Evaluation of minimal important difference and responder definition in the EORTC QLQ-PAN26 module for assessing health-related quality of life in patients with surgically resected pancreatic adenocarcinoma
Background
Although the European Organisation for Research and Treatment of Cancer (EORTC) QLQ-PAN26 is widely used to assess health-related quality of life (HRQoL), its group-level minimal important difference (MID) and individual-level responder definition (RD) are not established; we calculated MID and RD using HRQoL data from the APACT trial in patients with surgically resected pancreatic cancer who received adjuvant chemotherapy.
Methods
HRQoL was assessed using EORTC QLQ-C30 and QLQ-PAN26 at baseline, during treatment, at end of treatment, and during follow-up. Distribution-based MIDs were estimated using 0.5 × baseline standard deviation (SD) and reliability-based (intraclass correlation) standard error of measurement (SEM). Anchor-based MIDs and RDs (anchor, QLQ-C30 overall health) were estimated using a linear mixed model.
Results
Overall, 772 patients completed the baseline assessment. Distribution-based MIDs (0.5 × SD) for QLQ-PAN26 scales ranged from 12 to 13, except hepatic symptoms (≈8), pancreatic pain (≈10), and sexual dysfunction (≈17); those for stand-alone items ranged from 12 to 16. The SEM values were similar. Among scales/items sufficiently correlated (r > 0.30) with the anchor, MIDs ranged from 5 to 9. Within-patient QLQ-PAN26 RD estimates varied by direction (deterioration vs. improvement) and scale/item, but all values were lower than the true possible within-patient change (e.g. 16.7 points for a two-item scale) given a one-category change on the raw scale.
Conclusions
Compared with distribution-based MIDs, anchor-based MIDs were twice as sensitive in detecting group-level changes in QLQ-PAN26 scales/items. For interpreting clinically meaningful change, RDs cannot be less than the true minimum of the scale. The group-level MID may help clinicians/researchers interpret HRQoL changes.