基于儿童营养相关指标的反复呼吸道感染预测模型构建与验证

目的：探讨基于儿童营养相关指标的反复呼吸道感染预测模型构建及验证。方法：选取2018年1月至2020年3月温州医科大学附属第一医院收治的283例呼吸道感染患儿为研究对象，均接受规范治疗，根据呼吸道感染是否反复发作分为反复呼吸道感染组和非反复呼吸道感染组，比较患儿的临床资料，应用Logistic回归模型分析反复呼吸道感染相关因素，随机选取70%（198例）的病例作为训练集，30%（85例）作为测试集，并用R软件基于训练集建立列线图预测模型，采用决策曲线分析评估Nomogram模型临床净收益，并在测试集评估模型效能。结果：两组慢性呼吸系统疾病家族史比例、出生体质量、户外活动时间、母乳喂养时间、血红蛋白、维生素A、维生素D水平对比差异有统计学意义（P＜0.05）；性别、年龄、体质量指数、居住地、出生方式、是否窒息、睡眠时间、白细胞计数、血清锌、铁、钙、镁对比差异无统计学意义（P＞0.05）；多因素Logistic回归分析显示，出生体质量提高（OR=0.995，P=0.004）、户外活动时间延长（OR=0.630，P=0.003）、血红蛋白水平升高（OR=0.846，P＜0.001）、维生素A水平升高（OR=0.437，P＜0.001）、维生素D水平升高（OR=0.938，P=0.007）为患儿反复呼吸道感染的保护因素；构建的Nomogram模型预测反复呼吸道感染相关因素C-index为0.869（95%CI=0.818～0.920），决策曲线分析评估两组预测模型的临床收益，当含营养相关指标模型预测患儿反复呼吸道感染的值为0～0.59时，可提供附加临床收益。外部验证显示敏感度为92.00%，特异度为96.67%，准确率为95.29%。结论：出生体质量提高、户外活动时间延长、血红蛋白水平升高、维生素A水平升高、维生素D水平升高为儿童反复呼吸道感染的保护因素，含维生素A、维生素D营养指标构建的Nomogram模型可提高预测儿童反复呼吸道感染的准确性。

Construction of predictive model for recurrent respiratory tract infections based on children’s nutrition-related indicators and its validation analysis

Objective: To construct and validate a predictive model for recurrent respiratory tract infection based on nutrition-related indicators in children. Method: Altogether 283 children with respiratory infections admitted to the First Affiliated Hospital of Wenzhou Medical University from January 2018 to March 2020 were selected as the research subjects, and all received standardized treatment. They were divided into the recurrent respiratory infection group and the non-recurrent respiratory infection group according to whether the respiratory tract infection recurred or not, with 198 cases (70%) and 85 cases (30%) randomly selected as training set and test set respectively. Their clinical data were compared, and the logistic regression model was used to analyze the related factors of recurrent respiratory infection. And a nomogram prediction model was built on the training set, using decision curve analysis to evaluate the clinical net benefits of the Nomogram model, and the effectiveness of the model was evaluated on the test set. Results: The ratio of family history of chronic respiratory diseases, birth weight, outdoor activity time, breastfeeding time, hemoglobin, vitamin A, vitamin D levels were statistically different (P<0.05). There was no significant difference in gender, age, body mass index, place of residence, birth method, asphyxia, sleep time, white blood cell count, serum zinc, iron, calcium, and magnesium (P>0.05). Multivariate logistic analysis showed that high birth weight (OR=0.995, P=0.004), long time outdoor activity (OR=0.630, P=0.003), increased hemoglobin level (OR=0.846, P<0.001), increased vitamin A level (OR=0.437, P<0.001), increased vitamin D level (OR=0.938, P=0.007) were protective factors for children with repeated respiratory infections. The C-index of the Nomogram model constructed to predict recurrent respiratory tract infection was 0.869 (95%CI=0.818-0.920). Decision curve analysis made to evaluate the clinical benefits of the two prediction models showed that additional clinical benefits could be provided when the value of nutrition-related indicators to predict children with recurrent respiratory tract infection fell within the range 0-0.59. External verification showed that the sensitivity was 92.00% and the specificity 96.67%, with the accuracy rate being 95.29%. Conclusion: Higher birth weight, prolonged outdoor activities, elevated levels of hemoglobin, vitamin A and vitamin D were protective factors for children with recurrent respiratory infections. The Nomogram model constructed with vitamin A and vitamin D nutritional indicators can improve the predicting accuracy for recurrent respiratory tract infections in children.