基于RDW-CV与胆红素/白蛋白比值的列线图模型预测新生儿高胆红素血症颅脑超声异常的研究

Study on a nomogram model based on RDW-CV and bilirubin/albumin ratio for predicting cranial ultrasound abnormalities in neonatal hyperbilirubinemia

摘要:

目的 探讨红细胞分布宽度变异系数（RDW-CV）联合胆红素/白蛋白（B/A）比值对新生儿高胆红素血症颅脑超声异常的预测价值。方法 回顾性分析2022年3月—2025年3月在宣城市人民医院及南充市中心医院接受治疗的420例孕周≥35周的高胆红素血症新生儿的病历资料，根据颅脑超声诊断脑损伤结果分为颅脑超声异常组（115例）与颅脑超声正常组（305例），筛除非胆红素相关异常患儿64例，颅脑超声异常组最终纳入51例。比较两组RDW-CV、B/A比值。采用多因素逐步Logistic回归模型分析新生儿高胆红素血症颅脑超声异常的危险因素。基于RDW-CV、B/A比值建立列线图预测模型，评估两者对新生儿高胆红素血症颅脑超声异常的预测价值。结果 两组性别、出生日龄、出生体重、平均红细胞血红蛋白含量、白细胞计数、中性粒细胞数、淋巴细胞数、单核细胞数、血小板数、血小板压积、分娩方式、血型、产妇产次及病因比较，差异均无统计学意义（P >0.05）；颅脑超声异常组血小板分布宽度、总胆红素水平、B/A比值及RDW-CV均高于颅脑超声正常组（P <0.05）。总胆红素水平高［O^R=4.433（95% CI：1.778，11.051）］、B/A比值大［O^R=4.039（95% CI：1.621，10.069）］、RDW-CV高［O^R=4.267（95% CI：1.712，10.638）］、血小板分布宽［O^R=3.380（95% CI：1.349，8.467）］均为新生儿高胆红素血症颅脑超声异常的危险因素（P <0.05）。列线图预测模型的风险率为0.05～0.90，校正曲线趋近于理想曲线（P >0.05）。RDW-CV、B/A比值及二者联合预测新生儿高胆红素血症颅脑超声异常的曲线下面积（AUC）分别为0.721、0.704、0.826，二者联合时AUC更高。RDW-CV的敏感性为82.45%（95% CI：0.691，0.916），特异性为71.20%（95% CI：0.657，0.762）；B/A比值的敏感性为78.41%（95% CI：0.647，0.887），特异性为73.80%（95% CI：0.685，0.786）；二者联合预测的敏感性为88.21%（95% CI：0.761，0.956），特异性为79.66%（95% CI：0.747，0.840）。结论 RDW-CV联合B/A比值对新生儿高胆红素血症颅脑超声异常的预测效能显著优于任一单一指标，可作为早期识别高危患儿的更佳生物标志物组合。

Abstract:

Objective To investigate the predictive value of red blood cell distribution width coefficient of variation (RDW-CV) combined with the bilirubin/albumin (B/A) ratio for cranial ultrasound abnormalities in neonatal hyperbilirubinemia.Methods A retrospective analysis was conducted on the medical records of 420 neonates with hyperbilirubinemia (gestational age ≥ 35 weeks) treated at Xuancheng People's Hospital and Nanchong Central Hospital from March 2022 to March 2025. Based on cranial ultrasound findings for brain injury, they were divided into an abnormal cranial ultrasound group (n = 115) and a normal cranial ultrasound group (n = 305). After excluding 64 children with non-bilirubin-related abnormalities, 51 cases were finally included in the abnormal group. RDW-CV and the B/A ratio were compared between the two groups. Multivariate stepwise logistic regression was used to analyze risk factors for cranial ultrasound abnormalities in neonatal hyperbilirubinemia. A nomogram prediction model was established based on RDW-CV and the B/A ratio, and its predictive value was evaluated.Results There were no statistically significant differences between the two groups in terms of sex, postnatal age, birth weight, mean corpuscular hemoglobin (MCH), white blood cell count, neutrophil count, lymphocyte count, monocyte count, platelet count, plateletcrit, mode of delivery, blood type, maternal parity, or etiology (all P > 0.05). The abnormal cranial ultrasound group had significantly higher platelet distribution width (PDW), total bilirubin level, B/A ratio, and RDW-CV compared to the normal group (all P < 0.05). Multivariate analysis identified high total bilirubin level [O^R = 4.433 (95% CI: 1.778, 11.051) ], high B/A ratio [O^R = 4.039 (95% CI: 1.621, 10.069) ], high RDW-CV [O^R = 4.267 (95% CI: 1.712, 10.638) ], and wide platelet distribution width [O^R = 3.380 (95% CI: 1.349, 8.467) ] as risk factors for cranial ultrasound abnormalities in neonatal hyperbilirubinemia (all P < 0.05). The constructed nomogram prediction model had a risk rate range of 0.05 to 0.90, and its calibration curve was close to the ideal curve (P > 0.05). The area under the curve (AUC) for RDW-CV, B/A ratio, and their combination in predicting cranial ultrasound abnormalities was 0.721, 0.704, and 0.826, respectively, with the combination showing a higher AUC. The sensitivity and specificity of RDW-CV were 82.45% (95% CI: 0.691, 0.916) and 71.20% (95% CI: 0.657, 0.762), respectively. For the B/A ratio, sensitivity was 78.41% (95% CI: 0.647, 0.887) and specificity was 73.80% (95% CI: 0.685, 0.786). The combined prediction model showed a sensitivity of 88.21% (95% CI: 0.761, 0.956) and a specificity of 79.66% (95% CI: 0.747, 0.840).Conclusion The combination of RDW-CV and B/A ratio demonstrates significantly superior predictive performance for cranial ultrasound abnormalities in neonatal hyperbilirubinemia compared to either indicator alone, serving as a better biomarker combination for early identification of high-risk infants.