Received on April 4, 2003
Accepted on December 12, 2003

Evidence-based Laboratory Medicine and Test Utilization

Methodology in Diagnostic Laboratory Test Research in Clinical Chemistry and Clinical Chemistry and Laboratory Medicine

¹ Department of Clinic Analysis, General University Hospital of Alicante. Alicante, Spain
² Department of Internal Medicine, General University Hospital of Elche, Elche, Spain
³ Department of Public Health, University of Miguel Hernández, San Juan de Alicante, Spain

^* To whom correspondence should be addressed. E-mail: ihernandez{at}umh.es.

Background: The application of epidemiologic principles to clinical diagnosis has been less developed than in other clinical areas. Knowledge of the main flaws affecting diagnostic laboratory test research is the first step for improving its quality. We assessed the methodologic aspects of articles on laboratory tests.

Methods: We included articles that estimated indexes of diagnostic accuracy (sensitivity and specificity) and were published in Clinical Chemistry or Clinical Chemistry and Laboratory Medicine in 1996, 2001, and 2002. Clinical Chemistry has paid special attention to this field of research since 1996 by publishing recommendations, checklists, and reviews. Articles were identified through electronic searches in Medline. The strategy combined the Mesh term "sensitivity and specificity" (exploded) with the text words "specificity", "false negative", and "accuracy". We examined adherence to seven methodologic criteria used in the study by Reid et al. (JAMA 1995;274:645-51) of papers published in general medical journals. Three observers evaluated each article independently.

Results: Seventy-nine articles fulfilled the inclusion criteria. The percentage of studies that satisfied each criterion improved from 1996 to 2002. Substantial improvement was observed in the reporting of the statistical uncertainty of indices of diagnostic accuracy, in criteria that are based on clinical information from the study population (spectrum composition), and in the avoidance of workup bias. Analytical reproducibility was reported frequently (68%), whereas information about indeterminate results was rarely provided. The mean number of methodologic criteria satisfied showed a statistically significant increase over the 3 years in Clinical Chemistry but not in Clinical Chemistry and Laboratory Medicine.

Conclusions: The methodologic quality of the articles on diagnostic test research published in Clinical Chemistry and Clinical Chemistry and Laboratory Medicine is comparable to the quality observed in the best general medical journals. The methodologic aspects that most need improvement are those linked to the clinical information of the populations studied. The editorial actions aimed to increase the quality of reporting of diagnostic studies could have a relevant positive effect, as shown by the improvement observed in Clinical Chemistry.