Modeling Race/Ethnic Differences in Infant Mortality

随着我们进入21世纪，婴儿死亡率方面的种族/民族差异仍然是一个鲜为人知的现象。结果不确定的部分原因是使用了一个世纪以来以怀特标准为基础的出生结果指标，这“可能导致对其他族裔群体的风险评估无效和对需要干预服务的婴儿的错误识别”。一项中心任务是更好地记录出生体重和胎龄如何独立和联合地影响不同种族/民族人口的婴儿死亡风险。另一个一般性目标是确定其他危险因素的影响在多大程度上存在种族/族裔和性别差异，并通过或超出出生结果的中介影响来影响婴儿死亡率。具体地说，我们建议：(1)使用出生体重和胎龄的主效应和联合效应，为24个种族/民族和性别特定人口建立婴儿死亡率模型，以明确婴儿存活的最优出生结果，并评估“一刀切”假设；(2)基于包含种族/民族、性别、连续出生结果衡量标准和广泛的人口、社会经济、行为和母亲健康协变量的模型的逐步调整，确定解释婴儿死亡率种族/民族差异的可能性程度；(3)估计种族/民族和性别模型，以确定出生结果在多大程度上调节人口、社会经济状况、行为、产妇健康和其他协变量对婴儿死亡风险的影响；(4)利用较早的NCHS链接文件(1989-91)和最新的文件(1995-98)来检查在美国婴儿死亡率发生许多深刻变化的一段时间内，利益关系发生的变化；(5)由于婴儿死亡风险随婴儿年龄的不同而有很大差异，因此将婴儿年龄的非比例影响纳入分析。数据将来自-91年和1995年-98年NCHS与出生/婴儿死亡相关的队列档案，其中包含足够数量的病例来支持这一分析。

As we enter the 21st century, race/ethnic disparities in infant mortality remain a poorly understood phenomenon. Inconclusiveness of results is partially attributable to use of century-old indicators of birth outcomes based on White standards which "may result, for other ethnic groups, in invalid risk assessments and the misidentification of infants in need of intervention services". A central task is to better document how birth weight and gestational age, independently and conjointly, differentially affect infant mortality risk across race/ethnic populations. Another general objective is to determine the extent to which there are race/ethnic and sex differences in the way the effects of other risk factors for on infant mortality work through, or outside of, the mediating effects of birth outcomes. Specifically, we propose to: (1) formulate models of infant mortality for 24 race/ethnic- and sex-specific populations using both main and conjoint effects of birth weight and gestational age in order to specify birth outcome optimums for infant survival and evaluate the "one-size-fits-all" assumption; (2) determine the extent to which it is possible to explain race/ethnic differentials in infant mortality based on progressive adjustment of models that contain race/ethnicity, sex, the continuous birth outcome measures, and a wide range of demographic, SES, behavioral and maternal health covariates; (3) estimate race/ethnic- and sex-specific models to determine the extent to which birth outcomes mediate the effects of demographic, SES, behavioral, maternal health, and other covariates on infant mortality risk; (4) utilize earlier NCHS linked files (1989-91) in conjunction with the most recent files (1995-98) to examine changes that have occurred in the relationships of interest over a time period in which a number of profound alterations in infant mortality occurred in the US; (5) inasmuch as infant mortality risk varies substantially by age of infant, incorporate the nonproportional effect of infant age into the analyses. Data will be drawn from the 1989-91 and 1995-98 NCHS Linked Birth/Infant Death cohort files which contain a sufficient number of cases to support the analysis.