 |
 |
|
|
|
|
|
|
||||||||||||||||||
|
|
|||||||||||||||||||
|
|
|||||||||||||||||||
|
|||||
|
|
By Greg Lapin, N9GL
Chairman, ARRL RF Safety Committee
November 26, 2001
The "controlled experiment" is the hallmark of scientific study. So, why should we be skeptical about the outcomes of such studies?
What do you think when you learn from the news media that a scientist has exposed animals to RF energy, and they got an illness that other unexposed animals did not get? On its face, the logical conclusion is unassailable--that the RF energy caused the illness. This is an example of a controlled experiment, the hallmark of scientific study.
The logic of a controlled study is simple. You take identical subjects--such as strains of inbred rats that all have the same DNA--and divide them into two groups. You treat both groups the same in every way, except for the one variable that you are testing--in this case exposure to RF energy. In the end, you examine the two groups and observe any differences. It seems like a foolproof process, yet it has generated considerable confusion in the RF safety field.
A necessary requirement for a controlled scientific study is that the study and control groups are identical in every way, including how they are treated during the experiment. If it is necessary to stuff a rat into a waveguide in order to expose it to an RF source, the rats in the control group must be stuffed into the same waveguide for the same amount of time, though without turning on the RF power. Still, every detail of the exposure must be known, in order to reach a conclusion.
Where Controls Didn't Show the Answer
An example where this logic went awry was seen in a large group of experiments performed in the early 1970s. Exposure to a continuous RF carrier that was not strong enough to generate appreciable heat in the test subjects had not yielded any results to show that the RF was harmful. Most of the RF that we are exposed to is modulated, so experiments were designed to see if different forms of modulation affected the results. It was found that pulsing the RF carrier at relatively low rates caused rats to become ill. An entire body of theory was developed to explain how these interactions could affect biological organisms.
Many of the theories assumed that the modulated RF was detected in the body. In other words, biological tissues acted like an AM receiver and separated the modulating waveform from the carrier. This would imply the existence of a nonlinear process in tissue, much like a diode detector in a radio. Such a thing had never been shown and was illogical to engineers and physicists. Biologists had no such misgivings. To them, this made sense and, more to the point, offered an explanation for the results that they were seeing in the laboratory.
The final episode to this story came about in 1971 when it was demonstrated that pulsed microwaves evoked an auditory response in rats. In other words, rats could hear the pulses that modulated the microwaves, and it was subsequently shown that a nonlinear process of slight thermal expansion of brain tissue caused the AM detection that was necessary for the pulses to be "heard." The rats--and more importantly their organs--were not being directly affected by the modulated RF energy. Rather, the animals were being upset by phantom sounds in their heads that drove them crazy, leading to physiological changes that could be detected in several of the measures that were used to imply adverse health effects.
Some people at the time attempted to extend this conclusion to humans and were prepared to severely limit the exposure of humans to pulsed RF (effects on rats were seen for pulse frequencies between 200 and 400 Hz). If that had occurred, it would have been quite possible that CDMA cellular telephone technology would never have been developed. Fortunately, the actual reason for the experimental results on rats with pulsed RF was found.
In this example, the controlled experiment yielded the correct results. The rats that were exposed to pulsed RF did have a different response than those that were not exposed. However, the reasons postulated were not correct. If another control group had been used--one where the rats were forced instead to listen to loud clicking sounds--the results would have been similar to the experiment where the rats exposed to pulsed RF.
Multiple Types of Controls
The evoked auditory response in rats points to the need for a more complicated control system in most experiments, particularly when studying biological systems that are highly complex and easily perturbed. It may be enough to rely on only two experimental groups (exposed and unexposed) in relatively simple experiments, such as looking at effects on dishes of growing cells. A living animal is far more complex and there are many other physiological mechanisms that may be inadvertently drawn into the experiment.
There are three types of control that can be designed into an experiment. The one that was described above, the group of animals that is not exposed to the test stimulus is called a "sham control group." This group is set up exactly the same as the exposed group, but the exposure is a sham (such as never turning on the power).
Another type of control group is the "positive control group." In this group, a stimulus that is known to cause an effect is applied to a group of test animals that is otherwise treated the same as the other groups. With some effects, it is possible to define a "negative control group," such that the subjects in that group should be guaranteed not to show any effects.
When feasible, the best experiments use more than one control group. If the control group that is expected to show a result does so, and the control group that is not expected to show a result does not, the likelihood is lessened that there are additional factors that were not controlled.
Statistics
It is important to realize that there are virtually never absolutes in biological studies. In all of the experimental groups mentioned above, one would almost never see an experiment in which all of the subjects responded the same. There tend to be subjects in the control groups that respond as if they were stimulated, and subjects in the experimental group that do not respond. Experimental differences are concluded by seeing more responding subjects in the experimental group than in the control group.
