Pearce N E, Cryer P C
Am J Epidemiol. 1986 Jul;124(1):127-33. doi: 10.1093/oxfordjournals.aje.a114357.
The standard epidemiologic methods for evaluating trends in the prevalence or incidence of disease are reviewed, and a method is presented for assessing the contribution of a specific disease subgroup to the overall trend, based on the slope obtained by a grouped weighted linear regression of the proportions of persons experiencing the outcome of interest in each exposure level. The slopes for specific disease subgroups contribute to the overall slope in an additive manner, and measures based on the individual slopes can be used to assess the relative strengths of the trends for specific disease subgroups as well as their relative contributions to the overall trend. This approach is illustrated with data on social class patterns of mortality in New Zealand males aged 15-64 years during 1974-1978; it is shown that the strongest mortality gradients were for mental disorders, infectious diseases, respiratory diseases, and accidents. The latter two categories together accounted for approximately two thirds of the overall social class mortality gradient.
回顾了评估疾病患病率或发病率趋势的标准流行病学方法，并提出了一种方法，用于评估特定疾病亚组对总体趋势的贡献，该方法基于对每个暴露水平中经历感兴趣结局的人群比例进行分组加权线性回归所获得的斜率。特定疾病亚组的斜率以累加方式对总体斜率产生影响，基于个体斜率的测量方法可用于评估特定疾病亚组趋势的相对强度及其对总体趋势的相对贡献。以1974 - 1978年期间新西兰15 - 64岁男性社会阶层死亡率模式的数据为例进行说明；结果表明，死亡率梯度最强的是精神障碍、传染病、呼吸系统疾病和事故。后两类合起来约占总体社会阶层死亡率梯度的三分之二。
