Health Economics Unit, Centre for Health Policy, Melbourne School of Population and Global Health, The University of Melbourne, 207 Bouverie Street, Carlton, VIC, 3053, Melbourne, Australia.
School of Population Health, Curtin University, Perth, WA, Australia.
Pharmacoeconomics. 2022 Mar;40(3):297-307. doi: 10.1007/s40273-021-01109-3. Epub 2021 Nov 17.
The EQ-5D-5L and its value sets are widely used internationally. However, in the US and elsewhere, there is growing use of PROMIS, which has a value set (PROPr) based on the stated preferences of the US population. This paper aims to compare the characteristics of EQ-5D-5L and PROPr value sets and to highlight potential implications for users.
US, Australian and English value sets were used for EQ-5D-5L. PROPr utilities were calculated based on PROMIS-29 + 2. We examined, in each case, (i) the characteristics (e.g. range of values, number of unique values) and distribution of all possible 'theoretical' utilities; (ii) dimension/domain importance ranking by the utility of corner states (i.e. health states with the worst level in one domain and the best in all others); (iii) comparisons of utilities for health states hypothesised to be comparable in terms of severity across EQ-5D-5L descriptive systems and PROMIS-29 + 2 domain scores; (iv) the changes in values of adjacent states (i.e. a one-level change in one dimension for EQ-5D-5L and a four-point change in raw scores for PROMIS-29 + 2, with the other dimensions held constant) for dimensions hypothesised to overlap conceptually or be correlated between the two instruments.
EQ-5D-5L and PROPr utilities differ systematically. First, the US EQ-5D-5L utilities range from - 0.573 to 1, whereas PROPr values for PROMIS-29 + 2 range from - 0.022 to 0.954. Second, in the US (and English) EQ-5D-5L value sets, pain is the most important dimension whereas in PROPr pain is one of the least important (apart from sleep disturbance). Third, classified based on severity across EQ-5D-5L descriptive systems and PROMIS-29 + 2 domain scores, PROPr has substantially lower values than EQ-5D-5L values for comparable 'mild' health states, but higher values for more 'severe' health states. Last, when one dimension is considered across its best to worst levels and all other dimensions are held constant at their best or moderate level, in EQ-5D-5L value sets, the greatest changes in utility occur between levels 3 and 4 (moderate and severe) problems; in PROPr that occurred between the most severe states and their descriptively adjacent health states.
There are very considerable differences between US EQ-5D-5L and PROPr utilities, despite both in principle representing utility on the same scale anchored at 0 and 1 and both representing the preferences of the US general public. It is important for decision makers and clinical triallists to be aware of these differences. Further work is needed to assess the impact of these differences in value sets using population and patient data, and in longitudinal settings.
EQ-5D-5L 及其效标值在国际上得到了广泛应用。然而，在美国和其他地方，越来越多地使用 PROMIS，它有一个基于美国人群的偏好的效标值（PROPr）。本文旨在比较 EQ-5D-5L 和 PROPr 效标值的特点，并强调对用户的潜在影响。
使用美国、澳大利亚和英国的效标值对 EQ-5D-5L 进行评估。PROPr 效用是根据 PROMIS-29+2 计算的。我们在每种情况下都检查了（i）所有可能的“理论”效用的特征（例如，值的范围、唯一值的数量）和分布；（ii）通过角状态的效用（即一个域中最差水平和所有其他域中最佳水平的健康状态）对维度/域的重要性进行排名；（iii）对 EQ-5D-5L 描述系统和 PROMIS-29+2 域分数中假设具有相似严重程度的健康状态进行效用比较；（iv）对于假设在两个工具之间概念上重叠或相关的维度，相邻状态值的变化（即 EQ-5D-5L 的一个维度的一级变化和 PROMIS-29+2 的原始分数的四级变化，其他维度保持不变）。
EQ-5D-5L 和 PROPr 效用存在系统差异。首先，美国 EQ-5D-5L 效用值范围为-0.573 至 1，而 PROMIS-29+2 的 PROPr 值范围为-0.022 至 0.954。其次，在美国（和英国）EQ-5D-5L 效标值中，疼痛是最重要的维度，而在 PROPr 中，疼痛是最不重要的维度之一（除了睡眠障碍）。第三，根据 EQ-5D-5L 描述系统和 PROMIS-29+2 域分数的严重程度进行分类，PROPr 对可比“轻度”健康状态的效标值明显低于 EQ-5D-5L 值，但对更“严重”的健康状态的效标值较高。最后，当一个维度在最佳到最差水平之间考虑，并且所有其他维度在最佳或中度水平上保持不变时，在 EQ-5D-5L 效标值中，效用的最大变化发生在 3 级和 4 级（中度和严重）之间；在 PROPr 中，这种变化发生在最严重的状态与其描述性相邻健康状态之间。
尽管美国 EQ-5D-5L 和 PROPr 效标值在原则上都代表相同的 0 到 1 量表上的效用，并都代表美国公众的偏好，但两者之间存在很大差异。决策者和临床试验人员了解这些差异非常重要。需要进一步研究使用人群和患者数据以及纵向设置评估效标值差异的影响。
Zotero 插件