LIMITATIONS
Aside from the problems discussed above, other constraints limit the contribution that science can make in resolving problems of environmental hazard to human health. These limitations are related to the nature of science as an intellectual pursuit, the special problems involved in biology and medicine, and the personalities of individual scientists.
Pragmatic Nature of Science. Scientific facts and theories should combine information currently thought to be relevant and true. Such synthesis is always subject to revision as new information becomes available or as the alternative significance of old information is recognized. As a result, scientists tend to so qualify evidence that its value is diminished in aiding legal decisions about hazard evaluation. Some scientists function better than others as witnesses, but competence as a witness is not the same as competence as a scientist.
Probabilistic Nature of Scientific Conclusions in Biology. The scientific philosophy generally assumes that nothing is known with absolute certainty; rather, facts are known to some level of probability. In biology, the margin of uncertainty is usually larger than in the physical sciences because of the complexity of biological systems. In many experiments it is common practice to accept a 95% probability of a correct scientific hypothesis calculated from statistical models as the dividing line between acceptance and rejection. In expanding the field of knowledge, the 95% probability criterion is usually adequate. In hazard assessment, however, using conclusions based on 95% probability of correctness may involve serious problems, particularly when questions of serious human injury or considerable economic cost are involved. If the reliability of a result is questioned, it is possible to retest at a higher level of statistical certainty by using a larger number of animals.
Parametric Relationships in Biology. Biology has relatively few parametric (quantified) relationships among its variables, and many of those that exist are only approximate. Hence, problems in standard setting, when the level of a substance must be determined so that only a specified level of effect will occur, are often difficult to solve. For instance, the concept of a threshold concentration has been used regularly in hazard evaluation and standard setting for a number of years; yet no accepted formulation or set of formulations about the nature of definition of threshold exists. Thus, the determination of a threshold largely depends on empirical processes that rely on an evaluation of the probability that the effect will or will not occur in a randomly chosen human being.
Formulation of Scientific Hypotheses (the Null Hypothesis). In biological studies of the effect of a substance or agent on animals, the common experimental procedure divides the animals into two groups-one group exposed and the other unexposed. Then, for any effect observed or tested for in both groups, it is hypothesized that chance accounts for the differences in magnitude or frequency of occurrence of this effect between the exposed and unexposed groups and appropriate statistical methods are used to test this hypothesis. If the statistical probability indicates that the observed differences could not occur by chance more often than at some predetermined low frequency, the hypothesis is rejected. A residual observation is left: In the presence of substance or agent A, effect B is found. The assumption that agent A causes effect B is then made provisionally, subject to no other cause being found for effect B. When many related cause-effect conclusions of this type are linked by some rational theory of cause, it can be assumed for purposes of evidence that the conclusions are correct. However, when only the results of a single experiment are available, conclusions about cause and effect are generally weak and should be recognized as such in considering evidence.
Proving a Negative. In formulating scientific hypothesis about cause and effect, the basic hypothesis is to assume that there is no cause and effect relationship-in other words, form a negative statement about the effects; then all observed effects can be applied to disprove this negative statement. If the appropriate statistical tests applied to experimental data fail to disprove the statement, it does not thereby prove it. The only statement that can be made is that no effects were found. Hence, it is never possible to prove that no biological effects are produced by a given substance or agent. Regulatory agencies concerned with licensing, such as the FDA, sometimes resolve this dilemma by requiring the substance under consideration to demonstrate consistent negative results under a prescribed protocol of testing, but the agency recognizes that the results fall short of proving absolute safety of the substance.
Inconsistent Results. Not every individual is equally affected by substances or agents in the environment, and it is possible that only a few individuals will be affected by many substances or agents proposed as hazardous. Human beings differ, other factors contribute to effects, and pure chance intervenes. An example of the contribution of other factors is found among uranium miners, who are subject to a marginal increase in lung cancer that presumably results from their inhalation of radioactive material. However, miners who smoke cigarettes are subject to a much greater risk of lung cancer than either miners who do not smoke or smokers in the general population. In considering evidence on the potential hazard of a substance, it is thus necessary to inquire about how many people or what proportion of the population will be affected, what contribution individual behavior makes to the hazard, and whether it is possible to avoid hazardous consequences by identifying and protecting hypersensitive individuals.
Resource and Cost Problems in Biological Studies. Long-term biological testing is extremely expensive. It is estimated that currently each complete assay for carcinogenicity of a chemical under 1978-1979 National Cancer Institute protocols costs about $290,000 and takes about 3 years. Because of the requirements for tests on three or more animal species, long-term testing of drugs and food additives is probably more expensive. Costs aside, the nation's facilities and trained manpower are not and will not be adequate for complete and exhaustive testing of every substance or agent in the environment. very biological test required for a possible hazard results in a diversion of man power and facilities from other tests and scientific studies that might be more important for human welfare. For these reasons, decisions about what constitutes a hazard may have to be made in the absence of complete knowledge or test results, and the consequences of ordering additional testing should be weighed in analyzing the risks versus benefits to the general public.
Scientific Objectivity. Scientists, like others, have personal values, attitudes, beliefs, and goals that are incorporated in the work they report. Scientific "objectivity" applies to the scientists as a group, rather than to individuals. The maintenance of objectivity depends on the existence of numerous scientists who do not depend on political or partisan agencies for support. Fraud in reporting data is rare-and strongly condemned by scientists' ethics. Personal bias, on the other hand, can consciously or unconsciously affect how a scientists designs an experiment, what he observes and what he ignores in an experiment, how he interprets the data, and his belief about the significance of results. In adversary proceedings, the scientist has as much difficulty in maintaining objectivity as anyone else, and the possibility of bias in his testimony must be recognized.
Marino Home Page | Research Interests