The interplay between orthopedic events and dementia is complicated. A recent Japanese study by Matsumoto et al. revealed that individuals with early-onset dementia faced an increased likelihood of experiencing osteoporotic fractures compared to those without cognitive impairments.¹ As for the reverse direction of orthopedic events–dementia association, integrated evidence showed that osteoarthritis patients were at higher risk of developing dementia and cognitive impairments.² However, whether orthopedic surgical interventions, such as joint arthroplasty, influence the onset of dementia remains controversial. Cognitive dysfunction among individuals who underwent total joint arthroplasty was evaluated in recent studies.^{3, 4} A US study reported that individuals who had undergone total joint replacement demonstrate no significant disparity in the degree of cognitive decline compared to controls.⁴ Nonetheless, the existing body of literature predominantly comprises small-scale investigations; large-scale real-world studies investigating whether patients undergoing joint replacement surgery are susceptible to dementia based on the US population are lacking. Therefore, we performed an analysis in the TriNetX research network to evaluate the risk of new-onset dementia following total knee replacement (TKR) surgery among patients with osteoarthritis.

We utilized subsets of the TriNetX network, including the US collaborative network, the Global Collaborative Network, and the EMEA collaborative network. The US network comprises 64 institutions in the United States, covering data from over 18 million patients, and has been widely utilized in epidemiological studies.^{5, 6} The Global network includes 119 institutions across 19 countries, with data from over 120 million patients. The EMEA network consists of 25 institutions across nine European countries, with data from over 14 million patients. The TriNetX database undergoes monthly updates, ensuring up-to-date information. Participants diagnosed with osteoarthritis between January 1, 2005, and December 31, 2018, with more than two visit records were included. The TKR group comprised osteoarthritis patients with a TKR record. Propensity score matching in a 1:1 ratio (adjusted for covariates such as age, sex, race, socioeconomic status, comorbidities, medications, laboratory data, medical utilization status, and substance use) was done to determine controls. Exclusion criteria included individuals under 18, those with a history of neurocognitive disorders or neoplasms, and individuals deceased before or after the index date. The primary outcome was incident dementia, with participants followed from 3 months after the index date until dementia occurrence, last visit, or December 31, 2023, whichever came first. After matching, there were 38,358 TKR patients and the same number of non-TKR controls included for further analysis. All sensitivity analyses and stratification analyses were performed based on the US collaborative network. Given the significant influence of age and sex on dementia risk, we conducted stratification analyses based on these factors. Age subgroups included 18 to 64 years, 65 to 79 years, and over 80 years. Also, to enhance internal validity, sensitivity analyses were conducted using different approaches: (1) Varying wash-out periods: To mitigate the impact of reversed causation, we applied wash-out periods of 12, 24, and 36 months. Incidents within these periods were excluded. (2) Varying follow-up periods: As the main analysis had a 5-year follow-up, we examined shorter durations of 1 and 3 years after the index date to assess their effect on incident dementia. (3) Employing different matching algorithms: To counter potential overmatching bias, we explored models without propensity score matching and those with fewer covariates. Hazard ratios (HRs) and 95% confidence intervals (CIs) were calculated using Cox proportional hazards models via the TriNetX analytical platform. Ethical approval was exempted by the Institutional Review Board (IRB) of Tungs’ Taichung MetroHarbor Hospital (IRB TTMHH No.: 112208N). In the US population, TKR was associated with a 26% reduced risk of dementia (HR = 0.740; 95% CI: 0.647 to 0.846). This trend persisted across sensitivity analyses using different datasets, wash-out periods, and matching algorithms. Stratification by gender showed a 35.2% and 29.5% risk reduction in male and female TKR patients, respectively. Among patients over 80, a 30% reduction in dementia risk was observed (HR = 0.697; 95% CI: 0.596 to 0.814). In 1- and 3-year follow-ups, HRs were 0.473 (95% CI: 0.326 to 0.685) and 0.541 (95% CI: 0.448 to 0.654), respectively (Figure 1).

The results indicate a possible protective influence of TKR against dementia onset in osteoarthritis patients. Teipel et al. reported a decrease in dementia risk when follow-up began four quarters after joint replacement surgery.⁷ In our report, despite the limitation of residual confounders, the significance of reduced dementia risk persisted across different follow-up periods. Possible mechanisms for this relationship include TKR's pain relief and improved mobility, which could lead to increased physical activity and social interaction, known to reduce dementia risk.^{8, 9} However, further research is needed to confirm these mechanisms.

