J E Heffner1, D Feinstein, C Barbieri. 1. Department of Medicine, University of Arizona Health Sciences Center, St. Joseph's Hospital and Medical Center, Phoenix, USA.
Abstract
STUDY OBJECTIVES: To determine conformance with methodologic standards in the evaluation of diagnostic tests. DATA SOURCES: MEDLINE database search (1992 to 1997) of nine prominent general medicine and six subspecialty journals for articles that report discriminative properties of diagnostic tests in pulmonary medicine. STUDY SELECTION: Articles were eligible if they reported discriminative properties of diagnostic tests in humans, diagnostic tests were intended for the detection of existing conditions, and the target disorder was relevant to pulmonary medicine. DATA EXTRACTION: Each study was critically reviewed independently by two observers. DATA SYNTHESIS: Of the 1,029 retrieved articles, 41 met study inclusion criteria. The median number of the 12 major standards for design fulfilled by study articles was 6 (range, 1 to 12, 25th to 75th percentile, 5.0 to 8.5) and only 2 articles fulfilled all 12 standards. Seven (17%) articles did not report any standard measures of diagnostic accuracy and 7 (17%) provided data only for sensitivity and specificity. Only 4 of 17 articles (24%) that compared different tests used standard statistical methods. CONCLUSION: These results indicate that greater methodologic rigor is needed for studies that evaluate diagnostic tests in pulmonary medicine. Existing deficiencies in methodology risk the introduction of invalid tests into clinical practice.
STUDY OBJECTIVES: To determine conformance with methodologic standards in the evaluation of diagnostic tests. DATA SOURCES: MEDLINE database search (1992 to 1997) of nine prominent general medicine and six subspecialty journals for articles that report discriminative properties of diagnostic tests in pulmonary medicine. STUDY SELECTION: Articles were eligible if they reported discriminative properties of diagnostic tests in humans, diagnostic tests were intended for the detection of existing conditions, and the target disorder was relevant to pulmonary medicine. DATA EXTRACTION: Each study was critically reviewed independently by two observers. DATA SYNTHESIS: Of the 1,029 retrieved articles, 41 met study inclusion criteria. The median number of the 12 major standards for design fulfilled by study articles was 6 (range, 1 to 12, 25th to 75th percentile, 5.0 to 8.5) and only 2 articles fulfilled all 12 standards. Seven (17%) articles did not report any standard measures of diagnostic accuracy and 7 (17%) provided data only for sensitivity and specificity. Only 4 of 17 articles (24%) that compared different tests used standard statistical methods. CONCLUSION: These results indicate that greater methodologic rigor is needed for studies that evaluate diagnostic tests in pulmonary medicine. Existing deficiencies in methodology risk the introduction of invalid tests into clinical practice.