针对危重症患者谵妄的多组分非药物干预：一项随机对照试验的系统评价与荟萃分析。
全文（开放获取） · 中文译文
摘要 背景：有必要评估多组分非药物干预措施在预防和管理重症监护病房（ICU）谵妄方面的疗效，并确定最有效的干预组合。 方法：在PubMed、EMBASE、CINAHL、Cochrane Library以及ProQuest Dissertations and Theses A&I数据库中进行了从建库至2026年7月3日的系统性文献检索，并辅以对既往荟萃分析的手工检索。仅纳入随机试验。采用风险比（RR）或均数差（MD）及其95%可信区间（CI）进行荟萃分析。使用具有限制性最大似然估计和修正Hartung-Knapp可信区间的随机效应模型。使用Cochrane随机对照试验偏倚风险评估工具评估偏倚风险。使用GRADE方法评估证据确定性。 结果：检索共纳入13项随机对照试验，涵盖5525例患者。多组分干预显著降低了谵妄发生率（RR 0.53，95% CI [0.36，0.76]，I2 = 72.5%，低确定性证据）。在排除两项整群随机试验（RR 0.45，95% CI 0.30-0.67；I2 = 55.7%）、主动对照试验（RR 0.55，95% CI 0.38-0.79；I2 = 73.0%）以及同时排除全部三项试验（RR 0.46，95% CI 0.31-0.70；I2 = 57.0%）后，该效应仍然显著。在探索性亚组分析中，包含认知评估与刺激、重新定向、睡眠与环境管理、家庭或社区支持以及体力活动的组分模式B的亚组内合并估计值显示谵妄发生率降低（RR 0.43，95% CI [0.23，0.82]）；然而，亚组差异的正式检验不显著（p = 0.604）。在谵妄持续时间、ICU或住院时间、无谵妄天数或死亡率方面未发现统计学显著效应。 结论：本研究表明，多组分干预可有效降低危重症患者的谵妄发生率。证据未确定哪一种组分组合优于其他组合。 系统评价注册：https://www.crd.york.ac.uk/PROSPERO/view/CRD42024592821。 谵妄是一种急性起病且波动性的脑功能障碍，以注意力及意识受损为特征，是收入重症监护病房（ICU）患者的常见并发症（；）。谵妄在住院患者中的发病率约为20–30%（），但在ICU患者中显著升高至25–87%，具体取决于疾病严重程度和诊断方法（；；；）。ICU谵妄与多种不良结局相关，包括机械通气时间延长（）、长期认知功能下降（）、ICU及住院时间延长（），以及临床护士工作负荷增加（；；）。这些风险凸显了对危重症患者实施早期预防措施以改善预后的必要性。 药物干预措施，如右美托咪定、非典型抗精神病药物和氟哌啶醇，常用于ICU谵妄的预防和管理，因其起效迅速、可获得性强，并有可能缩短住院时间（；；）。然而，这些治疗可能带来显著风险，包括心律失常和呼吸窘迫，尤其是在脆弱患者中（；；）。此外，研究已表明药物干预在降低谵妄发生率方面的疗效有限（）。Bannon的研究指出，没有充分证据支持单一或多组分非药物干预措施能够降低谵妄发生率或缩短谵妄持续时间，但多组分干预措施可能为针对多种危险因素提供更可信的机会（）。当前的PADIS指南和近期研究提倡将多组分非药物策略作为预防和管理ICU谵妄的最有效方法（；；）。 迄今为止，多项荟萃分析已评估了单一组分和多组分非药物干预措施在预防ICU谵妄方面的有效性（；；；）。这些研究一致表明，多组分非药物干预措施能显著降低谵妄的发生率。然而，其在缩短谵妄持续时间方面的有效性仍存在争议。尽管三项研究（；；）报告多组分干预措施对谵妄持续时间无显著影响，但有研究提示，特定的干预措施组合可实现显著缩短。 确定了两种有效的多成分谵妄预防干预组合：（1）睡眠促进、认知刺激、早期活动、疼痛控制和评估；（2）睡眠促进和认知刺激。然而，他们的结论是基于对设计各异的研究进行的合并分析，这可能导致对非药物策略在降低ICU谵妄发生率方面相对疗效的估计不准确。鉴于已有大量随机对照试验（RCTs），仅关注RCTs的最新系统评价与荟萃分析具有临床相关性。 此外，尚无研究系统性地界定这些干预措施的具体实施细节。本研究的一个关键创新是全面列出并整合各种干预措施的实施规程，为临床实践提供精确且可操作的指导。 本研究旨在开展一项最新的系统评价与荟萃分析，以评估多成分非药物干预在预防和管理ICU谵妄中的疗效，识别最有效的干预组合，解决目前有关其对谵妄持续时间和ICU住院时间影响的争议，并提供循证建议以改进临床实践。 方法 研究设计与方案 本系统评价和荟萃分析遵循系统评价和荟萃分析首选报告条目（PRISMA）指南，并在国际前瞻性系统评价注册库（PROSPERO：CRD42024592821）（）注册。 纳入标准（PICOS） 我们根据以下标准纳入研究：（1）研究类型：随机对照试验（RCTs）；（2）研究对象：ICU成年患者；（3）以多组分非药物干预作为干预组；（4）以非药物干预或常规护理作为对照组；（5）研究内容：多组分非药物干预对谵妄的预防和治疗。 根据以下标准排除研究：（1）综述、信件、摘要、会议论文、重复发表或无法获得全文的文章。（2）非英语研究。（3）涉及新生儿或儿童ICU患者的研究。（4）使用药物干预或单一非药物干预的研究。 检索策略与数据来源 在五个电子数据库中进行了全面检索：PubMed、EMBASE、CINAHL、Cochrane Library 和 ProQuest Dissertations and Theses A&I，涵盖从建库至2026年7月3日的研究。检索使用了以下关键词组合：（"critical care" OR "intensive care" OR "ICU"）AND（"non-pharmacologic" OR "non-pharmacological" OR "multicomponent" OR "bundle" OR "mobilization" OR "daily sedation interruption" OR "music" OR "light therapy" OR "reorientation" OR "cognitive stimulation" OR "cognitive training" OR "eye mask" OR "earplugs" OR "physical therapy" OR "occupational therapy" OR "multicomponent"）AND delirium。此外，还对相关研究的参考文献列表进行了人工审查，以确定符合纳入标准的研究。 研究筛选与数据提取 两名研究者（Wang ST 和 Huang WJ）独立进行文献筛选和数据提取。如有分歧，通过讨论或咨询第三名研究者解决。筛选过程包括三个步骤：（1）审阅标题和摘要；（2）阅读全文；（3）审阅所选研究的参考文献。在每个步骤中，均一致应用纳入和排除标准。 使用标准化表格从纳入文章中提取相关数据。提取的项目包括第一作者、发表年份、研究地区、研究设计、样本量、平均年龄、研究场所、干预类型、中心类型及其他相关细节。 干预成分分类 数据提取后，对纳入试验的干预内容进行了详细审查。基于既有的ICU谵妄预防非药物方法和纳入研究报告的实际干预要素，将干预措施归纳为五个组成部分：认知评估与刺激、重新定向、睡眠与环境管理、家庭或社区支持，以及身体活动。随后根据这些组成部分的具体组合，将干预措施分为三种组成部分模式，如 ， 所示。此分类用于支持探索性亚组分析，不应被解释为对任何单一组成部分独立效应的评估。 用于探索性亚组分析的干预组成部分模式分类。 非药物干预的组成部分。 使用经过验证的ICU谵妄和镇静识别工具，如CAM-ICU、CAM、RASS等。 以增加患者警觉水平为目标的外部强烈刺激。 以激活心理功能为目标的运动库，涵盖以下领域，包括但不限于：警觉性、视觉感知、记忆、计算、问题解决、实践操作和语言。