定量流行病学和医学科学研究中的因果联系

Causality in quantitative epidemiology and medical science research

在医学研究中有一个永恒的挑战,就是确定影响因素(X)与疾病(Y)之间的因果联系。即便是目前被普遍接受的医学结论也常常遭到怀疑,比如吸烟与肺癌的关系,肥胖与高血压的关系、糖尿病和心脏病的关系等等。本文尝试提出一个新的理论模式来解释因果联系,并介绍一些推断因果联系的研究设计和分析方法。从哲学的角度,宇宙间万事万物相互关联,形成一个相互依存的网络系统。在研究X-Y关系时,如果X(新冠病毒)与Y(新冠肺炎)之间的关系距离很近,得到的结果大都很明确。若X(吸烟)与Y(肺癌)之间的关系存在较大距离,可能存在大量未知因素与X-Y相连接而形成关系网络系统,在这种情况下X-Y的关系变得复杂且难于确定。而在不同的研究中,相同含义的X或Y变量,可以用不同的方法来测量。比如,吸烟这个变量可以用体内尼古丁的代谢产物来测量,也可以使用调查数据,例如是否吸烟、吸烟的年数和每天吸烟支数来测量。用不同的方法测量相同的变量进一步增加了研究结果的不确定性。因此,对因果推断的研究结果产生质疑很正常,而且是必须的,这样才能促进更深入的研究。

An eternal challenge in medical research is to determine the causal link between influencing factors(X)and diseases(Y).Even those generally accepted medical conclusions are often questioned,such as the relationship between smoking and lung cancer,obesity and high blood pressure,diabetes and heart disease,and so on.This paper attempts to propose a new theoretical model to interpret causality and introduces some research designs and analytical methods for inferring causality.From a philosophical perspective,all things in the universe are interconnected,forming an interdependent network system.When studying an X-Y relationship,if the relationship between X(novel coronavirus)and Y(novel coronavirus pneumonia)is very close,the results are mostly clear.If there is a large distance between X(smoking)and Y(lung cancer),there may be many unknown factors connecting X with Y to form a relationship network.In this situation,the relationship between X and Y becomes complicated and difficult to determine.Moreover,in different studies,X or Y variables with the same meaning can be measured using different methods.For example,the variable smoking can be measured by the metabolites of nicotine in the body,or by survey data such as whether or not smoking,years of smoking and the number of cigarettes smoked.Measuring the same variable using different methods will add more uncertainty to study findings.Therefore,it is normal and necessary to question conclusions from causal inference studies in order to promote more in-depth research.