Exploratory research on developing lung cancer risk prediction model in female non-smokers
10.3760/cma.j.cn112150-20200805-01093
- VernacularTitle:非吸烟女性肺癌风险预测模型的构建研究
- Author:
Zhangyan LYU
1
;
Ni LI
;
Shuohua CHEN
;
Gang WANG
;
Fengwei TAN
;
Xiaoshuang FENG
;
Xin LI
;
Yan WEN
;
Zhuoyu YANG
;
Yalong WANG
;
Jiang LI
;
Hongda CHEN
;
Chunqing LIN
;
Jiansong REN
;
Jufang SHI
;
Shouling WU
;
Min DAI
;
Jie HE
Author Information
1. 国家癌症中心 国家肿瘤临床医学研究中心 中国医学科学院 北京协和医学院肿瘤医院癌症早诊早治办公室，北京 100021
- Keywords:
Neoplasms;
Lung;
Forecasting;
Female;
Non-smokers
- From:
Chinese Journal of Preventive Medicine
2020;54(11):1261-1267
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To develop a lung cancer risk prediction model for female non-smokers.Methods:Based on the Kailuan prospective dynamic cohort (2006.05-2015.12), a nested case-control study was conducted. Participants diagnosed with primary pathologically confirmed lung cancer during follow-up were identified as the case group, and others were identified as the control group. A total of 24 701 subjects were included in the study, including 86 lung cancer cases and 24 615 control population, respectively. Questionnaires, physical examinations, and laboratory tests were conducted to collect relevant information. Multivariable-adjusted logistic regressions were conducted to develop a lung cancer risk prediction model. Area Under the Curve (AUC) and Hosmer-Lemeshow tests were used to evaluate discrimination and calibration, respectively. Ten-fold cross-validation was used for internal validation.Results:Two sets of models were developed: the simple model (including age and monthly income) and the metabolic index model [including age, monthly income, fasting blood glucose (FBG), total cholesterol (TC) and high-density lipoprotein cholesterol (HDL-C)].The AUC (95%CI) [0.745 (0.719-0.771)] of the metabolic index model was higher than that of the simple prediction model [0.688 (0.660-0.716)] ( P=0.004). Both the simple model ( PHL=0.287) and the metabolic index model ( PHL=0.134) were well-calibrated. The results of ten-fold cross-validation indicated sufficient stability, with an average AUC of 0.699 and a standard error (SD) of 0.010. Conclusion:By incorporating metabolic markers, accurate and reliable lung cancer risk prediction model for female non smokers could be developed.
