臺灣博碩士論文加值系統

論文摘要： 目的
： 建立台灣地區新生兒（單胞胎，雙胞胎，三胞胎）之出生體重－懷孕週 數常模
方法：對內政
部出生登記民國82-85年間出生約120萬位新生兒的資料進行 統計分析，對不合理
的出生體重先以Tukey's boxplot approach 排除，再依性別 胎次別分層並以polynomia
l smoothing method, 與LMS method建立常模並計算 loss function, 以評估、比較
模式的goodness-of-fit. 結果：本研究較之前台灣地區同類型
的研究較具有代表性，並且因有充足的樣 本數而能得到較穩定的估計值。與1950年
代相比，台灣地區新生兒體重有明顯 的上升，但與美國白人相比，台灣地區新生兒
體重則顯著較低。兩種模式在常 模的建立上各有其優缺點。我們建議往後每10年應
更新常模。

Abstract


Objective:
To develop current birth wei
ght norms by gestational age for singleton , twin and triplet births in Taiwa
n.


Methods:
Birt
h weight data were obtained from ministry of the interior birth registrationsf
or over one million live births in Tawan from 1993-1996. Unlikely combinations
of birth weight and gestational age were defined by Tukey's boxplot approach
within each stratum of multiplicity, gender and gestational age. Birth weight
percentiles by gestational age, sex and parity were modeling by polynomial sm
oothing technique and LMS method. A loss function then was calculated for comp
arison of the goodness-of-fit between 2 models.


Results:

The large database assembled for this analysis provides cu
rrent, stable birthweight-gestational age percentiles. Compared with birth wei
ght distributions from the 1950s, these current norms are heavier. Compared wi
th the birth weight norms for white singleton births in the American populat
ion, our norms are lighter. The 2 modeling approaches show their own strength
in different ways. We recommend that birth weight norms be updated every 10 ye
ars.