产后慢性腰痛的预测模型建立及验证

Establishment and validation of prediction model for postpartum chronic low back pain

目的构建产后慢性腰痛的预测模型并进行验证。方法选取2019年3月~2021年3月于本院生产的产妇147例为建模组,2021年4月~2022年4月生产的产妇56例为验证组。对建模组的临床资料进行单因素和多因素Logistic回归分析,构建预测模型;采用受试者工作特征曲线(receiver operating characteristic,ROC)及Hosmer-Lemeshow检验验证模型的区分度和校准度,并在验证组中验证模型的预测价值。结果建模组产后慢性腰痛41例,发生率为27.89%。多因素Logistic回归分析结果显示,生育次数(OR=2.860,95%CI:1.050~7.778)、分娩方式(OR=2.844,95%CI:1.279~6.321)、孕期体质量增加(OR=2.604,95%CI:1.043~6.498)、妊娠期间腰痛史(OR=3.571,95%CI:1.608~7.933)为产后慢性腰痛的影响因素(P<0.05)。据此构建产后慢性腰痛的预测模型如下:Prob=1/e^(^-Y),Y=1.806-1.051×生育次数-1.045×分娩方式-0.957×孕期体质量增加-1.273×妊娠期间腰痛史。ROC曲线分析结果显示,该模型的敏感度为82.54%,特异度为76.19%,曲线下面积(area under the curve,AUC)为0.831(95%CI:0.761~0.888);在验证组中验证该模型,敏感度、特异度和AUC分别为79.37%、73.81%、0.801(95%CI:0.727~0.862),总体预测准确率为85.71%,Hosmer-Lemeshow检验显示该模型有较好的校准度(χ^(2)=7.531,P=0.406)。结论生育次数、分娩方式、孕期体质量增加、妊娠期间腰痛史均为影响产后慢性腰痛发生的影响因素,基于以上影响因素构建的预测模型具有良好的预测效能,可为临床早期识别产后慢性腰痛的高风险产妇提供参考。

Objective To establish a prediction model for postpartum chronic low back pain,and validate it.Methods A total of 147 parturients who gave birth in our hospital between March 2019 and March 2021 were selected as the modeling group,and 56 parturients who gave birth between April 2021 and April 2022 as the validation group.The general clinical data of the modeling group were analyzed by univariate analysis and multivariate logistic regression analysis.A prediction model was established based on the analysis results.The receiver operating characteristic(ROC)curve and Hosmer-Lemeshow test were used to validate the discrimination and calibration of the model.The predictive value of the model in the validation group was validated.Results There were 41 cases of postpartum chronic low back pain in the modeling group,with an incidence of 27.89%.Multivariate logistic regression analysis showed that parity(OR=2.860,95%CI:1.050~7.778),delivery mode(OR=2.844,95%CI:1.279~6.321),weight gain during pregnancy(OR=2.604,95%CI:1.043~6.498),and history of low back pain during pregnancy(OR=3.571,95%CI:1.608~7.933)were influencing factors for postpartum chronic low back pain(P<0.05).The established prediction model for postpartum chronic low back pain was as follows:Prob=1/e(^-Y),Y=1.806-1.051×parity-1.045×delivery mode-0.957×weight gain during pregnancy-1.273×history of low back pain during pregnancy.ROC curve analysis results showed that the sensitivity,specificity and the area under the curve(AUC)of the model was 82.54%,76.19%and 0.831(95%CI:0.761~0.888)respectively.The sensitivity,specificity and AUC of the model in the validation group was 79.37%,73.81%and 0.801(95%CI:0.727~0.862)respectively.The overall prediction accuracy was 85.71%,and the Hosmer-Lemeshow test showed that the model had good calibration(χ^(2)=7.531,P=0.406).Conclusion Parity,delivery mode,weight gain during pregnancy,and history of low back pain during pregnancy are influencing factors for postpartum chronic low back pain.The prediction model established based on these influencing factors has good predictive performance,which can provide reference for early clinical identification of high-risk parturients for postpartum chronic low back pain.