颅脑外伤后发生高血糖症的危险因素及预测模型的建立

Establishment of risk factors and predictive models for hyperglycemia after traumatic brain injury

摘要:

目的 探讨颅脑外伤后发生高血糖症的危险因素，并建立预测模型。方法 回顾性分析2022年1月—2023年6月南京医科大学附属无锡人民医院收治的90例颅脑外伤患者的临床资料，根据临床诊断应激性高血糖标准分为非高血糖症组65例和高血糖症组25例。比较两组患者的临床资料；采用单因素分析及多因素逐步Logistic回归模型分析颅脑外伤后发生高血糖症的危险因素；构建颅脑外伤后发生高血糖症预测模型；绘制受试者工作特征（ROC）曲线分析预测模型的诊断价值。结果 高血糖症组患者年龄≥45岁、重度损伤、使用葡萄糖、APACHE Ⅱ评分≥15分、有意识障碍、脑水肿占比高于非高血糖症组（P <0.05）。多因素逐步Logistic回归分析结果显示：年龄≥45岁［O^R =1.768（95% CI：1.129，2.769）］、重度损伤［O^R =1.773 （95% CI：1.307，2.405）］、使用葡萄糖［O^R =1.891 （95% CI：1.356，2.637）］、APACHE Ⅱ评分≥15分［O^R =1.232 （95% CI：1.053，1.441）］、有意识障碍［O^R =1.578 （95% CI：1.254，1.986）］、有脑水肿［O^R =517 （95% CI：1.224，1.880）］均是颅脑外伤后发生高血糖症的危险因素（P <0.05）。根据多因素Logistic回归分析结果，将年龄、损伤程度、使用葡萄糖、APACHE Ⅱ评分、意识障碍、脑水肿纳入预测模型，Logit（P）=-3.758+0.570×年龄+0.573×损伤程度+0.637×使用葡萄糖+0.209×APACHE Ⅱ评分+0.456×意识障碍+0.417×脑水肿。结果表明，预测模型预测颅脑外伤后发生高血糖症的正确率为85.6%（95% CI：0.826，0.906），敏感性为90.3%（95% CI：0.842，0.925），特异性为94.1%（95% CI：0.874，0.953）。结论 颅脑外伤后发生高血糖症受年龄、损伤程度、使用葡萄糖、APACHE Ⅱ评分、意识障碍、脑水肿等情况的影响，而建立预测模型可对高血糖症的发生进行有效评估。

Abstract:

Objective To explore the risk factors for the occurrence of hyperglycemia after traumatic brain injury (TBI) and establish a predictive model.Methods A retrospective analysis was conducted on the clinical data of 90 TBI patients admitted to Wuxi People's Hospital affiliated with Nanjing Medical University from January 2022 to June 2023. Patients were categorized into the non-hyperglycemia group (65 cases) and hyperglycemia group (25 cases) based on clinical diagnostic criteria for stress-induced hyperglycemia. Clinical data were compared between the two groups. Univariate and multivariate stepwise logistic regression analyses were employed to identify risk factors for hyperglycemia after TBI. A predictive model for hyperglycemia following TBI was constructed, and the diagnostic value of the model was analyzed using receiver operating characteristic (ROC) curve.Results In the hyperglycemia group, patients aged ≥ 45 years, severe injury, use of glucose, APACHE Ⅱ score ≥ 15, consciousness disorders, and higher brain edema percentage were significantly more common compared to the non-hyperglycemia group (P < 0.05). Multivariate stepwise logistic regression analysis revealed that age ≥ 45 years [O^R = 1.768 (95% CI: 1.129, 2.769) ], severe injury [O^R = 1.773 (95% CI: 1.307, 2.405) ], use of glucose [O^R = 1.891 (95% CI: 1.356, 2.637) ], APACHE Ⅱ score ≥ 15 [O^R = 1.232 (95% CI: 1.053, 1.441) ], consciousness disorders [O^R = 1.578 (95% CI: 1.254, 1.986) ], and brain edema [O^R = 517 (95% CI: 1.224, 1.880) ] were all risk factors for hyperglycemia after TBI (P < 0.05). Based on the multivariate logistic regression analysis, age, injury severity, use of glucose, APACHE Ⅱ score, consciousness disorders, and brain edema were included in the predictive model, Logit (P) = -3.758 + 0.570 × age + 0.573 × injury severity + 0.637 × use of glucose + 0.209 × APACHE Ⅱ score + 0.456 × consciousness disorders + 0.417 × brain edema. The predictive model demonstrated an accuracy of 85.6% (95% CI: 0.826, 0.906), sensitivity of 90.3% (95% CI: 0.842, 0.925), and specificity of 94.1% (95% CI: 0.874, 0.953) in predicting hyperglycemia after TBI.Conclusion The occurrence of hyperglycemia after TBI is influenced by factors such as age, injury severity, use of glucose, APACHE Ⅱ score, consciousness disorders, and brain edema. Establishing a predictive model can effectively assess the occurrence of hyperglycemia.