Lloyd Morgan's Column

[Return to column index]

A Comparison Between Two Interphone Studies
Risk of Parotid Gland Tumors from Cellphone Use

Introduction

The Interphone Study is a 13-country case-control study on the risk of brain and parotid gland tumors as a result of cellphone use. In order to roll-up the results from all 13 countries, each of the individual country studies are required to use the Interphone Protocol.

To date (March 2008) there have been two Interphone studies on the risk of parotid gland tumors (a salivary gland near the ear) from cellphone use. The two studies, using the same Interphone Protocol, reported diametrically different results. How could this be? The quick answer is that one study had far more tumors (cases) than the other. The longer answer is that one study restricted their study design to the highest levels of exposure, while the other study made no such effort.

What did each of these studies find?

The most recent study, an Israeli study by Sadetzki et al. reported, "[The] analysis [was] restricted to ... conditions that may yield higher levels of exposure (e.g., heavy use in rural areas) showed consistently elevated risks. For ipsilateral^[1] use, the odds ratios in the highest category of cumulative number of calls and call time ... were 1.58 (95% confidence interval: 1.11, 2.24) and 1.49 (95% confidence interval: 1.05, 2.13), respectively." Additionally, "A positive dose-response trend was found for these measurements." In other words, this study found a significant^[2] 58% increased risk of parotid gland tumors for the total number of cellphone calls and a significant 49% increased risk for the total number of cellphone hours. Perhaps of greater importance, they found the risk increases as the number of calls or the number of hours increases.

The earlier Swedish-Danish study, by Lönn et al., led by the prestigious Karolinska Institute, reported, "For regular^[3] mobile phone use, regardless of duration, the risk estimates for malignant and benign tumors were 0.7 (95% confidence interval: 0.4, 1.3) and 0.9 (95% confidence interval: 0.5, 1.5), respectively. Similar results were found for more than 10 years' duration of mobile phone use." In other words the study, in contrast to the Israeli study, reported no risk regardless of duration of use even when the duration exceeded 10 years. As we will see, though Lönn et al. reported no risk, they did find a risk.

Let's examine the Israeli study and compare it to the Swedish-Danish study.

¹ - Ipsilateral means the tumor is on the same side of the head where the cellphone was held.
² - Significant (as well as non-significant) is short had for statistically significant.
³ - "Regular" use is defined as use of a cellphone for at least once a week for 6 months or more (with tumor diagnosis at least a year prior to the cut-off date of the study). If "regular" smoking and risk of lung cancer replaces "regular" cellphone use and risk of brain tumors and/or parotid gland tumors, would a risk of lung cancers be reported?

Comparison of Studies

There are three reasons why the Sadetzki et al. study found a risk of parotid gland tumors where the Lönn et al. study did not find a risk. First, Israeli cellphone users are exceptionally heavy users (both in call time and in the number of calls). The result was a substantially greater number of tumors in the Sadetzki et al. study compared to the Lönn et al. study. Second, the Israeli study was "restricted to ... conditions that may yield higher levels of exposure" while the Lönn et al. study made no such effort. Third, the Lönn et al. study did not present any ipsilateral data for cumulative call hours or cumulative number of calls, much less for the highest category of time and number of calls as did Sadetzki et al.

Why is "heavy" cellphone use important when investigating tumor risk from cellphone use? Heavy cellphone use increases the total exposure.

Why is it important to compare rural cellphone use to urban cellphone use? The quick answer is that the power radiating from a cellphone is higher in rural areas than in urban areas. Simply stated, rural users are farther from a cell tower (base station, mask) than urban users. As a result the power of cellphone must he higher.

Why is it important to report ipsilateral cellphone use for the highest category of cumulative use time and cumulative number of calls? Because ipsilateral use (contralateral has no exposure) combined with the highest category of cumulative use time and/or number of calls results in the highest level of exposure. The Lönn et al. study made no such effort.

Below we present Table 1. It is a side-by-side comparison of the two studies.

Findings from Table 1

1. Number of tumors
   With a much smaller eligible population, Israel, for every parameter, had 2.7-fold more tumors. This difference goes a long way towards explaining why the studies have such diametrically different results.
    
2. Eligible study populations
   The eligible population in Israel (~3.2 million) is less than half the eligible population in Sweden and Denmark (6.7 million). Given the larger population in Sweden and Denmark we would expect there would be more tumors. But, it is just the opposite.
   
   Why is this? It suggests the prevalence of cellphone users was greater and the amount of exposure was larger in Israel than in Denmark and Sweden. The Lönn et al. study could have, but did not, resolve their dearth of tumors by using a longer eligibility range (first parameter in Table 1).
    
3. Ten or more years of cellphone use
   In an examination of cellphone use for 10 years or more there were 1.8-fold more tumors in Israel. The higher 1.4-fold risk reported in the Swedish-Danish study is meaningless because the confidence level is quite low (51%), a direct result of having too few tumors.

Table 1, Parameter Comparison

⁴ - OR is Odds Ratio is the Relative Risk of cases (people with tumors) compared to controls. For example, if an OR = 2.2, there relative 2.2-fold risk for cases compared to controls.
⁵ - Bold indicates an Odds Ratio is near-significant: >90% confidence equivalent to p<0.10.
⁶ - Bold italics indicates an Odds Ratio is statistically significant, >95% confidence (p<0.05).

