Time trends in summary estimates from meta-analyses of diagnostic accuracy studies

Article type
Authors
Cohen JF1, Korevaar DA1, Wang J1, Leeflang MM1, Bossuyt PM1
1 Department of Clinical Epidemiology, Biostatistics and Bioinformatics, Academic Medical Center, University of Amsterdam, The Netherlands
Abstract
Background: Summary estimates from meta-analyses of diagnostic accuracy studies can change over time, as evidence accumulates.
Objectives: To evaluate trends in changes over time in summary estimates of diagnostic accuracy.
Methods: We included 54 meta-analyses with at least five primary studies from 42 MEDLINE-indexed systematic reviews published in English between September 2011 and January 2012, reflecting 706 diagnostic accuracy studies, with 338,109 participants in total. Within each meta-analysis, we ranked studies by publication date and applied random-effects cumulative meta-analysis, to follow how summary estimates of diagnostic odds ratio (DOR), sensitivity and specificity would have evolved over time. Time trends were assessed by fitting a weighted linear regression model of the summary accuracy estimate against rank of publication. A negative slope indicates a decrease in summary accuracy estimates over time.
Results: Slopes ranged from -0.80 to 0.56 for DOR, -0.44 to 0.37 for sensitivity and -0.37 to 0.36 for specificity. We found a significant time trend in at least one accuracy measure in 42 (78%, 95% confidence interval (CI) 65% to 87%) meta-analyses. Among a total of 93 significant time trends in DOR, sensitivity or specificity, 62 (67%, 95% CI 57% to 75%) were negative, indicating that summary estimates of accuracy tended to decrease over time.
Conclusions: Changes in summary estimates are relatively frequent in meta-analyses of diagnostic accuracy studies, with estimates based on more studies typically reflecting weaker diagnostic performance. The results from early meta-analyses of diagnostic accuracy studies should be considered with caution before drawing conclusions about the performance of medical tests.