Department of Quantitative Health Sciences, Cleveland Clinic, Cleveland, OH, USA.
Center for Outcomes Research and Evaluation, Neurological Institute, Cleveland Clinic, Cleveland, OH, USA.
Med Decis Making. 2024 Nov;44(8):961-973. doi: 10.1177/0272989X241285038. Epub 2024 Oct 8.
Estimating change in health-related quality of life (HRQOL) from pre- to poststroke is challenging because HRQOL is rarely collected prior to stroke. Leveraging HRQOL data collected both before and after stroke, we sought to estimate the change in HRQOL from prestroke to early poststroke.
Stroke survivors completed the Patient-Reported Outcomes Measurement Information System Global Health (PROMIS-GH) scale at both pre- and early poststroke. Patient characteristics were compared for those who did and did not complete the PROMIS-GH. The mean change in PROMIS-GH T-score was estimated using complete case analysis, multiple imputation, and multiple imputation with delta adjustment.
A total of 4,473 stroke survivors were included (mean age 63.1 ± 14.1 y, 47.5% female, 82.6% ischemic stroke). A total of 993 (22.2%) patients completed the PROMIS-GH at prestroke while 2,298 (51.4%) completed it early poststroke. Compared with those without PROMIS-GH, patients with PROMIS-GH prestroke had worse comorbidity burden. Patients who completed PROMIS-GH early poststroke had better early poststroke clinician-rated function and shorter hospital length of stay. Complete case analysis and multiple imputation revealed patients' PROMIS-GH T-scores worsened by 2 to 3 points. Multiple imputation with delta adjustment revealed patients' PROMIS-GH T-scores worsened by 4 to 10 points, depending on delta values chosen.
Systematic differences in patients who completed the PROMIS-GH at both pre- and early poststroke suggest that missing PROMIS-GH scores may be missing not at random (MNAR). Multiple imputation with delta adjustment, which is better suited for MNAR data, may be a preferable method for analysis of change in HRQOL from pre- to poststroke. Given our study's large proportion of missing HRQOL data, future studies with less missing HRQOL data are necessary to verify our results.
Estimating the change in health-related quality of life from pre- to poststroke is challenging because health-related quality-of-life data are rarely collected prior to stroke. Previously used methods to assess the burden of stroke on health-related quality of life suffer from recall bias and selection bias.Using health-related quality-of-life data collected both before and after stroke, we sought to estimate the change in health-related quality of life after stroke using statistical methods that account for missing data.Comparisons of patients who did and did not complete health-related quality-of-life scales at both pre- and poststroke suggested that missing data may be missing not at random.Statistical methods that account for data that are missing not at random revealed more worsening in health-related quality of life after stroke than traditional methods such as complete case analysis or multiple imputation.
从卒中前到卒中后估计与健康相关的生活质量（HRQOL）的变化具有挑战性，因为很少在卒中前收集 HRQOL 数据。利用卒中前后都收集的 HRQOL 数据，我们试图估计从卒中前到卒中后早期的 HRQOL 变化。
卒中幸存者在卒中前和卒中后早期都完成了患者报告的结果测量信息系统全球健康量表（PROMIS-GH）。对于完成和未完成 PROMIS-GH 的患者，比较了患者特征。使用完全案例分析、多重插补和带有 delta 调整的多重插补来估计 PROMIS-GH T 评分的平均变化。
共纳入 4473 名卒中幸存者（平均年龄 63.1±14.1 岁，47.5%为女性，82.6%为缺血性卒中）。共有 993 名（22.2%）患者在卒中前完成了 PROMIS-GH，而 2298 名（51.4%）患者在卒中后早期完成了 PROMIS-GH。与未完成 PROMIS-GH 的患者相比，完成卒中前 PROMIS-GH 的患者合并症负担更重。完成卒中后早期 PROMIS-GH 的患者卒中后早期临床医生评定的功能更好，住院时间更短。完全案例分析和多重插补显示患者的 PROMIS-GH T 评分恶化了 2 到 3 分。带有 delta 调整的多重插补显示，根据选择的 delta 值，患者的 PROMIS-GH T 评分恶化了 4 到 10 分。
在卒中前和卒中后早期都完成 PROMIS-GH 的患者存在系统性差异，这表明缺失的 PROMIS-GH 评分可能是非随机缺失（MNAR）。带有 delta 调整的多重插补更适合 MNAR 数据，可能是分析从卒中前到卒中后 HRQOL 变化的首选方法。鉴于我们研究中 HRQOL 数据大量缺失，未来需要进行缺失 HRQOL 数据较少的研究来验证我们的结果。
从卒中前到卒中后估计与健康相关的生活质量的变化具有挑战性，因为很少在卒中前收集与健康相关的生活质量数据。以前用于评估卒中对与健康相关的生活质量的负担的方法存在回忆偏倚和选择偏倚。利用卒中前后都收集的与健康相关的生活质量数据，我们试图使用统计方法来估计卒中后的与健康相关的生活质量变化，这些方法考虑了缺失数据。对在卒中前和卒中后早期都完成和未完成与健康相关的生活质量量表的患者进行比较，表明缺失的数据可能是非随机缺失。考虑到非随机缺失的数据的统计方法比完全案例分析或多重插补等传统方法显示出卒中后与健康相关的生活质量恶化更多。