4. Tumor on the same side of head where cellphone was held
   When the tumor reported is on the same as where the cellphone was held (ipsilateral use), there were 1.4-fold more tumors in Israel. Interestingly, the Swedish-Danish study listed in a table within their study, a near-significant 2.6-fold risk for 10 years or more of ipsilateral cellphone use. Yet, other than listing this result in a table, the elevated risk was not reported within the text of the paper.
    
5. Cumulative number of calls
   When we look at the cumulative number of calls there were 3.0-fold more tumors in Israel compared to Sweden and Denmark Further, for the highest category of cumulative number of calls there were 2.6-fold more calls in Israel compared to Sweden and Denmark.
    
6. Cumulative number of hours
   We see in the Israeli study, for cumulative hours of use, 6.7-fold more tumors and 2.7-fold more cumulative hours.
    
7. Rural use compared to urban use
   Lastly, the Swedish-Danish study did not present any rural/urban data.

With every parameter, except two, the risk was higher in the Israeli study: regular cellphone use for >10 years (statistically meaningless), and ipsilateral use for >10 years. The latter finding was not reported in the text of the Swedish-Danish paper.

Table 2, compares all significant and near-significant risks for ipsilateral cellphone use. Why is this important? The highest level of exposure is on the side where the cellphone is held. There is no exposure on the opposite of the head (contralateral).

The Swedish-Danish study had only one near-significant finding. The equivalent risk in the Israeli study is nearly the same as the Swedish-Danish study's risk (2.53 and 2.6 respectively).

The results from Table 2 are not surprising. We know that ipsilateral use of a cellphone is the highest exposure, compared to the opposite side (contralateral) where there in no exposure. Indeed, if risk is reported without regard to the side of the head (which is the usual case in all studies), the result will be an underestimation of risk. For example, the 1.96-fold risk in Table 1 is an underestimation. In other words, the risk is greater than 1.96-fold.

Swedish-Danish Study's Reported Risk

Even with six differences between the Israeli study and the Swedish-Danish study, each strengthening the Israeli study, the Swedish-Danish study found a risk, but it was not discussed. For >10 years since first regular ipsilateral cellphone use a table reported a 2.6 fold near-significant risk (p=0.078, or 92% confidence). Why didn't the authors state, "There is a near-significant elevated risk for >10 years of ipsilateral use?"

Table 2, All Reported Ipsilateral Results: Significant and Near-Significant

How Benign are "Benign" Tumors

In both studies, the reported risk was only for "benign" tumors. Not to worry you say? After all, benign means harmless.

"Benign" tumors - are not benign!

For the researchers, patients, and other readers of science papers, "benign" creates an expectation that there is nothing to be concerned about.

Let me describe the experience of having a "benign" parotid gland tumor. First, it must be surgical removed. If not removed it continues to grow until a facial nerve is damaged. The result is partial or complete paralysis of one side of the face. Though the surgery is relative simple (one to two nights in the hospital), some surgeons will leave part of the parotid gland in place for fear of cutting a facial nerve (with the possibility of tumor re-growth). A common side effect is a loss of facial feeling. In most cases the feeling returns after a couple of years. Is this harmless (AKA "benign")?

Interestingly, if the "benign" parotid gland tumor is larger than a certain size, it is arbitrarily called malignant.

"Benign" tumors, just like "malignant" tumors, are the result of genetic mutations. These mutations cause the tumor cells to grow without limit (unless treated). It would be far better to refer to these tumors as "non-malignant", or even, "low-grade cancer." The Central Brain Tumor Registry of the United States will soon publish its 2007-2008 Primary Brain Tumors In The United States, Statistical Report. This edition compared to previous editions, changes the term "benign" to "non-malignant."^[14]

The Swedish-Danish study states, "Benign parotid gland tumors are not reported to the cancer registries in any of the countries." This was also true in the United States for "benign" brain tumors until the US Congress unanimously passed a law mandating data collection of "benign" brain tumors. Similar legislation for "benign" brain^[15], parotid gland, and facial nerve tumors is necessary in all countries, in order to learn, delayed by a latency time, if the incidence of these tumors increases as a result of cellphone use.

⁸ - Benign and malignant tumors combined.
⁹ - Only benign tumors.
¹⁰ - Bold indicates near-significant (p<0.10).
¹¹ - Since first regular use.
¹² - Anomalous finding: Ten % of near-significant results will show a protective effect due to chance.
¹³ - Bold-italics indicates statistical significance (p<0.05)
¹⁴ - For a copy of this Report go to www.cbtrus.org.
¹⁵ - Acoustic neuromas and meningiomas are "benign" brain tumors exposed to cellphone radiation.

Summation

The Israeli study, using the same Interphone Protocol as the Swedish-Danish study was designed to emphasize the subjects with the highest exposure (rural users and ipsilateral cellphone use.

Why wasn't a similar design used in the Swedish-Danish study? There are two answers to this question. First, there was no effort made to design an approach that would emphasize highest cellphone exposures (rural users and ipsilateral use). Second, there were too few tumors. If there had been a design emphasis for the highest levels of exposure, then it would have been clear that there would be too few tumors. The solution? Extend the diagnoses eligibility time in order to capture a sufficient number of tumors.

The rhetorical question must be asked. What was the reason for choosing an approach that, as reported, found no risk?

[Return to column index]