Supplementary Material for: Feasibility and Diagnostic Accuracy of Early Mood Screening to Diagnose Persisting Clinical Depression/Anxiety Disorder after Stroke

<b><i>Background:</i></b> Depression/anxiety disorders are common after stroke and have a negative impact on outcomes. Guidelines recommend that all stroke survivors are screened for these problems. However, there is no consensus on timing or method of assessment. We investigated the feasibility and accuracy of a very early screening strategy and the diagnostic accuracy this has for depression/anxiety disorders at 1 month. <b><i>Methods:</i></b> Screening tools were Hospital Anxiety and Depression Scale (HADS) and Depression Intensity Scale Circles (DISCs); we also assessed cognition using the Montreal Cognitive Assessment (MoCA). Screening was offered to sequential stroke admissions. At 1 month we assessed for clinical depression/anxiety disorder using Mini-International Neuropsychiatric Interview (MINI) and retested screening tools. We described test accuracy of acute depression/anxiety screening for clinical diagnosis of depression/anxiety disorder at 1 month and described temporal change in screening test scores. We assessed feasibility by describing proportions that were able, agreed to and completed the screening tests. <b><i>Results:</i></b> Over 4 months, 102/146 admissions were suitable for screening following initial medical assessment, 69 (68%) agreed to screening, of whom 33 (48%) required researcher assistance to complete the screening test battery. Median time to assessment was 2 days (IQR: 1-4). Early HADS suggested n = 9 (13%) with depression; DISCs n = 25 (37%). Median acute MoCA was 21/30. At 1 month, n = 61 (88%) provided data. Repeat scores showed improvement over time; HADS (anxiety) mean difference: 2.5 (95% CI: 1.2-3.7), HADS (depression) mean difference: 1.6 (95% CI: 0.3-2.9). MINI defined n = 12 (20%) with depression and n = 6 (10%) with anxiety disorder. Comparing baseline screening to 1-month clinical diagnosis, HADS sensitivity was 0.25 (95% CI: 0.09-0.53) and specificity 0.94 (95% CI: 0.84-0.98); DISCs sensitivity was 0.92 (95% CI: 0.65-0.99) and specificity 0.78 (95% CI: 0.64-0.87). <b><i>Conclusions:</i></b> Even amongst ‘medically stable' stroke patients, depression/anxiety screening at the acute stage may not be feasible or accurate. Half of participants required assistance from the researcher to complete assessments. The poor predictive accuracy of HADS for depression/anxiety disorder at 1 month may be due in part to the high prevalence of cognitive impairment in our sample. Screening in the first few days after stroke does not appear useful for detecting clinically important and sustained depression/anxiety problems.

**背景**：卒中后抑郁/焦虑障碍较为常见，且会对患者预后产生不良影响。现有指南建议对所有卒中幸存者开展此类问题的筛查，但目前学界对于筛查的时机与评估方式尚未达成共识。本研究旨在探讨极早期筛查策略的可行性与准确性，以及该策略在卒中后1个月时对抑郁/焦虑障碍的诊断效能。
**方法**：本研究采用的筛查工具包括医院焦虑抑郁量表（Hospital Anxiety and Depression Scale, HADS）与抑郁强度圆圈量表（Depression Intensity Scale Circles, DISCs），同时使用蒙特利尔认知评估量表（Montreal Cognitive Assessment, MoCA）评估患者认知功能。研究对象为连续收治的卒中患者。在患者卒中后1个月时，采用迷你国际神经精神访谈（Mini-International Neuropsychiatric Interview, MINI）评估其临床抑郁/焦虑障碍患病情况，并再次实施筛查量表测评。本研究分析了急性期抑郁/焦虑筛查对卒中后1个月临床确诊抑郁/焦虑障碍的检验效能，同时描述了筛查量表得分的时间变化趋势。可行性通过能够参与、同意参与并完成筛查测试的受试者占比进行评估。
**结果**：在4个月的研究周期内，经初始医学评估后共有102/146例入院患者符合筛查纳入标准，其中69例（68%）同意参与筛查，该69例受试者中有33例（48%）需要研究人员协助才能完成整套筛查测试。首次评估的中位时间为2天（四分位距：1~4天）。急性期HADS筛查显示9例（13%）受试者存在抑郁症状，DISCs筛查则显示25例（37%）存在相关症状。急性期MoCA评分的中位数为21/30。卒中后1个月时，共有61例（88%）受试者提供了随访数据。复测评分显示随时间推移得分有所改善：HADS焦虑维度的平均差值为2.5（95%置信区间：1.2~3.7），HADS抑郁维度的平均差值为1.6（95%置信区间：0.3~2.9）。经MINI评估，共12例（20%）受试者确诊抑郁障碍，6例（10%）确诊焦虑障碍。将基线筛查结果与卒中后1个月的临床诊断结果进行对比后发现，HADS的灵敏度为0.25（95%置信区间：0.09~0.53），特异度为0.94（95%置信区间：0.84~0.98）；DISCs的灵敏度为0.92（95%置信区间：0.65~0.99），特异度为0.78（95%置信区间：0.64~0.87）。
**结论**：即便针对“病情稳定”的卒中患者，急性期开展抑郁/焦虑筛查也可能既不可行也不准确。近半数受试者需要研究人员协助才能完成评估。HADS对卒中后1个月抑郁/焦虑障碍的预测效能较差，这可能部分归因于本研究样本中认知障碍的高患病率。卒中后最初数天内开展的筛查，似乎无法有效检出具有临床意义且持续存在的抑郁/焦虑问题。