Purpose: Patient-reported outcomes (PROs) are used increasingly for individual patient management. Identifying which PRO scores require a clinician's attention is an ongoing challenge. Previous research used a needs assessment to identify EORTC-QLQ-C30 cutoff scores representing unmet needs. This analysis attempted to replicate the previous findings in a new and larger sample.
Methods: This analysis used data from 408 Japanese ambulatory breast cancer patients who completed the QLQ-C30 and Supportive Care Needs Survey-Short Form-34 (SCNS-SF34). Applying the methods used previously, SCNS-SF34 item/domain scores were dichotomized as no versus some unmet need. We calculated area under the receiver operating characteristic curve (AUC) to evaluate QLQ-C30 scores' ability to discriminate between patients with no versus some unmet need based on SCNS-SF34 items/domains. For QLQ-C30 domains with AUC ≥ 0.70, we calculated the sensitivity, specificity, and predictive value of various cutoffs for identifying unmet needs. We hypothesized that compared to our original analysis, (1) the same six QLQ-C30 domains would have AUC ≥ 0.70, (2) the same SCNS-SF34 items would be best discriminated by QLQ-C30 scores, and (3) the sensitivity and specificity of our original cutoff scores would be supported.
Results: The findings from our original analysis were supported. The same six domains with AUC ≥ 0.70 in the original analysis had AUC ≥ 0.70 in this new sample, and the same SCNS-SF34 item was best discriminated by QLQ-C30 scores. Cutoff scores were identified with sensitivity ≥0.84 and specificity ≥0.54.
Conclusion: Given these findings' concordance with our previous analysis, these QLQ-C30 cutoffs could be implemented in clinical practice and their usefulness evaluated.