Protocol: 30-59 years
– Some studies reduce minimum age to 20 years
• Underestimates risk
L. Lloyd Morgan [bilovsky@aol.com] 14
Flaw 5Flaw 5Young AdultsYoung Adults andand ChildrenChildren ExcludedExcluded
0
1
2
3
4
20-29 years 30-39 years 40-49 years 50-59 years
Age Range
OR
P<0.01Source: J.W. Choi el al.Case-control Studies on Human Effects of Wireless Phone RF in Korea, BEMS 2006
Increased Risk of Brain Tumor
0
1
2
3
4
5
6
7
8
20-80 years 20-29 years 20-80 years 20-29 years
Analog cellphone Cordless phone
OR
Source: Hardell et al.Arch Environ Health. 2004 Mar;59(3):132-7.
Israel: Ionizing Radiation
Korea: Cellphone Rad.
Sweden: Cellphone Rad.
Excess Relative Risk (ERR) per Gray (Gy)Malignant Brain Tumors
by Age from Ionizing Radiation Exposure
356%
224%
47%0%
100%
200%
300%
400%
<5 5-9 10+Age at Exposure
ERR/GY
Mean estimated dose: 1.5 Gy (range 1.0 to 6.0 Gy)
Source: Sadetzki et al., RADIATION RESEARCH 163, 424–432 (2005)
L. Lloyd Morgan [bilovsky@aol.com] 15
Interphone Protocol Design FlawsInterphone Protocol Design Flaws
Flaw 6: Comparison cellphone radiated power: higher vs lower Analog Vs Digital phones
• No longer possible
Rural Vs Urban users Underestimates risk
L. Lloyd Morgan [bilovsky@aol.com] 16
Interphone Protocol Design FlawsInterphone Protocol Design Flaws Flaw 7: Cordless phone, walkie-talkie, Ham, and
proximity to TV & radio transmitters Treated as unexposed Underestimation of risk
Flaw 8: Exclusion of brain tumor types Includes only acoustic neuroma, glioma & meningioma Other brain tumor types are excluded
• For example lymphoma and neuroepithelial brain tumors Underestimates risk
Flaw 9: Exclusion of brain tumor cases because of death Underestimates risk of most deadly brain tumors
L. Lloyd Morgan [bilovsky@aol.com] 17
Interphone Protocol Design Interphone Protocol Design FlawsFlaws
Flaw 10: Recall bias Light users underestimate use Heavy users overestimate use Result: Large underestimation of risk
L. Lloyd Morgan [bilovsky@aol.com] 18
How to Resolve How to Resolve FlawsFlaws Increase diagnosis eligibility time
Nine Interphone studies: weighted-average 2.6 years• Hardell et al. eligibility time: 6 years
Lower age range to <15 years Pay controls (and cases?) for participation in study
Do not tell controls what is the purpose of the study
Interview proxies in case of death Separately report both case and proxy interview results
Treat unexposed tumors as unexposed Etc., Etc., Etc., …
L. Lloyd Morgan [bilovsky@aol.com] 19
Conflict-of-InterestConflict-of-Interest Cellphone Industry
Interphone funding is inadequate to resolve flaws• More funding, greater potential of substantial revenue
loss
Researchers’ conflict-of-interest (unconscious?) Source of funds is known in spite of “Firewall” Honest, but “Don’t bite the hand that feeds you”
• 90 significant protective results – Ignored by authors (no commentary in the text)
L. Lloyd Morgan [bilovsky@aol.com] 20
Potential Brain Tumor RiskPotential Brain Tumor Risk30-year Latency30-year Latency
Poisson Distribution Calculation
0%
20%
40%
60%
80%
100%
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40
Years Since First Exposure
Risk
19851st Use
199510 Yrs
200520 Yrs
201530 Yrs
0.002%
4%
55%
97%
L. Lloyd Morgan [bilovsky@aol.com] 21
Potential Public Health RiskPotential Public Health RiskPotential Brain Tumor Cases From Use of a Cellphone
Assuming a 30-Year Latency Time and 10% of Users1
Diagnosed with a Brain Tumor
0
200,000
400,000
600,000
800,000
1,000,000
1,200,000
1,400,000
1,600,000
1,800,000
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
Potential Cases of Brain Tumors
per Year
0
50
100
150
200
250
300
350
400
450
CellphoneSubscribers
millions
Year 200444,447 Dx
~3,600 fromcellphone use
Year 20191,590,513
1 Based on 10% of long-term smokers are diagnosed with lung cancer
Source Cellphone Subscribers: CTIASource brain tumor diagnosed in 2004: CBTRUS
L. Lloyd Morgan [bilovsky@aol.com] 22
ConclusionsConclusions Interphone results substantially underestimate the risk of brain
tumors Great majority of results have OR<1.0
• Either cellphone use is protective, or the study has major flaws
Ratio is lowest for highest exposures: ipsilateral use or >10 years of use
Significant risk found for >10 years and ipsilateral use
Without design flaws Odds Ratios would increase substantially Cellphone industry’s conflict-of-interest is obvious Government: ignores potential epidemic (see no evil)
Public health impact is enormous
Industry independent studies are required
L. Lloyd Morgan [bilovsky@aol.com] 23
I Pray I’m Wrong!
