应激指数（血糖钾比值）作为创伤性脑损伤结局的预测指标：系统评价与荟萃分析。
完整摘要 · 中文译文
背景：创伤性脑损伤（TBI）是死亡和残疾的主要原因。葡萄糖钾比值（GPR）或应激指数（SI）反映应激性高血糖和低钾血症，可能改善超越临床和影像学评分的早期预后评估。 方法：我们按照PRISMA指南对成人（≥18岁）CT确诊的TBI观察性队列研究进行了系统综述和荟萃分析，这些研究报告了入院时血糖和血钾及其与死亡率或功能结局/"不良结局"的关联。随机效应模型汇总了二分类结局的比值比（OR）和连续GPR值的均数差（MD）；异质性采用I2评估。预设亚组分析按样本量（>500 vs. <500）进行，并通过敏感性分析（逐一剔除法；固定效应 vs. 随机效应）和漏斗图检查来探索稳健性和小样本研究效应。 结果：三项研究（4,550例患者）纳入了高GPR与低GPR对比及死亡率的主要比较。高GPR与死亡几率增加相关（汇总OR 11.97，95% CI 2.68-53.37；p = 0.001；I2 = 88%）。在四项研究中，不良结局患者的入院GPR高于良好结局患者（汇总MD 16.82，95% CI 12.70-20.94；p < 0.00001；I2 = 61%）。在五项评估GPR作为预后预测指标的研究中，不良结局的汇总校正OR为2.10（95% CI 1.51-2.93；p < 0.0001；I2 = 37%）。效应在小型单中心队列中较大，在大型注册登记研究中较为温和但仍显著；样本量解释了大部分 研究间存在异质性。漏斗图不对称提示可能存在小样本效应/发表偏倚。 结论：入院时GPR/SI升高与成人TBI死亡率和不良结局增加相关，并可能反映早期全身应激。作为一项源自常规实验室检查的床旁指标，它是早期风险分层的有前景的辅助工具。然而，异质性、小样本效应和观察性证据需要谨慎解读；需要大型多中心前瞻性研究来验证标准化阈值及其相对于现有预后模型的增量价值。
完整摘要 · 英文原文
BACKGROUND: Traumatic brain injury (TBI) is a major cause of death and disability. The glucose-to-potassium ratio (GPR), or stress index (SI), reflects stress hyperglycemia and hypokalemia and may improve early prognostication beyond clinical and radiologic scores. METHODS: We performed a PRISMA-guided systematic review and meta-analysis of observational cohort studies in adults (≥18 years) with CT-confirmed TBI reporting admission glucose and potassium and their association with mortality or functional/"poor outcome." Random-effects models pooled odds ratios (ORs) for dichotomous outcomes and mean differences (MDs) for continuous GPR values; heterogeneity was assessed with I2. Prespecified subgroup analyses were conducted by sample size (>500 vs. < 500), with sensitivity analyses (leave-one-out; fixed vs. random) and funnel-plot inspection to explore robustness and small-study effects. RESULTS: Three studies (4,550 patients) contributed to the primary comparison of high versus low GPR and mortality. High GPR was associated with increased odds of death (pooled OR 11.97, 95% CI 2.68-53.37; p = 0.001; I2 = 88%). Across four studies, patients with poor outcomes had higher admission GPR than those with good outcomes (pooled MD 16.82, 95% CI 12.70-20.94; p < 0.00001; I2 = 61%). In five studies evaluating GPR as a prognostic predictor, the pooled adjusted OR for adverse outcome was 2.10 (95% CI 1.51-2.93; p < 0.0001; I2 = 37%). Effects were larger in small single-center cohorts and more moderate, but significant, in the large registry-based study; sample size explained much of the between-study heterogeneity. Funnel-plot asymmetry suggested possible small-study effects/publication bias. CONCLUSION: Elevated admission GPR/SI is associated with increased mortality and poor outcome in adult TBI and may capture early systemic stress. As a bedside metric derived from routine laboratory tests, it is a promising adjunct for early risk stratification. However, heterogeneity, small-study effects, and observational evidence warrant cautious interpretation; large multicenter prospective studies are needed to validate standardized thresholds and incremental value over existing prognostic models.