Systematic oserver variation in trachoma studies.

There has been increasing awareness in recent years among trachomatologists, as well as among workers in other fields of medical research, of the frequency and importance of observer variation in epidemiological studies and clinical trials. In trachoma, the lack of simple definitive laboratory diagnostic procedures suitable for wide application has placed the onus largely, and usually exclusively, on clinical observation.The study reported is based on the recorded observations of two skilled ophthalmologists in an epidemiological survey covering more than 35 000 persons in Taiwan. Observer differences were found to lie not only in deciding whether a case is trachomatous or not but also in assigning cases diagnosed as trachoma to the appropriate evolutive stage of the WHO trachoma classification.THE CONCLUSIONS REACHED ARE THAT: (a) inter- and intra-observer variations of some degree are inevitable if dependence is placed on clinical examination alone; (b) it is possible by preliminary testing of observers' interpretation of clinical signs to determine the nature of these differences, to assess their importance, and to reduce them; also to set base-lines for the detection of subsequent divergences over time; and (c) it is better to have two observers than one in any trachoma survey or clinical trial.