Epidemiology is the study of the relationship between possible determining causes and the distribution of the frequency of diseases in human populations. Epidemiological research has been one of the mainstays of our understanding of nutrition.

There are three objectives for epidemiology:

• to describe the distribution and extent of disease problems in human populations
• to elucidate the aetiology of diseases
• to provide the information necessary to manage and plan services for the prevention, control and treatment of disease

What is being studied must be carefully defined and the interpretation of the results of the study be within the framework of the original definition.

Nutritional epidemiology, in addition to the study of the nutritional determinants of disease, measures nutritional status in relation to environment and dietary intake. Most clinical measurements of nutritional status are used to identify deficiencies in overall intake or intake of individual nutrients.

Historical nutritional studies

The great nutritional studies of the 19th and early part of the 20th century led to the discovery of dietary deficiencies in the form of beri-beri, goitre, pellagra, rickets, scurvy and xerophthalmia. These spectacular discoveries led to the concept of single dietary factors in the aetiology of disease. There is an understandable desire to identify simple and singular causes of contemporary diseases. However, it is now apparent that the system is complex, with groups vulnerable to multiple other predisposing factors in addition to nutritional factors, for example infection, activity levels and individual genetic variation. Furthermore, only a proportion of individuals who are at risk develop the particular nutritional problems under study.

Reaching rapid conclusions on the basis of epidemiological associations all too frequently mars the reputation of nutrition as a science, and delays identification of processes. Many recorded associations may not be causative of the condition; instead, they are markers for the net effect of many variables that influence morbidity and mortality ( post hoc, non ergo propter hoc : after this, not therefore on account of this, the tendency to confuse sequence with consequences)

Epidemiological research methodology

A systematic approach to epidemiological research involves:

• identifying the area of study
• reviewing the literature on the topic

A thorough read of a comprehensive review by an acknowledged expert and discussion with trusted colleagues is often sufficient. The detailed reading is best devoted to methodology and statistical support.

• developing a specific study question
• writing a research protocol
• consultation with a statistician

Intelligent and understanding professional statistical advice is essential. It is mandatory that this advice be taken before the trial rather than as an after-thought or in response to the comments of the editor of a journal.

• ethic’s committee approval

The completed study protocol must be reviewed by an ethics committee. This committee must be expert but also include independent members.

• completing the research described by the protocol
• analysis of the data with emphasis on (i) problems with the protocol; and (ii) statistics
• correct interpretation of the results examining the relationship between exposure and outcome

Discussion with the statistician and the ethics committee should be regarded as one of the most rewarding parts of the whole study.

The best epidemiological studies include large numbers of variables carefully and independently measured. Examples of such studies are the Framingham study; the British Cancer Register; The British doctors smoking study, and the US Nurses Study.

Variables which are easy to identify but are complicated in interpretation are: smoking; gender, alcohol intake; social class; coffee consumption, and racial and national differences.

Epidemiological studies can be seen as being experimental or observational. In experimental investigations a particular set of conditions are established for subjects by the investigator. In observational studies the investigator observes and measures but has no control over the actions of the subjects.

Examples of experimental studies are community trials and clinical or field trials. Community trials use populations whose disease status is unknown but most of whom will be healthy, that is, therapeutic trials, or studies of subjects who are known to be free from the disease at the time of the study (prevention studies). Clinical or field trials work with subjects who already have a disease In both instances, the general design is to divide randomly the population into a treatment or exposure group and a control group and to compare the effect of a treatment to a control regime given at the same time under the same conditions. This design enables the effects of the treatment to be established. It is important that there is perfect randomisation between the control and the treatment groups. Both populations should be studied for an equivalent and biologically relevant length of time. The observer and the subject should be blind to the treatment regime but it is important that there is constant contact, warmth and support with the subjects during the study and when relevant to watch and test whether there is compliance to the regime in question. Subjects do get bored and silently alter the protocol.

1. Epidemiology is a study of the relationship between possible determining factors and the distribution of the frequency of disease in human populations. Nutritional epidemiology requires the measurement of nutritional status in relation to environment and dietary intake.

2. The great epidemiology studies led to the discovery of beri-beri, goitre, pellagra, rickets, scurvy and xerophthalmia.

3. The science of epidemiology demands a precise identification of the area of study, a specific study questionnaire, a research protocol, and effective analysis and interpretation of the resulting data.

4. Epidemiological studies can be experimental or observational. In experimental investigations a particular set of conditions are established for subjects by the investigator. In observational studies, the investigator observes and measures but has no control of the actions of the subjects. Studies may be community trials, or clinical or field trials.

5. Sampling is required to obtain a manageable number of individuals who are truly representative of the population to be studied. A system of random sampling is used.

6. An outcome must be measured with appropriate validity, sensitivity and specificity. It is also important to evaluate the role of chance and also to test a hypothesis. Bradford Hill has identified nine points which indicate causality between exposure and outcome.

7. Meta-analysis is a precise statistical tool in which the results of all relevant trials, even those which are small and may have given inconsistent results, can be analysed together.

Berry DA ( 1996) Statisistics, a Bayesian perspective, CA: Duxbury Press

Bland JM and Altman DG ( 1996) Measurement error and correlation coefficients. British Medical Journal 313, 41-2

Britten N ( 1995 ) Qualitative interviews in medical research . British Medical Journal 311, 251-3.

Clayton D and McKeigue PM ( 2001) Epidemiological methods for studying genes and environmental factors in complex diseases. Lancet 358, 1356-60

Doll, R. (1994) The use of meta-analysis in epidemiology. Diet and cancer of the breast and colon. Nutrition Reviews, 52, 233–7.

Editorial (1990) Volunteering for research. Lancet, 340, 823–4.

Egger M, Davey-Smith G ( 1997) Meta-analysis , potentials and promise British Medical Journal 315, 1371-4

Edwards SJL, Braunholz DA, Lilford RJ , Stevens AJ ( 1999) Ethical issues in the design and conduct of cluster controlled trials British Medical Journal 318, 1407-9.

Hawkes CH (1997 ) Twin studies in medicine – what do they tell us? Q.J. Medicine 90, 311-321

Higginson IJ and Carr AJ ( 2001) Using quality of life measures in the clinical setting . British Medical Journal. 322, 1297-1300

Horgan GW ( 2001) Statistical analysis of nutritional studies. British Journal of Nutrition. 86, 141-144.

Margetts, B.M. and Nelson, M. (1997) Design Concepts in Nutritional Epidemiology. Oxford Medical Publications, 2nd edition, Oxford University Press, Oxford.

Marr JW ( 1971) Individual dietary surveys : purposes and methods. World review of nutrition and dietetics. 13, 105-164

Neiderhiser JM ( 2001) Understanding the roles of genome and envirome : methods in genetic epidemiology. British Journal of Psychiatry 178 ( suppl 40) s12-17.

Wright P, Jansen C and Wyatt JC ( 1998 ) How to limit clinical errors in interpretation of data. Lancet 352 1539-43