Wednesday, 14 September 2011

Relative risk: Letters regarding risk and the Beauly–Denny power lines

Correspondence between me and Gemma from the Energy Department in the Scottish Government in connection with proposed pylons between Beauly and Denny. Previous posts on this are here, here and here. Gemma advises going to the Health Protection Agency, which advises her Department. 

From Gemma to me (January 2011):
You continue to express your particular concern about the relationship between exposure to electric and magnetic fields (EMFs) from power lines and an increased risk of childhood leukaemia.  In particular you state that the 'relative risk of childhood leukaemia in areas close to high voltage power lines can be up to 4', and you compare this number to the 'relative risks of lung cancer and heart disease resulting from exposure to secondary smoke commonly given as 1.2-1.3'.  Your assertion therefore is that to public health action taken by government in respect of environmental tobacco smoke is inconsistent with an approach that does not require undergrounding of power lines. 
Regarding the possibility of an increased risk of childhood leukaemia, in 2001 a Working Group of the World Health Organization's International Agency for Research on Cancer (IARC) reviewed health effects of static and Extremely Low Frequency EMFs (IARC, 2002).  They concluded that the pooled analysis of data from a number of well conducted studies (Ahlbom, 2000) had shown a fairly consistent statistical association between a doubling of the risk of childhood leukaemia and prolonged exposure to power frequency, residential ELF magnetic field strengths above 0.4 microtesla (T).  However, unlike for smoke, lung cancer and heart disease, there is no known biological mechanism for this association which has not yet been demonstrated to be causal. 
Based on this finding, the UK Health Protection Agency have estimated that, given the distribution of magnetic field exposures for the UK population and assuming that the association between magnetic fields and the risk of cancer was causal, of about 500 new cases of childhood leukaemia occurring in the United Kingdom each year, 2 cases might result from exposure to magnetic fields at exposures above 0.4 microtesla at home. 
In 2004 the Health Protection Agency reviewed their advice and introduced a note of precaution because of the persistence of the finding of an association between power frequency electric and magnetic field exposure and childhood leukaemia:
“The government should consider the need for further precautionary measures in respect of exposure of people to EMFs. In doing so it should note that the overall evidence for adverse effects of EMFs on health at levels of exposure normally experienced by the general public is weak. The least weak evidence is for the exposure of children to power frequency magnetic fields and childhood leukaemia.” 
A subsequent study by Draper et al (2005) on 'Childhood cancer in relation to distance from high voltage power lines in England and Wales: published in the British Journal of Medicine, considered the risk of developing childhood cancer in relation to residential proximity to the power lines rather than electric or magnetic field strengths.  The study involved 29,081 cases of cancer in children aged 0-14 years who were diagnosed during 1962-95. For each case, a matched control was selected. For each case and control, the distance of residence at the time of birth to the nearest high-voltage overhead power line was calculated.  The study concerned all 275 and 400 kV overhead lines in the UK and a small fraction of 132 kV lines.  Draper et al found that, compared with children born more than 600 m from a line, the risk of leukaemia was greater by a factor of 1.69 for those born within 200 m and by a factor of 1.23 for those born between 200 and 600 m; both these elevations were statistically significant*. 
However, with regard to this apparent increase in the relative risk, Draper et al reported that the increased risk seems to extend to at least 200 m, beyond the distance at which there would be elevated magnetic fields from the local high voltage power line. They commented: 
“Thus our results do not seem to be compatible with the existing data on the relation between magnetic fields and risk”.
This would suggest, therefore, that at least some of the increased leukaemia risk may be associated with factors other than EMFs, and the authors did not regard their findings as establishing a definite causal association between childhood leukaemia and exposures to EMFs from power lines.
From their results, Draper et al estimated that, of the 400-420 cases of childhood leukaemia that occur annually in England and Wales, up to five cases might be associated with living in proximity to high voltage power lines (though not necessarily caused by the electric and magnetic fields). 
This evidence, and other new studies, have not caused the Health Protection Agency to change its advice that the epidemiological evidence is not strong enough to justify the firm conclusion that electric and magnetic fields from power lines cause leukaemia in children. 
In contrast to this weak evidence of a risk and small number of potentially attributable cases, the evidence for an association between 'passive' smoking and ill health is strong and convincing, and the estimated burden of associated disease is high.  For example, Cancer Research UK note that
·        Non-smoking spouses of people who smoke at home have 27% higher risks of lung cancer.·        People exposed to second-hand smoke in the workplace have 24% higher risks of lung cancer.·        One study estimates that passive smoking may kill over 11,000 people every year in the UK from cancer, heart disease, strokes and other diseases.·        A study by the Royal College of Physicians showed that about 17,000 children in the UK are admitted to hospital every year because of illnesses caused by second-hand smoke.·        children in households where both parents smoke have a 72% higher risk of respiratory diseases.·        Childhood exposure to second-hand smoke had also been linked to a wide range of other conditions including asthma, Sudden Infant Death Syndrome (or cot death), childhood meningitis and mental disabilities.
I hope that this comparison demonstrates that the approach taken by the Scottish Government Health Directorates to the protection of public health is consistent with the best available scientific evidence and authoritative advice.
AGNIR (2001). ELF electromagnetic fields and the risk of cancer. Report of an Advisory Group on Non-ionising Radiation. Doc NRPB, 12(1), 3-179.
Ahlbom A, Day N, Feychting M, Roman E et al (2000). A pooled analysis of magnetic fields and childhood leukaemia. Br J Cancer, 83(5), 692-8.
Advice on Limiting Exposure to Electromagnetic Fields (0–300 GHz). Documents of the NRPB: Volume 15, No. 2 2004
Draper GJ, Vincent T, Kroll ME and Swanson J (2005). Childhood cancer in relation to distance from high-voltage power lines in England and Wales: a case-control study. BMJ, 330(7503), 1290.
*A statistically significant association is denoted by a confidence interval that does not span unity. For example a relative risk estimated to be 1.02 with a confidence interval of 1.01–1.04 would be considered statistically significant.  A relative risk estimate of 4.02 with a confidence interval of 0.78–34.2 would not be considered significant because of the much wider uncertainly of the estimate.

