Greene Meridith E, Rader Kevin A, Garellick Göran, Malchau Henrik, Freiberg Andrew A, Rolfson Ola
Harris Orthopaedic Laboratory, Massachusetts General Hospital, 55 Fruit Street, GRJ 1125, Boston, MA, 02114, USA.
Swedish Hip Arthroplasty Register, Gothenburg, Sweden.
Clin Orthop Relat Res. 2015 Nov;473(11):3383-90. doi: 10.1007/s11999-014-4091-y.
The EQ-5D is a generic health survey that can be used to compare improvement across different interventions, measure changes in health-related quality of life over time, or to explore cost-effectiveness among treatments, hospitals, or providers. The original EQ-5D survey has three response options for each of five health dimensions; however, with so few response options, ceiling and floor effects are problematic in some populations. A new version, called the EQ-5D-5L, was developed, which gives respondents five answer options (the "5L" refers to five response levels, which is in contrast to the original survey's three levels). However, the validity of this version has not, to our knowledge, been evaluated in patients undergoing total hip arthroplasty (THA).
QUESTIONS/PURPOSES: The purposes of this study were (1) to characterize the redistribution of responses using the new version; (2) to describe the ceiling and floor effects in the current three-level version and identify whether the new EQ-5D-5L survey diminished these effects; and (3) to understand the convergent validity of the new version with the old and the EQ visual analog scale.
Both versions of the survey were administered either preoperatively or 1 to 6 years after THA, allowing at least 2 weeks between administrations. Responses to the two versions were compared to determine response redistribution properties, ceiling and floor effects, and convergent validity. Sample sizes were determined so that the study would have 90% power to detect a Spearman correlation over 0.7 when comparing the responses of the three-level survey with the five-level survey and allowing for a rate of 25% loss to followup.
Most patients before surgery used the new responses in the majority of dimensions, whereas the patients taking the test after surgery used the new responses predominantly for the pain dimension. The five-level diminished ceiling effects in both groups by up to 30% and diminished floor effects in the pain dimension for patients taking the assessment before surgery by 14%, which was the only dimension in either patient group that had high rates of floor effects. The correlation between the surveys' visual analog scale and response patterns was not different for the two versions of the survey in these populations.
The EQ-5D five-level survey appears able to discriminate new health states indistinguishable in the original, which may allow more sensitive measurements of change in patients undergoing THA. The five-level survey should be considered for implementation in local and national registry monitoring of health-related quality of life in patients undergoing THA.
EQ-5D是一项通用健康调查，可用于比较不同干预措施的改善情况、衡量健康相关生活质量随时间的变化，或探讨不同治疗方法、医院或医疗服务提供者之间的成本效益。原始的EQ-5D调查在五个健康维度上各有三个回答选项；然而，由于回答选项如此之少，在某些人群中天花板效应和地板效应成为问题。一个名为EQ-5D-5L的新版本被开发出来，它为受访者提供了五个答案选项（“5L”指五个回答级别，与原始调查的三个级别形成对比）。然而，据我们所知，该版本在接受全髋关节置换术（THA）的患者中的有效性尚未得到评估。
问题/目的：本研究的目的是：（1）使用新版本来描述回答的重新分布情况；（2）描述当前三级版本中的天花板效应和地板效应，并确定新的EQ-5D-5L调查是否减少了这些效应；（3）了解新版本与旧版本以及EQ视觉模拟量表之间的收敛效度。
两个版本的调查均在术前或THA术后1至6年进行，两次调查之间至少间隔2周。比较两个版本的回答，以确定回答的重新分布特性、天花板效应和地板效应以及收敛效度。确定样本量，以便在比较三级调查和五级调查的回答并考虑25%的失访率时，该研究有90%的把握检测到Spearman相关性超过0.7。
大多数术前患者在大多数维度上使用了新的回答，而术后接受测试的患者主要在疼痛维度上使用了新的回答。五级调查使两组的天花板效应最多降低了30%，并使术前接受评估的患者在疼痛维度上的地板效应降低了14%，这是两个患者组中唯一地板效应发生率较高的维度。在这些人群中，两个版本的调查在视觉模拟量表和回答模式之间的相关性没有差异。
EQ-5D五级调查似乎能够区分原始版本中无法区分的新健康状态，这可能使对接受THA的患者的变化进行更敏感的测量成为可能。在对接受THA的患者进行与健康相关生活质量的地方和国家登记监测时，应考虑采用五级调查。