L. Lloyd Morgan [bilovsky@aol.com] 24
Now What?Now What?Based on CBTRUS Incidence Data
Window closed for case-control studies No unexposed cases remain
Cohort studies Unable to know users of company owed cellphones Unable to interview cellphone users Requires enormous numbers
• 1,000,000 user-years will find (assuming cellphones do not increase risk)
– ~6 acoustic neuromas
– ~54 gliomas
– ~45 meningiomas
Requires ~1 billion user-years to analyze by• Gender, SES, Years of use ,Exposed tumors only
Requires 30 year cohort study
L. Lloyd Morgan [bilovsky@aol.com] 25
Interphone Protocol Design FlawsInterphone Protocol Design Flaws Flaw 11: Recall bias
Interview cases immediately after diagnosis and 6 months after surgery
• Improved memory and cognition 6 months after surgery
Flaw 12: Observational bias Interviewer not blinded with face-to-face interviews
• Mailed questionnaires provide blindness– Supplement by phone as necessary
Flaw 13: Too few cases for statistical power Nine Interphone Brain Tumor Studies: Use for >10 years
• Average 18 cases per study• At minimum requires 2-fold more cases and controls for
sufficient statistical power
L. Lloyd Morgan [bilovsky@aol.com] 26
Design Changes to Resolve FlawsDesign Changes to Resolve Flaws
Treat unexposed tumors as unexposed Tumors outside radiation plume
• Data was available, but to date not used, or even discussed– Too few cases?
Treat RF/MW exposures and exposed Cordless phone, walkie-talkie radios, Ham transmitters
Overweight rural users or increase eligibility time Compare risk of brain tumor with rural and urban users
• Requires sufficient number of cases and controls
Use questionnaires not face-to-face interviews
L. Lloyd Morgan [bilovsky@aol.com] 27
Design Changes to Resolve FlawsDesign Changes to Resolve Flaws
Reporting “regular” use Do not publish “regular” use data At minimum report “regular” use for >5 years,
or >10 years• Assumes >3-fold increase in case eligibility range
Latency time: initiation or promotion? Some researchers assume cellphone can only be
promoters• What is evidence for initiation vs promotion?
Follow cases & controls for a longer period
L. Lloyd Morgan [bilovsky@aol.com] 28
Design Changes to Resolve FlawsDesign Changes to Resolve Flaws
Increase eligibility time to 9 years (for sufficient statistical power)
• >3-fold increase in cases and controls
• Publish results every 3 years
• Provides longer latency time
• Resolves whether cellphones use initiates or promotes tumors
L. Lloyd Morgan [bilovsky@aol.com] 29
Flaw 2Tumors Outside Radiation Plume Are UnexposedTumors Outside Radiation Plume Are Unexposed
900 MHz European Phone: Depth by Structure
0%
10%
20%
30%
40%
50%
60%
15-24 25-34 35-44 45-54 55-64 65-74 75-84Depth (mm)
RelativeSAR
Temporal Frontal Parietal Occipital Cerebellum
50% 19%Total Raditaion Absobed by Structure
9% 5% 12%
L. Lloyd Morgan [bilovsky@aol.com] 30
Flaw 5Children ExcludedChildren Excluded
Ionizing Radiation ExampleIonizing Radiation ExampleExcess Relative Risk (ERR) per Gray (Gy)
Malignant Brain Tumorsby Age from Ionizing Radiation Exposure
47%
224%
356%
0%
100%
200%
300%
400%
<5 5-9 10+Age at Exposure
ERR/GY
Mean estimated dose: 1.5 Gy (range 1.0 to 6.0 Gy)
Source: Sadetzki et al., RADIATION RESEARCH 163, 424–432 (2005)
L. Lloyd Morgan [bilovsky@aol.com] 31
Flaw 5Young Adults ExcludedYoung Adults ExcludedKorean Cellphone StudyKorean Cellphone Study
0
1
2
3
4
20-29 years 30-39 years 40-49 years 50-59 years
Age Range
OR
P<0.01Source: J.W. Choi el al.Case-control Studies on Human Effects of Wireless Phone RF in Korea, BEMS 2006
L. Lloyd Morgan [bilovsky@aol.com] 32
Flaw 5Young Adults ExcludedYoung Adults Excluded
Swedish Cellphone StudySwedish Cellphone Study
Increased Risk of Brain Tumor
0
1
2
3
4
5
6
7
8
20-80 years 20-29 years 20-80 years 20-29 years
Analog cellphone Cordless phone
OR
Source: Hardell et al.Arch Environ Health. 2004 Mar;59(3):132-7.
L. Lloyd Morgan [bilovsky@aol.com] 33
Flaw 2Tumors Outside Radiation Plume Are UnexposedTumors Outside Radiation Plume Are Unexposed
Source: http://serendip.brynmawr.edu/bb/kinser/Glossary.html ~10.4
cm
~5.6 cm
Surface area ~162 cm2
Ipsilateral temporal lobe’s volume to total brain’s volume ~15%
Worst case: ~62% of ipsilateral radiation is absorbed in ~20% of brain
Ipsilateral cerebellum’s volume to total volume ~5%
top related