烧伤后慢性神经痛:解剖学方法与风险预测模型的开发与验证
研究比较11个解剖部位的差异,采用改良德尔菲法筛选14个潜在风险变量,通过多变量回归等方法构建预测模型。最终模型包含8个变量,能较准确预测烧伤后6个月慢性神经痛风险。
要点
- 烧伤后6个月慢性神经痛风险可通过8变量模型预测。
- 上臂烧伤、大腿烧伤、年龄、吸烟等是较强预测因素。
- 模型区分度AUC 0.75,拟合良好,适用于成人烧伤患者。
核心结论
- 研究识别了11个解剖部位烧伤后神经痛的风险因素。
- 构建的模型能准确预测成人烧伤后6个月慢性神经痛风险。
- 年龄、吸烟、物质滥用、酒精滥用、上臂烧伤、大腿烧伤、烧伤手术次数和住院时长是最强预测因素。
研究设计与样本
- 回顾性分析2014至2019年一家烧伤中心收治的1862名成人。
- 采用改良德尔菲法选择14个潜在风险变量,使用多变量回归、Brier评分、AUC、Hosmer-Lemeshow拟合优度及分层K折交叉验证开发模型。
- 慢性疼痛定义为出院后持续6个月及以上的疼痛。
主要发现
- 开发队列神经痛患病率为6.1%,验证队列为5.4%。
- Brier评分=0.15;分层K折交叉验证(K=10)AUC=0.75(95%CI 0.68-0.81);Hosmer-Lemeshow拟合优度p=0.73(n=10组)。
- 最终方程包含8个变量:风险评分=-6.3+0.02(年龄)+1.77(吸烟)+1.04(物质滥用)+0.67(酒精滥用)+0.84(上臂烧伤)+1.28(大腿烧伤)+0.21(烧伤手术次数)+0.01(住院时长)。
- 预测概率=1-1/[1+exp(风险评分)],随着风险因素数量增加,疼痛概率增加。
关键数值与推荐
- 慢性疼痛定义为出院后持续6个月及以上的疼痛。
- 风险评分公式及预测概率计算方法如上。
- 证据等级为风险,III级。
临床怎么用
- 该模型可用于指导烧伤患者的医疗和/或手术管理决策。
- 对于具有上臂烧伤、大腿烧伤、吸烟、物质滥用等高危因素的患者,可加强疼痛监测和预防干预。
局限性
- 回顾性单中心研究,样本来自一家烧伤中心。
- 模型需外部验证。
- 原文未报告其他局限性。
以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。
原文
BACKGROUND: A model that predicts a patient's risk of developing chronic, burn-related nerve pain may guide medical and/or surgical management. This study determined anatomy-specific variables and constructed a mathematical model to predict a patient's risk of developing burn-related nerve pain. METHODS: A retrospective analysis was conducted from 1862 adults admitted to a burn center from 2014 to 2019. One hundred thirteen patients developed burn-related nerve pain. Comparisons were made using 11 anatomy-specific locations between patients with and without burn-related nerve pain. The modified Delphi technique was used to select 14 potential risk variables. Multivariate regression techniques, Brier scores, area under the curve, Hosmer-Lemeshow goodness-of-fit, and stratified K-fold cross-validation was used for model development. Chronic pain was defined as pain lasting 6 or more months after release from the Burn Center. RESULTS: Prevalence rates of burn-related nerve pain were similar in the development (6.1 percent) and validation (5.4 percent) cohorts [Brier score = 0.15; stratified K-fold cross-validation (K = 10): area under the curve, 0.75; 95 percent CI, 0.68 to 0.81; Hosmer-Lemeshow goodness-of-fit, p = 0.73; n = 10 groups]. Eight variables were included in the final equation. Burn-related nerve pain risk score = -6.3 + 0.02 (age) + 1.77 (tobacco use) + 1.04 (substance abuse) + 0.67 (alcohol abuse) + 0.84 (upper arm burn) + 1.28 (thigh burn) + 0.21 (number of burn operations) + 0.01 (hospital length-of-stay). Burn-related nerve pain predicted probability = 1 - 1/[1 + exp(burn-related nerve pain risk score)] for 6-month burn-related nerve pain risk score. As the number of risk factors increased, the probability of pain increased. CONCLUSIONS: Risk factors were identified for developing burn-related nerve pain at 11 anatomical locations. This model accurately predicts a patient's risk of developing burn-related nerve pain at 6 months. Age, tobacco use, substance abuse, alcohol abuse, upper arm burns, thigh burns, the number of burn operations, and hospital length of stay represented the strongest predictors. CLINICAL QUESTION/LEVEL OF EVIDENCE: Risk, III.
引用
- Klifto KM, Yesantharao PS, Lifchez SD, et al. Chronic Nerve Pain after Burn Injury: An Anatomical Approach and the Development and Validation of a Model to Predict a Patient's Risk. Plast Reconstr Surg. 2021. PMID: 34550938. doi:10.1097/PRS.0000000000008315
供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。