Department of Preventive Medicine, Seoul National University College of Medicine, Seoul, Korea.
ABSTRACT
It is necessary for medical readers or reviewers to assess critically the methodological and statistical validity of medical articles before accepting their results or conclusion. The authors develope d a validity-assessing checklist of 21 items. Among them 14 items for methodological validity included the followings : clear statement of research hypothesis or specific aims, suitable focus, definition of study population (or subjects), eligibility criteria, exclusion criteria, appropriateness of samples, description of methods in detail, desconclusion of accuracy, description of reliability, presence of control, susceptability bias, performance bias, detection bias, transfer bias. The last 5 items are applicable only to analytic study. And 7 errors to statistical validity : incomplete description of basic data, statistical test performed yet not defined, incomplete description of power or confidence interval, inadequate description of measures of central tendency or dispersion, incorrect analysis, multiplicity on hypothesis testing, unwarranted conclusion.
The first 3 items are 'errors of omission', and the other are 'errors of commission'. The authors suggest the checklist be very helpful, but not perfect. Critical mind is needed, which enables someone to distinguish minor errors from major fallacies.
PDF Links
Full text via DOI
Download Citation 
