Mixing babies, Popper, power lines and this blog’s Comments Section

Posted on February 4, 2014


Unfortunately, I missed a January blog post, just by a couple of days. Fortunately, that was to wait for the publication of a paper, which is now available for download online…and which is the topic of this particular blog post. It’s one where I am going to try something new; I am going to ask you to get your science heads on, your falsification heads more specifically, and see if the power of the internet can be used for the good (or alternatively, if it stays eerily quiet to abandon this plan for future posts) .

So, back to the paper in question.


It’s entitled “Maternal residential proximity to sources of extremely low frequency electromagnetic fields and adverse birth outcomes in a UK cohort and it was published in the journal Bioelectromagnetics online on January 31. Early View, or hot of the press so to speak. The pdf can, if you have access, be downloaded here (link), but just so you know what I am talking about this is the abstract:

Studies have suggested that exposure to extremely low frequency electromagnetic fields (ELF-EMF) may be associated with increased risk of adverse birth outcomes. This study tested the hypothesis that close proximity to residential ELF-EMF sources is associated with a reduction in birth weight and increased the risk of low birthweight (LBW), small for gestational age (SGA) and spontaneous preterm birth (SPTB). Closest residential proximity to high voltage cables, overhead power lines, substations or towers during pregnancy was calculated for 140356 singleton live births between 2004 and 2008 in Northwest England. Associations between proximity and risk for LBW, SGA and SPTB were calculated, as well as associations with birth weight directly. Associations were adjusted for maternal age, ethnicity, parity and for part of the population additionally for maternal smoking during pregnancy. Reduced average birth weight of 212 g (95% confidence interval (CI): −395 to −29 g) was found for close proximity to a source, and was largest for female births (−251 g (95% CI: −487 to −15 g)). No statistically significant increased risks for any clinical birth outcomes with residential proximity of 50 m or less were observed. Living close (50 m or less) to a residential ELF-EMF source during pregnancy is associated with suboptimal growth in utero, with stronger effects in female than in males. However, only a few pregnant women live this close to high voltage cables, overhead power lines, substations or towers, likely limiting its public health impact. 


So living close to power lines and other residential sources of ELF seems to have an effect on fetal growth. This association has been investigated previously, but so far itbungalow_pylon_big_469x300 has been quite difficult to establish an association because not many people live close to power lines so a large population is required, while also for such large populations generally not much information about confounders is available. We were lucky enough to have a very large dataset (140,356 births) that enabled us to specifically look at women living very close (<50 meters) away from a source, which generally speaking has above background exposure levels, while also we had information on important confounders was available; at least for part of the cohort. I only just realized that, in addition to the confounders mentioned in the abstract, we could also adjust for socio-economic status using the Index of Multiple Deprivation (IMD). Maternal smoking information during pregnancy was only available for about a third of the cohort.

For clinical outcomes we couldn’t really say anything definite because there were just not enough “exposed cases”, but the effect on birthweight was unexpectedly large (212 grams lighter on average). It is also not because of preterm birth….we checked that.

Maybe too large?

There are a couple of obvious points for caution here that we discussed in the paper.

With respect to the exposure assessment it is important to realize that exposure is not measured but inferred from distance to the residential sources while, moreover, distance was based on the centroid of the postal code area and not the residential location itself. We also did not take into account other (indoor) sources of ELF. And of course, women will not always have stayed indoors (although, presumably, for longer periods than their non-pregnant counterparts; especially in later pregnancy).

Maternal_smokingMaternal smoking during pregnancy was only available for about a third of the cohort, and while we didn’t find any evidence of (expected) differential missingness (i.e. more women who smoked during pregnancy did not report this compared to reported by women who didn’t smoke), this  should not be dismissed.

Residual confounding cannot be excluded. IMD is an ‘area variable’ and we did not have socio-economic information at an individual level. Generally speaking however, women living closer to power lines were from more affluent, higher socio-economic groups. This goes against the idea that it is lower socio-economic groups live close to power lines and towers…unless of course, we missed some small low SES pockets within affluent areas, which I don’t think is the case.

We also don’t really know by which biological mechanism this would work. This is an argument that pops up repeatedly in EMF research (usually in relation to cancer);  often not as part of a scientific debate, but more in a baseball bat-approach to kill any further discussion. In this case however, there some possible mechanisms. Not definitely, 100%, beyond doubt proven (i.e. they may all be true…or none), but anyway: we postulated a possible indirect mechanism in which disruption in the balance between plasma and vascular cell Ca2+  results in placental vascular function changes, while other proposed a similar idea of modulation of calcium binding to cell surface, and yet others have described disruption of cell-to-cell electric communication, direct alteration of growth, estrogen-mediated, and impaired growth because of enhanced DNA synthesis.

As with any epidemiological study therefore, it is not perfect. In the mean time we have done some further work and residual confounding as discussed above may be responsible for some of the observed difference in birthweight, but by no means all. A statistically significant difference remains. And that’s where Popper comes out of the wood work.

Philosopher Sir Karl Popper (1902-1994)Most people are familiar with the important idea of his “critical rationalism” about “falsification; that, in a nutshell, no number of positive outcomes at the level of experimental testing can confirm a scientific theory, but a single counterexample is logically decisive: it shows the theory, from which the implication is derived, to be false. The term “falsifiable” does not mean something is made false, but rather that, if it is false, it can be shown by observation or experiment. Anyway, this comes from good old Wikipedia, so if you want to know more (link).

And that’s where you come in!

My reader…or well, the proportion who sat through it all until the end!

This blog comes with a comments section underneath and I am very interested in your input. In other words…tell me why this doesn’t make sense ?!

I, or better we (the authors), have come up with some possible reasons which I have outlined above. Further work on this does not seem to fully explain the association away, so what have we missed?

If you are of a shy inclination than of course, just post anonymously. I don’t mind…I am just very interested in whether we can find that one black swan amongst the white ones (as a figure of speech, obviously) (link).

Or well, maybe, just maybe, if we can’t find that swan, maybe, just maybe this could be a true association.

………..Now there is a thought…….