Investigation of under-ascertainment in epidemiological studies based in general practice

One of the aims of the Study of Infectious Intestinal Disease (IID) in England is to estimate the incidence of IID presenting to general practice. This sub-study aims to estimate and correct the degree of under-ascertainment in the national study.

Cases of presumed IID which presented to general practice in the national study had been ascertained by their GP. In 26 general practices, cases with computerized diagnoses suggestive of IID were identified retrospectively. Cases which fulfilled the case definition of IID and should have been ascertained to the coordinating centre but were not, represented the under-ascertainment. Logistic regression modelling was used to identify independent factors which influenced under-ascertainment.

The records of 2021 patients were examined, 1514 were eligible and should have been ascertained but only 974 (64%) were. There was variation in ascertainment between the practices (30% to 93%). Patient-related factors independently associated with ascertainment were: i) vomiting only as opposed to diarrhoea with and without vomiting (OR 0.37) and ii) consultation in the surgery as opposed to at home (OR 2.18). Practice-related factors independently associated with ascertainment were: i) participation in the enumeration study component (OR 1.78), ii) a larger number of partners (OR 0.3 for 7-8 partners); iii) rural location (OR 2.27) and iv) previous research experience (OR 1.92). Predicted ascertainment percentages were calculated according to practice characteristics.

Under-ascertainment of IID was substantial (36%) and non-random and had to be corrected. Practice characteristics influencing variation in ascertainment were identified and a multivariate model developed to identify adjustment factors which could be applied to individual practices. Researchers need to be aware of factors which influence ascertainment in acute epidemiological studies based in general practice.