Postoperative delirium (POD) remains one of the most prevalent neurological complications after surgery [1] and pre-existing cognitive impairment is a leading risk factor [1, 2]. Consequently, early identification of high-risk patients through pre-operative cognitive assessment is essential. Usually, the evaluation of cognitive status is performed by trained neuropsychologists, using a standardised battery of neuropsychological tests. However, integrating this type of evaluation in a busy surgical context is challenging as it requires expertise, availability and time. The objective of this secondary analysis was to evaluate whether focusing on one instead of a battery of neuropsychological tests would predict POD and ease pre-operative screening by targeting specific cognitive domains.

This study is part of a prospective observational project in which the primary objective was to evaluate intra-operative frontal electroencephalogram α band power as a predictor of POD in patients having cardiac surgery [3]. From May 2019 to December 2021, we included 220 adult patients who underwent an elective cardiac surgery using cardiopulmonary bypass. The day before surgery, enrolled patients were submitted to a battery of pre-operative neuropsychological tests (Fig. 1): 16-item free and cued selective reminding test (FCSRT); modified visual reproduction test from the Wechsler memory scale; digit span test from the revised version of the Wechsler adult intelligence scale (WAIS-R); trail making test; and digit symbol test from the WAIS-R. Sample-specific z-scores ([individual result − mean of the cohort]/standard deviation of the cohort) were computed from each test result. For the FCSRT z-score, only the sum of the results of the three free recalls was considered. For the trail making test, only the difference (time part B minus time part A) was considered. After surgery, patients were screened for POD until discharge from the hospital using the confusion assessment method for intensive care unit (CAM-ICU) and confusion assessment method (CAM) on the ICU and ward, respectively, and a chart review. Comparisons between patients with and without POD were performed using Mann–Whitney U-test or Student's t-test, depending on normality assumption. Categorical variables were compared using a χ² test. Univariable and multivariable logistic regressions were performed to determine the predictive ability of the neuropsychological tests.

Patient characteristics and peri-operative clinical data of patients with and without POD are shown in Table 1. Sixty-five (29.5%) patients had at least one episode of POD. Patients with delirium were older (mean (SD) 71 (10) y vs. 65 (11) y, p < 0.001). There was no difference regarding educational level between groups. Patients who experienced POD had lower pre-operative cognitive score results, except for the digit span test (Table 1). Univariable regression analyses confirmed that the 16-item FCSRT (odds ratio (OR) (95%CI) 0.49 (0.35–0.68), p < 0.001); modified visual reproduction test (OR (95%CI) 0.45 (0.32–0.62), p < 0.001); trail making test (OR (95%CI) 0.63 (0.47–0.84), p = 0.002); and digit symbol test (OR (95%CI) 0.54 (0.39–0.75), p < 0.001) predicted POD. In multivariable logistic regression analysis, only the modified visual reproduction test predicted POD (adjusted OR (95%CI) 0.60 (0.38–0.95), p = 0.030) (online Supporting Information Table S1).

Although recent guidelines on peri-operative brain health recommend the systematic assessment of pre-operative cognitive status, there is currently no consensus on which screening test should be implemented [4-6]. Among the standardised neuropsychological tests performed in this study, our results showed that the modified visual reproduction test might be the most promising instrument to predict POD [7]. The efficient completion of this test requires the integrity of several brain regions and draws on not only visual memory but also perception and constructional skills. Practically, the patient observes for 10 s, then draws five geometric elements of increasing complexity from immediate recall. Twenty minutes later, the patient is asked to remember and draw the same pictures without prior visual reminder. A significant advantage of this test is that it does not depend on literacy or language proficiency, contrary to the 16-item FCSRT or the trail making test. Additionally, most participants found it enjoyable and less stressful than other tools in the battery, making it suitable for pre-operative assessment. However, results should be interpreted cautiously, especially regarding the patient's educational level. Indeed, it has been shown that older patients with < 8 y of scholarship had less ability to reproduce geometric elements with high complexity [8].

This study confirms that pre-operative cognitive impairment is associated with the occurrence of POD. The modified visual reproduction test stands out as a promising screening tool, provided the patient's level of education is considered.

