Table A4.
Analysis of the JAMRIS-TMJ scores by multiway ANOVA, where all effects are modeled as random. In the J:P × R × A design (top half), the effects of three factors, patient, aid, and reader on the joint score are analyzed by combining the 1A and 1B datasets under the atlas column, and the 2A and 2B datasets under the tutorial column from the radiologists. In the J:P × R:S design, the effect of patient, reader and specialty are analyzed by combining the 1B and 2A datasets for the atlas row, and 2B and 3 datasets for the tutorial row, with 1A as baseline. Bolded p values are those which remained significant after applying Bonferroni correction for multiple testing.
| J:P × R × A Design | Inflammatory Domain | Damage Domain | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Atlas Radiologists | Atlas Non-Radiologists | Tutorial Radiologists | Atlas Radiologists | Atlas Non-Radiologists | Tutorial Radiologists | |||||||
| Variance Component | Var Comp | p | Var Comp | p | Var Comp | p | Var Comp | p | Var Comp | p | Var Comp | p |
| Joint (patient) | 44% | <0.0001 | 27% | 0.00 | 46% | <0.0001 | 40% | <0.0001 | 24% | 0.00 | 46% | <0.0001 |
| Patient | 20% | 0.05 | 14% | 0.11 | 27% | 0.02 | 38% | 0.00 | 23% | 0.01 | 34% | 0.01 |
| Reader | 8% | 0.03 | 0% | 0.30 | 3% | 0.06 | 0% | 0.91 | 19% | 0.00 | 4% | 0.05 |
| Aid | 0% | 0.41 | 2% | 0.40 | 3% | 0.04 | 1% | 0.19 | 1% | . | 0% | 0.44 |
| Reader*joint (patient) | 4% | <0.0001 | 16% | 0.00 | 1% | 0.16 | 0% | 0.66 | 4% | 0.14 | 0% | 0.32 |
| Aid*joint (patient) | 0% | 0.75 | 5% | 0.09 | 0% | 0.34 | 0% | 0.86 | 2% | 0.29 | 0% | 0.93 |
| Reader*Patient | 2% | 0.13 | 1% | 0.45 | 5% | 0.00 | 0% | 0.43 | 3% | 0.25 | 3% | 0.00 |
| Aid*patient | 0% | 0.63 | 0% | 0.91 | 0% | 0.34 | 1% | 0.19 | 0% | 0.74 | 0% | 0.33 |
| Aid*reader | 2% | 0.00 | 0% | 0.34 | 1% | 0.00 | 1% | 0.00 | 0% | 0.98 | 2% | <0.0001 |
| Aid*reader*patient | 13% | <0.0001 | 18% | 0.00 | 8% | <0.0001 | 6% | <0.0001 | 7% | 0.06 | 4% | <0.0001 |
| Residual | 8% | . | 17% | . | 7% | . | 12% | . | 18% | . | 7% | . |
| J:P × R:S Design | Baseline | Atlas | Atlas+Tutorial | Baseline | Atlas | Atlas+Tutorial | ||||||
| Variance Component | Var Comp | p | Var Comp | p | Var Comp | p | Var Comp | p | Var Comp | p | Var Comp | p |
| Joint (patient) | 38% | <0.0001 | 35% | <0.0001 | 38% | <0.0001 | 34% | <0.0001 | 36% | <0.0001 | 43% | <0.0001 |
| Patient | 11% | 0.13 | 24% | 0.01 | 27% | 0.01 | 35% | 0.00 | 32% | 0.01 | 29% | 0.01 |
| Reader (specialty) | 5% | <0.0001 | 7% | <0.0001 | 6% | <0.0001 | 5% | <0.0001 | 5% | <0.0001 | 4% | <0.0001 |
| Specialty | 9% | 0.07 | 1% | 0.21 | 0% | 0.84 | 0% | 0.79 | 0% | 0.33 | 1% | 0.19 |
| Patient*specialty | 0% | 0.88 | 0% | 0.53 | 1% | 0.31 | 3% | 0.08 | 0% | 0.84 | 1% | 0.15 |
| Patient*reader (specialty) | 19% | <0.0001 | 15% | <0.0001 | 11% | <0.0001 | 9% | <0.0001 | 7% | <0.0001 | 8% | <0.0001 |
| Specialty*joint (patient) | 2% | 0.07 | 5% | <0.0001 | 3% | 0.01 | 2% | 0.05 | 7% | <0.0001 | 2% | 0.00 |
| Residual | 16% | . | 13% | . | 15% | . | 12% | . | 14% | . | 11% | . |