Are Cell Phones a Possible Carcinogen? An Update on the IARC Report

EDITOR’S NOTE: Because I am at the annual meeting of the American Association for Cancer Research in Chicago, between the meetings, working on a policy statement, working on a manuscript, and various other miscellaneous tasks, I alas was unable to produce a post worthy of the quality normally expected by SBM readers. Fortunately, Lorne Trottier, who’s done a great job for us twice before, was able to step in again with this great post about “safe” cell phone cases. Speaking of the manufactroversy over whether cell phone radiation causes brain cancer, there’s a session at the AACR that I’ll have to try to attend entitled Do Cell Phones Cause Brain Cancer? Who knows? It might be blogging material. I also might post something later that those of you who know of my not-so-super-secret other blog might have seen before. However, I often find it useful to see how a different audience reacts. Now, take it away, Lorne…

In May of last year, the International Agency for Research on Cancer (IARC) issued a press release (1) in which it classified cell phones as Category 2B, which is “possibly carcinogenic to humans“. This ruling generated headlines world wide. Alarmist groups seized on it and now regularly cite this report to justify their concerns for everything ranging from cell phones to WiFi and smart meters.

IARC maintains a list of 269 substances in the 2B category, most of which are chemical compounds. A number of familiar items are also included in this list: coffee, pickled vegetables, carbon black (carbon paper), gasoline exhaust, talcum powder, and nickel (coins). The IARC provides the following definition of the 2B category (2  P 23): “This category is used for agents for which there is limited evidence of carcinogenicity in humans and less than sufficient evidence of carcinogenicity in experimental animals. It may also be used when there is inadequate evidence of carcinogenicity in humans but there is sufficient evidence of carcinogenicity in experimental animals“.

The Category 2B “possible carcinogen” classification does not mean that an agent is carcinogenic. As Ken Foster of the University of Pennsylvania pointed out to me. “Their conclusion is easy to misinterpret.” “Saying that something is a “possible carcinogen” is a bit like saying that someone is a “possible shoplifter” because he was in the store when the watch was stolen. The real question is what is the evidence that cell phones actually cause cancer, and the answer is — none that would persuade a health agency.”

None the less this ruling was highly controversial. Expert groups of most of the world’s major public health organizations have taken the same position as the European Commission’s Scientific Committee on Emerging and Newly Identified Health Risks (SCENIHR) which had stated that (3  P 8): “It is concluded from three independent lines of evidence (epidemiological, animal and in vitro studies) that exposure to RF fields is unlikely to lead to an increase in cancer in humans“. The representative of the US National Cancer Institute walked out of the IARC meeting before the voting. The NCI issued a statement (4) quoting other studies stating that: “overall, cell phone users have no increased risk of the most common forms of brain tumors — glioma and meningioma“.

Immediately following the IARC decision the WHO issued a reassuring new Fact Sheet (5) on mobile phones and public health: “A large number of studies have been performed over the last two decades to assess whether mobile phones pose a potential health risk. To date, no adverse health effects have been established as being caused by mobile phone use”. Since this controversial IARC classification, several new papers have been published that substantially undermine the weak evidence on which the IARC based its assessment.

The evidence that IARC cited to support its assessment was poor to begin with. Their initial press release (1) was followed by a more complete report that was published in the July 1, 2011 issue of the Lancet Oncology as well as online (6). In this article, I will review the evidence cited by IARC in support of its conclusion. I will also review updates from new papers published over the past year that cast further doubt on IARC’s conclusion.

Brain Cancer Incidence Rates

Brain cancer is one of the rarer forms of cancer. In Canada, it ranks as #15, well below the leading cancers such as lung, breast, and prostate. There are more than 5 billion cell phones in the world. The fact that brain cancer incidences rates have remained flat  in the US and elsewhere is one of the simplest and strongest indicators that cell phones do not cause cancer. Yet in the Lancet article, IARC dismisses the value of studies of brain cancer incidence rates with the statement that they “have substantial limitations because most of the analyses examined trends until the early 2000s only“. This is surprising, since at the time of IARC’s assessment there were already a couple of studies that found flat cancer incidence rates up to the end of 2006 in the US (7) and 2007 in the UK (8). These important papers are certainly past the “early 2000s”.

In the same month as the IARC assessment, a new study was published online in Environmental Health Perspectives that examined brain cancer incidence in Sweden up to the end of 2009 (9). Fig 1 of the Supplemental Material (see below) shows that brain cancer incidence rates in Sweden were essentially flat up to 2009. Furthermore, in the past year two more studies have been released which looked at cancer incidence rates up to 2008 in the Nordic countries (10) and the US (11) respectively. All of these papers show no increase in brain cancer. Thus the IARC statement dismissing the flat brain cancer incidence rate statistics because they: “examined trends until the early 2000s only” is simply not accurate. The fact that brain cancer rates are unchanged to date continues to be the simplest and strongest indicator that cell phones do not cause cancer.

