Reporting bias in imaging diagnostic accuracy studies: Are studies with positive conclusions or titles submitted and published faster?

Tags: Oral
Treanor L1, Frank R1, Atyani A1, Dehmoobad Sharifabadi A1, Hallgrimson Z1, Fabiano N1, Salameh J1, McGrath T1, Korevaar D2, Bossuyt P3, McInnes M1
1Department of Radiology, University of Ottawa, 2Department of Respiratory Medicine, University of Amsterdam, 3Clinical Epidemiology and Biostatistics and Bioinformatics, University of Amsterdam

Background: Time-lag bias is a form of reporting bias resulting from the delayed publication of studies with ‘negative’ findings. If imaging diagnostic test accuracy (DTA) studies with negative conclusions have a delayed publication course, this may impede the ability to evaluate all evidence on a topic (e.g. in a systematic review)—potentially leading to the misallocation of healthcare resources towards diagnostic tests and detrimental downstream effects on patient outcomes.

Objective: To evaluate whether imaging DTA studies with positive conclusions or titles have a shorter time to publication than those with non-positive (i.e. negative or neutral) conclusions or titles.

Methods: We included primary imaging DTA studies from systematic reviews published in 2015. Conclusion and title positivity were extracted independently in duplicate for each study, based on a previously published classification scheme. Time from study completion to publication was extracted and calculated. A Cox regression model was used to evaluate associations of conclusion and title positivity with time to publication, adjusting for potentially confounding variables.

Results: 774 imaging DTA studies were included; time from study completion to publication could be calculated for 516 studies. Median time from completion to publication was 18 months (interquartile range [IQR]:13-26; 413 studies) for studies with positive conclusions, 23 months (IQR:16-33; 63 studies) for those with neutral conclusions, and 25 months (IQR:15-38; 40 studies) for those with negative conclusions (as seen in Table 1). Conclusion positivity was associated with a shorter time from completion to publication for studies with positive conclusions, compared to those with non-positive conclusions (hazard ratio [HR] 1.31; 95% confidence interval [CI]: 1.02-1.68), as seen in Figure 1. Positive titles were not significantly associated with a shorter time to publication (HR 1.12; 95% confidence interval: 0.75-1.69). When subdividing time from study completion to publication, conclusion positivity was associated with shorter time from study completion to submission (HR 1.31; 95% CI:1.00-1.74), but no association was identified between conclusion positivity and time from submission to publication (HR 0.95; 95% CI: 0.61-1.48).

Conclusions: Positive author conclusions (but not titles) were associated with a shorter time to publication. Imaging DTA studies with positive conclusions may be over-represented in the literature, potentially leading to overly-optimistic perceptions of the performance of diagnostic imaging tests.

Patient or healthcare consumer involvement: There were no patients or healthcare consumers involved in this project.