[Research on diagnostic tests in Medicina Clinica. A methodological assessment]

Research on diagnostic tests have not reached the same methodological rigour as other areas of clinical research. Identification of more frequent and important methodological flaws could contribute to raise the quality of diagnostic test studies. The aim of this study was to evaluate the methodology of diagnostic test studies published in MEDICINA CLINICA.

Forty two articles about diagnostic test studies published in MEDICINA CLINICA from 1992 through 1995 were selected. Two set of methodological criteria were applied (JAMA 1995; 274: 645-651 and Revisiones en Salud Pública 1993; 3: 243-262). Each article was independently evaluated by two observers.

Of the 42 articles evaluated, 33 determined sensitivity and specificity, and 9 only sensitivity. There was a well-defined gold-standard in the majority of the studies (90%) but it was not applied to all patients in 21% of them. Description of the diagnostic test evaluated was present in 79% and test reproducibility was assessed in only 19%. Source of patients, eligibility criteria for study subjects and spectrum composition were considered in 31, 36 and 31%, respectively. Avoidance of work-up bias and review bias was done in 69 and 36% of the articles. Only 12% considered indeterminate results, 50% reported test indexes for relevant clinical subgroups and statistical precision was provided in 17% of the studies.

Methodological quality of the research on diagnostic tests published in MEDICINA CLINICA is similar to that observed in the best world clinical journals. Nevertheless, there are opportunities to improve several aspects of study design and presentation that will facilitate its clinical applicability and general utility.