- 學位論文
以經皮式黃疸測定預測新生兒黃疸
Prediction of Neonatal Hyperbilirubinemia with Transcutaneous Bilirubin Measurement
摘要
研究背景： 新生兒黃疸近年來重新被重視。一部份是因為核黃疸及其他較輕微的神經損傷的病例報告出現，另一部份是因為保險早期出院的新生兒照顧政策。 研究目的： 本研究之目的：1)建立台灣嬰兒之nomogram; 2)以conventional models及five-state transition modl預測新生兒黃疸; 3)分析相關之危險因子。 研究對象與方法： 一個包括531個新生嬰兒的醫院基礎的前瞻性世代研究，嬰兒的經皮式黃疸測定值(TcB)至少每八小時被紀錄一次。出生體重，性別，ABO血型不合，餵食方式，懷孕週數，體重變化都同時被紀錄。所有嬰兒均受邀回來門診追蹤。Nomogram根據百分位數的數值建立，以四小時為一單位。分析出生後24及48小時測量的敏感性，特異性，以及ROC(Receiver Operator Characterisitc)曲線。建立五階段預測模式，來分析黃疸值的升高或降低。危險因子以linear mixed model及generalized linear mixed model分析。晚期發生(late-onset)的新生兒黃疸的危險因子以邏輯回歸方式分析。 結果： 根據nomogram，嬰兒的黃疸的達到高峰的時間，平均及中位數分別是出生後79.5及76小時，最高值的平均及中位數均是13.7 mg/dL，嬰兒接受照光的平均及中位數時間是出生後80及76小時。於出生後48小時，預測照光治療，ROC曲線下面積是0.873，而預測臨床顯著的黃疸是0.844。ROC曲線下面積於出生後24小時，預測照光治療及臨床顯著黃疸都是是0.773。transition model以48小時切割呈現不均勻的分佈顯示48小時可能是一個重要的時間點。初產婦，ABO血型不合是可歸因於生物因素的危險因子。 結論： 本研究採用一系列經皮式黃疸測定值，以conventional models或five-state transition model來預測需要照光治療或黃疸。結果顯示產後48小時的經皮式黃疸測定值是一個好的預測者。初產婦，ABO血型不合，蠶豆症是顯著的危險因子。其他的顯著因子可能都受到健保新生兒照顧政策的影響。所有新生兒於出院前均應該接受非侵入性黃疸測定，高危險群應該於出院數天內追蹤。
並列摘要
Background: Early prediction of hyperbilirubinemia has regained attention in recent years partly because kernicterus and minor neurological abnormalities has occasionally been reported, and partly because of early discharge due to the change of managed care policy. Objectives: The aims of this thesis were (1) to build up the nomogram of Taiwanese infants, (2) to predict neonatal hyperbilirubinemia by conventional models and a five-state transition model, and (3) to analyze their related risk factors. Materials and Methods: A hospital-based prospective cohort study consisted of 531 infants were term or near-term infants eligible for caring at the well-baby nursery. All infants’ transcutaneous bilirubin (TcB) were recorded regularly at least every 8 hours. Birth weights, maternal age, sex, presence of ABO incompatibility, feeding method, type of delivery, gestational history, and body weight change were recorded. All babies were invited to come back to follow-up clinic after discharge. Nomogram was plotted according to percentile data at a four-hour interval. Sensitivity, specificity, and receiver operator characteristic (ROC) curve were analyzed at age of 24 and 48 hours old. A five-state transition model was built to predict the progression or regression of bilirubin level. Risk factors were analyzed by both linear mixed model , generalized linear mixed model, proportional hazards transitional model. Later-onset neonatal hyperbilirubinemia was analyzed by logistic regression model. Results: According to hour-specific nomogram, infants reached their peak bilirubin level at mean and median age of 79.5 and 76 hours old. Both mean and median peak bilirubin level were about 13.7 mg/dL. Infants received phototherapy at a mean and median age of 80 and 76-hour-old. The areas under ROC cure were 0.873 for predicting need for phototherapy and 0.844 for clinically significant jaundice at age of 48-hour-old. The areas under ROC cure were both 0.773 at 24-hour-old. Non-homogenous process by the division of 48 hours may be a critical point for differentiating transition rate in the transition model. First pregnancy and ABO incompatibility were two risk factors attributed to biological reasons. Conclusion: The present study used a series of TcB measurement to predict hyperbilirubinemia or the need for phototherapy using conventional method or a novel five-state transition model. The results found 48-hour-old TcB level may be a good predictor. Primiparity, incompatibility of ABO blood type, and G6PD deficiency were significant risk factors. Other established significant factors were largely influenced by the neonatal care policy of national health insurance. All infants should receive non-invasive bilirubin measurement before discharge. High risk infants should be followed up within several days after discharge.