Abstract Infant mortality rate is considered as one of the most important indicators of development of the socio-economic and health status of any country. For Bangladesh, the estimates of infant mortality on 64 districts of the country is available, but these regionally disaggregated estimates, based on sample surveys, may give misleading pictures because in reality they do not capture the neighborhood and contextual influences. If any spatial pattern exists in the characteristics of interest, estimates obtained for a particular district must be adjusted for any spatial effect of the neighboring districts, otherwise the unadjusted estimates alone may give a confounded meaning. Policy implication at regional level on the basis of such estimates may lead to both under coverage and redundancy. In this paper, we utilize a small area estimation approach by intrinsic conditional autoregressive model isolating the effect of the key covariates. Our analysis reveals that there exist spatial autocorrelation in infant mortality over the districts. We obtain revised district level estimates of the infant mortality from available data on infant mortality and other covariates from different sources. We believe our estimates would be more accurate than the nationally available infant mortality dataset. This in the long run, can be supportive in policy measures related to reducing infant mortality.

中文翻译：

---

孟加拉国婴儿死亡率的空间修订估计

摘要婴儿死亡率被认为是任何国家社会经济和健康状况发展的最重要指标之一。对于孟加拉国，可以获得该国64个地区的婴儿死亡率的估算值，但是根据抽样调查得出的这些按地区分类的估算值可能会产生误导性的图片，因为实际上它们并不能反映邻里和环境的影响。如果感兴趣的特征中存在任何空间模式，则必须针对特定区域获得的估计值针对邻近区域的任何空间效应进行调整，否则仅未调整的估计值可能会产生混淆的含义。根据这些估计数，区域一级的政策影响可能导致覆盖不足和裁员。在本文中，我们通过内在条件自回归模型利用小区域估计方法来隔离关键协变量的影响。我们的分析表明，各地区婴儿死亡率存在空间自相关。我们从婴儿死亡率和不同来源的其他协变量的可用数据中获得了婴儿死亡率的经修订的地区级估计值。我们相信我们的估算将比全国可用的婴儿死亡率数据集更为准确。从长远来看，这可以为降低婴儿死亡率的政策措施提供支持。我们从婴儿死亡率和不同来源的其他协变量的可用数据中获得了婴儿死亡率的经修订的地区级估计值。我们相信我们的估算将比全国可用的婴儿死亡率数据集更为准确。从长远来看，这可以为降低婴儿死亡率的政策措施提供支持。我们从婴儿死亡率和不同来源的其他协变量的可用数据中获得了婴儿死亡率的经修订的地区级估计值。我们相信我们的估算将比全国可用的婴儿死亡率数据集更为准确。从长远来看，这可以为降低婴儿死亡率的政策措施提供支持。