Statistical methods are used to ascertain the probability that the experimental groups are the same. When that probability is very low, the scientists conclude that the experiment shows an effect. In biological experiments, statistical significance is usually defined as a probability of less than 5% that the groups are the same.
How Can We Distinguish What's Right From What's Wrong?
In 1994, three authors set out to determine if the scientific community was capable of discerning fraudulent scientific papers from those that were performed correctly (WP Whitely, D Rennie, AW Hafner: The Scientific Community's Response to Evidence of Fraudulent Publication: The Robert Slutsky Case. JAMA 1994;272:170-173). They chose publications by a scientist who had been shown to fabricate some of his scientific results and compared how often this man's work was referenced by other scientists before and after the public disclosure that some of his work was not trustworthy. What they found was surprising. They concluded that scientists are incapable of recognizing most faulty work. Even after the revelation of fraud, there was still some subsequent scientific work that used the fabricated science as a basis for future work.
Investigator Bias
I don't mean to imply that all faulty laboratory work is intentionally misleading. It takes more than a control group to ensure that the effect being studied is truly responsible for differences seen in a scientific study. Unintentional bias is a potential problem in all studies. If an investigator believes that a cause will, or won't, lead to an effect, a subconscious bias in the way that the study is performed can affect the experimental results without the investigator realizing that such a bias exists.
In the studies that determine the efficacy of the drugs that we take, the FDA requires that investigator bias be mitigated by disguising subjects that are being used to test the drug from the ones that are in the control group. Such a study is called "double-blind," and the investigators give drugs to patients without knowing if the bottle contains the real drug or a placebo. These things are revealed only after the experiment is completed and the results are recorded.
Most scientists are honest and in search of the truth. However, the temptation for the results to come out a certain way is great, particularly in the RF bioeffects field. The scientist who reports a harmful effect from RF exposure is almost certain to become a household name, and, in many cases, publication of these results leads to additional research money--the lifeblood of a scientist's career. How often have we heard a phrase such as "these results require further study?" In contrast, the scientist who does not find an effect is much less likely to be heard of.
So, Can We Trust Controlled Experiments?
Standard scientific method presents a powerful set of techniques that permit us to understand the cause and effect of many things. The use of a control group is an integral part of this method and helps convince us that we are really seeing what we think. However, differences between experimental and control groups do not necessarily prove cause and effect. Other factors, such as unrecognized physical processes and experimental bias, can play large parts in the final results, particularly in work that is very visible to the world, as are studies of RF effects.
After more than 30 years of research in the RF safety field, when I see experimental results that purport to show something different, I wait for independent replication of the experiment before I am convinced that something new has been shown. One problem with getting scientific information from the press is that you will probably not see everything. It is exciting to report about someone who claims to have shown a new danger of RF energy, and you are likely to see mention of it on the evening news and in the newspapers. However, when attempts at replicating the results by independent laboratories fail to confirm the original results, the news media often do not find this as stimulating and often fails to report it. As a result, people tend to remember the shocking reports of the dangers of RF energy and rarely get the final story.
Safety Standards
We use RF energy under the restrictions of several safety standards that attempt to set safe exposure limits. The standards are based on more than one thousand scientific studies and take into account the reports of dangers and whether there were successful or failed attempts to replicate them. The studies are also examined in detail by a number of expert reviewers who specifically look for evidence of experimental errors or bias.
My recommendation is that when you see the earth-shattering new research that tells you something new about the dangers of RF in your local newspaper or on TV, take a deep breath, sit back, and wait to see if confirmation from other independent researchers is forthcoming. If the reported dangers are real, you are sure to hear more about them.
Editor's note: Greg Lapin, N9GL, started working in the RF safety world after spending many years first studying cardiac function imaging and then brain tumor kinetics. He serves as chairman of the ARRL RF safety Committee and as a member of the FCC Technological Advisory Council. A former professor of biomedical engineering and neurology at Northwestern University, Lapin now works as a consulting professional engineer in the electronics industry. He was first licensed while a teenager in 1969 and continues to be fascinated by virtually all aspects of Amateur Radio. One of his many interests is electronic design, and he is the author of Chapter 8, "Analog Signal Theory and Components" in The ARRL Handbook for Radio Amateurs. His non-ham interests include making things grow in his garden and serving as commissioner of the local children's baseball and softball league. At other times--when he is not working or helping his kids with their homework--you might find him with the local emergency services agency, climbing his tower, building a new QRP rig, playing with his APRS setup, responding to QSL cards, going off on a DXpedition, or trying to get that "new one." You can reach him by email at g.lapin@ieee.org. The ARRL RF Safety Committee page contains a link to archives of previously posted editions of N9GL's RF Safety Column.
