Citation bias in imaging research: are studies with higher diagnostic accuracy estimates cited more often?

ID: 

166

Session: 

Poster session 1

Date: 

Sunday 16 September 2018 - 12:30 to 14:00

All authors in correct order:

Frank R1, Dehmoobadsharifabadi A1, Salameh J2, McGrath T1, Kraaijpoel N3, Dang W4, Li N1, Gauthier I1, Wu M4, Bossuyt P5, Levine D6, McInnes M7
1 Department of Radiology, Faculty of Medicine, University of Ottawa, Canada
2 Clinical Epidemiology Program, Ottawa Hospital Research Institute, University of Ottawa, Canada
3 Clinical Epidemiology and Biostatistics and Bioinformatics, Amsterdam Public Health Research Institute, Academic Medical Center, University of Amsterdam, The Netherlands
4 Department of Radiology, The Ottawa Hospital, Canada
5 Clinical Epidemiology and Biostatistics and Bioinformatics, Amsterdam Public Health Research Institute, Academic Medical Center, University of Amsterdam, Canada
6 Department of Radiology, Beth Israel Deaconess Medical Center, and Radiology Editorial Office, USA
7 Department of Radiology, The Ottawa Hospital and University of Ottawa, and Clinical Epidemiology Program, Ottawa Hospital Research Institute, Canada
Presenting author and contact person

Presenting author:

Robert Frank

Contact person:

Robert Frank
Matthew McInnes
Abstract text
Background:
Citation bias (preferential citation of articles with favorable results) inflates the perceived value of certain studies, which can influence the evidence clinicians choose to inform their clinical decisions. Citation bias is well-documented in therapeutic trials, but has not been explored among diagnostic accuracy studies.

Objective:
To evaluate whether imaging diagnostic accuracy studies with higher accuracy estimates are cited more often than those with lower estimates.

Methods:
We searched MEDLINE for diagnostic accuracy meta-analyses published in imaging journals from January 2005 to April 2016. We screened primary studies from the meta-analyses; those that assessed the diagnostic accuracy of an imaging test and reported sensitivity (SN) and specificity (SP) were eligible for inclusion. We excluded studies not indexed in Web of Science, duplicates, and inaccessible articles. We extracted data on study topic (modality/subspecialty), study design, sample size, impact factor, publication date, times cited, highest SN and highest SP for each study. Negative binomial regression analysis was performed to evaluate for association of citation rate (times cited per month since publication) with Youden's index, highest SN, and highest SP, controlling for the potential confounding effects of modality, subspecialty, impact factor, study design, sample size, and source meta-analysis.

Results:
We included 1016 primary studies (Figure 1). Youden's index demonstrated a moderate positive association with citation rate, yielding a regression coefficient of 0.33 (P = 0.016). Regression coefficients for association of SN and SP with citation rate were 0.41 (P = 0.034) and 0.32 (P = 0.15), respectively (Table 1).

Conclusions:
A moderate positive association exists between diagnostic accuracy estimates and citation rates, indicating that citation bias is present in imaging diagnostic accuracy literature.

Patient involvement:
Citation bias in diagnostic accuracy research could lead clinicians to over-estimate the accuracy of imaging tests; this may contribute to patient harm from incorrect interpretation of test results. Identification of this risk is critical in improving the quality of the literature that guides patient care.

Attachments: 

PDF icon Cochrane Abstract Table 1.pdf
PDF icon Cochrane Abstract Figure 1.pdf

Relevance to patients and consumers: 

Due to over-representation of highly cited data, and the common use of citation rate as a proxy for study quality, citation bias in diagnostic accuracy research could drive clinicians to over-estimate the diagnostic accuracy of imaging tests. Over-estimating sensitivity, for instance, can lead to false reassurance and delayed diagnosis of patients' diseases, contributing to adverse health outcomes. By identifying this risk, our study represents an important step toward improving the quality of the literature that guides patient care.