Podcast #128: We chat with Kent C Dodds about why he loves React and discuss what life was like in the dark days before Git. Listen now.
2 typos and other small fixes
source | link

While the Swedish study brings new data, it is inconsistent with other recent studies and suffers from similar flaws to many other studies reporting positive results

The Swedish study referred to tinin the question (full text available here as a pdf) is typical of many studies on the effect of mobile phone use. It uses case-control retrospective studies starting with the population of people observed to have gliomas in a particular area. It uses sellsself-administered questionnaires on groups of diagnosed patients and on a "matching" group of non-sufferers selected to be demographically similar.

There are at least two key issues that affect the reliability of this sort of study. One is that the matching does not adequately eliminate some special characteristic of the group suffering cancer. theThe other is that questionnaires, especially when asking about behaviour a long time in the past, are unreliable and subject to recall bias especially in those patients seeking to blame something other than random chance for their condition. It is extremely hard to eliminate such bias and just as hard to estimate how big the bias might be. For relatively rare conditions the error bars are already large and the possibility of bias is also large.

So it is worth finding ways to test the resulting statistics for credibility in ways that avoid the particular potential sources of error. For example, what would we expect to see in a different population if the estimated effects were applied there? This has been done in a recent BMJ paper.

The BMJ paper built models of the expected rates of gliomas in the population of the USA using estimated effects based on the swedishSwedish study and the Interphone study (which reported much lower risks than the swedishSwedish study). They then compared the estimated incidence of gliomas to those predicted by various versions of the models (allowing for a variety of possible latency periods and so on.) Note, critically, there is no obvious trend in the population incidence of gliomas since before mobile phones became popular.

They conclude (my emphasis):

Age specific incidence rates of glioma remained generally constant in 1992-2008 (−0.02% change per year, 95% confidence interval −0.28% to 0.25%), a period coinciding with a substantial increase in mobile phone use from close to 0% to almost 100% of the US population. If phone use was associated with glioma risk, we expected glioma incidence rates to be higher than those observed, even with a latency period of 10 years and low relative risks (1.5). Based on relative risks of glioma by tumour latency and cumulative hours of phone use in the Swedish study, predicted rates should have been at least 40% higher than observed rates in 2008. However, predicted glioma rates based on the small proportion of highly exposed people in the Interphone study could be consistent with the observed data. Results remained valid if we used either non-regular users or low users of mobile phones as the baseline category, and if we constrained relative risks to be more than 1.

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.

The meaning of their comment on the Interphone results is that the scale of the effect seen there is not large enough to be clearly seen in the whole US population. So even if it has avoided all the biases possible in these studies, the effect is so small as to be hard to confirm and probably not large enough to worry about.

So the overall conclusion is that while some studies produce estiatesestimates of significant (though small) effects, these are not consistent with the flat incidence of gliomas in large populations where mobile phone use has rocketedskyrocketed.

While the Swedish study brings new data, it is inconsistent with other recent studies and suffers from similar flaws to many other studies reporting positive results

The Swedish study referred to tin the question (full text available here as a pdf) is typical of many studies on the effect of mobile phone use. It uses case-control retrospective studies starting with the population of people observed to have gliomas in a particular area. It uses sells-administered questionnaires on groups of diagnosed patients and on a "matching" group of non-sufferers selected to be demographically similar.

There are at least two key issues that affect the reliability of this sort of study. One is that the matching does not adequately eliminate some special characteristic of the group suffering cancer. the other is that questionnaires, especially when asking about behaviour a long time in the past, are unreliable and subject to recall bias especially in those patients seeking to blame something other than random chance for their condition. It is extremely hard to eliminate such bias and just as hard to estimate how big the bias might be. For relatively rare conditions the error bars are already large and the possibility of bias is also large.

So it is worth finding ways to test the resulting statistics for credibility in ways that avoid the particular potential sources of error. For example, what would we expect to see in a different population if the estimated effects were applied there? This has been done in a recent BMJ paper.

The BMJ paper built models of the expected rates of gliomas in the population of the USA using estimated effects based on the swedish study and the Interphone study (which reported much lower risks than the swedish study). They then compared the estimated incidence of gliomas to those predicted by various versions of the models (allowing for a variety of possible latency periods and so on.) Note, critically, there is no obvious trend in the population incidence of gliomas since before mobile phones became popular.

They conclude (my emphasis):

Age specific incidence rates of glioma remained generally constant in 1992-2008 (−0.02% change per year, 95% confidence interval −0.28% to 0.25%), a period coinciding with a substantial increase in mobile phone use from close to 0% to almost 100% of the US population. If phone use was associated with glioma risk, we expected glioma incidence rates to be higher than those observed, even with a latency period of 10 years and low relative risks (1.5). Based on relative risks of glioma by tumour latency and cumulative hours of phone use in the Swedish study, predicted rates should have been at least 40% higher than observed rates in 2008. However, predicted glioma rates based on the small proportion of highly exposed people in the Interphone study could be consistent with the observed data. Results remained valid if we used either non-regular users or low users of mobile phones as the baseline category, and if we constrained relative risks to be more than 1.

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.

The meaning of their comment on the Interphone results is that the scale of the effect seen there is not large enough to be clearly seen in the whole US population. So even if it has avoided all the biases possible in these studies, the effect is so small as to be hard to confirm and probably not large enough to worry about.

