ABSTRACT
OBJECTIVE: 1. To describe the frequency of adherence to methodologic criteria proposed for assessment of diagnostic tests. 2. To identify potential bias. 3. To construct Receiver Operating Characteristic (ROC) curves based on data published on the assessed papers. DESIGN: Descriptive cross-sectional survey. STUDY MATERIAL: All papers listed under diagnosis and immunological headings of amebiasis in the Index Medicus from 1970 to 1988. STUDY UNITS: 80 papers. MEASUREMENT: a: Adherence to methodologic criteria proposed for assessment of diagnostic test. b: presence of potential bias. MAIN RESULTS: The adherence to methodologic criteria varied 1 to 55% of the papers. The comparison with a "gold standard" occurred in 40% the independent "blind" assessment occurred in 6% the setting for the study was described 1%. The study of an appropriate spectrum of disease was done in 11%. We were able to construct ROC curves with data from 37 papers and differences were observed among studies evaluating the same test. A high proportion of papers had potential bias. CONCLUSIONS: The adherence to methodologic standards is poor in the papers analyzed in this series. Consequently the frequency of potential bias was high.