Fig 1: Incidence Rates of Glioma in Sweden 1970 – 2009 (from 9)

Epidemiologic Studies

In making its assessment, IARC relied primarily on two epidemiological studies including Interphone (12), and the Swedish group of Hardell et al (13). It is important to examine these studies in greater detail in order to assess the strengths and weaknesses of the evidence on which IARC based its assessment. Some new studies also shed important new light on this evidence.

There are two basic kinds of such studies: case control and cohort studies. In the case control studies on cell phones, subjects diagnosed with brain cancer respond to a questionnaire about their past history of cell phone use. Their responses are compared with those of controls who do not have brain cancer. The results are tabulated to produce an OR or “odds ratio” of developing brain cancer as a function of past use. An OR of 1 means no risk. An OR of 2 means a 2X risk. Case control studies suffer from a number of limitations, of which the most serious is “recall bias”. The results are entirely dependent on the accuracy of memory of brain cancer patients and control subjects regarding their past cell phone use.

The occurrence of recall bias, in which brain cancer patients overestimate past cell phone use in comparison to controls, has been documented in a recent study (14). “In conclusion, there was little evidence for differential recall errors overall or in recent time periods. However, apparent overestimation by cases in more distant time periods could cause positive bias in estimates of disease risk associated with mobile phone use.”

Cohort studies are another type of epidemiologic study. Cohort studies follow a population of cell phone users over a period of time and monitor how many develop brain cancer. Cohort studies are generally more robust than case control studies because exposure is more reliably assessed, using records rather than individual memory. A series of “retrospective” cohort studies were done in Denmark (15, 16). The Danish studies are “retrospective” because they used existing cell phone records to establish the number of years subjects had been using their phones. Accessing government medical records, they determined which cell phone users developed brain cancer. From this they were able to assess if the risk of developing brain cancer increased with the number of years of cell phone use. These studies have the advantage of not depending on memory. They are limited however by the uncertainty of whether the subscriber was the actual user, and the necessary elimination of business phones (the names of individual business users were not available). The initial studies (15) involved 420,095 subjects, some of whom had used cell phones for up to 21 years. No cancer risk was found.

The Danish group released another study (16) in Oct 2011 after the IARC assessment (1 ,6). In this follow up study the number of person years of long term users was much higher than any previous study. “We followed up the mobile phone subscriber study to 2007, with a focus on tumours of the central nervous system. Longer follow-up increased the numbers of person years for subscribers, particularly in long term subscribers (≥10 years), in whom the total of person years under risk increased from 170,000 to 1.2 million”. Their conclusion for long term users was: “When restricted to individuals with the longest mobile phone use—that is, ≥13 years of subscription—the incidence rate ratio was 1.03 (95% confidence interval 0.83 to 1.27) in men and 0.91 (0.41 to 2.04) in women

Because this series of studies was based on hard data, it is generally considered quite solid. Surprisingly, IARC dismissed these studies with the statement that: “In this study, reliance on subscription to a mobile phone provider, as a surrogate for mobile phone use, could have resulted in considerable misclassification in exposure assessment”. The NCI (4) took the opposite view of IARC by saying this about these studies: “Furthermore, a large population-based cohort study in Denmark has found no evidence of increased risk of brain tumors”.

Hardell et al. Studies

In making its assessment, IARC (1 & 6) relied primarily on case control studies including Interphone (12), and the Swedish group of Hardell et al (13). “A Swedish research group did a pooled analysis of two very similar studies of associations between mobile and cordless phone use and glioma, acoustic neuroma, and meningioma”. Most epidemiologists consider that Hardell is an outlier in the scientific literature (20).
The Hardell study (13) which IARC used in its risk assessment appears to be a rehash of a pair of similar papers published in 2006 (17 & 18). It is curious the Hardell timed the release of this “new” study just before the IARC meeting. In other words, Hardell repackaged a couple of old papers based on data collected in 2003, originally published in 2006, and then republished in 2011. At the time of the original publication, Hardell’s “pooled analysis” studies generated something of a media sensation. A number of expert groups felt compelled to comment on and critique these studies.
The FDA issued the following statement: “The FDA received numerous media inquires about a recently published paper (Pooled analysis of two case–control studies on use of cellular and cordless telephones and the risk for malignant brain tumors diagnosed in 1997–2003by Hardell et al.)…. The results reported by Hardell et al. are not in agreement with results obtained in other long term studies. Also, the use of mailed questionnaire for exposure assessment and lack of adjustments for possible confounding factors makes the Hardell et al. study design significantly different from other studies. These facts along with the lack of an established mechanism of action and absence of supporting animal data make it difficult to interpret Hardell et al. findings.
The same studies were also criticized by the EMF-NET which was established by the Institute for Health and Consumer Protect of the European Commission (19). Their report states that: “The results of the pooled study by Hardell et al. differ from most previously published studies, including the recently published studies with large numbers of long-term users”. The SCENIHR also criticized these studies (3, P 17): “However, both studies are non-informative because of inappropriate exclusion criteria and combination of studies”.

