Woolhandler S, Himmelstein D U
Lancet. 1985 Jun 15;1(8442):1375-8. doi: 10.1016/s0140-6736(85)91795-7.
Examination of data from 141 countries showed that infant mortality rates for 1979 were positively correlated with the proportion of gross national product devoted to military spending (r = 0.23, p less than 0.01) and negatively correlated with indicators of economic development, health resources, and social spending. In a multivariate analysis controlling for per caput gross national product, arms spending remained a significant positive predictor of infant mortality rate (p less than 0.0001), while the proportion of the population with access to clean water, the number of teachers per head, and caloric consumption per head were negative predictors. The multivariate model accounted for much of the observed variance in infant mortality rate (R2 = 0.78, p less than 0.0001), and showed good fit to similar data for the year 1972 (R2 = 0.80, p less than 0.0001). The model was also predictive of infant mortality rates in subgroup analysis of underdeveloped, middle developed, and developed nations. Analysis of time trends confirmed that an increase in military spending presages a poor record of improvement in infant mortality rate. These findings support the hypothesis that arms spending is causally related to infant mortality.
对141个国家的数据调查显示，1979年的婴儿死亡率与国民生产总值中用于军事开支的比例呈正相关（r = 0.23，p小于0.01），与经济发展、卫生资源及社会开支指标呈负相关。在一项控制人均国民生产总值的多变量分析中，军备开支仍然是婴儿死亡率的一个显著正向预测指标（p小于0.0001），而能获得清洁水的人口比例、人均教师数量及人均热量消耗则为负向预测指标。该多变量模型解释了婴儿死亡率中很大一部分观测到的方差（R2 = 0.78，p小于0.0001），并且对1972年的类似数据显示出良好的拟合度（R2 = 0.80，p小于0.0001）。该模型在对欠发达国家、中等发达国家及发达国家的亚组分析中也能预测婴儿死亡率。对时间趋势的分析证实，军事开支的增加预示着婴儿死亡率改善记录不佳。这些发现支持了军备开支与婴儿死亡率存在因果关系这一假设。
