Möhner M
Zentralinstitut für Krebsforschung der Akademie der Wissenschaften der DDR, Berlin-Buch.
Arch Geschwulstforsch. 1988;58(3):191-5.
Analysing the regional distribution of cancer incidence a graphic presentation of the data in form of cancer maps often give a clear idea of the underlying distribution. Nevertheless, if the cancer map is variegated, an answer to the question of regional dependence of the incidence is difficult to find. Statistical methods as the test of Kemp et al. (3), which is based on the mean absolute difference in rank between all possible pairs of adjacent regions with a common border, or outlier tests can be helpful in such cases. An easy calculable approximation of Kemp's method will be presented and demonstrated on lung cancer incidence in districts of the GDR over the five-year period 1978-1982. Moreover statistical methods of generating new hypotheses about possible risk factors will be discussed.
分析癌症发病率的区域分布时，以癌症地图形式呈现的数据图表通常能清晰地展现潜在分布情况。然而，如果癌症地图呈现出多样化，就很难找到发病率区域依赖性问题的答案。像肯普等人（3）所做的检验那样的统计方法，该方法基于所有有共同边界的相邻区域对之间排名的平均绝对差异，或者异常值检验在这种情况下可能会有所帮助。本文将给出肯普方法的一个易于计算的近似方法，并以1978 - 1982年五年间民主德国各地区肺癌发病率为例进行说明。此外，还将讨论关于可能的风险因素生成新假设的统计方法。
