First, it was difficult to understand how each group was actually treated. The authors tabulated “surgery” (n = 105), “chemotherapy” (n = 34), and “both” (n = 104). However, these groups are described as follows: “All participants were receiving medical treatment: 209 participants had surgery (86.8%), 138 had received chemoradiotherapy (49.3%), and 34 of them had received both treatments.”
response: Of the 209 participants who underwent surgery, 105 had surgery as their primary treatment, and the remaining 104 received both treatments. Likewise, 34 of 138 patients adopted chemoradiotherapy as their primary treatment modality. Due to limitations in health conditions, it is normal for many patients to only receive one type of treatment. Using both types of treatment means fewer restrictions and better health. Indeed, the settings of these known groups were discussed with the patients’ clinicians in the analysis.
The first hypothesis was that patients with advanced tumor stages would have lower health utility. The QLU-C10D results are consistent with this hypothesis, comparing stage I versus stages II, III, and IV. However, EQ-5D-5L results showed the opposite direction, with lowest utility often in stage I and often highest utility in the most advanced tumor stages. Therefore, these results actually argue against the validity of the EQ-5D-5L in these known group tests.
In summary, two of the five groups used (family history and treatment modality) did not have a sufficient evidence base or supporting evidence to be considered “known groups.” Of the remaining three known groups (tumor stage, metastases and time since treatment), each was tested using a set of nine country-specific values from the EQ-5D-5L (N = 27), only 56% (N = 15) The test results showed results consistent with the hypothesis. This is well below the threshold of the COSMIN guidelines, which require that at least 75% of tests match the hypothesis to support construct validity (4). Evaluated using 10 country-specific value sets, QLU-C10D had 29 out of 30 (97%) results consistent with hypothesis.
Answer: We disagree on this point. For example, for patients with different tumor stages, stage III patients always had higher QLU-C10D utility values than stage II patients, according to the 9 value sets (Appendix 4).