A 2009 paper by the Ahlbom et al on behalf of ICNIRP (International Commission for Non-Ionizing Radiation Protection) Standing Committee on Epidemiology (20) did a systematic review of the epidemiologic studies of cell phones and cancer. This paper had this to say about Hardell’s studies: “The studies by Hardell et al. are particularly problematic because of variation across their publications in the exact constitution of case groups, criteria for exclusion, exposure definitions, and the selection of results for presentation in the multiple overlapping publications….all studies found risk estimates close to or below unity, except the 2  most recent studies by Hardell et al., where up to 4-fold risk increases were reported. It is interesting to note that at the meeting where the IARC made its assessment, Alhbom was removed because he served as a board member of his brother’s telecommunications company. But Hardell remained and was permitted to vote despite the fact that he has testified as an “expert witness” in litigation by brain cancer patients against cell phone companies and has probably earned fees doing so. In any case, he got to vote on the merits of his own paper.

Despite all these criticisms, IARC used Hardell’s rehashed paper as primary evidence in its assessment. “Although both the INTERPHONE study and the Swedish pooled analysis are susceptible to bias—due to recall error and selection for participation—the Working Group concluded that the findings could not be dismissed as reflecting bias alone, and that a causal interpretation between mobile phone RF-EMF exposure and glioma is possible”. IARC could not dismiss Hardell’s results on the basis of recall bias even though studies such as (14) have shown that this is a likely source of error.

Over the years Hardell et al. have published more epidemiologic studies on cell phones than any other researcher. Virtually all his studies have used the same methodology of mailed questionnaires to establish the level of cell phone use. He has presented his findings in many different ways including by type of telephone (analog and digital cell phones as well as cordless phones), brain cancer type, cumulative hours of use, and number of years of use. Virtually every one of his studies has concluded that cell phones (and wireless and cordless phones) increase the risk for brain cancer.

Testing Hardell

If Hardell is correct in his finding that cell phone use causes brain cancer, this should cause an increase in brain cancer incidence rates. The fact that these rates have remained flat over the 25 years since cell phones were introduced and the roughly 15 years since use became widespread and common in the population can now be used to test some of his findings. In the paper cited by IARC (13), the Hardell finding that cancer risk increases with the number of years is easy to test. Table II (reproduced below) of Hardell’s paper (13) shows cancer risk increasing over the three latency periods listed. The highest risk is for >10-year latency with an OR of 2.1, 2.5 and 1.6 for wireless phones, mobile phones and cordless phones respectively.

Table II  From (13): Pooled analysis of case-control studies on malignant brain tumours and the use of mobile phone; Hardell et al.;  Intl Journal Oncology 2011 38: 1465-1474

Fig 1 of Swerdlow’s (9) Supplementary Data that I referred to earlier shows that brain cancer incidence rates in Sweden were essentially flat up to 2009. Fig 2 below shows the history of cell phone subscriptions in Sweden from 1987 — 2010. This chart shows that in 1999 (10 years before 2009), approximately 50% of the population of Sweden had a cell phone. Using Hardell’s finding that >10 year mobile cell phone users have a 2.5 OR, simple arithmetic predicts that brain cancer incidence should increase by 75%:
(50% pop nonuser x OR 1) + (50% pop cell phone users x OR 2.5) = Average OR 1.75.
But there is no change. It is interesting to note that Hardell also conducted his studies in Sweden – the same country as Swerdlow’s data. Hardell’s finding on users >10 years is flatly contradicted.

Hardell’s case control studies as INTERPHONE were entirely dependent on the memory of people suffering from brain cancer. The IARC report states that Hardell used “Self-administered mailed questionnaires were followed by telephone interviews to obtain information on the exposures and covariates of interest.” As we have noted “recall bias” (14) is a major potential source of error in this these studies. The hard data from the Swedish cancer incidence rates and cell phone subscriptions is reliable and it flatly contradicts Hardell.

