Table 3.
MCID based on the One SEM Criterion Using Physician-rated Change in cSLE as External Standard †.
| Disease activity measure | ICC‡ | MCID Definition¶ | MCID for‘Clinically Important Improvement’ | MCID for ‘Clinically Important Worsening’ | Detection rate* for ‘Clinically Important Improvement’ | Detection rate for ‘No Change’ | Detection rate for‘Clinically Important Worsening’ | Overall Correct Detection Rate of ALL cSLE Courses |
|---|---|---|---|---|---|---|---|---|
| VASMD | 0.76 | 63%CI of SEM | − 0.7 | + 0.7 | 71% | 61% | 65% | 63% |
| 90%CI of SEM | − 1.2 | + 1.2 | 29% | 90% | 35% | 71% | ||
| ECLAM | 0.47 | 63%CI of SEM | − 0.9 | + 0.9 | 49% | 51% | 55% | 52% |
| 90%CI of SEM | − 1.5 | + 1.5 | 38% | 75% | 39% | 63% | ||
| SLEDAI | 0.63 | 63%CI of SEM | − 1.9 | + 1.9 | 58% | 52% | 45% | 52% |
| 90%CI of SEM | − 3.1 | + 3.1 | 31% | 77% | 33% | 62% | ||
| SLAM | 0.57 | 63%CI of SEM | − 3.8 | + 3.8 | 29% | 74% | 36% | 60% |
| 90%CI of SEM | − 6.3 | + 6.3 | 18% | 84% | 17% | 62% | ||
| BILAGLiang | 0.62 | 63%CI of SEM | − 4.2 | + 4.2 | 40% | 72% | 40% | 61% |
| 90%CI of SEM | − 6.9 | + 6.9 | 30% | 86% | 21% | 66% | ||
| BILAGGladman | 0.77 | 63%CI of SEM | − 1.4 | + 1.4 | 40% | 70% | 40% | 60% |
| 90%CI of SEM | − 2.3 | + 2.3 | 30% | 84% | 29% | 66% | ||
| BILAGStoll | 0.58 | 63%CI of SEM | − 2.0 | + 2.0 | 51% | 65% | 49% | 60% |
| 90%CI of SEM | − 3.3 | + 3.3 | 34% | 86% | 27% | 67% | ||
ICC: intraclass correlation coefficient. The ICC presented considers random effects in mixed effect models (Shrout PE, Fleiss JL.; Psychol Bulletin 1979;86:420-427). High ICC values (range 0 to 1) indicate a high level of chance-corrected agreement of the changes of disease activity scores among patients with stable disease course (or unchanged disease) between consecutive visits.
SEM: Standard error of measurement, SEM= square root of WMS, with WMS defined as ‘within-subject variability or within-subject mean square error’ (Wyrwich KW, Tierney WM, Wolinsky FD; J Clin Epidemiology 1999; 861-873). The 1-SEM-Criterion assumes that the MCID for worsening and improvement have the same absolute values. Besides the 1-SEM-Criterion, i.e. the 63% confidence interval (CI) at ±1 SEM around the mean change of the scores of disease activity measures in of patients with stable disease, an alternative 90% CI at ±1.645 SEM is shown that more accurately detects patients with stable disease course.
The detection rate for improved is defined as the number of improved patients as per the MCID value considered divided by the number of the truly improved patients as per the criterion standard (physician assessment of change in disease). The detection rates for worsening and unchanged disease courses were calculated accordingly.