25/07/2007 - Essex Provocation Study finds no effect

Summary

The latest double blind provocation study by Essex University claims to have found that there is no significant difference between the subjective (self-assessed symptoms) and objective (measured symptoms - e.g heart rate) responses between sensitive and control participants, regardless of the system to which they were exposed.

Edited (03/August/2007): Since this news article was first posted, we have now had time to properly study the published paper. We have found some serious problems with the way that the results from this study were collected, analysed and presented and we will be posting a further critical response on the Powerwatch website in due course. We also have real problems with the way the sudy results were press released through the Science Media Centre (SMC) and we will be posting critical comments about this, too.

Study Flaws

Firstly, of the original 56 selected to take part in the study, 12 withdrew before the double-blind exposures, of which some withdrew very specifically because of the severity of symptoms from the open exposure. To lose 20% of the "sensitive" category in this way may well remove the most sensitive, or "truly" sensitive participants in the study.

Secondly, when addressing a subject such as this, where there is so much general public concern, there is bound to be a high level of nocebo effect amongst those that believe they may have a problem. This would give rise to a much higher proportion of self-assessed sensitive participants that were not in fact sensitive, and this would have the effect of "hiding" those that were actually responding to the provocations. This is a potentially very serious confounder that has simply not been addressed in this study.

Thirdly, anecdotal reports of sensitivity (useful when deciding how an experiment must be designed) show that many of the effects experienced only manifest themselves a few hours after exposure, and have often not fully subsided until the following day. This also seems to vary from person to person, but it is hard to evaluate what the reason for this would be. Both of these points are hard to properly incorporate into the design of the study.

Flaw counter-arguments

There is very little the study authors can do about the withdrawals, aside from comment on the potential damage that they can do to the statistical resolving power of the data (i.e. the chance that all 12 would be 100% "sensitive" participants and the chance that all 12 would be 0% "sensitive" participants would both greatly skew the data). It would be useful perhaps to keep a record of those withdrawing due to excessive symptoms, to create a "pool" of people that seem to exhibit highest sensitivity (see next point for reasoning).

Edit (25/07/2007): Actually, the above point is largely incorrect. Whilst there are 2 who withdrew prior to completion of the open provocation (lowering the number of sensitive participants from 58 to 56), this is a much smaller proportion of the total participants. The open provocation results are very similar to the double blind results, and there is no evidence to suggest that these 12 reacted any more or less strongly than the other 44 who remained in the study.

Nocebo responders are also very hard to sensibly deal with as a confounder, due to having no prior knowledge of who they are - this is especially true when cause is not established and it is a theoretical possibility that all fall into this category. However, there are certain steps that can be taken to separate out those with highly successful responses, perhaps with the aim of adding them to the "highest sensitivity pool" (see point above). This gives the opportunity to test those that seem to be the most sensitive in a given study without the possible confounders of nocebo sufferers. This was not done here, but in fairness to the authors this would require another round of tests and may not have the required participation from the selected group (who may not want to go through another round of tests).

The problem of effects experienced after a few hours (but not immediately) was addressed by the study, which tested for symptoms very shortly after the end of the exposure and allowed the participants to fill in a questionnaire to return the following week. However, there is a good argument for it not being necessary as the sensitive group have reported much higher levels of perceived symptoms than the control group anyway. This gives reasonable support to the idea that the authors have designed it in a way that should find effect if there is one.

There is the added possible confounder that the sensitive group were suffering from stress responses based on the pressure of believing they might be exposed, and again this is something that it is not reasonable to expect the authors to eliminate. However, it is a very important issue in these double-blind provocation studies that can not be ignored.

The problem of responses not subsiding until the following day has been very well catered for by the research team, as the tests were separated by a period of one week in each case.

Discussion

All in all, the Essex team have carried out one of the best designed and executed studies to date. They consulted widely during the planning stages and took notice of a wide variety of sources (including Powerwatch and some individuals who claim to be electrically sensitive). We were present when Sir William Stewart opened their testing laboratory. Their laboratory is well designed and uses non-toxic paints and DC lighting to help eliminate possible confounders.

Whilst there were distinct problems early on in the realism of the exposure system, by the time the actual provocation tests were performed it was the best we have seen in an experiment to date, including a well designed simulation of real network traffic. Measured background EMFs in the lab are negligible (the rooms were screened to a 60 dB reduction in field strength), and thus one of the biggest flaws of Rubin's work has been dealt with effectively.

As with all double-blind studies, there is no way in which the effects of chronic exposure can be addressed, so this is not a failing of the study but another point for consideration with regards to final conclusions.

Conclusion

Following this paper, it is becoming increasingly likely that there is a significant proportion of self-assessed Electrically Sensitive people whose response is entirely nocebo - whether induced by fear of harm or other reasons, the cause is not radiofrequency EMFs.

There are fundamental flaws in the idea of evaluating subjective responses in a double blind system that simply cannot be avoided regardless of the care taken on experiment design (such as the potentially large proportion of nocebo responses). There is also a chance that chronic exposure may give rise to genuine symptoms whereas acute exposure may not, but that cannot be addressed by this study.

