Magnification Effects on Visual Ratings by Clinical Research Investigators

Click on images to view full size.

Introduction: An open-access online calibration system for clinical research investigators (http://www.dent.umich.edu/CER/) was launched in 2006.  A vexing question persisted about potential on-screen magnification effects on visual rating assignments.    Objective:  Evaluate magnification effects on visual ratings of 5 USPHS categories by 5 trained clinical evaluators using color images of restorations.      Methods:  Clinical evaluators (full-time calibrated clinical research faculty; n=5) were challenged with visual ratings (A= clinically excellent; B= clinically acceptable; C= clinically unacceptable) for 6 color printed images for each of 5 USPHS categories at 3 magnifications (1.0x [none, see attachment], 3.4x [loupes], 6.4x [large computer-screen image]) with 24h elapsed times between magnification tests.  Results (N=30x3x5) were analyzed in terms of (1) evaluator agreement with historical ratings as “shifts/directions” (e.g., A-to-B= +1, C-to-B= -1) and (2) pooled agreement (evaluators and categories) among magnifications (ANOVA, p≤0.05).                                       Evaluator Agreement (n=30/category)  Pooled (n=180)  Agreement (%)                                            Pooled Rating Shifts (X±sd) Mag  Anatomic  Form  Color  Match  Margin Dis-  coloration  Marginal  Integrity  Surface  Texture  All USPHS  Categories 1.0x  –0.07±0.45a  +0.13±0.35a –0.37±0.56a –0.07±0.69a –0.30±0.53a  70±10A 3.3x  –0.13±0.35a  +0.07±0.52a –0.10±0.61a –0.03±0.67a +0.13±0.35a  75±11B 6.4x  –0.07±0.37a  +0.23±0.43a +0.10±0.55a +0.17±0.46a +0.20±0.41a  78±15BC Results:  No differences arose for evaluators (not shown) or categories (see small-letter superscripts across rows), so all USPHS values were pooled.  Minor shift ratings occurred among categories at each magnification but were not statistically different (p>0.05).  Pooled USPHS results were small but different (p=0.02, see capital-letter superscripts).  The largest gain in agreement between different magnifications was achieved with loupes (3.3X).   Discussion:  The principal advantage for clinical research is that ratings are only “marginally influenced” by magnification.  This facilitates clinical research in large networks where clinicians vary in trusted magnification level.   Conclusion:  Within boundaries of trained/calibrated clinical faculty for 5 USPHS categories selected, only minor magnification effects occurred.    Acknowledgment:  Supported in part by CRLT Whitaker Grant (U-M) and Dentsply-Caulk.