Investigation of Causation
Cross-Sectional, Case-Control, and Cohort Studies
Epidemiologic methods used to determine the incidence and prevalence of infectious diseases, the relationships between cause and effect, and the evaluation of risk factors include the cross-sectional study, the case-control study, and the cohort study. A cross-sectional study can be carried out relatively quickly and provides data on the prevalence of particular diseases in the population. A case-control study, the most common kind of investigation, starts after the disease has occurred and attempts to identify the cause; it is thus a retrospective study, going back in time to determine causative events. Although it does not require the creation of new data or records, a case-control study does require careful selection of the control group, matched to the test group so as to avoid bias. The advantages of the retrospective study are that it lends itself to quick analysis, is relatively inexpensive to carry out, and is the only practicable method of investigating rare occurrences.
In cohort studies, also called prospective studies, investigation starts with a presumed cause, and a population exposed to the causative virus is followed into the future to identify correlated resulting effects This type of study requires the creation of new data and records, and careful selection of the control group to be as similar as possible to the exposed group, except for the absence of contact with the presumed causative virus. It does not lend itself to quick results as groups must be followed until disease is observed, which makes such studies expensive. However, when cohort studies are successful, proof of the cause-effect relationship is often incontrovertible.
The discovery of the causation of congenital defects by rubella virus provides examples of both retrospective and prospective studies. Dr. N. M. Gregg, an ophthalmologist working in Sydney, Australia, was struck by the large number of cases of congenital cataract he saw in 1940-1941, and by the fact that many of the children also had cardiac defects. By interviewing the mothers he found that the great majority of them had experienced rubella early in the related pregnancy. His hypothesis that there was a causative relation between maternal rubella and congenital defects quickly received support from other retrospective studies, and prospective studies were then organized. Groups of pregnant women were sought who had experienced an acute exanthematous disease during pregnancy, and the subsequent occurrence of congenital defects in their children was compared with that in women who had not experienced such infections. Gregg's predictions were confirmed and the parameters defined more precisely.
Epidemiologic aspects of several specifically human diseases that have not been reproduced in other animals have been studied in human volunteers; for example, early work with yellow fever, hepatitis viruses, rhino-viruses, and a range of other respiratory viruses involved human volunteer studies. Many major discoveries that have led to the control of viral diseases were possible only with the use of human volunteers. An absolute requirement is that the investigators obtain informed consent from the subjects or, in the case of minors, from their parents. It is essential in such work that careful consideration be given to any short- or long-term risks that may be involved, including the possibility of transferring other agents that may be present in the inoculum as contaminants.
Proving a Causal Relationship between Virus and Disease
One of the great landmarks in the scientific study ol infectious diseases was the development of the Henle-Koch postulates of causation. They were originally drawn up with bacteria and protozoa in mind, not viruses, and were revised in 1937 by Rivers, who developed another set of criteria to cover viruses. With the advent of tissue culture for viral diagnosis many new viruses were discovered—"viruses in search of disease"—and Huebner further revised the Koch and Rivers postulates. Later the problem arose of determining whether viruses were causally involved in various chronic diseases and in cancers, a question that is still of major concern to medical virologists. Because the relevant disease cannot be reproduced by inoculation of experimental animals, scientists have to evaluate the probability of "guilt by association," a difficult procedure that relies on an epidemiologic approach. Two tools that are of central importance in assisting the epidemiologist are immunologic investigations and demonstration of the presence of the viral genome in tumor cells by the use of nucleic acid probes or the polymerase chain reaction.
The immunologic criteria were first formulated by Evans in an assessment of the relationship of Epstein-Barr (EB) virus to infectious mononucleosis, at a time when there was no method of isolation of the virus. Subsequently the same approach was applied to investigating the role of EB virus in Burkitt's lymphoma. A large prospective study carried out on 45,000 children in an area of high incidence of Burkitt's lymphoma in Africa showed that (1) EB jyrusJnfection preceded development of the tumors by 7-54 months; (2) exceptionally high EB virus antibody titers often preceded the appearance of tumors; and (3) antibody titers to other viruses were not elevated. In addition, it was demonstrated that the EB virus genome is always present in the cells of Burkitt's lymphomas in African children, and that a malignant lymphoma can be induced in certain primates with EB virus or EB virus-infected lymphocytes (see Chapter 11). There is thus a very strong causal association between EB virus and Burkitt's lymphoma in African children.
