"On the methodology of investigations of etiologic factors in chronic diseases"--further comments.

Iv1 UCH of our knowledge concerning etiologic factors in chronic diseases has been derived from uncontrolled, or poorly controlled, observations. Experimentation on human beings is difficult if not impossible in investigations of causes of diseases and conditions of long duration, especially those requiring strict randomization. Occasionally a set of circumstances occurs which enables the alert investigator to utilize observations in such a manner as to approximate and simulate planned experimentation. The classic and oft-quoted observations on cholera by Snow represent an example of an occurrence of this type. Such an advantageous combination of circumstances, however, is indeed rare. The search for meaningful associations, in the main, will have to be guided by observations of phenomena as they occur, with little, if any, interference on the part of the observer. Progress in the identification of causative factors in the important chronic diseases will depend to a great extent on our ability to utilize such observations and on the development of more rigorous methods and rules of analysis which will increase the likelihood of correct interpretation and will minimize the effect of extraneous factors inherent in uncontrolled observations. The multiple approach, i.e., the investigation of the effect of a specific factor on a disease through a number of different independent types of observations, if each independently indicates the same relationship, reduces the chance that results are due to extraneous factors and increases the significance of the findings, However, the observations must be indeed different in type and not merely different manifestations of the same set of phenomena all subject to the same limi.. tations. Moreover, each observation must be accurate and reliable, and the relationship implied in each of the different types of observation must be valid in itself.