Table 2.
Elaboration and evaluation of the external validity of the different potential thresholds defining a relevant improvement in the rheumatoid arthritis impact of disease (RAID) score
| ELABORATION | Time of evaluation during the studyc | EVALUATION | ||||||
|---|---|---|---|---|---|---|---|---|
| Proposed thresholda | Methodological techniqueb | Patient's perspectived | Physician's perspectivee | |||||
| Se | Spe | LR+ (95% CI) | Se | Spe | LR+ (95% CI) | |||
| ABSOLUTE CHANGE | ||||||||
| a. ≥ 0.2 | 75th percentile at week 4 | Week 4 | 89.6 | 54.2 | 2.0 (1.3; 3.0) | 100.0 | 29.6 | 1.4 (1.2; 1.7) |
| a. ≥ 0.2 | 75th percentile at week 4 | Week 12 | 96.1 | 40.9 | 1.6 (1.1; 2.3) | 93.3 | 17.6 | 1.1 (1.0; 1.3) |
| b. ≥ 1 | Empirical/Correct classification at week 4 | Week 4 | 75.3 | 75.0 | 3.0 (1.5; 6.1) | 89.3 | 47.9 | 1.7 (1.3; 2.2) |
| b. ≥ 1 | Empirical/Correct classification at week 4 | Week 12 | 90.8 | 54.5 | 2.0 (1.3; 3.2) | 88.9 | 27.5 | 1.2 (1.0; 1.5) |
| c. ≥ 1.3 | SDC/75th percentile at week 12 | Week 4 | 75.3 | 79.3 | 3.6 (1.6; 8.0) | 85.7 | 47.9 | 1.6 (1.3; 2.2) |
| c. ≥ 1.3 | SDC/75th percentile at week 12 | Week 12 | 88.2 | 59.1 | 2.2 (1.3; 3.6) | 84.4 | 29.4 | 1.2 (1.0; 1.5) |
| d. ≥ 1.6 | ROC at week 4 | Week 4 | 72.7 | 87.5 | 5.8 (2.0; 16.9) | 85.7 | 53.5 | 1.8 (1;4; 2.5) |
| d. ≥ 1.6 | ROC at week 4 | Week 12 | 84.2 | 63.6 | 2.3 (1.3; 4.1) | 82.2 | 33.3 | 1.2 (1.0; 1.6) |
| e. ≥ 2 | Empirical/Correct classification at week 4 | Week 4 | 67.5 | 87.5 | 5.4 (1.9; 15.7) | 78.6 | 56.3 | 1.8 (1.3; 2.5) |
| e. ≥ 2 | Empirical/Correct classification at week 4 | Week 12 | 78.9 | 63.6 | 2.2 (1.2; 3.8) | 77.8 | 37.3 | 1.2 (1.0; 1.6) |
| f. ≥ 2.5 | Correct classification at week 12 | Week 4 | 59.7 | 95.8 | 14.3 (2.1; 98.5) | 71.4 | 64.8 | 2.0 (1.4; 3.0) |
| f. ≥ 2.5 | Correct classification at week 12 | Week 12 | 69.7 | 68.2 | 2.2 (1.2; 4.1) | 77.8 | 52.9 | 1.7 (1.2; 2.3) |
| g. ≥ 3 | ROC at week 12 | Week 4 | 41.6 | 95.8 | 10.0 (1.4; 69.2) | 57.1 | 77.5 | 2.5 (1.5; 4.3) |
| g. ≥ 3 | ROC at week 12 | Week 12 | 63.2 | 90.9 | 6.9 (1.8; 26.3) | 71.1 | 64.7 | 2.0 (1.3; 3.1) |
| RELATIVE CHANGE | ||||||||
| a. ≥ 6% | 75th percentile at week 4 | Week 4 | 88.3 | 54.2 | 1.9 (1.2; 3.0) | 100.0 | 31.0 | 1.4 (1.2; 1.7) |
| a. ≥ 6% | 75th percentile at week 4 | Week 12 | 96.1 | 45.5 | 1.6 (1.2; 2.6) | 93.3 | 19.6 | 1.2 (1.0; 1.4) |
| b. ≥ 17% | ROC at week 4 | Week 4 | 80.5 | 75.0 | 3.2 (1.6; 6.5) | 100.0 | 46.5 | 1.9 (1.5; 2.3) |
| b. ≥ 17% | ROC at week 4 | Week 12 | 94.7 | 59.1 | 2.3 (1.4; 3.8) | 91.1 | 25.5 | 1.2 (1.0; 1.5) |
| c. ≥ 20% | Empirical | Week 4 | 76.6 | 75.0 | 3.1 (1.5; 6.2) | 96.4 | 49.3 | 1.9 (1.5; 2.4) |
| c. ≥ 20% | Empirical | Week 12 | 90.8 | 59.1 | 2.2 (1.3; 3.7) | 88.9 | 27.5 | 1.2 (1.0; 1.5) |
| d. ≥ 25% | Correct classification at week 4/75th percentile at week 12 | Week 4 | 76.6 | 75.0 | 3.1 (1.5; 6.2) | 96.4 | 49.3 | 1.9 (1.5; 2.4) |
| d. ≥ 25% | Correct classification at week 4/75th percentile at with 12 | Week 12 | 88.2 | 63.6 | 2.4 (1.4; 4.2) | 88.9 | 33.3 | 1.3 (1.1; 1.7) |