Fig 2: Cell phone subscriptions in Sweden 1987 – 2010 (from 9)

Since the publication of the paper by Swerdlow et al. (9) other major studies have been published that used brain cancer incidence data to explicitly test the findings of Hardell’s case control studies. A study (10) by Deltour et al. did an extensive analysis of brain cancer incidence rates against cell phone subscription history using data from the Nordic countries. It tested various values of OR and latency, including those Hardell claims to have found. The conclusion states: “Several of the risk increases seen in case-control studies appear to be incompatible with the observed lack of incidence rate increase in middle-aged men. This suggests longer induction periods than currently investigated, lower risks than reported from some case-control studies, or the absence of any association”.

Another rigorous study which tested Hardell’s results was authored by scientists from the National Cancer Institute Little et al. (11). “In this study, we compared the observed patterns for glioma incidence trends in the US in 1992-2008 with projected incidence rates for the same period based on relative risks reported by the two epidemiological studies forming the basis of the IARC Working Group classification”. Their conclusion states: “Raised risks of glioma with mobile phone use, as reported by one (Swedish) study forming the basis of the IARC’s re-evaluation of mobile phone exposure, are not consistent with observed incidence trends in US population data, although the US data could be consistent with the modest excess risks in the Interphone study”. Their results are best illustrated with one of the graphs from the study (Fig 3). From this it is clear that Hardell’s findings (dashed red line) diverge sharply from the actual cancer incidence trend. It is also clear why they could draw no conclusion concerning the Interphone results, since modest risks implied by some of the Interphone results produced a flat cancer incidence curve.

Fig 3: NCI Brain cancer incidence rate projections vs. observed (From 11)

Interphone Study

It is important to remember that even though the Interphone study (12) found a modest increase in cancer risk in a subset of the study, the overall conclusion was: “Overall, no increase in risk of glioma or meningioma was observed with use of mobile phones. There were suggestions of an increased risk of glioma at the highest exposure levels, but biases and error prevent a causal interpretation. The possible effects of long-term heavy use of mobile phones require further investigation”. It is interesting to note that while the authors of Interphone understood that “biases and error prevent a causal interpretation”, IARC none the less cited Interphone to support its “possibly carcinogenic” ruling.

The only part of Interphone with a significant “positive” result was an OR of 1.4 for glioma in the 10th decile of call time (> 1640 cumulative hours of use). However this result was contradicted by the results for cumulative number of calls where the OR was less than 1 even for the highest decile. A close look at the “positive” call time data shows another glaring inconsistency. Among users with the highest call time the OR is 3.77 for people with 1 – 4 years of use, OR of 1.28 for 5 – 9 years, and OR of 1.34 for > 10 years of use. There is no logical progression with time — no dose response relationship. These inconsistent results are most likely due to recall bias and other errors. As a result, the authors of Interphone refrained from drawing any conclusion. For a more complete analysis of the Interphone study see (* 21).

Other Items in IARC’s report

The IARC report in the Lancet (6) included a reference to a study (22) by Cardis et al. which attempts to correlate the physical location of gliomas in the brain with the area of maximum radiation from the cell phone. IARC makes the following statement in connection with this study: “Associations between glioma and cumulative specific energy absorbed at the tumour location were examined in a subset of 553 cases that had estimated RF doses. The OR for glioma increased with increasing RF dose for exposures 7 years or more before diagnosis, whereas there was no association with estimated dose for exposures less than 7 years before diagnosis”. In other words, this study claims to have found a higher risk of glioma in parts of the brain receiving the highest energy.

It is interesting to note that the original IARC press release (1), contained a reference to another paper by Larjavaara et al (23) that examined the same issue. The Larjavaara paper (23) is listed in note 4d of the press release. However, this study reached the opposite conclusion of the Cardis paper (22). The Larjavaara abstract included the following conclusion: “These results do not suggest that gliomas in mobile phone users are preferentially located in the parts of the brain with the highest radio-frequency fields from mobile phones“. At the time, I found it curious that the IARC would reference a study that seems to contradict their precautionary 2B categorization of cell phone EMF. But what is most surprising is that in its report one month later in the Lancet (6), IARC had dropped any reference to this paper!! Again it is worth noting that Elizabeth Cardis also served on the IARC committee and voted to have her own paper count as evidence for the “possible carcinogen” ruling.


