Uncontrolled Variables


When the long bones are immobilized, (e.g., by casting) a frequent physiological response is a loss of bone material - a condition known as osteoporosis. As we have seen in chapter 2, bone is a piezoelectric material and, consequently, it exhibits the converse piezoelectric effect (mechanical deformation under the influence of an applied electric field). McElhaney et al. (52) hypothesized that an electric field could simulate the naturally present mechanical stresses in bone via the converse piezoelectric effect, and thereby eliminate the osteoporosis associated with disuse. The theory was tested by immobilizing the hind limbs of rats and then observing the effect of an electric field; it was found that the osteoporosis caused by immobilization was reduced by exposure to 7 kv/m, 3-30 Hz. However, in addition, 44% of the EMF-exposed animals developed bone tumors; none were seen in the sham-irradiated rats. Martin and Gutman (53), (Martin worked with McElhaney et al. on the original study) performed a replicate study and confirmed the observation that the EMF ameliorated the immobilization-induced osteoporosis. No tumors, however, or other malformations were observed either by gross or microscopic examination.

The two studies were done under essentially identical conditions, but tumors were seen only in the first study. Statistically, it is unlikely that they were unrelated to the field and developed only in the exposed group by chance. This suggests that an uncontrollable variable (UV) capable of inducing tumors in conjunction with an EMF was present in the McElhaney study.

We too observed an EMF-related biological effect that was not seen in a replicate study; in our case, however, it was possible to preselect the animals in the second study and thereby gain information about the UV associated with the biological effect. In the initial study, we found secondary glaucoma in 10 of 60 rats that had been exposed for 30 days to 0.6-19.7 kv/m vertical electric fields; the glaucoma was not seen in 43 rats exposed to horizontal fields (0.3 -9.7 kv/m) or in 72 controls (43). None of the rats had been subjected to an ophthalmic examination prior to field exposure because the appearance of eye diseases had not been anticipated. It was, therefore, not possible to determine whether the glaucoma resulted from a worsening of an already existing defect, or was caused solely by the EMF. These alternatives were examined in two vertical-field studies (2.8 kv/m, 19.7 kv/m) in which all animals were subjected to a pretest eye examination with the bimicroscope and the indirect ophthalmoscope. Rats that exhibited any identifiable disorder (iris hemorrhage, anterior synechia, dacyroadenitis, keratitis) were destroyed, and only defect-free animals were placed on study. Following 30-day exposures, no cases of secondary glaucoma were seen in either the exposed or sham-exposed rats 50 in each group). It seems to us, therefore, that our initial observations of secondary glaucoma most likely stemmed from an exacerbation of preexisting eye defects by the EMF. The EMF, in any event, could not have been the sole cause of the glaucoma.

The clearest example of the operation of a UV may be the multigeneration study done at the Battelle Laboratories (54). Following the publication of our first multigeneration study (34), Battelle was commissioned to replicate the work. The investigators first developed an exposure system that was unexcelled with regard to field homogeneity and reproducibility of electrical environment. Every aspect of the animals' physical environment-light, temperature, humidity, presence of pathogens in the air, air flow, for example-was rigorously monitored and controlled by automatic equipment. The investigators then constructed two complete exposure facilities: each consisted of a completely characterized exposure unit, an identical unit for sham-irradiation, and a completely controlled environment suitable for housing both units.

The multigeneration study was begun in the first exposure facility, and 3 weeks later a replicate study was begun in the second facility; both replicates were done double blind. The body-weight data for the males and females of each of the three generations in each replicate is shown in table 8.10. Despite the fact that the maximum level of human intervention and control was exercised, and despite the unprecedented resources devoted to the study, it was obviously not possible to eliminate the role of a W: at the end the study, the males and females in the first replicate were statistically significantly smaller than the controls, but in the second replicate they were significantly larger.




When an experiment is replicated and different results are observed, there is no general rule by which it can be decided whether the first or the second replicate (or both or neither) are the true descriptions of nature. In each case an analysis must be made of the details of the studies and their relation to other studies. Only in this manner can it be decided whether an UV likely was present (in which case both experiments would correctly describe nature, but under different circumstances), or whether a Type I or Type II error was made in one of the replicates.

Chapter 8 Index