每位患者获得一本认知练习笔记本，以及排序卡片和游戏（卡片、多米诺骨牌和记忆游戏，以及视觉空间构建游戏）等工具。 纠正感觉缺陷，例如鼓励患者使用眼镜和助听器，保护角膜免受干燥，并清洁耳道。 鼓励积极沟通，并使用适当的沟通技巧（言语/文字/图片）来补偿感觉丧失并克服语言障碍。 在可能的情况下，允许患者看电视、听喜欢的音乐或阅读报纸/杂志。 对机械通气的患者进行每日自主觉醒试验（SAT）或每日自主呼吸试验（SBT）。 为患者提供可见的时钟、日历、手机和时间安排表。 告知患者所在科室、医院以及入院原因。 采取了维持良好睡眠-觉醒周期的具体措施。例如，白天采用自然光照明，夜间采用柔和灯光。 夜间尽量减少所有护理操作。 尽可能降低噪音。例如，将报警系统的各种提示音调得更加柔和。 为患者提供眼罩和隔音耳塞，以保证良好睡眠。 限制患者转移到其他病床，个性化其床旁区域，并允许携带家中喜爱的物品（枕头/家人照片）。 工作人员邀请患者参观ICU，以熟悉环境。 尽可能延长家属探视时间，并鼓励家属参与患者的日常护理。 使用有效的沟通设备与家属交谈。 识别患者的需求和可能的压力来源，并在家庭成员帮助下予以处理。 鼓励早期活动并制定活动计划。 基本日常生活活动（BADLs），是指基于BADL促进独立生活的一组活动，包括个人卫生、仪容修饰和进食。 每个班次为患者安排活动。若RASS评分为-2至-4，采取Fowler体位，进行按摩、床上运动，每2小时更换体位，做蹬车样关节运动。若RASS评分为-1至2，采取Fowler体位，坐在椅子上，站立，在帮助下小步行走。 尽量减少并优化放置约束带。 偏倚风险评估 采用Cochrane随机对照试验偏倚风险工具（），由两名评审者（Wang ST和Huang WJ）独立评估所纳入研究的方法学质量。偏倚风险从七个领域进行评估：随机序列生成、分配隐藏、参与者和人员盲法、结局评估盲法、结局数据不完整、选择性报告以及其他潜在偏倚来源。每个领域均判定为低偏倚风险、不清楚或高偏倚风险。分歧通过讨论或咨询第三名评审者解决。 证据确定性 每个结局的证据确定性使用推荐意见分级的评估、制定与评价（GRADE）方法（）进行评估。由于所有纳入的研究均为随机对照试验，证据确定性最初被评定为高，随后在适当时根据偏倚风险、不一致性、间接性、不精确性和发表偏倚进行降级。最终的证据确定性被分为高、中、低或极低。 数据提取自原始研究，并使用R 4.4.2版进行了更新后的荟萃分析。风险比（RR）用于二分类结局，均数差（MD）用于连续结局。以均数和标准差之外格式报告的数据仅在临床和统计学上适当时进行转换。效应估计值以95%置信区间（CI）呈现。采用限制最大似然估计和修正的Hartung–Knapp法进行随机效应荟萃分析。使用χ²检验评估异质性，并使用I²统计量量化。对于两项整群随机试验，使用了针对整群效应进行校正的效应估计值。一项试验报告了校正比值比，使用对照组的观察事件风险将其转换为近似风险比；其标准误通过delta法估计。通过分别排除两项整群随机试验和活性对照试验，然后同时排除全部三项试验进行分析，来检验主要分析的稳健性。通过漏斗图目视检查和Egger回归检验评估谵妄发生率的潜在发表偏倚和小研究效应。 结果 通过电子数据库检索共识别出13,086篇文章，另通过手动检索策略获得4篇文章。经自动化工具去除156篇重复记录并标记62篇不符合条件的记录后，对12,868篇文章的标题和摘要进行了筛选，排除了12,831篇不符合纳入标准的文章。其余37篇全文文章进行了合格性评估，其中24篇因纳入非ICU患者、非随机对照试验、非多组分非药物干预或未报告谵妄发生率等原因被排除。在审查了符合条件研究的参考文献列表后，13篇文章符合标准并被纳入系统评价（见图）。 PRISMA 2020流程图：多组分非药物干预在减少ICU谵妄中的比较。 纳入研究的特点 本评价纳入了13项随机对照试验，包括2项阶梯楔形整群随机试验，样本量从40例到2,618例不等，共5,525名参与者。这些研究发表于2015年至2026年。所有研究均报告了谵妄发生率的数据。纳入研究的基本特征详见【表格】。 研究特征表。 ICU，重症监护病房；RCT，随机对照试验；I，干预；C，对照；SICU，外科重症监护病房；MICU，内科重症监护病房；NICU，神经科重症监护病房；CSICU，心血管外科重症监护病房。①认知评估与刺激；②重新定向；③睡眠与环境管理；④家庭与社区支持；⑤体力活动。 偏倚风险评估 偏倚风险评估确定6项研究为低偏倚风险，3项为风险不清楚，4项为高偏倚风险（见 ）。主要问题与非药物干预中参与者和人员盲法的困难、部分试验分配程序报告不完整，以及少数研究在结局评估或选择性报告方面的潜在问题有关。缺失结局数据总体上被判定为低偏倚风险。 纳入研究的偏倚风险评估。 综合发现 合并分析 包含5,525例ICU患者的13项试验的合并分析显示，多组分非药物干预显著降低了谵妄发生率（RR 0.53，95% CI [0.36, 0.76]；I2 = 72.5%）（ ）。敏感性分析结果一致。在排除两项整群随机试验（11项试验；RR 0.45，95% CI [0.30, 0.67]；I2 = 55.7%）、排除Dong等人的主动对照试验（12项试验；RR 0.55，95% CI [0.38, 0.79]；I2 = 73.0%），以及同时排除所有三项试验（10项试验；RR 0.46，95% CI [0.31, 0.70]；I2 = 57.0%）后，该效应仍然显著。 多组分非药物干预与对照比较对谵妄发生率影响的森林图。 未观察到谵妄持续时间（MD -1.49天，95% CI [−3.67, 0.69]；I2 = 96.3%）、ICU住院时间（MD -2.90天，95% CI [−6.83, 1.03]；I2 = 97.8%）、住院时间（MD -1.09天，95% CI [−4.27, 2.10]；I2 = 42.8%）、ICU死亡率（RR 0.77，95% CI [0.48, 1.23]；I2 = 0%）、28天死亡率（RR 1.09，95% CI [0.32, 3.71]；I2 = 0%）、90天死亡率（RR 1.13，95% CI [0.38, 3.36]；I2 = 0%）或无谵妄天数（MD 0.10天，95% CI [−0.71, 0.91]；I2 = 0%）存在统计学显著效应。 住院死亡率以叙述性方式汇总。报告干预组4/60例死亡，对照组13/63例死亡。报告实施后146/1434例死亡，实施前145/1184例死亡（校正RR 0.81，95% CI 0.61–1.07）。 亚组分析 亚组分析显示，组分模式B干预显著降低谵妄发生率（RR 0.43，95% CI [0.23, 0.82]）。组分模式A（RR 0.62，95% CI [0.29, 1.33]）或组分模式C（RR 0.58，95% CI [0.31, 1.07]）的效应不显著。然而，亚组差异的正式检验无统计学显著性（p = 0.604），因此上述结果不能确立组分模式B的优效性（见 ）。 根据干预组分模式对谵妄发生率进行的亚组分析。 发表偏倚 发表偏倚评估针对纳入的13项研究中谵妄发生率这一主要结局进行。漏斗图的目视检查提示分布不对称（见 ），Egger随机效应回归：斜率 −2.756（95% CI -3.794 至 −1.718），p < 0.001。这些发现提示不能排除潜在的发表偏倚。 谵妄发生率的漏斗图。 证据确定性 谵妄发生率的证据确定性因偏倚风险和显著不一致性而被评定为低。次要结局的确定性范围为低至极低，主要原因是偏倚风险、不一致性和不精确性（）。 主要和次要结局的汇总结果与证据确定性。 讨论 本荟萃分析的目的是评估非药物干预对ICU谵妄的有效性，并确定特定的干预组合。这项更新研究纳入了13项随机对照试验，共5,525名参与者。与对照条件相比，多组分非药物干预
全文（开放获取） · 英文原文