From all this it would appear that the decision making process in which IARC classed cell phones under Category 2B as “possibly carcinogenic” was flawed. IARC was ill informed on the true state of studies of brain cancer incidence rates, claiming that (6) “most of the analyses examined trends until the early 2000s only“. It ignored numerous warnings by expert groups who had determined that the case control studies by Hardell et al. were outliers in the literature. It accepted a reincarnation of an old study (13, 17, 18) that had been roundly criticized. It undervalued studies which established the existence of recall bias (14) with the statement that: “the Working Group concluded that the findings could not be dismissed as reflecting bias alone, and that a causal interpretation between mobile phone RF-EMF exposure and glioma is possible”. IARC undervalued the solid cohort studies (15, 16) by Schutz et al. of Denmark. “In this study, reliance on subscription to a mobile phone provider, as a surrogate for mobile phone use, could have resulted in considerable misclassification in exposure assessment

Since the IARC published its report the evidence against its conclusion has grown stronger. Three new studies (9, 10, 11) of cancer incidence rates have shown that rates have remained flat up to at least 2009. The findings of Hardell et al. have been put to the test in these same three studies. A large portion of Hardell’s results (13) have been shown to be wrong. This undermines one of the main pillars for IARC’s finding. The authors of Interphone (12), the other main pillar have stated that biases and error prevent a causal interpretation of their results. Finally, IARC appears to have been disingenuous in removing reference to a study (23) that contradicted one of the papers cited in their final report in the Lancet.

As the WHO stated in their updated Fact Sheet of June 2011 (5): “A large number of studies have been performed over the last two decades to assess whether mobile phones pose a potential health risk. To date, no adverse health effects have been established as being caused by mobile phone use”. While it is still cannot be ruled out that brain cancer incidence rates might increase in the future — say if latency is 20 years or more – this appears to be increasingly improbable. It is unfortunate that the IARC decision making process was flawed. But it is unlikely that IARC will ever change its Category 2B classification. Like coffee and pickled vegetables, cell phones may well remain in this category forever. IARC lists a grand total of one substance under category 4 which is “The agent is probably not carcinogenic to humans”. Alarmist groups will continue to exploit and misrepresent this flawed ruling.
See our website for more science based information on the issue of EMF and health (


  1. Open article: IARC Press Release May 31, 2011: Return to text
  2. Open Article: IARC Document on Carcinogens: Return to text
  3. Open document: European Commission Scientific Committee on Emerging and Newly Identified Health Risks (SCEHINR): Report Health Effects of Exposure to EMF Jan 2009: Return to text
  4. Open article: NCI Statement: International Agency for Research on Cancer Classification of Cell Phones as “Possible Carcinogen” May 31, 2011: Return to text
  5. Open article: WHO Fact Sheet: Electromagnetic fields and public health: mobile phones June 2011: Return to text
  6. Open article: IARC follow up report July 1, 2011 Lancet Oncology and online: Return to text
  7. Open Access:  Brain cancer incidence trends in relation to cellular telephone use in the United States; Inskip et al.; Neur-Oncology July 2010: Return to text
  8. Time Trends (1998-2007) in Brain Cancer Incidence Rates in Relation to Mobile Phone Use in England; Vocht et al.; Bioelectromagnetics July 2011, first published online Jan 28, 2011: Return to text
  9. Open access: Mobile Phones, Brain Tumors and the Interphone Study: Where Are We Now?; Swerdlow et al.; Environ Health Perspect; published online July 2011: Return to text
  10. Mobile Phone Use and Incidence of Glioma in the Nordic Countries 1979–2008; Deltour et al.; Epidemiology Vol 23 No 2 Mar 2012: Return to text
  11. Open Access: Mobile phone use and glioma risk: comparison of epidemiological study results with incidence trends in the United States; Little et al.; BMJ 2012; 344: Return to text
  12. Interphone study; Cardis et al.; International Journal of Epidemiology May 2010; 39; 675–694 Return to text
  13. Pooled analysis of case-control studies on malignant brain tumours and the use of mobile phone; Hardell et al.; Intl Journal Oncology 2011 38: 1465-1474 Return to text
  14. Recall bias in the assessment of exposure to mobile phones; Vrijheid M J et al.; Expo Sci Environ Epidemiol 2009; 19: 369-81: Return to text
  15. Open access: Danish cohort studies: Cellular Telephone Use and Cancer Risk: Update of a Nationwide Danish Cohort; Schuz et al.; JCNI J Natl Cancer Inst 2006 Vol 98 Issue 23: Return to text
  16. Open access: Use of mobile phones and risk of brain tumours: update of Danish cohort study; Schuz et al.; BMJ Oct 2011; 343: Return to text
  17. Pooled analysis of two case–control studies on use of cellular and cordless telephones and the risk for malignant brain tumours diagnosed in 1997–2003; Hardell et al.; Int Arch Occup Environ Health 2006 79: 630–639 Return to text
  18. Pooled analysis of two case-control studies on the use of cellular and cordless telephones and the risk of benign brain tumours diagnosed during 1997-2003; Hardell et al.; Intl Journal Oncology 2006 28: 509-518 Return to text
  19. EMF-NET: Comments on the study by Hardell et al.: Pooled analysis of two case-control studies: Return to text
  20. Epidemiologic Evidence on Mobile Phones and Tumor Risk: A Review Ahlbom et al. (Epidemiology 2009; 20: 639–652; Return to text
  21. Commentary on the Interphone Study: Return to text
  22. Risk of brain tumours in relation to estimated RF dose from mobile phones – results from five Interphone countries;  Cardis et al.; Occup Environ Med 2011; 68: 631-640; Return to text
  23. Location of gliomas in relation to mobile telephone use: a case-case and case-specular analysis; Larjavaara et al.; Am J Epidemiol 2011 Jul 1; 174(1): 2-11; Return to text

