Dose-response relationship between active smoking and lung cancer mortality/prevalence in the Chinese population: a meta-analysis

BackgroundThe dose-response relationship-based relative risk (RR) of smoking exposure could better predict the risk of lung cancer than the dichotomous RR. To date, there is a lack of large-scale representative studies illustrating the dose-response relationship between smoking exposure and lung cancer deaths, and no study has systematically pooled the current evidence in the Chinese population.ObjectivesTo elucidate the dose-response relationship of smoking and the risk of lung cancer mortality in the Chinese population.MethodsData were derived from studies on dose-response relationships of smoking exposure and the risk of lung cancer among Chinese adults published before June 30(th), 2021. Based on smoking exposure indicators and RR of lung cancer mortality, a series of dose-response relationship models were developed. For smokers, 10 models were built to fit the dose-response relationships between pack-years and RR of lung cancer deaths. For quitters, quit-years and corresponding RRs were used, and the pooled dichotomous RR value was used as the starting point to avoid overestimation. Finally, the results were compared with the estimates from 2019 Global Burden of Disease (GBD) study.ResultsA total of 12 studies were included. Among 10 dose-response relationship models of pack-years with the RR of lung cancer mortality, the integrated-exposure-response (IER) model achieved the best fit. In all models, less than 60 pack-years presented RRs below 10. For former smokers, the RR decreased to 1 when quit-years reached up to 7 years. Both smokers and quitters had much lower RRs than that of the global level estimated by GBD.ConclusionThe risk of lung cancer mortality rose with pack-years and decreased with quit-years among Chinese adults, and both values were far below global level. The results suggested that the dose-response RR of lung cancer deaths associated with smoking in China should be estimated separately.