Bross I D, Natarajan N
Invest Radiol. 1980 Jan-Feb;15(1):52-64. doi: 10.1097/00004424-198001000-00009.
Using a mathematical model and newly developed computer software, the data from the Tri-State Leukemia Survey involving different combinations of radiation exposures to the father and mother prior to conception and to the mother during pregnancy were analyzed. The hypothesis that radiation exposure produces genetic damage which may be expressed in the child both as indicator disease and as leukemia was tested. The genetic damage was estimated in terms of the proportion affected by a given exposure. The relative risk of leukemia and certain other indicator diseases among those "affected" could then be estimated. The results were obtained by direct minimization of the total chi-square. The results show that there are at least two distinguishable risk groups, one group with lower (one or two exposures), and the other group with higher (two or three) radiation exposures. The estimates for the proportion affected are 0.13 and .04 for the lower and higher exposure groups, respectively. The relative risks of leukemia and indicator diseases are 25.0 and 5.0, respectively. The 95% confidence intervals for the proportion affected in the lower and higher exposure categories are (.0075, .02) and (.028, .055), respectively, and for the relative risk of leukemia are (16.0, 35.0). These results have confirmed and extended the findings of our preliminary report published in the Journal of the American Medical Association in May 1977.
利用一个数学模型和新开发的计算机软件，对三州白血病调查的数据进行了分析，该调查涉及受孕前父亲和母亲接受不同辐射组合以及母亲孕期接受辐射的情况。对辐射暴露会产生遗传损伤这一假设进行了检验，这种遗传损伤可能在孩子身上表现为指示性疾病和白血病。遗传损伤是根据给定暴露所影响的比例来估计的。然后可以估计那些“受影响”人群中患白血病和某些其他指示性疾病的相对风险。结果是通过直接最小化总卡方得到的。结果表明，至少有两个可区分的风险组，一组辐射暴露较低（一次或两次暴露），另一组辐射暴露较高（两次或三次）。较低和较高暴露组受影响比例的估计值分别为0.13和0.04。白血病和指示性疾病的相对风险分别为25.0和5.0。较低和较高暴露类别中受影响比例的95%置信区间分别为(0.0075, 0.02)和(0.028, 0.055)，白血病相对风险的95%置信区间为(16.0, 35.0)。这些结果证实并扩展了我们于1977年5月发表在美国医学协会杂志上的初步报告的研究结果。