Posted in: Cancer, Public Health, Science and Medicine

Leave a Comment (21) ↓

21 thoughts on “Are Cell Phones a Possible Carcinogen? An Update on the IARC Report

  1. chaos4zap says:

    Great article. I have a question that maybe someone can elaborate on. It seems pretty clear that the widespread use of cell-phones has not demonstrated an increase in brain cancers, but it seems like more information is needed for any long-term conclusions. This is the first generation that will see children starting to use cell phones at a very early age and they will likely continue to do so throughout their lives. It’s seems to me that cell phones have not been in widespread use long enough to account for the population that begins using cell phones from an early age to their 30’s, 40’s and 50’s. I know that no one is making the claim that the case is closed on cell phones, I just wanted to make that point. The fact that children are more likely to use cell phones, and more often than the adults who’s plans they are on, could also factor into the use to cell phone records and applying them to adults (as was mentioned in your article). Are there any studies in progress to track young children throughout their lives for long-term trending?

  2. cervantes says:

    Arrggh. You’ve done it again. An odds ratio of 2 does not mean a 2X risk. If I have a 80% risk my odds are 4:1 (80:20). If I have a 66.6% risk my odds are 2:1 (2/3:1/3). The odds ratio is 2, but the relative risk is 1.2.

    At the very least we should expect basic statistical literacy here.

  3. cervantes says:

    Okay, to be fair: When risk is very low, odds ratios and relative risk converge and become very similar. In a case control study, unless you know the population risk, you can’t actually compute a relative risk from the odds ratio. But still, technically, I’m just sayin’, they aren’t the same thing.

  4. Lorne Trottier says:


    My article was reviewed by an epidemiologist. I suggest that you recheck the definition of odds ratio: My interpretation and calculations are correct.

    Lorne Trottier

  5. cervantes says:

    No it isn’t. The odds ratio is not the same as the relative risk. Your interpretation and calculations are not correct. My original comment is correct. Just read it.

  6. cervantes says:

    And here’s what it says at the link you gave:

    The odds ratio is a function of the cell probabilities, and conversely, the cell probabilities can be recovered given knowledge of the odds ratio and the marginal probabilities P(X = 1) = p11 + p10 and P(Y = 1) = p11 + p01. If the odds ratio R differs from 1, then

    p_{11} = \frac{1 + (p_{1\cdot}+p_{\cdot 1})(R-1) – S}{2(R-1)}

  7. Carpus says:

    Cervantes is right about this, technically. When dealing with a rare disease (which brain cancer is), the odds ratio approaches the corresponding risk ratio that one would get in a corresponding cohort study. When not rare, the odds ratio overestimates the corresponding risk ratio. But they are not actually the same, and it is technically incorrect to say that the ‘risk’ is two times greater in a case-control study. However, most people would not spend too much time worrying about this – even most epidemiologists would probably talk about ‘risk’ rather than ‘odds’, even when talking about c-c study results. Technically wrong, but …

  8. BillyJoe says:

    Forget odds ratios and relative risk. I want to know the absolute risk. If you double your risk of brain tumour by using a mobile phone daily for twenty years but the absolute risk is only one in a million, I’ll continue to use my mobile phone.

  9. KenFoster says:

    The term “relative risk” occurs only once in Trottier’s important essay, in a quote by an eminent epidemiologist.

    I would prefer to focus discussion on his main point, that the IARC “possible carcinogen” (2B) designation for cell phone radiation was strongly influenced by Hardell’s studies, which are clearly inconsistent with a recent analysis of incidence trends of brain cancer by Deltour et al. I agree with his conclusion that it is time to revisit 2B.