| e. ≥ 30% | 75th percentile at week 12 | Week 4 | 67.5 | 87.5 | 5.4 (1.9; 15.7) | 92.9 | 62.0 | 2.4 (1.8; 3.3) |
| e. ≥ 30% | 75th percentile at week 12 | Week 12 | 84.2 | 63.6 | 2.3 (1.3; 4.1) | 86.7 | 37.3 | 1.4 (1.1; 1.8) |
| f. ≥ 35% | ROC at week 12 | Week 4 | 63.6 | 91.7 | 7.6 (2.0; 29.1) | 85.7 | 64.8 | 2.4 (1.7; 3..5) |
| f. ≥ 35% | ROC at week 12 | Week 12 | 81.6 | 72.7 | 3.0 (1.5; 6.0) | 86.7 | 43.1 | 1.5 (1.2; 2.0) |
| g. ≥ 42% | Correct classification at week 12 | Week 4 | 59.7 | 95.8 | 14.3 (2.1; 98.5) | 82.1 | 69.0 | 2.7 (1.8; 3.9) |
| g. ≥ 42% | Correct classification at week 12 | Week 12 | 72.4 | 72.7 | 2.7 (1.3; 5.3) | 82.2 | 52.9 | 1.7 (1.3; 2.4) |
| h. ≥ 50% | Empirical | Week 4 | 53.2 | 100.0 | ND | 78.6 | 76.1 | 3.3 (2.1; 5.2) |
| h. ≥ 50% | Empirical | Week 12 | 67.1 | 86.4 | 4.9 (1.7; 14.3) | 75.6 | 60.8 | 1.9 (1.3; 2.8) |
aThe values are those resulting from analysis of the data according to a specific methodology (for example, empiric smallest detectable difference, correct classification probabilities, 75th percentile, ROC curve). bMethodological technique used in the analysis to propose a potential threshold and (in the case of the ROC, 75th percentile and correct classification probabilities) the time point during the study (for example, 4 or 12 weeks after initiation of etanercept therapy) such evaluation has been performed. cThe visit during the study (for example, either 4 or 12 weeks after initiation of etanercept therapy) during which the data collected were used to evaluate the external validity of the proposed thresholds. dPatient's perspective based on the PASS question (X, Y and Z patients answered 'yes' at baseline, week 4 and week 12 respectively: Se, % patients with an absolute RAID score below the proposed cutoff who considered their condition to be acceptable (answering 'yes' to the patient acceptable symptom state (PASS) question) among all patients considering their condition to be acceptable; Spe, % patients with an absolute RAID score above the proposed cutoff who considered their condition to be unacceptable (answering' no' to the PASS question) among all patients considering their condition to be unacceptable. ePhysician's perspective based on the low disease activity defined by the disease activity score-erythrocyte sedimentation rate (DAS-ESR) < 3 (X, Y, and Z patients had a DAS < 3.2 at baseline, week 4 and week 12 respectively): Se, % patients with an absolute RAID score below the proposed cutoff and in low disease activity (for example, DAS28-ESR ≤ 3.2) at week 12 among all patients with low disease activity at week 12: Spe, % patients with an absolute RAID score above the proposed cutoff and with at least moderately active disease (for example, DAS28-ESR ≥ 3.2) at week 12 among all patients with at least moderately active disease at week 12. ROC, receiver operating characteristic; LR+, positive likelihood ratio.