H—3l

A possible source of inaccuracy was the fact that in most cases information was obtained from the housewife, who might be expected to have a better recollection of her own illnesses than of those of the remainder of the household. It is possible, therefore, that the incidence of illnesses amongst adult females might be somewhat exaggerated under this method of inquiry. Y. THE BACKGKOUND OF THE EPIDEMIC : THE EAELY STAGES As has already been explained, during their first visits to test and control areas the investigators collected information about the health of persons in all types of families, including those consisting only of adults. The epidemic itself did not get properly under way until the third week in November, and it was decided, therefore, to separate illnesses in which the onset occurred before Ist November from those commencing later. To simplify an assessment of the results of this part of the inquiry, the populations concerned were reduced to equal numbers. This was done by ignoring the excess persons in each age group, care being taken to ensure that the resulting populations would have the same composition in respect of each type of family.* The following comparisons are therefore concerned with equal test and control populations, composed as follows: —

Considering, then, these equal populations of similar composition, we found that 164 persons in the test areas had a recent history of suspicious illness and 114 in the control areas —i.e., 15 per cent, of the test population and 10-4 per cent, of the controls. These illnesses varied enormously in severity, and their analysis presented a difficult problem. To classify them as either " possibly "or " probably " related to poliomyelitis (the first method which was tried) would be too open to personal bias. It was finally decided that the only way to achieve an objective analysis was to classify them according to the number of suggestive symptoms actually recorded by the investigator. For this purpose — (a) The following were designated "cardinal" symptoms: Fever, vomiting, diarrhoea, headache, sore throat. (b) The following were regarded as suggestive : Pain or stiffness in the neck, abdominal pain, drowiness, delirium. All illnesses recorded were then classified according to the number of " cardinal " symptoms mentioned, but if two or more of the features listed under (b) were also present, the illness was placed in the next category above. All illnesses were further subdivided into those of less than four days' duration and those lasting for four days or more. The results are shown in Fig. IV (next page). Figures will be found in the Appendix, Table 111.

* It was necessary, for example, to reduce the number of adults in the test areas from 732 to 717 in order to make them equal to the total in the control areas. This was done by ignoring the last 2 adults in families without children, the last 10 in families including only pre-school children, and the last 3in families with school-children. This made the number of adults in each type of family the same in both populations.

70

• Total. Type of Family. Adults Only. Only Pre-school Children. With Schoolchildren. Adults School children Pre-school children Persons 717 246 128 295 79 47 343 246 81 1,091 295 126 670