    For what it is worth, there is a lot of discussion in the literature about relative risk vs odds ratio as a measure of effect size. While these measures are defined differently, the practical consequence of interpreting odds ratios as if they were relative risks can be moot — and completely nonexistent if the disease in question is rare.

    I quote from the paper by Davies, “When can odds ratios mislead?” (BMJ. 1998 March 28; 316(7136): 989–991):

    “Odds ratios may be non-intuitive in interpretation, but in almost all realistic cases interpreting them as though they were relative risks is unlikely to change any qualitative assessment of the study findings. The odds ratio will always overstate the case when interpreted as a relative risk, and the degree of overstatement will increase as both the initial risk increases and the size of any treatment effect increases. However, there is no point at which the degree of overstatement is likely to lead to qualitatively different judgments about the study. Substantial discrepancies between the odds ratio and the relative risk are seen only when the effect sizes are large and the initial risk is high.”

    Brain cancer (the topic of Trottier’s essay) is rare, and the two numbers are the same at the level of precision of the quoted ORs.

    The more important question is how valid were Hardell’s studies that reported odds ratios above 1 for brain cancer as related to use of mobile phones — and the answer at present seems to be — not sufficiently to detect reliably an effect of the rather small size that he reported. At least his reports are inconsistent with current incidence data. In the future we will all be dead, but at present it seems that the risks of our being carried away by brain cancer are not increasing, despite the fact that nearly everybody uses cellphones.

    People who text and drive have much bigger risks to worry about.

    Kenneth R. Foster
    Prof. of Bioengineering
    University of Pennsylvania

  10. David Gorski says:

    OK. I attended the cell phones/cancer session, and it was as bad as I feared. First off, it was all epidemiologists, and all they looked at was epidemiological data. The session was formatted as an introductory talk to give the background, an argument for a connection between cell phone radiation and cancer, and finally a talk arguing that cell phones do not cause cancer. All of this was followed by a Q&A. I may blog about this for next week or some other time (or maybe I’ll write about it for my not-so-super-secret other blog). In the meantime, I’ll give the CliffsNotes version.

    Basically, this session was about as great an illustration as I can think of of the shortcomings of evidence-based medicine compared to science-based medicine. All three speakers were epidemiologists. The speaker introducing the topic barely mentioned just how utterly implausible on basic physics grounds it is to assert that RF radiation can cause molecular changes that could result in cancer, and the speaker supporting the link didn’t mention it at all. (She did, however, use a bunch of dubious post hoc subgroup analyses of the Interphone study to argue that there is a link. Even the speaker who argued against a link concentrated mainly on the epidemiology. He did mention that on basic science grounds such a link is implausible, as well as the fact that, taken as a whole, animal studies don’t support a link, but he spent 99% of his talk discussing only epidemiology. It’s like discussing homeopathy based only on clinical evidence and completely ignoring that for homeopathy to be true vast swaths of modern physics would have to be overthrown. Oh, and the pro-cell phone/cancer link person pulled out the hoary old gambit of “just because science doesnt know the mechanism doesn’t mean it doesn’t happen” in order to it dismiss basic science evidence, comparing the state of the data to that of smoking in the 1960s and 1970s, which was before the mechanisms by which tobacco smoke causes lung cancer began to be dissected.

    The Q&A session was somewhat interesting. One researcher who was apparently in charge of the IARC subcommittee analyzing the animal data (whose name I didn’t catch, unfortunately) commented. He pointed out that it was mainly the IARC rules that forced the committee to conclude that cell phone radiation is a “possible carcinogen.” Basically, he pointed out that, of all the chemicals and potential carcinogens that the IARC has ever evaluated, only one has ever been declared to be not carcinogenic. That should tell you all you need to know about the process. The Hardell studies were enough to force the committee to assign cell phone radiation to category 2B. Unfortunately, the media spun it as a validation by the IARC that people should be worried about cell phone radiation. I wanted to get up and ask a question, but quite frankly what I wanted to ask had pretty much been asked and all I could think of to do would have been to ask them why they concentrated only on epidemiological data or to castigate the one who made the comparison to smoking and pulled out the “just because there’s no mechanism doesn’t mean something isn’t going on” whine.

  11. icarus4me says:

    Came in to nitpick about odds ratios, stayed to comment on the poor response to constructive criticism.

  12. Lorne Trottier says:

    I did blur the distinction between odds ratio and risk ratio. But as Ken Foster and Billy Joel point out, in this case they are practically the same. Thank you all for your comments on this point.

