Has the Objective Quality of Evidence in Imaging Papers Changed Over the Last 20 Years?

We aimed to determine if both evidence level (EL) as well as clinical efficacy (CE) of imaging manuscripts have changed over the last 20 years.

With our review of medical literature, Institutional Review Board approval was waived, and no informed consent was required. Using Web of Science, we determined the 10 highest impact factor imaging journals. For each journal the 10 most cited and 10 average cited papers were compared for the following years: 1994, 1998, 2002, 2006, 2010, and 2014. EL was graded using the same criteria as the Journal of Bone and Joint Surgery (Wright et al., 2003). CE was graded using the criteria of Thornbury and Fryback (1991). Statistical software R and package lme4 were used to fit mixed regression models with fixed effects for group, year, and a random effect for journal.

EL has improved -0.03 every year on average (P < .001). The more cited papers had better ELs (group effect = -0.23, SE 0.09, P = .011). CE is lower in top cited compared to average cited articles, although the differences were not statistically significant (group effect = -0.14, SE = 0.09, P = .16). CE level increased modestly in both groups over this 20-year time period (0.06 per year, SE = 0.007, P < .001).

Over the last 20 years, imaging journal articles have improved modestly in quality of evidence, as measured by EL and CE.