bems presentation v14_6-1-08-3

L. Lloyd Morgan [[email protected]] 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-ESTIMATION

of the Risk of Brain Tumorsof the Risk of Brain Tumors

L. Lloyd MorganBEMS, San Diego, 12 June 2008


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


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)


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%


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


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=0.0097 p=0.16

p=0.097

HighestExposure

Lowest Ratio

Non-significantNear-significant


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


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


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%)


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


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


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


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


Flaw 5Flaw 5Young AdultsYoung Adults andand ChildrenChildren ExcludedExcluded

0

1

2

3

4

20-29 years 30-39 years 40-49 years 50-59 yearsAge 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

012345678

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)


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


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


Interphone Protocol Design Interphone Protocol Design FlawsFlaws

Flaw 10: Recall bias Light users underestimate use Heavy users overestimate use Result: Large underestimation of risk


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., …


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)


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%


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

19

86 19

87 19

88 19

89 19

90 19

91 19

92 19

93 19

94 19

95 19

96 19

97 19

98 19

99 20

00 20

01 20

02 20

03 20

04 20

05 20

0620

0720

0820

0920

1020

1120

1220

1320

1420

1520

1620

1720

1820

19

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


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


I Pray I’m Wrong!


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


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


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


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


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


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 Cerebellum50% 19%

Total Raditaion Absobed by Structure9% 5% 12%


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)


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 yearsAge Range

OR

P<0.01Source: J.W. Choi el al.Case-control Studies on Human Effects of Wireless Phone RF in Korea, BEMS 2006


Flaw 5Young Adults ExcludedYoung Adults Excluded

Swedish Cellphone StudySwedish Cellphone Study

Increased Risk of Brain Tumor

012345678

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.


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%

bems presentation v14_6-1-08-3

Data & Analytics