Edited (03/08/2007):So whilst a minority of people maybe very sensitive, the results from this study suggest that the proportion of sensitive people is likely to be lower than the 3% - 35% that has been quoted. However, serious questions are now being raised about how the results from this study were analysed and we will be posting a further critical response in due course.

Links

[View paper abstract]

Also in the news

Korean study finds increased leukaemia risk for those living near an AM radio transmitter

This text is an intro to an excellent article by Louis Slesin of Microwave News. Please read the full article for a complete analysis and further implications.

In the largest and most detailed study of AM radio radiation to date, a team led by Mina Ha of South Korea's Dankook University in South Korea has found that children living within 2km of an AM transmitter had more than twice the risk of developing leukaemia, compared to those living more than 20km away. The study, which included 36 cases of children with leukaemia living within 2km of an AM station, will appear in the August 1 issue of the American Journal of Epidemiology and is already available on the journal's website. The abstract reads as follows:

Cancer risks were estimated using conditional logistic regression adjusted for residential area, socioeconomic status, and community population density. The odds ratio for all types of leukemia was 2.15 (95% confidence interval (CI): 1.00, 4.67) among children who resided within 2 km of the nearest AM radio transmitter as compared with those resided more than 20 km from it. For total RFR exposure from all transmitters, odds ratios for lymphocytic leukemia were 1.39 (95% CI: 1.04, 1.86) and 1.59 (95% CI: 1.19, 2.11) for children in the second and third quartiles, respectively, versus the lowest quartile.

This supports their earlier findings in 2004 and 2003, and the findings of Italian researchers findings in 2002, and starts to lend some genuine epidemiological weight to the possibility of increased cancer risk for those living within a few kilometres from radio transmitters.

When Ha compared cases and controls relative to estimated RF exposures, she found that the risk was significantly higher for those in the second and third exposure quartiles, but not in the 25% most exposed children. Regarding the lack of an association among those who are most exposed to RF, Ha suggested that it might be due to "decreasing statistical power" or to a "bystander effect."

Ha estimates that the electric field at 2km from the AM transmitters ranged from 1 V/m to 3 V/m.

[Full Microwave News article]
[Abstract of featured 2007 paper]
[Abstract of previous paper from 2004]
[Abstract of previous paper from 2003]
[Abstract of Italian paper from 2002]

Two more studies find effects on rats from GSM mobile phone radiation

The first of which, from INERIS in France, found a temporary change in brain activity (specifically, glial evolution, suggesting neuronal damage) from exposure to 15 minutes of GSM 900 Mhz mobile telephony radiation. This temporary effect was found to be statistically significant at 2 and 3 days, and to have disappeared after 6 to 10 days from exposure. This supports an earlier paper from France in 2004 finding the same effect (abstracts linked at the bottom of news entry). Whether this effect could pose long term harm remains to be studied.

The Medical College of Wisconsin in the US has found that rats exposed to cellular phone emissions (2 x 3 hour periods per day for 18 weeks) exhibited "a significantly higher incidence of sperm cell death than control group rats through chi-squared analysis. In addition, abnormal clumping of sperm cells was present in rats exposed to cellular phone emissions and was not present in control group rats." The authors conclude that "These results suggest that carrying cell phones near reproductive organs could negatively affect male fertility".

[2007 rat brain activity paper]
[Earlier 2004 rat brain activity paper]
[2007 rat sperm motility paper]

New Study from Imperial College finds increased infection risk from Electric Fields

Electrical fields generated by everyday electrical equipment such as computers, and excess static charge created by many modern materials, could be bad for your health, says new research published by Imperial scientists.

Keith Jamieson of Imperial's Centre for Environmental Policy, lead author of the paper, says: "Many of the factors that can cause high electric fields and increased deposition and contamination are often found in hospital ward environments and in buildings where incidents of sick building syndrome are noted."

The researchers propose, however, that adopting a number of simple guidelines in home, office and hospital environments could reduce the size of electrical fields generated, and therefore reduce the levels of potentially unhealthy particles deposited on the skin and in the lungs as well as making surface contamination far easier to remove.

Keith Jamieson explains: "In the case of electrical equipment, particularly laptops, ensuring they are earthed can often greatly reduce fields. In terms of the electrostatic charge generated by people themselves, careful selection of materials and humidity levels can significantly reduce problems as can balanced bipolar air ionisation. Trying to avoid spending time in areas where high fields are created, and unplugging electrical equipment when not in use, are also good options - so there are a number of easy actions which can already be implemented in the workplace and the home to help reduce the toxic load our bodies have to deal with and the risk of illness and infection being transmitted in this way."

[Taken from this Imperial College news summary]

Lloyd Morgan on non-thermal biological effects from EMFs

Lloyd Morgan has raised an excellent point regarding the continuous denial that there is good evidence for non-thermal effects from Electromagnetic Radiation as follows:

[View as published on EMFacts]