2.1 Convergent Validity: was assessed with the Disability (ODI), Quality of Life (EQ-5D-5L), Pain (NRS), and Function (PSFS) using correlation analyses. Moreover, we conducted relevant analyses for each sub-scale. We set a-priori hypotheses for the correlation analysis (Table 1), and then we conducted hypothesis testing.
2.2 Responsiveness: we used 3 approaches: 1) distribution-based with effect sizes (ES) and a t-test; 2) construct with correlations and hypothesis testing; 3) criterion with ROC curves. We set the cut-off for the AUC at 0.7.
2.3 MCID: we used the criterion approach to determine the MCID. We opted for a point where both sensitivity and specificity are maximized.
2.4 Sample size: To estimate sample size, we used Fisher's z test, so we required to include 161 participants.
3.1. Convergent Validity: Correlation between Impact PROMIS and ODI=0.78, NRS=0.77; PSFS -0.56, EQ-5D-5L VAS=-0.62, index=-0.75. The one-tailed analysis demonstrated that identified correlations were significantly higher than a-priori hypotheses for NRS, ODI and EQ-5D-5L but not for PSFS. We met 4 out of 5 hypotheses for total score and 3 out of 4 for subscale analyses.
3.2 Responsiveness: Effect size was calculated large, and t-test showed a statistically significant improvement at follow-up. For construct approach, the correlation was conducted and one-tailed analysis showed that we met 3 out of 5 hypotheses. Lastly, ROC curves showed an AUC of 0.76 (95% CI: 0.69 to 0.82) (Figure 1).
3.3 MCID was estimated 7 points with a sensitivity of 0.61 and specificity of 0.8.
Fig. 1 Receiver operating characteristic curves for the Impact PROMIS questionnaire for ‘improved’ and ‘unchanged’ on the GPE. Impact PROMIS change scores (area under curve = 0.76). Abbreviations: GPE, Global Perceived Effect of change. The highlighted point is the OCP
4. Conclusion: Impact PROMIS showed high convergent validity and acceptable responsiveness in individuals with chronic non-specific LBP, providing further support the use of the Impact PROMIS. However, we strongly recommend using a legacy outcome measure for function