Imperial College study

An academic group (the Small Area Health Statistics Unit, SAHSU, led by Professor Paul Elliott) at Imperial College has been conducting an epidemiological study of adult cancers and power lines in England and Wales.  This can be seen as a parallel to the “Draper” or Oxford study of childhood cancers that reported findings on childhood leukaemia in 2005.

The study was published in the journal Epidemiology in January 2013. It is broadly negative.

Design of the study

The study investigated certain types of cancer that have been implicated with power lines or EMFs: leukaemia, brain tumours, malignant melanoma, and female breast cancer.  It compared incidence of these cancers to certain "control" cancers that are not implicated with EMFs.  It calculated the distance of each case to the nearest National Grid overhead line (275 kV and 400 kV in England and Wales, plus a very few 132 kV lines) that existed in the year of diagnosis (and also five years before diagnosis).  It also calculated the magnetic field exposure of each case from the distance, information about the line, and an estimate of the current carried by the line in the year of interest.

The methods have been described in detail in a previous published paper.

The results in summary

The study is broadly negative.  For some of the cancer types (most notably breast cancer), there is a suggestion of an association with proximity to the power lines in the unadjusted risks.  But any associations largely disappear after adjustment for a number of factors including deprivation.  This could mean that there is a difference in the socioeconomic status of the people who live close to power lines compared to the rest of the population (see what we already know about this), and that difference means that they contract some cancers at different rates, but there is no direct effect of the power lines or the magnetic fields they produce.

Whatever the explanation, the authors conclude:

"Our results do not support an epidemiologic association of adult cancers with proximity to residential magnetic fields from high-voltage overhead power lines."

They actually go on to say:

"Unless new biological hypotheses emerge, our findings should help to settle a long-standing debate on the safety of residential exposures to extremely low-frequency magnetic fields from high-voltage overhead power lines and adult cancers."

This last statement is something different people will have different opinions on.  But the study is a negative one, even if there may be debate about exactly how much weight to place on it.

The results in more detail

Results for distance from power lines

 leukaemiabrain / central nervous systemmalignant melanomafemale breast cancer
total numbers
cases78236781915358404
controls23469203432745956516
Odds ratios and 95 confidence intervals by distance
0-49 m1.11 (0.83-1.48)1.22 (0.88-1.69)0.82 (0.61-1.11)1.07 (0.93-1.24)
50-99 m1.19 (0.98-1.45)1.03 (0.83-1.29)1.07 (0.88-1.31)1.03 (0.93-1.14)
100-199 m0.93 (0.82-1.05)1.04 (0.91-1.19)0.98 (0.87-1.10)1.04 (0.98-1.11)
200-299 m1.11 (1.00-1.23)0.90 (0.81-1.01)0.97 (0.88-1.08)0.99 (0.94-1.05)
300-399 m1.00 (0.90-1.10)1.08 (0.98-1.20)0.98 (0.90-1.08)1.02 (0.98-1.07)
400-499 m1.02 (0.93-1.11)1.00 (0.91-1.10)0.95 (0.87-1.04)1.00 (0.96-1.05)
500-599 m1.03 (0.95-1.12)1.03 (0.94-1.13)1.02 (0.94-1.10)0.97 (0.93-1.01)
600-1000 m1.001.001.001.00
P value for trend
categories0.2810.6570.4060.275
inverse rank of distance0.2950.6960.3110.240
inverse of distance squared0.3140.7900.2900.308

The Odds Ratios given are the adjusted ones (the paper also gives unadjusted), for year of diagnosis (the paper also gives 5 years prior to diagnosis), and, for female breast cancer, are for 100% of cases (the paper also gives a 50% sample of cases).

Results for magnetic field calculated from power lines

 leukaemiabrain / central nervous systemmalignant melanomafemale breast cancer
total numbers
cases74796454869927843
controls22380194092621527946
Odds ratios and 95 confidence intervals by magnetic field
>1 µT1.03 (0.57-1.86)1.02 (0.47-2.22)0.68 (0.39-1.17)1.08 (0.77-1.51)
0.4-1 µT0.91 (0.57-1.46)0.92 (0.54-1.55)1.17 (0.75-1.81)1.06 (0.78-1.43)
0.3-0.4 µT1.19 (0.60-2.36)1.36 (0.66-2.81)0.87 (0.47-1.59)0.86 (0.57-1.30)
0.2-0.3 µT1.22 (0.74-2.02)0.98 (0.53-1.78)0.74 (0.42-1.33)0.80 (0.58-1.12)
0.1-0.2 µT1.19 (0.83-1.70)1.10 (0.76-1.59)0.98 (0.67-1.44)0.97 (0.77-1.24)
<0.1 µT1.001.001.001.00
P value for trend
categories0.8180.9020.1770.832
continuous measure0.6620.8100.0870.420

