Evaluation of the research methodology in genetic, molecular and proteomic tests

Gac Sanit. 2006 Sep-Oct;20(5):368-73. doi: 10.1016/s0213-9111(06)71522-9.


Introduction: The advances in genomic analysis technologies have conducted to the development of new diagnostic tests in the clinical practice. As well as it happened in other diagnostic fields, the knowledge of the main flaws affecting genetic investigation will facilitate the application of the results.

Methods: We included 44 original articles that evaluate diagnostic exactitude of genetic and molecular tests (including proteomic), published from 2002 to June 2005 in five international publications: JAMA, Lancet, New England Journal of Medicine, Cancer Research and Clinical Cancer Research. We examined adherence to 24 methodological criteria included in the guide STARD (Standards for Reporting of Diagnostic Accuracy.

Results: The mean number of methodological criteria satisfied was 9.8 (95% CI 8.8-10.6); the greater deficiencies were in aspects related to the description of patient's selection 9 (20%), the treatment of indeterminate results 5 (11%) and the determination of test reproducibility 6 (13%). It was observed a high fulfillment in the description of the gold standard 39 (87%) and in the methodology of the test 28 (62%).

Discussion: The methodologic quality of the evaluated articles is lower than the quality observed in other research fields. The methodologic aspects that most need improvement are those linked to the clinical information of the populations studied and the reproducibility of the tests. The research and development of new genetic-molecular technologies requires a better fulfillment of the epidemiological and clinical criteria already followed by other diagnostic fields.

Publication types

  • Evaluation Study
  • Research Support, Non-U.S. Gov't
  • Review

MeSH terms

  • Genetic Testing / statistics & numerical data*
  • Humans
  • Molecular Diagnostic Techniques / statistics & numerical data*
  • Proteomics / statistics & numerical data*
  • Reproducibility of Results