Table 2.

Receiver operating characteristic analyses of fatigue NRS—SPIRIT-P1

Change in fatigue NRS	Sensitivity	Specificity	Positive predictive value	Negative predictive value	Phi coefficient
A: NRS—SPIRIT-P1
−8	0.009	0.995	0.667	0.464	0.022
−7	0.042	0.995	0.900	0.472	0.116
−6	0.111	0.995	0.960	0.491	0.218
−5	0.213	0.973	0.902	0.516	0.279
−4	0.310	0.914	0.807	0.533	0.276
−3	0.454	0.828	0.754	0.566	0.300
−2	0.653	0.634	0.675	0.611	0.287
−1	0.806	0.478	0.642	0.679	0.302
0	0.912	0.290	0.599	0.740	0.262
B: NRS—SPIRIT-P2
−8	0.007	1.000	1.000	0.587	0.063
−7	0.062	1.000	1.000	0.601	0.192
−6	0.123	1.000	1.000	0.617	0.276
−5	0.226	0.966	0.825	0.638	0.298
−4	0.342	0.937	0.794	0.668	0.359
−3	0.425	0.879	0.713	0.683	0.346
−2	0.541	0.723	0.581	0.690	0.268
−1	0.678	0.490	0.485	0.682	0.168
0	0.870	0.291	0.465	0.759	0.190

Receiver operating characteristic analyses suggested –3-point fatigue NRS score improvement is the minimum clinically important difference after 24 weeks.

NRS, numeric rating scale; SPIRIT, Standard Protocol Items: Recommendations for Interventional Trials.