The Odds Ratios given are the adjusted ones (the paper also gives unadjusted), for year of diagnosis (the paper also gives 5 years prior to diagnosis), and, for female breast cancer, are for a 50% sample of cases.

Results presented graphically

 

by distance

by magnetic field

Leukaemia

p=0.280

p=0.818

 graph of imperial results

 

 graph of imperial results
brain / central nervous system cancers

p=0.657

p=0.902

 

 graph of imperial results

 graph of imperial results
malignant melanoma

p=0.406

p=0.177

 graph of imperial results

 

 graph of imperial results
female breast cancer

p=0.275

p=0.832

 

 graph of imperial results

 graph of imperial results

Note: the graphs and tables are reconstructed by us from the data in the published paper, they are not the originals from the paper.

Adjusted and unadjusted results

For breast cancer, the findings before the adjustments for socioeconomic status etc were slightly higher (see also what we already know about socioeconomic status near power lines):

 graph of imperial results

Strengths and weaknesses of the study

Strengths

  • The size of the study - over 80,000 cancer cases, over a 35 year period, with tens of cases at the higher exposure levels
  • The objectivity of the exposure assessments, all performed from records and blind to case-control status

Weaknesses

  • The study is based on the address of the subjects at diagnosis, with no information about how long or short a time they had lived there, and no information on cumulative exposure
  • The study had to use selected other cancers as controls
  • The magnetic field calculations are only approximate (principally because of the limited historical load information)

Interestingly, each of these main weaknesses would act so as to dilute, in the study findings, the size of any true effect.

Conduct of the study

The study came about as an initiative of National Grid, who considered there ought to be a parallel study on adult cancers to the CCRG or "Draper" study on childhood cancers.

National Grid provided the data on the locations of the power lines and performed the calculations of proximity and exposure, using the same methods as for the Draper study.  Unlike that study, National Grid was not involved in the analysis or in writing the paper, which were the sole responsibility of the Imperial College group.  The study was formally under the auspices of the Department of Health (DH). Most of the funding came from the Energy Networks Association (ENA), but ENA provided that funding to DH and Imperial College's contract was with DH.

The abstract in full

Extremely Low Frequency Magnetic Fields From High Voltage Overhead Power Lines and Adult Cancers

Paul Elliott, Gavin Shaddick, Margaret Douglass, Kees de Hoogh, David J. Briggs, Mireille B. Toledano

Background: Extremely low-frequency magnetic fields are designated as possibly carcinogenic in humans, based on an epidemiologic association with childhood leukemia. Evidence for associations with adult cancers is weaker and inconsistent.
Methods: We conducted a case-control study to investigate risks of adult cancers in relation to distance and extremely low-frequency magnetic fields from high-voltage overhead power lines using National Cancer Registry Data in England and Wales, 1974–2008. The study included 7823 leukemia, 6781 brain/central  nervous system cancers, 9153 malignant melanoma, 29,202 female breast cancer cases, and 79,507 controls frequency-matched on year and region (three controls per case except for female breast cancer, one control per case) 15–74 years of age living within 1000 m of a high-voltage overhead power line.
Results: There were no clear patterns of excess risk with distance from power lines. After adjustment for confounders (age, sex [except breast cancer], deprivation, rurality), for distances closest to the power lines (0–49 m) compared with distances 600–1000 m, odds ratios (ORs) ranged from 0.82 (95% confidence interval = 0.61–1.11; 66 cases) for malignant melanoma to 1.22 (0.88–1.69) for brain/ central nervous system cancer. We observed no meaningful excess risks and no trends of risk with magnetic field strength for the four cancers examined. In adjusted analyses at the highest estimated field strength, ≥1000 nanotesla (nT), compared with <100 nT, ORs ranged from 0.68 (0.39–1.17) for malignant melanoma to 1.08 (0.77–1.51) for female breast cancer.