    Dave’s points about a lack of any plausible physical mechanism, and the negative animal studies are also very strong arguments against any cell phone cancer link. The goal of my article was to point out that the already weak epidemiological evidence on which IARC based its classification, has become much weaker over the past year.

    Lorne Trottier

  13. KenFoster says:

    IARC decision rules include both animal and epi studies, and as far as I can tell the IARC committee’s decision about cell phone radiation (2B – “possible carcinogen”) was consistent with its rigid set of decision rules.

    The epi data on the whole are negative, but there are small increases in odds ratio in the Interphone studies for long-term use (but of course the studies were never designed to be able to reliably detect small increased in risk after long term use, since they employed questionnaires to assess past use of mobile phones – who remembers?) And of course there are the larger odds ratios reported by Hardell, as well as his strong advocacy for a cancer connection as a member of the IARC panel.

    The thing that is missing from the discussion is the failure, after a half century of research, to find a plausible mechanism by which RF fields at cellphone frequencies and ordinary exposure levels can produce any change in biological systems except through heating. Bob Adair and before that Herman Schwan have pointed this out very forcefully. Epidemiologists are generally reluctant to dismiss causality on the basis of biophysical arguments, but there is something to be said for factoring such arguments into assessment of causation when faced with weak epi results.

    The basic problem is that the IARC classifications indicate level of suspicion, and there is a big difference between raising suspicions and drawing substantive conclusions that something does, in fact, cause cancer. And proving that something does not cause cancer is mission impossible.

    A second problem is that the scientific literature related to health effects of RF energy is a real mess, with thousands of studies of all description that point in all directions and which vary greatly in quality. There are very high quality and very expensive studies, and lots of fishing expeditions of variable quality reporting small effects close to the level of statistical significance. Anybody inclined to worry can pick and choose data to argue that a real health problem exists, while at the same time health agencies conduct massive reviews of the literature and fail to find clear evidence of a hazard.

    I got into this field in 1971 when I was a new PhD and assigned to study health effects of microwaves for the Navy, in part because of public concerns about possible health hazards of military radar and communications installations. Very similar questions were being raised then as are being raised now, except that many more technologies now exist that use RF energy. More research is needed.

    Kenneth R. Foster
    University of Pennsylvania

  14. Lorne Trottier says:

    Ken points out that there are “thousands of studies of all description that point in all directions and which vary greatly in quality”. But “after a half century of research” and “thousands of studies”, the only effect that has ever been demonstrated is heating. To the best of my knowledge, not one of of the “small effects” other than heating has ever been reproduced in high quality studies. One would think that any real effect would have been detected and confirmed by now.

    Lorne Trottier

  15. KenFoster says:

    “One would think that any real effect would have been detected and confirmed by now” I agree completely. For that reason (among others) I am skeptical that any serious, previously unknown, health effect will be uncovered in the future. But nobody can predict what might be discovered in the future.

    I have spent a lot of time in my career wondering why there has been so much confusion in this literature. To a large extent it is a consequence of too many exploratory studies in search for small effects (as opposed to hypothesis-driven studies designed to follow up previous findings, or standardized assays). Science is not very reliable when it comes to interpreting the results of fishing expeditions that find tiny effects that are just above the level of statistical significance. The bioeffects literature is filled with such findings.

    Also, the subject of biological effects of weak electromagnetic fields seems to generate strong feelings in many people, and perhaps scientists who strongly believe that exposures at low levels will produce effects might be less skeptical of their findings of such effects than otherwise. There is a large body of evidence that says that scientists tend to find in their research what they expected to find. The standard ways to counter investigator bias – double blinded study design, etc – have been seldom used in the bioeffects field, and they are imperfect remedies in any event.

    Also, there is a strong tendency towards confirmation bias in this field — paying a lot of attention to a recent positive finding but ignoring other negative results that cast doubt on the more sensational interpretations of the positive results.

    No doubt these factors are universal in science. In any event they call for some caution in interpreting science, even in science-based medicine.

    – Kenneth R. Foster
    University of Pennsylvania

  16. stanmrak says:

    Why has there been so much confusion in this literature? I’ll tell you why. The telecommunications industry knows that they don’t have to win this argument; they can get by just by creating doubt and dragging the whole process out. Paying for bad studies is one way. By the time conclusive evidence is found, we’ll have our evidence in the statistics. It’s called ‘tobacco’ science.

  17. BillyJoe says:

    stanmrak: “It’s called ‘tobacco’ science.”

    What you need is evidence that this is true.
    But all you have is a sort of Galileo fallacy.

Comments are closed.