Whitrow Melissa J, Smith Brian J, Pilotto Louis S, Pisaniello Dino, Nitschke Monika
Department of Medicine, University of Adelaide, South Australia, Australia.
Respirology. 2003 Dec;8(4):513-21. doi: 10.1046/j.1440-1843.2003.00497.x.
In 2000 there were 1.1 million lung or bronchial cancer deaths worldwide, with relatively limited evidence of causation other than for smoking. We aimed to search and appraise the literature regarding evidence for a causal relationship between air pollution and lung cancer according to the 10 Bradford Hill criteria for causality.
A MEDLINE search was performed using the following key words: 'lung neoplasm', 'epidemiology', 'human', 'air pollution'and 'not molec*'. The criteria for inclusion was: cited original research that described the study population, measured environmental factors, was of case control or cohort design, and was undertaken after 1982.
Fourteen papers (10 case control, four cohort studies) fulfilled the search criteria, with a sample size ranging from 101 cases and 89 controls, to a cohort of 552 cases and 138 controls. Of the 14 papers that fulfilled the search criteria the number of papers addressing each of the Bradford Hill criteria were as follows: Strength of association: eight studies demonstrated significant positive associations between environmental exposure and lung cancer with a relative risk range of 1.14-5.2. One study found a negative association with relative risk 0.28. Consistency: eight of 14 studies found significant positive associations and one of 14 a significant negative association. Specificity: tobacco smoking and occupational exposure were addressed in all studies (often crudely with misclassification). Temporality: exposure prior to diagnosis was demonstrated in nine studies. Dose-response relationship: evident in three studies. Coherence, analogy: not addressed in any study.
Evidence for causality is modest, with intermediate consistency of findings, limited dose-response evidence and crude adjustment for important potential confounders. Large studies with comprehensive risk factor quantification are required to clarify the potentially small effect of air pollution given the relatively large effects of tobacco smoking and occupational carcinogen exposure.
2000年全球有110万人死于肺癌或支气管癌，除吸烟外，因果关系的证据相对有限。我们旨在根据布拉德福德·希尔因果关系的10条标准，检索和评估有关空气污染与肺癌因果关系证据的文献。
使用以下关键词在MEDLINE数据库中进行检索：“肺肿瘤”“流行病学”“人类”“空气污染”和“非分子*”。纳入标准为：引用的原始研究描述了研究人群、测量了环境因素、为病例对照或队列设计，且研究在1982年之后进行。
14篇论文（10篇病例对照研究、4篇队列研究）符合检索标准，样本量从101例病例和89例对照到552例病例和138例对照不等。在符合检索标准的14篇论文中，涉及每条布拉德福德·希尔标准的论文数量如下：关联强度：8项研究表明环境暴露与肺癌之间存在显著正相关，相对风险范围为1.14 - 5.2。1项研究发现负相关，相对风险为0.28。一致性：14项研究中有8项发现显著正相关，14项中有1项发现显著负相关。特异性：所有研究都涉及吸烟和职业暴露（通常粗略且存在错误分类）。时间顺序：9项研究证明了诊断前的暴露情况。剂量反应关系：3项研究中有体现。连贯性、类推法：任何研究均未涉及。
因果关系的证据力度适中，研究结果一致性一般，剂量反应证据有限，对重要潜在混杂因素的调整粗略。鉴于吸烟和职业致癌物暴露的影响相对较大，需要开展综合危险因素量化的大型研究，以阐明空气污染可能存在的微小影响。
