Reporting and concordance of methodologic criteria between abstracts and articles in diagnostic test studies

Abstract

Objective: To evaluate the quality and concordance of methodologic criteria in abstracts versus articles regarding the diagnosis of trichomoniasis.

Study design: Survey of published literature.

Data sources: Studies indexed in MEDLINE (1976-1998).

Study selection: Studies that used culture as the gold or reference standard.

Data extraction: Data from abstract and articles were independently abstracted using 4 methodologic criteria: (1) prospective evaluation of consecutive patients; (2) test results did not influence the decision to do gold standard; (3) independent and blind comparison with gold standard; and (4) broad spectrum of patients used. The total number of criteria met for each report was calculated to create a quality score (0-4).

Measurements and main results: None of the 33 abstracts or full articles reported all 4 criteria. Three criteria were reported in none of the abstracts and in 18% of articles (95% confidence interval [95% CI] 8.6% to 34%). Two criteria were reported in 18% of abstracts (95% CI, 8.6% to 34%) and 42% of articles (95% CI, 27% to 59%). One criterion was reported in 42% of abstracts (95% CI, 27% to 59%) and 27% of articles (95% CI, 15% to 44%). No criteria were reported in 13 (39%) of 33 abstracts (95% CI, 25% to 56%) and 4 (12%) of 33 articles (95% CI, 4.8% to 27%). The agreement of the criteria between the abstract and the article was poor (kappa -0.09; 95% CI, -0.18 to 0) to moderate (kappa 0.53; 95% CI, 0.22 to 0.83).

Conclusions: Information on methods basic to study validity is often absent from both abstract and paper. The concordance of such criteria between the abstract and article needs to improve.