【Abstract】 BACKGROUND: It is necessary to evaluate the efficacy of multicomponent non-pharmacological interventions in preventing and managing intensive care unit (ICU) delirium and identify the most effective intervention combinations. METHODS: A systematic literature search was conducted in PubMed, EMBASE, CINAHL, Cochrane Library, and ProQuest Dissertations and Theses A&I databases from inception to July 3, 2026, supplemented by a hand search of prior meta-analyses. Only randomized trials were included. Meta-analysis was performed using risk ratios (RR) or mean differences (MD) with 95% confidence intervals (CI). Random-effects models with restricted maximum likelihood estimation and modified Hartung-Knapp confidence intervals were used. Risk of bias was assessed using the Cochrane risk-of-bias tool for randomized controlled trials. The certainty of evidence was assessed using the GRADE approach. RESULTS: The search identified 13 randomized controlled trials encompassing 5,525 patients. Multicomponent interventions significantly reduced delirium incidence (RR 0.53, 95% CI [0.36, 0.76], I2 = 72.5%, low-certainty evidence). The effect remained significant after excluding the two cluster-randomized trials (RR 0.45, 95% CI 0.30-0.67; I2 = 55.7%), the active-control trial (RR 0.55, 95% CI 0.38-0.79; I2 = 73.0%), and all three trials simultaneously (RR 0.46, 95% CI 0.31-0.70; I2 = 57.0%). In exploratory subgroup analysis, the within-subgroup pooled estimate for component pattern B, comprising cognitive assessment and stimulation, reorientation, sleep and environmental management, family or community support, and physical activity, showed a reduced incidence of delirium (RR 0.43, 95% CI [0.23, 0.82]); however, the formal test for subgroup differences was not significant (p = 0.604). No statistically significant effects were found for delirium duration, ICU or hospital length of stay, delirium-free days, or mortality. CONCLUSION: This study demonstrated that multi-component interventions effectively reduce delirium incidence in critically ill patients. The evidence did not establish that one component combination was superior to another. SYSTEMATIC REVIEW REGISTRATION: https://www.crd.york.ac.uk/PROSPERO/view/CRD42024592821. 【Introduction】 Delirium is an acute-onset and fluctuating brain dysfunction characterized by impaired attention and awareness, and is a common complication among patients admitted to the intensive care unit (ICU) (; ). While the incidence of delirium in hospitalized patients is approximately 20–30% (), it increases significantly to 25–87% among ICU patients, depending on illness severity and diagnostic methods (; ; ; ). ICU delirium is associated with various adverse outcomes, including prolonged mechanical ventilation (), long-term cognitive decline (), extended ICU and hospital stays (), and increased workload for clinical nurses (; ; ). These risks highlight the necessity of implementing early preventive measures to improve prognoses in critically ill patients. Pharmacologic interventions, such as dexmedetomidine, atypical antipsychotics, and haloperidol, are commonly used for ICU delirium prevention and management due to their rapid onset, widespread availability, and potential to reduce hospital length of stay (; ; ). However, these treatments may carry