Questions and Answers About EMF Electric and Magnetic Fields Associated with the Use of Electrical Power.
January 1995.

Human Health Studies

Q. How do scientists study possible effects of EMFs on people?

A. They use a type of research called epidemiology—the study of patterns and possible causes of diseases in human populations. Epidemiologists study short-term epidemics such as outbreaks of food poisoning and long-term diseases such as cancer and heart disease. Results of these studies are reported in terms of statistical associations between various factors and disease. The challenge is to discover whether the statistical results indicate a true causal association. This includes assessing possible effects of other factors ("confounders") that could affect study results. A "statistically significant" finding is one in which researchers are 95% confident that an association exists. However, a "statistically significant" finding does not necessarily prove a cause-effect association. Usually, supplemental data are needed from studies of laboratory animals before scientists can conclude that a given factor is a cause of disease.

ESTIMATES AND ODDS RATIOS

The language of epidemiology can appear, to the uninitiated, more precise, than it actually is. An odds ratio (see example in the "Examples" section below) is an estimate. Epidemiologists must calculate, along with odds ratio, the range over which they are confident that this estimate is reliable. Sample size is a key factor in this calculation. The smaller the sample, the less reliable the information.

HOW EPIDEMIOLOGISTS CONDUCT CASE- CONTROL STUDIES

THE PROCESS

1. A list of people with a particular disease is assembled. These are the cases.
2. A list is assembled of people who are similar to the cases, but who do not have the disease. They are the controls.
3. The numbers of cases and controls who were previously exposed to factor X are estimated. This is often one of the most difficult parts of the study because exposures have often occurred many years in the past.
4. The exposure ratio of the cases is compared to that of the controls. If the ratios are the same, there is no association between factor X and the disease. If cases have a higher ratio, there is a positive association, and factor X may be cause of the disease. If the cases have a lower exposure ratio than the controls, there is a negative association. This would suggest that factor X may help protect people from the disease.

Here are some examples of possible outcomes of a study of potential risk factor X, based on 300 cancer cases and 300 controls:

• If 71 cases were exposed to factor X and 229 were not exposed, the case exposure ratio = 71/229 = 0.31. If 71 controls were also exposed, the control exposure ratio is also 0.31. Dividing the case exposure ratio by the control ratio gives the odds ratio (OR), sometimes called relative risk (0.31/0.31 = 1.00). An OR of 1.00 means the odds that the cases were exposed to factor X was the same as for the controls. Therefore, in the example, there is no association between factor X and cancer.
• Now suppose 110 of the total 300 cases were exposed (ratio = 110/190 = 0.58), and 71 controls were exposed (ratio = 0.31). The OR is 0.58/0.31 = 1.87. If the OR is above 1.00, there is a positive association between factor X and the disease. With certain assumptions, this means that in the example, people exposed to factor X had an 87% increased risk of having cancer.
• Even when the OR is above 1.00, calculations must be done to see whether it is statistically significant (more than just chance.) In the example, the OR of 1.87 is statistically significant. Suppose another study was done also with 300 cases and 300 controls. In this study, however, there were only 11 exposed cases and 6 exposed controls. Although the OR = 1.90, it is not statistically significant because of the small numbers of exposed subject.