Sources of Variation in Indirect Wear Evaluations of Clinical Results

Introduction:  Clinical wear measurement techniques can produce wide-ranging standard deviations (e.g., ±15-90µm) and make detection of small changes for newer composites difficult. While evaluator calibration (level=85%) for USPHS clinical ratings is strongly encouraged, inter and intra-evaluator calibration for indirect wear assessment is not.
 
Objective:  Examine inter-evaluator agreement (consensus) associated with clinical wear assessment using Leinfelder method for three long-term (5-10y) posterior composite trials.
 
Methods:  Three previously reported studies (S1=FulFil, 1981-1991, Dentsply, n=65; S2=Occlusin, 1987-1992, ICI, n=80; S3=SureFil; 1998-2008, Dentsply, n=60) were included. Each monitored wear using Leinfelder method (impressions; casts; 3 evaluators/trial; ratings equal to or between standards [0-46-92-152-221-272-322-352-382-438-493 µm]).  Wear results ranged from 0±15 to 240±73µm.  Numerical ratings for each evaluator-restoration-recall-composite combination (N=3564) were transformed (C=central, H=higher, L=lower) (e.g., 156µm-92µm-92µm=H-C-C=2C) and then assessed for consensus (3C=all agree, 2C= two agree, 1C= none agree) for individual evaluators (e1, e2, e3).  Average consensus (3C+2C+1C) for individual evaluators (E=[e1+e2+e3]/3) were statistically analyzed (1-way ANOVA, p≤0.05, Bonferroni correction).  Restoration averages (R) of evaluator consensus (3C+2C) per restoration were compared to an 85% target.
 
Results:
Surprisingly, R-consensus values showed no statistically significant differences (C1, p=0.44; C2, p=0.42; C3, p=0.22) and  most R-consensus values (15/21=71%) exceeded 85% calibration targets.  H and L variations were almost exclusively limited to a single rating step (3C=0.75; 2C+1C with +1H and/or -1L =0.23). 

Conclusion:  Strong consensus existed among experienced evaluators despite absence of regular calibration. 
 
Acknowledgment: Dentsply-Caulk; ICI.