The immunogenicity, potency, safety, and efficacy of vaccines are first studied in laboratory animals, followed by small-scale closed trials in normal adults to determine safety (Phase 1 trials) and the immune response (Phase 2 trials), followed by large-scale field trials in a place where efficacy can be tested (Phase 3 trials). Sometimes these trials can be combined. The Phase 3 trials employ epidemiologic methods to analyze the results, rather like those of the cohort study just described. Perhaps the best-known vaccine trial was the famous "Francis field trial" of the Salk inactivated poliovirus vaccine, carried out in 1954, but similar trials were necessary for live-virus poliovirus vaccines, for yellow fever vaccines before this, and for measles, mumps, and rubella vaccines since. There is no alternative way to evaluate new vaccines, and the design of field trials has now been developed so that they yield maximum information with minimum risk and cost. Even with this system, however, a serious problem may be recognized only after a vaccine has passed into commercial use. Similar considerations apply to trials of new antiviral chemo-therapeutic agents (see Chapter 16).
Another kind of epidemiologic investigation that can provide etiologic information and data on the value of vaccines or therapeutic agents is the long-term study of a closed population, such as "Junior Village" in Washing ton, or the long-term family studies that have been carried out in several cities in the United States. Because of the present advanced state of diagnostic virology, such studies now yield a much greater array of valuable data than was possible a few years ago, but they are very expensive and require long-term dedication of both personnel and money. Conducted with careful attention to the ethical problems involved, these studies have greatly augmented our understanding of viral infections and accelerated progress toward the control of several of them. When used for the estimation of the value of vaccines or therapeutic agents, long-term population studies have the exceptional advantage that they include all of the variables occurring in a natural population.
Population studies may also be used for the early detection of the first appearance of virus in a region. Such sentinel studies are widely used for assessing the seasonal prevalence of arbovirus infections; for example, sentinel chickens are used for the early detection of eastern equine encephalitis and St. Louis encephalitis viruses in the southern United States and for the detection of Murray Valley encephalitis virus in Australia.
From the time of William Farr, who studied infectious diseases in the 1840s, mathematicians have been interested in "epidemic curves" and secular trends in the incidence of infectious diseases. With the development of mathematical modeling using the computer there has been a resurgence of interest in the dynamics of infectious diseases within populations. Since modeling involves predictions about future occurrences of diseases, models carry a degree of uncertainty; it is sometimes said that "for every model there is an equal and opposite model." However, models are being developed to predict patterns of disease transmission, critical population sizes required to support the continuous transmission of viruses with short and long incubation periods, the dynamics of endemicity of viruses that establish persistent infection, and the important variables in age-dependent viral pathogenicity.
Computer modeling has provided useful insights into the effects of different vaccination regimes and different levels of acceptance of vaccination against rubella and measles on important complications of these two diseases, congenital rubella and subacute sclerosing panencephalitis, respectively. One effect of vaccination programs is to raise the average age at which unim-munized individuals contract the infections against which that vaccine protects. Up to a certain level, increasing herd immunity may actually increase the risk in especially vulnerable (older) individuals, by increasing the age of first exposure to the virus. Thus, unless the acceptance rate is high, vaccination of infants against rubella may paradoxically increase the risk of the most important complication of the natural disease, namely, congenital rubella. A second conclusion from these studies is that because of the cyclic incidence of the viral exanthemata, with intervals between peaks that increase as vaccination coverage improves, the evaluation of vaccination programs that stop short of countrywide elimination must be carried out over a prolonged period of time.
Was this article helpful?