significant risks, including arrhythmias and respiratory distress, particularly in vulnerable patients (; ; ). Additionally, studies have demonstrated limited efficacy of pharmacological interventions in reducing delirium incidence (). Bannon’s study state that insufficient evident to support single or multicomponent non-pharmacological interventions in reducing delirium rate or duration, but multi-component interventions may present a more credible opportunity to target several risk factors (). Current PADIS guidelines and recent research advocate for multicomponent non-pharmacological strategies as the most effective approach to prevent and manage ICU delirium (; ; ). To date, multiple meta-analyses have evaluated the effectiveness of single-component and multicomponent non-pharmacological interventions in preventing ICU delirium (; ; ; ). These studies consistently demonstrated that multicomponent non-pharmacological interventions significantly reduce the incidence of delirium. However, their effectiveness in shortening delirium duration remains a topic of debate. While three studies (; ; ) reported no significant impact of multi-component interventions on delirium duration, study suggested that specific intervention combinations could achieve a significant reduction. identified two effective multi-component intervention combinations for delirium prevention: (1) sleep promotion, cognitive stimulation, early activity, pain control, and assessment, and (2) sleep promotion and cognitive stimulation. However, their conclusions were based on a combined analysis of studies with varying designs, which may have resulted in inaccurate estimates of the comparative effectiveness of non-pharmacological strategies in reducing ICU delirium incidence. Given the availability of numerous randomized controlled trials (RCTs), an updated systematic review and meta-analysis focusing exclusively on RCTs is clinically relevant. Moreover, no study has systematically defined the specific implementation details of these interventions. A key innovation of this study is the comprehensive listing and integration of the implementation protocols for various interventions, providing precise and actionable guidance for clinical practice. This study aimed to conduct an updated systematic review and meta-analysis to evaluate the efficacy of multicomponent non-pharmacological interventions in preventing and managing ICU delirium, identify the most effective intervention combinations, address existing controversies regarding their impact on delirium duration and ICU stay, and offer evidence-based recommendations to enhance clinical practice. 【Methods】 【Study design and protocol】 This systematic review and meta-analysis adhered to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines and was registered with the International Prospective Register of Systematic Reviews (PROSPERO: CRD42024592821) (). 【Eligibility criteria (PICOS)】 We included studies based on the following criteria: (1) Types of studies: randomized controlled trials (RCTs); (2) Participants: ICU adult patients; (3) using multicomponent non-pharmacological interventions as the intervention group; (4) using either non-pharmacological interventions or routine care as the control group; (5) Research content: multicomponent non-pharmacological interventions prevention and treatment delirium. Studies were excluded based on the following criteria: (1) Reviews, letters, abstracts, conference proceedings, duplicate publications, or articles without full-text availability. (2) Non-English studies. (3) Studies involving neonatal or pediatric ICU patients. (4) Use of pharmacological interventions or single non-pharmacological interventions. 