Liu Lin, Lo Kenneth, Huang Cheng, Feng Ying-Qing, Zhou Ying-Ling, Huang Yu-Qing
Department of Cardiology, Guangdong Provincial Key Laboratory of Coronary Heart Disease Prevention, Guangdong Cardiovascular Institute, Guangdong Provincial People's Hospital, Guangdong Academy of Medical Sciences, Guangzhou, China.
Ann Palliat Med. 2021 Feb;10(2):1167-1179. doi: 10.21037/apm-20-580. Epub 2020 Nov 2.
A simple clinical model that can predict all-cause mortality in the middle-aged and older adults in general population based on demographics and physical measurement indicators. The aim of this study was to develop a simple nomogram prediction model for all-cause mortality in middle-aged and elderly general population based on demographics and physical measurement indicators.
This was a prospective cohort study. We used data from the 1999-2006 National Health and Nutrition Examination Survey (NHANES), which included adults aged ≥40 years with mortality status updated through 31 December 2015. Cox proportional hazards regression, nomogram and least absolute shrinkage and selection operator (LASSO) binomial regression model were performed to evaluate the prediction model in the derivation and validation cohort.
A total of 13,026 participants (6,414 men, mean age was 61.59±13.80 years) were included, of which 6,671 (3,263 men) and 6,355 (3,151 men) were included in the derivation cohort and validation cohort, respectively. During an average follow-up period of 129.23±9.62 months, 4,321 died. We developed a 9-item nomogram mode included age, gender, smoking, alcohol intake, diabetes, hypertension, marriage status, education and poverty to income ratio (PIR). The area under the curve (AUC) was 0.842 and had good calibration. Internal validation showed good discrimination of the nomogram model with AUC of 0.849 and good calibration. Application of the LASSO regression model in the validation cohort also revealed good discrimination (AUC =0.854) and good calibration. A time-dependent and optimism-corrected AUC value for the model showed no significant relationship with the change of follow-up time.
A simple nomogram model, including age, gender, smoking, alcohol intake, diabetes, hypertension, marriage, education and PIR, could predict all-cause mortality well in middle-aged and elderly general population.
一种简单的临床模型，可基于人口统计学和身体测量指标预测普通人群中老年人的全因死亡率。本研究的目的是基于人口统计学和身体测量指标，开发一种用于中老年普通人群全因死亡率的简单列线图预测模型。
这是一项前瞻性队列研究。我们使用了1999 - 2006年国家健康和营养检查调查（NHANES）的数据，其中包括年龄≥40岁的成年人，其死亡状态更新至2015年12月31日。采用Cox比例风险回归、列线图和最小绝对收缩和选择算子（LASSO）二项回归模型在推导队列和验证队列中评估预测模型。
共纳入13,026名参与者（6,414名男性，平均年龄为61.59±13.80岁），其中推导队列和验证队列分别纳入6,671名（3,263名男性）和6,355名（3,151名男性）。在平均129.23±9.62个月的随访期内，4,321人死亡。我们开发了一个包含年龄、性别、吸烟、饮酒、糖尿病、高血压、婚姻状况、教育程度和贫困收入比（PIR）的9项列线图模型。曲线下面积（AUC）为0.842，校准良好。内部验证显示列线图模型具有良好的区分度，AUC为0.849，校准良好。LASSO回归模型在验证队列中的应用也显示出良好的区分度（AUC = 0.854）和校准良好。该模型的时间依赖性和乐观校正AUC值与随访时间的变化无显著关系。
一个简单的列线图模型，包括年龄、性别、吸烟、饮酒、糖尿病、高血压、婚姻、教育程度和PIR，能够很好地预测中老年普通人群的全因死亡率。