术后谵妄 （POD） 仍然是术后最普遍的神经系统并发症之一 [1]，既往认知障碍是一个主要的危险因素 [1， 2]。因此，通过术前认知评估及早识别高危患者至关重要。通常，认知状态的评估由训练有素的神经心理学家使用标准化的神经心理学测试系列进行。然而，在繁忙的手术环境中整合这种类型的评估是具有挑战性的，因为它需要专业知识、可用性和时间。这项二次分析的目的是评估专注于一个而不是一系列神经心理学测试是否会预测 POD 并通过针对特定的认知领域来简化术前筛查。

这项研究是一个前瞻性观察项目的一部分，该项目的主要目的是评估术中额叶脑电图 α 带功率作为心脏手术患者 POD 的预测因子 [3]。从 2019 年 5 月到 2021 年 12 月，我们纳入了 220 名接受体外循环择期心脏手术的成年患者。手术前一天，入组患者接受一系列术前神经心理学测试（图 1）：16 项游离和提示选择性提醒测试 （FCSRT）;韦氏记忆量表的改良视觉再现测试;韦氏成人智力量表 （WAIS-R） 修订版的数字跨度测试;Trail Making 测试;以及 WAIS-R 的数字符号测试。根据每个测试结果计算样本特异性 z 分数（[个体结果 − 队列的平均值]/队列的标准差）。对于 FCSRT z 评分，仅考虑了 3 次免费召回的结果之和。对于跟踪测试，仅考虑了差异（时间部分 B 减去时间部分 A）。手术后，分别在 ICU 和病房使用重症监护病房意识模糊评估方法 （CAM-ICU） 和意识模糊评估方法 （CAM） 对患者进行 POD 筛查直至出院，并进行图表审查。根据正态性假设，使用 Mann-Whitney U 检验或 Student t 检验对有 POD 和无 POD 的患者进行比较。使用 χ² 检验比较分类变量。进行单变量和多变量 logistic 回归以确定神经心理学测试的预测能力。

有和没有 POD 的患者特征和围手术期临床数据见 表 1。65 例 （29.5%） 患者至少有一次 POD 发作。谵妄患者年龄较大 （平均值 （SD） 71 （10） y vs. 65 （11） y，p < 0.001）。组间教育水平没有差异。经历过 POD 的患者术前认知评分结果较低，除了数字跨度测试（表 1）。单变量回归分析证实，16 项 FCSRT （比值比 （OR） （95%CI） 0.49 （0.35–0.68），p < 0.001）;改良视觉再现试验 （OR （95%CI） 0.45 （0.32–0.62），p < 0.001）;Trail Making Test （OR （95%CI） 0.63 （0.47–0.84），p = 0.002）;和数字符号检验 （OR （95%CI） 0.54 （0.39–0.75），p < 0.001） 预测 POD。在多变量 logistic 回归分析中，只有改良的视觉再现测试预测了 POD（调整后的 OR （95%CI） 0.60 （0.38–0.95），p = 0.030）（在线支持信息表 S1）。

尽管最近的围手术期脑健康指南推荐对术前认知状态进行系统评估，但目前对于应该实施哪种筛查测试尚未达成共识 [4-6]。在本研究进行的标准化神经心理学测试中，我们的结果表明，改良的视觉再现测试可能是预测 POD 的最有前途的工具 [7]。高效完成这项测试需要多个大脑区域的完整性，不仅需要视觉记忆，还需要感知和构建技能。实际上，患者观察 10 秒，然后从即时回忆中绘制出五个复杂程度不断增加的几何元素。20 分钟后，要求患者在没有事先视觉提醒的情况下记住并绘制相同的图片。该测试的一个显着优势是它不依赖于识字或语言能力，这与 16 项 FCSRT 或试探测试相反。此外，大多数参与者发现它比电池中的其他工具更有趣且压力更小，使其适合进行术前评估。然而，应谨慎解释结果，尤其是关于患者的教育水平。事实上，已经表明，具有 < 8 y 学术研究的老年患者再现高度复杂几何元素的能力较差 [8]。

本研究证实术前认知障碍与 POD 的发生有关。如果考虑到患者的教育水平，改良的视觉再现测试是一种很有前途的筛查工具。