【Search strategy and data sources】 A comprehensive search was conducted across five electronic databases: PubMed, EMBASE, CINAHL, Cochrane Library, and ProQuest Dissertations and Theses A&I, covering studies from inception to July 3, 2026. The search used the following keyword combinations: (“critical care” OR “intensive care” OR “ICU”) AND (“non-pharmacologic” OR “non-pharmacological” OR “multicomponent” OR “bundle” OR “mobilization” OR “daily sedation interruption” OR “music” OR “light therapy” OR “reorientation” OR “cognitive stimulation” OR “cognitive training” OR “eye mask” OR “earplugs” OR “physical therapy” OR “occupational therapy” OR “multicomponent”) AND delirium. Additionally, the reference lists of relevant studies were manually reviewed to identify eligible studies for inclusion in this review. 【Study selection and data extraction】 Two researchers (Wang ST and Huang WJ) independently conducted the literature screening and data extraction. Any disagreements were resolved through discussion or consultation with a third researcher, if necessary. The screening process involved three steps: (1) reviewing titles and abstracts, (2) examining full texts, and (3) reviewing references of selected studies. At each step, the inclusion and exclusion criteria were applied consistently. Relevant data were extracted from the included articles using a standardized form. The extracted items included the first author, year of publication, study region, study design, sample size, average age, setting, types of interventions, type of center, and other relevant details. 【Intervention component classification】 After data extraction, the intervention content of each included trial was reviewed in detail. Based on established non-pharmacological approaches to ICU delirium prevention and the actual intervention elements reported in the included studies, the interventions were summarized into five components: cognitive assessment and stimulation, reorientation, sleep and environmental management, family or community support, and physical activity. The specific combination of these components was then used to classify interventions into three component patterns, as shown in , . This classification was used to support the exploratory subgroup analysis and should not be interpreted as an assessment of the independent effect of any single component. Classification of intervention component patterns used for exploratory subgroup analysis. Components of non-pharmacological interventions. Use proven ICU delirium and sedation identification tools such as CAM-ICU, CAM, RASS etc. Intense external stimulation with an objective of increasing the patient’s level of alertness. Pool of exercises with an objective of activating mental functions, which include the following areas, among others: alertness, visual perception, memory, calculus, problem solving, praxis, and language. Each patient received a notebook for cognitive