 From me to Gemma (1 September 2011):
Thank you for your email from January below. I apologise for the delay in replying. You have shown me that the relative risk of leukaemia from living within 200m of a pylon is 1.69 but the risk of lung cancer from exposure to secondary smoke (from CRUK's work) is 1.27 or 1.24. Then you state (using CRUK's evidence) that the risk from secondary smoke is well established. Cancer Research primarily funds work that supports further regulation of tobacco – a bias that is surely fundamentally unscientific. You consider the possibility of confounding factors for leukaemia near electricity pylons, but not for exposure from tobacco smoke, where the statistical probability of lung cancer is lower and the likelihood of confounding factors higher. 
You define statistical significance as an elevated risk with a confidence interval that doesn't span zero, a definition that I don't find satisfactory. How can a risk factor of 1.1 clearly show causation when the elevation of risk is so small? You also bring no evidence to bear that the confidence intervals with lung cancer incidence don't straddle zero (here's one that does straddle zero).
It seems to me that there is no evidence to suggest that there is greater risk of lung cancer to adults from secondary smoke than there is of leukaemia to children from pylons. In addition the emphasis should surely be on protecting younger lives from illness at home, when adults are not permitted to take risks with their own lives at work and in recreation by exposing themselves to secondary smoke, voluntarily or otherwise.

 From Gemma to me (14 September):
Thank you for your email of 1 September 2011. I note your concerns however our position has not changed since our last correspondence with you earlier in January. I work in the Energy Consents and as you will know  we have taken our advice from the Health Protection Agency for Scotland and the UK Health Protection Agency. Last January I contacted both of these organisations personally for an up to date position citing the reports you mentioned and when they replied I included this in my previous response to you. As this is out-with our area of expertise, I can only suggest you contact the Health Protection Agency direct if you have continuing concerns. Here is a link to their website: 
I hope this is helpful.
I think I'll take her advice.


Anonymous said...

Funny thing about a RR 1.25 or so. It always includes the 1.

Thus;since 1 is 80% of 1.25, there is a 80% probability that whatever you are considering was caused by whatever is the 1 and only a 20% probability of it being caused by something else.

This explains the insistence on a RR of at least 2 before considering the possibilty of 'causation'.

Since 1 is 50% of 2, a RR of 2 implies a 50-50 probality that something other than what you are blaming was the cause of the problem.

Gary K.

Anonymous said...

Reference Guide on Epidemiology, Michael D. Green, D. Mical Freedman & Leon Gordis$file/sciman06.pdf

Page 384
When the relative risk reaches 2.0, the agent is responsible for an equal number
of cases of disease as all other background causes.

Thus, a relative risk of 2.0 implies a 50% likelihood that an exposed individual’s disease was caused by the agent.
(NOTE: That means an RR=2 implies a 50% likelihood the disease was NOT caused by the agent!!-GK)
An RR of 1.5 implies that 66.7% of a disease was caused by other background causes and only 33.3% was caused by the agent.(1 is 2/3rds of 1.5)

Gary K.