So the overall conclusion is that while some studies produce estiates of significant (though small) effects, these are not consistent with the flat incidence of gliomas in large populations where mobile phone use has rocketed.

While the Swedish study brings new data, it is inconsistent with other recent studies and suffers from similar flaws to many other studies reporting positive results

The Swedish study referred to in the question (full text available here as a pdf) is typical of many studies on the effect of mobile phone use. It uses case-control retrospective studies starting with the population of people observed to have gliomas in a particular area. It uses self-administered questionnaires on groups of diagnosed patients and on a "matching" group of non-sufferers selected to be demographically similar.

There are at least two key issues that affect the reliability of this sort of study. One is that the matching does not adequately eliminate some special characteristic of the group suffering cancer. The other is that questionnaires, especially when asking about behaviour a long time in the past, are unreliable and subject to recall bias especially in those patients seeking to blame something other than random chance for their condition. It is extremely hard to eliminate such bias and just as hard to estimate how big the bias might be. For relatively rare conditions the error bars are already large and the possibility of bias is also large.

So it is worth finding ways to test the resulting statistics for credibility in ways that avoid the particular potential sources of error. For example, what would we expect to see in a different population if the estimated effects were applied there? This has been done in a recent BMJ paper.

The BMJ paper built models of the expected rates of gliomas in the population of the USA using estimated effects based on the Swedish study and the Interphone study (which reported much lower risks than the Swedish study). They then compared the estimated incidence of gliomas to those predicted by various versions of the models (allowing for a variety of possible latency periods and so on.) Note, critically, there is no obvious trend in the population incidence of gliomas since before mobile phones became popular.

They conclude (my emphasis):

Age specific incidence rates of glioma remained generally constant in 1992-2008 (−0.02% change per year, 95% confidence interval −0.28% to 0.25%), a period coinciding with a substantial increase in mobile phone use from close to 0% to almost 100% of the US population. If phone use was associated with glioma risk, we expected glioma incidence rates to be higher than those observed, even with a latency period of 10 years and low relative risks (1.5). Based on relative risks of glioma by tumour latency and cumulative hours of phone use in the Swedish study, predicted rates should have been at least 40% higher than observed rates in 2008. However, predicted glioma rates based on the small proportion of highly exposed people in the Interphone study could be consistent with the observed data. Results remained valid if we used either non-regular users or low users of mobile phones as the baseline category, and if we constrained relative risks to be more than 1.

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.

The meaning of their comment on the Interphone results is that the scale of the effect seen there is not large enough to be clearly seen in the whole US population. So even if it has avoided all the biases possible in these studies, the effect is so small as to be hard to confirm and probably not large enough to worry about.

So the overall conclusion is that while some studies produce estimates of significant (though small) effects, these are not consistent with the flat incidence of gliomas in large populations where mobile phone use has skyrocketed.

1
source | link

While the Swedish study brings new data, it is inconsistent with other recent studies and suffers from similar flaws to many other studies reporting positive results

The Swedish study referred to tin the question (full text available here as a pdf) is typical of many studies on the effect of mobile phone use. It uses case-control retrospective studies starting with the population of people observed to have gliomas in a particular area. It uses sells-administered questionnaires on groups of diagnosed patients and on a "matching" group of non-sufferers selected to be demographically similar.

There are at least two key issues that affect the reliability of this sort of study. One is that the matching does not adequately eliminate some special characteristic of the group suffering cancer. the other is that questionnaires, especially when asking about behaviour a long time in the past, are unreliable and subject to recall bias especially in those patients seeking to blame something other than random chance for their condition. It is extremely hard to eliminate such bias and just as hard to estimate how big the bias might be. For relatively rare conditions the error bars are already large and the possibility of bias is also large.

So it is worth finding ways to test the resulting statistics for credibility in ways that avoid the particular potential sources of error. For example, what would we expect to see in a different population if the estimated effects were applied there? This has been done in a recent BMJ paper.

The BMJ paper built models of the expected rates of gliomas in the population of the USA using estimated effects based on the swedish study and the Interphone study (which reported much lower risks than the swedish study). They then compared the estimated incidence of gliomas to those predicted by various versions of the models (allowing for a variety of possible latency periods and so on.) Note, critically, there is no obvious trend in the population incidence of gliomas since before mobile phones became popular.

They conclude (my emphasis):

Age specific incidence rates of glioma remained generally constant in 1992-2008 (−0.02% change per year, 95% confidence interval −0.28% to 0.25%), a period coinciding with a substantial increase in mobile phone use from close to 0% to almost 100% of the US population. If phone use was associated with glioma risk, we expected glioma incidence rates to be higher than those observed, even with a latency period of 10 years and low relative risks (1.5). Based on relative risks of glioma by tumour latency and cumulative hours of phone use in the Swedish study, predicted rates should have been at least 40% higher than observed rates in 2008. However, predicted glioma rates based on the small proportion of highly exposed people in the Interphone study could be consistent with the observed data. Results remained valid if we used either non-regular users or low users of mobile phones as the baseline category, and if we constrained relative risks to be more than 1.

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.

The meaning of their comment on the Interphone results is that the scale of the effect seen there is not large enough to be clearly seen in the whole US population. So even if it has avoided all the biases possible in these studies, the effect is so small as to be hard to confirm and probably not large enough to worry about.

So the overall conclusion is that while some studies produce estiates of significant (though small) effects, these are not consistent with the flat incidence of gliomas in large populations where mobile phone use has rocketed.