exercises, as well as tools such as sequencing cards and games (cards, dominoes and memory, and visuospatial construction games). Correction of sensory deficits, such as encouraging patients to use glasses and hearing aids, protects the cornea from dryness and cleans the ear canal. Encourage positive communication and use an appropriate communication technique (verbal/written/pictures) to compensate sensory loss and overcome language barriers. Allow patients watch TV and listen to preferred music or read a newspaper/magazine when possible. Daily spontaneous awakening trial (SAT) or Daily spontaneous breathing trial (SBT) was given for the patient with mechanical ventilated. Provide visible clock, calendar, cell phone, and schedule for patients. Inform the department, the hospital and the reason for admission. Specific measures to maintain a good sleep–wake cycle were adopted. For example, the room was illuminated by natural light during the day and by downy light at night. All nursing procedures were minimized in the night. Noise was decreased as much as possible. For example, various alert tones of the alarm system were modulated to be gentler. Provide an eyeshade and acoustic earplugs for patients in sleeping to a good sleep. Limited movement to other beds, individualized their bedside area, and allow to bring home favorites (pillow/family photo). The staff invited the patients to visit the ICU to become acquainted with the environment. Extend family visits as long as possible and encourage family involvement in the patients’ daily care. Use effective communication devices to talk to family members. Identify patients’ needs and possible causes of stress and address them with the help of family members. Encourage early mobilization and plan mobility schedule. Basic activities of daily living (BADLs), group of activities that promotes independent living based on BADL that include hygiene, personal grooming, and eating. Mobilize the patient each shift. If −2 to −4 RASS fowler position, massage, exercise on bed, position changes each 2 h, moving joints pedaling. If −1 to 2 RASS, fowler position, chairs sitting, standing, small steps with help. Minimizing and optimally locating restraining lines. 【Risk-of-bias assessment】 The methodological quality of the included studies was assessed independently by two reviewers (Wang ST and Huang WJ) using the Cochrane risk-of-bias tool for randomized controlled trials (). Risk of bias was evaluated across seven domains: random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective reporting, and other potential sources of bias. Each domain was judged as low, unclear, or high risk of bias. Disagreements were resolved through discussion or consultation with a third reviewer. 【Certainty of evidence】 The certainty of evidence for each outcome was assessed using the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) approach (). As all included studies were randomized controlled trials, the certainty of evidence was initially rated as high and then downgraded, where appropriate, for risk of bias, inconsistency, indirectness, imprecision, and publication bias. The final certainty of evidence was classified as high, moderate, low, or very low. 