What is epidemiology?
This is the study of a disease within a population. It determines how and why diseases may occur within specific populations, and is central to the understanding of disease mechanisms, to theories of possible risk factors and causes, to development of public health strategies, and to the prevention and treatment of disease.
Types of epidemiology
Epidemiological studies can be described as descriptive, analytical, or experimental. Descriptive epidemiological studies are "The first stage of epidemiologic investigation...[focussing] on describing disease distribution by characteristics relating to time, place, and person." (Mosby's Medical Dictionary, 8th edition. © 2009, Elsevier.). Therefore descriptive epidemiology establishes where relationships may exist between a disease and the characteristics of a population (sex, age, ethnicity), how the incidence and mortality of the disease may change over time, and may help to identify where syndromes or risk factors may exist.
Analytical epidemiology is the use of statistical and laboratory methods to analyse the relationship between factors and diseases, in order to develop theories for cause-effect or pathogenesis. Lastly, experimental epidemiology may be synonymous with clinical trials. Often experimental epidemiological studies are used to determine whether the theory for a cause-effect relationship is supported y evidence. Statistics are used to compile epidemiological information about populations from a sample.
The Bradford Hill criteria
These were proposed by Austin Bradford Hill in 1965, and although not intended for this use, are often referenced when attempting to determine whether a relationship is causal or simply consequential. (Hill, A.B. (1965). "The environment and disease: association or causation?". Proceedings of the Royal Society of Medicine 58: 295–300.)
Hill used the example of Potts to illustrate his point. In the 18th century, Potts established that a great number of men with scrotal cancer had histories that included an occupation as a chimney sweep. Due to the overwhelming proportion of cases of scrotal cancer patients with this history, Potts concluded that there was a link between substances encountered in chimney sweeping and development of scrotal cancer. Indeed, later chimney soot was found to contain tumorigenic agents. However, Hill cautions that the existence of a slight association does not necessarily suggest a weak relationship between factor and disease.
If the results have been repeated by different people, in different locations, using different samples, then the relationship established is likely to be causal.
If, for example, a specific disease is found in workers at a particular location (i.e. a small population), and there appears to be no particular link between the population characteristic and other unrelated diseases, it is likely that the relationship is causal.
This is the 'chicken and egg' question. Was the factor causal, or did it develop as a consequence of the disease? Are people who are more likely to have the disease also more likely to have a specific occupation?
5. Biological gradient
This is best demonstrated by a dose-response curve. Hill uses the illustration of smoking and lung cancer. The incidence of lung cancer is linearly correlated to the number of cigarettes smoked per day; consequently this is a strong indication that cigarettes are linked to the development of lung cancer.
It is more likely that the theory regarding a relationship is true if the cause-effect can be explained by a biological understanding of disease mechanisms. However, Hill states that this should not be a necessary criteria, as current knowledge of biological mechanisms may be inadequate to understand the biological basis of the cause-effect relationship.
The data and theory should not particularly disagree with previous accepted knowledge concerning the biological mechanisms of the disease.
This is where clinical trials become useful; does a drug expected to prevent the disease actually prevent the disease? This would indicate that theories explaining the trend between factor and disease are true.
This is the consideration of the effects of similar factors on the contraction and course of the disease.
Types of epidemiological study
These can be divided into several different types of study. The first is the descriptive analysis of the experience of a single patient; whilst these cannot indicate the experiences of a population as a whole, they can provide opportunity for the development of new theories. Alternatively, the study may analyse a group of patients with the same condition. Lastly, the study may be a self-controlled case series study, in which patients are subjected to peroids of exposure and non-exposure.
Case control studies
The recruited participants are either part of the 'case' group (i.e. they have the specific disease) or the 'control' group (i.e. they are healthy and disease negative). A comparison between the two groups' potential exposure to possible causative factors is used to detemine links between factors and disease.
Here participants are selected based on their exposure; for example, heavy smokers, light smokers, and non-smokers. These normally take more time to undertake than case control studies.