Anchor-based estimation of minimal clinically important difference (MCID)
| Anchor | p-value# | Subjects n | Mean change in DTA score (95% CI) | Effect size | ICC |
| FVC¶ | |||||
| Much worse (≤ −10%) | 45 | 9.1 (7.0–11.3) | 0.75 | ||
| Slightly worse (> −10% to ≤ −5%) | 33 | 3.4 (0.5–6.3) | 0.28 | ||
| Same (> −5% to ≤5%) | <0.0001 | 47 | 1.8 (0.1–3.6) | 0.13 | 0.91 |
| Slightly better (>5% to ≤10%) | 13 | 0.7 (−2.0–3.5) | 0.06 | ||
| Much better (>10%) | 3 | 1.1 (−12.2–14.4) | 0.15 | ||
| DLCO¶ | |||||
| Much worse (≤ −15%) | 47 | 7.6 (5.6–9.6) | 0.53 | ||
| Slightly worse (> −15% to ≤ −10%) | 10 | 5.1 (−1.6–11.8) | 0.48 | ||
| Same (> −10% to ≤10%) | 0.0028 | 66 | 3.1 (1.5–4.7) | 0.28 | 0.84 |
| Slightly better (>10% to ≤15%) | 6 | 3.4 (−10.8–17.6) | 0.21 | ||
| Much better (>15%) | 12 | −1.1 (−5.0–2.8) | −0.09 | ||
| 6MWD m | |||||
| Much worse (≤ −60) | 33 | 7.7 (5.1–10.1) | 0.54 | ||
| Slightly worse (> −60 to ≤ −30) | 25 | 5.3 (3.0–7.5) | 0.40 | ||
| Same (> −30 to ≤30) | 0.0002 | 47 | 2.7 (0.3–5.1) | 0.23 | 0.78 |
| Slightly better (>30 to ≤60) | 18 | 5.1 (1.4–8.8) | 0.53 | ||
| Much better (>60) | 15 | 0.0 (−2.7–2.7) | 0.0 | ||
| SGRQ total score | |||||
| Much worse (>12) | 31 | 7.4 (4.4–10.3) | 0.51 | ||
| Slightly worse (>7 to ≤12) | 24 | 5.4 (2.5–8.2) | 0.45 | ||
| Same (> −7 to ≤7) | 0.0090 | 57 | 3.7 (1.6–5.9) | 0.31 | 0.78 |
| Slightly better (> −12 to ≤ −7) | 11 | 2.2 (−1.1–5.5) | 0.12 | ||
| Much better (≤ −12) | 12 | −0.3 (−3.0–2.5) | −0.24 |
DTA: data-driven texture analysis; ICC: intraclass correlation coefficient; FVC: forced vital capacity; DLCO: diffusing capacity of the lung for carbon monoxide; 6MWD: 6-min walk distance; SGRQ: St George's Respiratory Questionnaire. Mean change in DTA score (95% CI) and effect sizes stratified by changes in other measures. Stratification levels were selected to represent groups that were “much worse”, “slightly worse”, “same”, “slightly better” and “much better”. Cut-off values for each anchor were selected based on published MCIDs. Mean change in DTA in each “slightly worse” group (shown in italics) is an anchor-based estimate of MCID. ICC of baseline and follow-up DTA score in each “same” group is an estimate of reliability. #: one-way ANOVA; ¶: percentage change relative to baseline (follow-up value minus baseline divided by baseline).