【Data analysis】 Data were extracted from the original studies, and the updated meta-analysis was conducted using R version 4.4.2. Risk ratios (RRs) were used for dichotomous outcomes, and mean differences (MDs) for continuous outcomes. Data reported in formats other than means and standard deviations were converted only when clinically and statistically appropriate. Effect estimates are presented with 95% confidence intervals (CIs). Random-effects meta-analyses were performed using restricted maximum likelihood estimation and the modified Hartung–Knapp method. Heterogeneity was assessed using the χ2 test and quantified using the I2 statistic. For the two cluster-randomized trials, effect estimates adjusted for clustering were used. One trial reported an adjusted odds ratio, which was converted to an approximate risk ratio using the observed event risk in the control group; its standard error was estimated using the delta method. The robustness of the primary analysis was examined by separately excluding the two cluster-randomized trials and the active-control trial, followed by an analysis excluding all three trials simultaneously. Potential publication bias and small-study effects for delirium incidence were assessed through visual inspection of the funnel plot and Egger’s regression test. 【Results】 【Study selection】 A total of 13,086 articles were identified through electronic database searches, and an additional four articles were retrieved through hand-searching strategies. After the removal of 156 duplicate records and marked 62 ineligible Records by automation tools, the titles and abstracts of 12,868 articles were screened, resulting in the exclusion of 12,831 articles that did not meet the inclusion criteria. The remaining 37 full-text articles were assessed for eligibility, with 24 excluded due to reasons such as inclusion of non-ICU patients, non-randomized controlled trials, non-multicomponent non-pharmacological interventions, or failure to report the incidence of delirium. After reviewing reference lists of the eligible studies, 13 articles met the criteria and were included in the systematic review (see ). PRISMA 2020 flow diagram for multicomponent non-pharmacological interventions comparison in reducing ICU delirium. 【Characteristics of the included studies】 The review incorporated 13 randomized controlled trials, including two stepped-wedge cluster randomized trials, with sample sizes ranging from 40 to 2,618 participants, totaling 5,525 individuals. These studies were published from 2015 to 2026. All studies reported data on delirium incidence. The basic characteristics of the included studies are detailed in . Characteristics of studies. ICU, Intensive Care Unit; RCT, randomized controlled trial; I, Intervention; C, Control; SICU, Surgical Intensive Care Unit; MICU, Medical ICU; NICU, Neurological ICU; CSICU, cardiovascular surgery ICU. ① Cognitive assessment and stimulation; ② Reorientation; ③ Management of sleep and environment; ④ Support from family and community; ⑤ Physical activity. 【Risk-of-bias assessment】 The risk-of-bias assessment identified six studies at low risk of bias, three at unclear risk, and four at high risk (see ). The main concerns were related to the difficulty of blinding participants and personnel in non-pharmacological interventions, incomplete reporting of allocation procedures in some trials, and potential concerns regarding outcome assessment or selective reporting in a small number of studies. Missing outcome data was generally judged to have a low risk of bias. Risk-of-bias assessment of the included studies. 