Jason M. Guattery MS, Agnes Z. Dardas BA, Michael Kelly MD, Aaron Chamberlain MD, MSc, Christopher McAndrew MD, MSc, Ryan P. Calfee MD, MSc, Department of Orthopedic Surgery, Washington University in St Louis School of Medicine, St Louis, MO, USA.
Clin Orthop Relat Res. 2018 Apr;476(4):696-703. doi: 10.1007/s11999.0000000000000076.
The Patient Reported Outcomes Measurement Information System (PROMIS) was developed to provide valid, reliable, and standardized measures to gather patient-reported outcomes for many health domains, including depression, independent of patient condition. Most studies confirming the performance of these measures were conducted with a consented, volunteer study population for testing. Using a study population that has undergone the process of informed consent may be differentiated from the validation group because they are educated specifically as to the purpose of the questions and they will not have answers recorded in their permanent health record.
QUESTIONS/PURPOSES: (1) When given as part of routine practice to an orthopaedic population, do PROMIS Physical Function and Depression item banks produce score distributions different than those produced by the populations used to calibrate and validate the item banks? (2) Does the presence of a nonnormal distribution in the PROMIS Depression scores in a clinical population reflect a deliberately hasty answering of questions by patients? (3) Are patients who are reporting minimal depressive symptoms by scoring the minimum score on the PROMIS Depression Computer Adaptive Testing (CAT) distinct from other patients according to demographic data or their scores on other PROMIS assessments?
Univariate descriptive statistics and graphic histograms were used to describe the frequency distribution of scores for the Physical Function and Depression item banks for all orthopaedic patients 18 years or older who had an outpatient visit between June 2015 and December 2016. The study population was then broken into two groups based on whether they indicated a lack of depressive symptoms and scored the minimum score (34.2) on the Depression CAT assessment (Floor Group) or not (Standard Group). The distribution of Physical Function CAT scores was compared between the two groups. Finally, a time-per-question value was calculated for both the Physical Function and Depression CATs and was compared between assessments within each group as well as between the two groups. Bivariate statistics compared the demographic data between the two groups.
Physical Function CAT scores in musculoskeletal patients were normally distributed like the distribution calibration population; however, the score distribution of the Depression CAT in musculoskeletal patients was nonnormal with a spike in the floor score. After excluding the floor spike, the distribution of the Depression CAT scores was not different from the population control group. Patients who scored the floor score on the Depression CAT took slightly less time per question for Physical Function CAT when compared with other musculoskeletal patients (floor patients: 11 ± 9 seconds; normally distributed patients: 12 ± 10 seconds; mean difference: 1 second [0.8-1.1]; p < 0.001 but not clinically relevant). They spent a substantially shorter amount of time per question on the Depression CAT (Floor Group: 4 ± 3 seconds; Standard Group: 7 ± 7 seconds; mean difference: 3 [2.9-3.2]; p < 0.001). Patients who scored the minimum score on the PROMIS Depression CAT were younger than other patients (Floor Group: 50 ± 18 SD; Standard Group: 55 ± 16 SD; mean difference: 4.5 [4.2-4.7]; p < 0.001) with a larger percentage of men (Floor Group: 48.8%; Standard Group 40.0%; odds ratio 0.6 [0.6-0.7]; p < 0.001) and minor differences in racial breakdown (Floor Group: white 85.2%, black 11.9%, other 0.03%; Standard Group: white 83.9%, black 13.7%, other 0.02%).
In an orthopaedic surgery population that is given PROMIS CAT as part of routine practice, the Physical Function item bank had a normal performance, but there is a group of patients who hastily complete Depression questions producing a strong floor effect and calling into question the validity of those floor scores that indicate minimal depression.
Level II, diagnostic study.
患者报告的结果测量信息系统（PROMIS）旨在提供有效的、可靠的和标准化的措施，以收集许多健康领域的患者报告结果，包括抑郁症，而不受患者病情的影响。大多数证实这些措施性能的研究都是在同意的、志愿者的研究人群中进行测试的。使用经历了知情同意过程的研究人群可能与验证组有所不同，因为他们专门接受过关于问题目的的教育，并且他们的答案不会记录在他们的永久健康记录中。
问题/目的：（1）当作为常规实践的一部分提供给骨科人群时，PROMIS 身体功能和抑郁量表银行产生的分数分布是否与用于校准和验证量表银行的人群产生的分数分布不同？（2）在临床人群中，PROMIS 抑郁评分中存在非正态分布是否反映了患者故意匆忙回答问题？（3）根据人口统计学数据或其他 PROMIS 评估的分数，报告最小抑郁症状的患者（通过 PROMIS 抑郁计算机自适应测试（CAT）获得最低分数）与其他患者有何不同？
使用单变量描述性统计和图形直方图来描述 2015 年 6 月至 2016 年 12 月期间接受门诊就诊的所有 18 岁或以上的骨科患者的身体功能和抑郁量表银行的分数分布。然后，根据他们是否表示没有抑郁症状且在抑郁 CAT 评估中获得最低分数（34.2）（地板组）或没有（标准组），将研究人群分为两组。比较两组之间的身体功能 CAT 分数分布。最后，计算了身体功能和抑郁 CAT 的每个问题的时间价值，并比较了每个组内以及两个组之间的评估。双变量统计比较了两组之间的人口统计学数据。
肌肉骨骼患者的身体功能 CAT 分数呈正态分布，与校准人群的分布相似；然而，肌肉骨骼患者的抑郁 CAT 评分分布是非正态的，有一个地板分数的峰值。排除地板峰值后，抑郁 CAT 分数的分布与对照组人群没有差异。与其他肌肉骨骼患者相比，在抑郁 CAT 上获得地板分数的患者在身体功能 CAT 上每个问题花费的时间略少（地板患者：11 ± 9 秒；正态分布患者：12 ± 10 秒；平均差异：1 秒[0.8-1.1]；p < 0.001，但无临床意义）。他们在抑郁 CAT 上每个问题花费的时间明显短（地板组：4 ± 3 秒；标准组：7 ± 7 秒；平均差异：3 [2.9-3.2]；p < 0.001）。在 PROMIS 抑郁 CAT 上获得最低分数的患者比其他患者年轻（地板组：50 ± 18 SD；标准组：55 ± 16 SD；平均差异：4.5 [4.2-4.7]；p < 0.001），男性比例较大（地板组：48.8%；标准组：40.0%；优势比 0.6 [0.6-0.7]；p < 0.001），种族分布略有差异（地板组：白人 85.2%，黑人 11.9%，其他 0.03%；标准组：白人 83.9%，黑人 13.7%，其他 0.02%）。
在接受 PROMIS CAT 作为常规实践一部分的骨科手术人群中，身体功能量表银行表现正常，但有一组患者匆忙完成抑郁问题，产生强烈的地板效应，并质疑那些表明最小抑郁的地板分数的有效性。
二级，诊断研究。
