Interphone Studies To Date An Examination of Poor Study Design Resulting in an UNDERESTIMATION of th

1
Interphone Studies To DateAn Examination of
Poor Study Design Resulting in an
UNDER-ESTIMATIONof the Risk of Brain Tumors L.
Lloyd MorganBEMS, San Diego, 12 June 2008
2
MethodologyWhat If There Is No Risk of Brain
Tumors?

• ORs lt1.0 would be equal ORsgt1.0
• Think coin tossing
• OR1.0 are excluded
• 5 of ORs would be significant
• ORs lt1.0 would be equal ORsgt1.0
• Calculate ratio ORlt1.0/ORgt1.0
• 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

3
MethodologyCalculate 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)

4
ResultsPercent Significant Findings By
CategoryExpectation 5
5
ResultsRatio (ORlt1.0/ORgt1.0) Exposures gt10
Year and lt10 Year
6
ResultsRatio by Category
7
ResultsRatio by Brain Tumor Type
8
Interphone 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

9
Flaw 2Tumors Outside Radiation Plume Are
Unexposed

• Radiation plumes volume is small of brains
  volume
• Ipsilateral exposed
  Contralateral unexposed
• Absorbed radiation decreases rapidly with plume
  penetration depth
• Half-way to the brains mid-line, gt90 of energy
  is absorbed
• Percentage of absorbed cellphone radiation
• Ipsilateral temporal lobe 50-60 (wt. av.53)
• 15 of brains volume
• Ipsilateral cerebellum 12-25 (wt. av.19)
• 5 of brains volume
• 62-85 of absorbed radiation is in 20 of the
  brains volume
• Plume decreases rapid with depth (actual exposed
  brains volume lt20, perhaps 15)

10
Flaw 2Absorbed Radiation Decreases Rapidly w
Depth
11
Interphone 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

12
Flaws 3 and 4Latency Time and the Definition of
Regular Users
13
Interphone 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

14
Flaw 5Young Adults and Children Excluded
Korea Cellphone Rad.
Sweden Cellphone Rad.
Israel Ionizing Radiation
15
Interphone 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

16
Interphone 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

17
Interphone Protocol Design Flaws

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

18
How to Resolve Flaws

• Increase diagnosis eligibility time
• Nine Interphone studies weighted-average 2.6
  years
• Hardell et al. eligibility time 6 years
• Lower age range to lt15 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.,

19
Conflict-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
• Dont bite the hand that feeds you
• 90 significant protective results
• Ignored by authors (no commentary in the text)

20
Potential Brain Tumor Risk30-year Latency
Poisson Distribution Calculation
21
Potential Public Health Risk
22
Conclusions

• Interphone results substantially underestimate
  the risk of brain tumors
• Great majority of results have ORlt1.0
• Either cellphone use is protective, or the study
  has major flaws
• Ratio is lowest for highest exposures
  ipsilateral use or gt10 years of use
• Significant risk found for gt10 years and
  ipsilateral use
• Without design flaws Odds Ratios would increase
  substantially
• Cellphone industrys conflict-of-interest is
  obvious
• Government ignores potential epidemic (see no
  evil)
• Public health impact is enormous
• Industry independent studies are required

23
I Pray Im Wrong!
24
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

25
Interphone 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 gt10
  years
• Average 18 cases per study
• At minimum requires 2-fold more cases and
  controls for sufficient statistical power

26
Design 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

27
Design Changes to Resolve Flaws

• Reporting regular use
• Do not publish regular use data
• At minimum report regular use for gt5 years, or
  gt10 years
• Assumes gt3-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

28
Design Changes to Resolve Flaws

• Increase eligibility time to 9 years (for
  sufficient statistical power)
• gt3-fold increase in cases and controls
• Publish results every 3 years
• Provides longer latency time
• Resolves whether cellphones use initiates or
  promotes tumors

29
Flaw 2Tumors Outside Radiation Plume Are
Unexposed
30
Flaw 5Children ExcludedIonizing Radiation
Example
31
Flaw 5Young Adults ExcludedKorean Cellphone
Study
32
Flaw 5Young Adults ExcludedSwedish Cellphone
Study
33
Flaw 2Tumors Outside Radiation Plume Are
Unexposed