【Synthesized findings】 【Pooled analysis】 The pooled analysis of 13 trials involving 5,525 ICU patients showed that multicomponent non-pharmacological interventions significantly reduced delirium incidence (RR 0.53, 95% CI [0.36, 0.76]; I2 = 72.5%) (). Sensitivity analyses yielded consistent results. The effect remained significant after excluding the two cluster-randomized trials (11 trials; RR 0.45, 95% CI [0.30, 0.67]; I2 = 55.7%), the active-control trial by Dong et al. (12 trials; RR 0.55, 95% CI [0.38, 0.79]; I2 = 73.0%), and all three trials simultaneously (10 trials; RR 0.46, 95% CI [0.31, 0.70]; I2 = 57.0%). Forest plot of multicomponent non-pharmacological interventions versus control for delirium incidence. No statistically significant effects were observed for delirium duration (MD -1.49 days, 95% CI [−3.67, 0.69]; I2 = 96.3%), ICU length of stay (MD -2.90 days, 95% CI [−6.83, 1.03]; I2 = 97.8%), hospital length of stay (MD -1.09 days, 95% CI [−4.27, 2.10]; I2 = 42.8%), ICU mortality (RR 0.77, 95% CI [0.48, 1.23]; I2 = 0%), 28-day mortality (RR 1.09, 95% CI [0.32, 3.71]; I2 = 0%), 90-day mortality (RR 1.13, 95% CI [0.38, 3.36]; I2 = 0%), or delirium-free days (MD 0.10 days, 95% CI [−0.71, 0.91]; I2 = 0%). Hospital mortality was summarized narratively. reported 4/60 deaths in the intervention group and 13/63 in the control group. reported 146/1434 deaths after implementation and 145/1184 before implementation (adjusted RR 0.81, 95% CI 0.61–1.07). 【Subgroup analysis】 Subgroup analyses revealed that component pattern B interventions significantly reduced delirium incidence (RR 0.43, 95% CI [0.23, 0.82]). The effects were not significant for component pattern A (RR 0.62, 95% CI [0.29, 1.33]) or component pattern C (RR 0.58, 95% CI [0.31, 1.07]). However, the formal test for subgroup differences was not significant (p = 0.604), and therefore the findings do not establish superiority of component pattern B (see ). Subgroup analysis of delirium incidence according to intervention component patterns. 【Publication bias】 Publication bias was assessed for the primary outcome of delirium incidence across the 13 included studies. Visual inspection of the funnel plot indicated an asymmetric distribution (see ), and Egger random-effects regression: slope −2.756 (95% CI -3.794 to −1.718), p < 0.001. These findings suggest that potential publication bias could not be excluded. Funnel plot of delirium incidence. 【Certainty of evidence】 The certainty of evidence for delirium incidence was rated as low because of risk of bias and substantial inconsistency. Certainty for the secondary outcomes ranged from low to very low, mainly because of risk of bias, inconsistency, and imprecision (). Summary of findings and certainty of evidence for primary and secondary outcomes. 【Discussion】 The objective of this meta-analysis was to evaluate the effectiveness of non-pharmacological interventions for ICU delirium and identify specific intervention combinations. This updated study included 13 randomized controlled trials with 5,525 participants. Compared with control conditions, multicomponent non-pharmacological intervent