Musculoskeletal Statistics Unit, The Parker Institute, Bispebjerg and Frederiksberg Hospital, Copenhagen, Denmark; Department of Rehabilitation, Municipality of Guldborgsund, Nykoebing F, Denmark.
David Geffen School of Med. Division Rheumatology. UCLA, USA.
Semin Arthritis Rheum. 2021 Jun;51(3):607-617. doi: 10.1016/j.semarthrit.2020.09.023. Epub 2021 Jan 9.
Underreporting of harms in randomized controlled trials (RCTs) may lead to incomplete or erroneous assessments of the perceived benefit-to-harm profile of an intervention. To compare benefit with harm in clinical practice and future clinical studies, adverse event (AE) profiles including severity need to be understood. Even though patients report harm symptoms earlier and more frequently than clinicians, rheumatology RCTs currently do not provide a reporting framework from the patient's perspective regarding harms. Our objective for this meta-research project was to identify AEs in order to determine harm clusters and whether these could be self-reported by patients. Our other objective was to examine reported severity grading of the reported harms.
We considered primary publications of RCTs eligible if they were published between 2008 and 2018 evaluating pharmacological interventions in patients with a rheumatic or musculoskeletal condition and if they were included in Cochrane reviews. We extracted data on harms such as reported AE terms together with severity (if described), and categorized AE- and severity-terms into overall groups. We deemed all AEs with felt components appropriate for patient self-reporting.
The literature search identified 187 possible Cochrane reviews, of which 94 were eligible for evaluation, comprising 1,297 articles on individual RCTs. Of these RCTs, 93 pharmacological trials met our inclusion criteria (including 31,023 patients; representing 20,844 accumulated patient years), which reported a total of 21,498 AEs, corresponding to 693 unique reported terms for AEs. We further sub-categorized these terms into 280 harm clusters (i.e., themes). AEs appropriate for patient self-reporting accounted for 58% of the AEs reported. Among the reported AEs, we identified medical terms for all of the 117 harm clusters appropriate for patient reporting and lay language terms for 86%. We intended to include severity grades of the reported AEs, but there was no evidence for systematic reporting of clinician- or patient-reported severity in the primary articles of the 93 trials. However, we identified 33 terms suggesting severity, but severity grading was discernible in only 9%, precluding a breakdown by severity in this systematic review.
Our results support the need for a standardized framework for patients' reporting of harms in rheumatology trials. Reporting of AEs with severity should be included in future reporting of harms, both from the patients' and investigators' perspectives.
PROSPERO: CRD42018108393.
随机对照试验（RCT）中危害的漏报可能导致对干预措施的感知获益-危害特征的不完全或错误评估。为了在临床实践和未来的临床研究中比较获益与危害，需要了解包括严重程度在内的不良事件（AE）概况。尽管患者比临床医生更早、更频繁地报告危害症状，但风湿病 RCT 目前并未从患者的角度提供关于危害的报告框架。我们这个元研究项目的目的是确定 AE，以确定危害群集，以及这些群集是否可以由患者自我报告。我们的另一个目标是检查报告的危害严重程度分级。
我们认为，如果 RCT 的主要出版物在 2008 年至 2018 年之间发表，评估了风湿性或肌肉骨骼疾病患者的药物干预措施，并且包含在 Cochrane 综述中，则符合纳入标准。我们提取了有关危害的信息，例如报告的 AE 术语以及严重程度（如果有描述），并将 AE 和严重程度术语分类为总体组。我们认为所有具有感觉成分的 AE 都适合患者自我报告。
文献检索确定了 187 项可能的 Cochrane 综述，其中 94 项可进行评估，包括 93 项药理学试验，共涉及 1297 项 RCT 文章。这些 RCT 中，有 93 项药理学试验符合我们的纳入标准（包括 31023 名患者；代表 20844 名患者的累计就诊年限），共报告了 21498 例 AE，相当于 693 例 AE 的独特报告术语。我们进一步将这些术语细分为 280 个危害群集（即主题）。适合患者自我报告的 AE 占报告的 AE 的 58%。在报告的 AE 中，我们确定了适合患者报告的 117 个危害群集的所有医学术语和 86%的通俗语言术语。我们打算包括报告 AE 的严重程度等级，但在 93 项试验的主要文章中，没有系统地报告临床医生或患者报告的严重程度的证据。然而，我们确定了 33 个表示严重程度的术语，但只有 9%的术语可以辨别严重程度，这使得在这个系统评价中无法按严重程度进行分类。
我们的结果支持在风湿病试验中为患者报告危害制定标准化框架的必要性。应从患者和研究者的角度报告具有严重程度的 AE，包括严重程度。
PROSPERO：CRD42018108393。
