建立影响新生儿肺出血病死率危险因素量化评分体系的研究

Risk factors for mortality in neonatal pulmonary hemorrhage: a prediction score model

目的探讨影响肺出血新生儿预后的危险因素并建立相应量化评分体系。方法回顾性分析244例肺出血新生儿临床资料,用Logistic回归分析等方法筛选影响病死率的危险因素,进行量化评分并绘制受试者工作曲线(receiver operating characteristic curve,ROC),确定诊断界点。结果在244例患儿中,病死率为74.59%(182/244),治愈组平均胎龄、日龄中位数与病死组比较,差异无统计学意义(P>0.05),治愈组双胎或三胎发生率、出生体质量与病死组比较,差异有统计学意义(P<0.05)。治愈组呼吸窘迫综合征、呼吸衰竭、寒冷损伤综合征患病率与病死组比较,差异无统计学意义(P>0.05)。治愈组颅内出血、心力衰竭、败血症的患病率显著低于病死组(P<0.01)。对上述差异有统计学意义的变量进行Logistic回归分析,出生体质量、颅内出血、心力衰竭、败血症4个变量入选回归方程,方程预测准确率为80.7%,建立的ROC曲线下面积为0.786,界点为9分时该评分体系预测对肺出血预后灵敏度为84.6%,特异度为66.1%,阳性预测值为88.0%,阴性预测值为59.4%。9分以下组(低度危险组)病死率显著小于10分以上组(高度危险组)(P<0.01)。结论Ⅲ度或Ⅳ度颅内出血、出生体质量低、败血症、心力衰竭是影响肺出血新生儿病死率的危险因素,量化评分9分以上则高度危险。

Objective To explore the risk factors that affect the mortality haemorrhage （PH） and establish a mortality prediction score model. Methods of neonates with pulmonary Totally 244 PH infants were analyzed retrospectively. The risk factors were screened out by logistic regression analysis and the scores were developed according to regression coefficients. Then the receiver operating characteristic curve was constructed and the cutoff was determined. Results The mortality was 74.59% （182/244） , the mean gestational age and median age were （36. 513±3. 805） weeks and 0.24 days in the survived infants, （35. 746±3. 929） weeks and 0.33 days in the died infants respectively, without significance between two groups （P 〉0.05）. The rate of twins or triplets and birth weight of the survived infants were 3.2% （2/62）, （2 786. 866± 606. 991） g, while 20.1% （25/182）, （2 338. 812±756. 790） g in the died infants （P 〈 0.05）. The survived infants had no difference from those died in the rate of neonatal respiratory distress syndrome （43.5% vs 34.6% , X^2=1. 585, P = 0. 208）, respiratory failure （77.4% vs 78.6%, X^2 = 0. 036, P = 0.849）, or cold injury syndrome （35.5% vs 44.5% , X^2 =0. 396, P=0. 529）. The survived infants had significant difference from those died in the rate of IVH （25.8% vs 53.8%, X^2 = 14. 607, P =0. 000）, heart failure （22.6% vs 48.9%, X^2 = 13.134, P = 0. 000） , sepsis （3.2% vs 16.5 %, X^2 = 7.134, P = 0. 008 ）. We further used logistic regression analysis to test above significantly different variables, then the four variables （birth weight, IVH, heart failure and sepsis） were in the logistic regression equation. The predicted percentage of correct was 80.7%. According to regression coefficients, a score ranging from 4 to 40 was developed, the area under the receiver operating characteristic curve was 0. 786. At a cutoff point of 9, the pulmonary hemorrhage was predicted with the sensitivity of 0, 846, the specificity of 0,661, the positive predictive value of 0.88 and the negative predictive value of 0.594. The mortality was higher in the ＂low risk group＂ （ the score ≤9） than that in the ＂high risk group＂ （the score 〉9） （40.6% vs 88%, X^2 = 58. 714, P = 0. 000）. Conclusion Low birth weight, IVH, heart failure and sepsis are the risk factors in the prognosis of neonates with pulmonary haemorrhage. The infants with the score higher than 9 are assessed to be at ＂high risk＂ for death.