骨科事件和痴呆症之间的相互作用很复杂。 Matsumoto 等人最近的一项日本研究。研究表明，与没有认知障碍的人相比，早发性痴呆症患者发生骨质疏松性骨折的可能性更高。 ¹至于骨科事件的反向-痴呆关联，综合证据表明骨关节炎患者发生痴呆和认知障碍的风险较高。 ²然而，关节置换术等骨科手术干预措施是否会影响痴呆症的发病仍存在争议。最近的研究对接受全关节置换术的个体的认知功能障碍进行了评估。 ^{3, 4}美国的一项研究报告称，与对照组相比，接受全关节置换术的人认知能力下降的程度没有显着差异。 ⁴尽管如此，现有文献主要包括小规模调查；目前还缺乏以美国人口为基础调查接受关节置换手术的患者是否容易患痴呆症的大规模现实世界研究。因此，我们在 TriNetX 研究网络中进行了一项分析，以评估骨关节炎患者全膝关节置换 (TKR) 手术后新发痴呆的风险。

我们利用了 TriNetX 网络的子集，包括美国协作网络、全球协作网络和 EMEA 协作网络。美国网络由美国64家机构组成，覆盖超过1800万患者的数据，已广泛应用于流行病学研究。 ^{5, 6}全球网络包括 19 个国家的 119 个机构，拥有超过 1.2 亿患者的数据。 EMEA 网络由 9 个欧洲国家的 25 个机构组成，拥有超过 1400 万患者的数据。 TriNetX 数据库每月更新一次，确保信息最新。纳入2005年1月1日至2018年12月31日期间诊断患有骨关节炎且有两次以上就诊记录的参与者。 TKR 组由有 TKR 记录的骨关节炎患者组成。以 1:1 的比例进行倾向评分匹配（根据年龄、性别、种族、社会经济地位、合并症、药物、实验室数据、医疗利用状况和物质使用等协变量进行调整）以确定对照。排除标准包括 18 岁以下的个体、有神经认知障碍或肿瘤病史的个体以及在索引日期之前或之后死亡的个体。主要结局是痴呆症事件，参与者从索引日期后 3 个月起进行随访，直至痴呆症发生、最后一次就诊或 2023 年 12 月 31 日（以先到者为准）。匹配后，共有 38,358 名 TKR 患者和相同数量的非 TKR 对照纳入进一步分析。所有敏感性分析和分层分析均基于美国协作网络进行。 鉴于年龄和性别对痴呆风险的显着影响，我们根据这些因素进行了分层分析。年龄分组包括 18 至 64 岁、65 至 79 岁和 80 岁以上。此外，为了增强内部有效性，使用不同的方法进行敏感性分析：（1）不同的清除期：为了减轻反向因果关系的影响，我们采用了 12、24 和 36 个月的清除期。这些时期内发生的事件被排除在外。 (2) 不同的随访期：由于主要分析进行了 5 年随访，因此我们检查了索引日期后 1 年和 3 年的较短持续时间，以评估其对痴呆事件的影响。 (3)采用不同的匹配算法：为了应对潜在的过度匹配偏差，我们探索了没有倾向得分匹配和协变量较少的模型。通过 TriNetX 分析平台，使用 Cox 比例风险模型计算风险比 (HR) 和 95% 置信区间 (CI)。获得同氏台中港湾医院机构审查委员会 (IRB) 的伦理批准（IRB TTMHH 编号：112208N）。在美国人群中，TKR 与痴呆风险降低 26% 相关（HR = 0.740；95% CI：0.647 至 0.846）。这种趋势在使用不同数据集、清除期和匹配算法的敏感性分析中持续存在。按性别分层显示，男性和女性 TKR 患者的风险分别降低了 35.2% 和 29.5%。在 80 岁以上的患者中，痴呆风险降低了 30%（HR = 0.697；95% CI：0.596 至 0.814）。在 1 年和 3 年随访中，HR 分别为 0.473（95% CI：0.326 至 0.685）和 0.541（95% CI：0.448 至 0.654）（图 1）。

结果表明 TKR 对骨关节炎患者痴呆症的发生可能具有保护作用。泰佩尔等人。报告称，在关节置换手术后四个季度开始随访时，痴呆症风险有所下降。 ⁷在我们的报告中，尽管存在残余混杂因素的限制，但痴呆风险降低的重要性在不同的随访期间仍然存在。这种关系的可能机制包括 TKR 缓解疼痛和改善活动能力，这可能会导致身体活动和社交互动的增加，从而降低痴呆风险。 ^{8, 9}然而，需要进一步研究来证实这些机制。