Article type
Year
Abstract
Background: Ideally, diagnostic studies should examine a population comprised of symptomatic patients, of which some do and some do not have the disease. The patients receive a reference standard and comparison test, a 2x2 contingency table is derived, and the resulting sensitivities and specificities are calculated. However, some studies include only patients that have tested positive to the reference standard and assess the comparison test in patients who truly have the disease. Consequently, a 2x1 contingency table is generated and only sensitivity is calculated. This can result in spectrum bias and an over-estimated sensitivity. In a recent systematic review examining the diagnosis of occupational asthma, we found results from 2x2 and 2x1 tables to be dissimilar. Among the seven main comparisons, sensitivity from 2x1 tables was either lower or similar to those obtained from 2x2 tables in six instances.
Objectives: To conduct a systematic review of diagnostic systematic reviews to determine if data derived from 2x1 tables is included and if so, whether it is combined with data obtained from 2x2 tables.
Methods: Medline will be searched from 2000-2005 to retrieve meta-analyses of diagnostic tests. Systematic reviews will be included if there is a comparison of two or more diagnostic tests and a pooled sensitivity. The methodological quality will be assessed. The following data will be extracted: clinical topic; review characteristics (e.g., sample size, study designs included); description of reference standard and comparison test(s); inclusion criteria regarding 2x2 and/or 2x1 data and associated rationales; and, pooling methods and rationales (i.e., were the results of 2x1 and 2x2 tables combined?). A descriptive analysis will be performed.
Results: Results will be available at the Cochrane Colloquium.
Conclusions: The findings of this review of diagnostic systematic reviews will contribute to the understanding of spectrum bias and the effect of including data derived from 2x1 tables in diagnostic systematic reviews.
Objectives: To conduct a systematic review of diagnostic systematic reviews to determine if data derived from 2x1 tables is included and if so, whether it is combined with data obtained from 2x2 tables.
Methods: Medline will be searched from 2000-2005 to retrieve meta-analyses of diagnostic tests. Systematic reviews will be included if there is a comparison of two or more diagnostic tests and a pooled sensitivity. The methodological quality will be assessed. The following data will be extracted: clinical topic; review characteristics (e.g., sample size, study designs included); description of reference standard and comparison test(s); inclusion criteria regarding 2x2 and/or 2x1 data and associated rationales; and, pooling methods and rationales (i.e., were the results of 2x1 and 2x2 tables combined?). A descriptive analysis will be performed.
Results: Results will be available at the Cochrane Colloquium.
Conclusions: The findings of this review of diagnostic systematic reviews will contribute to the understanding of spectrum bias and the effect of including data derived from 2x1 tables in diagnostic systematic reviews.