不同肥胖症评价指标联合应用对不同性别人群新发胆石症预测价值的多中心回顾性研究（附88947例报告）

The predictive value of combined application of the different obesity measures on incident gallstone diseases： a multicenter retrospective study （A report of 88 947 cases ）

目的：探讨不同肥胖症评价指标联合应用对不同性别人群新发胆石症的预测价值，寻找最优组合。方法：采用回顾性队列研究方法。收集2006年7月至2015年12月华北理工大学附属开滦总医院、开滦林西医院、开滦赵各庄医院、开滦唐家庄医院、开滦范各庄医院、开滦荆各庄医院、开滦吕家坨医院、开滦林南仓医院、开滦钱家营医院、开滦马家沟医院、开滦医院分院行健康体检的88 947例受试者的体检资料。由固定医师团队于2006、2008、2010、2012、2014年在相同地点按相同健康体检顺序对受试者进行5次健康体检。收集流行病学调查内容、人体测量学指标、生化指标。观察指标：（1）不同性别受试者的一般资料比较。（2）受试者胆石症的发病情况。（3）不同肥胖症评价指标影响受试者新发胆石症的危险因素分析。（4）不同肥胖症评价指标联合应用对新发胆石症模型的拟合情况和预测价值的比较。正态分布的计量资料以±s表示，组间比较采用独立样本t检验。偏态分布的计量资料采用M（P25，P75）表示，组间比较采用秩和检验。计数资料比较采用x2检验。采用人年发病率计算不同性别受试者胆石症的发病情况。采用COX比例风险模型分析各肥胖症评价指标对新发胆石症的风险比（HR）和95%可信区间（CI）。采用似然比检验和赤池信息量准则（AIC）计算不同组合对新发胆石症模型的拟合情况。结果：（1）不同性别受试者的一般资料比较：88 947例受试者中，男性受试者的年龄、体质量指数（BMI）、腰围、收缩压、舒张压、总胆固醇（TC）、三酰甘油（TG）、空腹血糖、糖尿病、高血压病、吸烟、饮酒、体育锻炼人数分别为（51±12）岁、（25±3）kg/m2、（88±10）cm、（132±20）mmHg（1 mmHg=0.133 kPa）、（84±12）mmHg、（4.95±1.16）mmol/L、1.18 mmol/L（0.81 mmol/L，1.74 mmol/L）、（5.5±1.6）mmol/L、6 223例、31 816例、26 993例、15 779例、11 063例，女性受试者分别为（49±11）岁、（25±4）kg/m2、（83±11）cm、（124±21）mmHg、（79±11）mmHg、（4.98±1.08）mmol/L、1.30 mmol/L（0.92 mmol/L，2.00 mmol/L）、（5.3±1.6）mmol/L、1 409例、5 866例、248例、87例、2 450例，两组上述指标比较，差异均有统计学意义（t=587.20，894.27，1 064.97，813.49，986.22，630.97，H（x2）=642.39，t=452.87， x2=35.10，1 205.40，9 619.42，4 901.75，84.82，P〈0.05）。（2）受试者胆石症的发病情况：88 947例受试者总随访时间为713 345人年，共有新发胆石症4 291例，总人年发病率为6.02千人/年。其中男性受试者总随访时间为562 821人年，共有新发胆石症3 268例，人年发病率为5.81千人/年；女性受试者总随访时间为150 524人年，新发胆石症1 023例，人年发病率为6.80千人/年。（3）不同肥胖症评价指标影响受试者新发胆石症的危险因素分析：COX比例风险模型分析结果显示：在男性受试者中，校正年龄、TC、TG、糖尿病、高血压病、吸烟、饮酒、体育锻炼因素对新发胆石症的影响后，与BMI正常受试者比较，超重、肥胖受试者新发胆石症的风险增加（HR=1.35,1.63，95%CI：1.24～1.46，1.48～1.80，P〈0.05）；与腰围低受试者比较，腰围中、高受试者新发胆石症的风险增加（HR=1.27,1.53，95%CI：1.15～1.40，1.39～1.67，P〈0.05）；与WHtR低受试者比较，WHtR中、高受试者新发胆石症的风险增加（HR=1.20,1.44，95%CI：1.09～1.32，1.31～1.58，P〈0.05）。在女性受试者中，与BMI正常受试者比较，超重、肥胖受试者新发胆石症的风险增加（HR=1.35，1.77，95%CI：1.16～1.56，1.49～2.10，P〈0.05）；与腰围低受试者比较，腰围中、高受试者新发胆石症的风险增加（HR=1.38，1.72，95%CI：1.15～1.66，1.44～2.07，P〈0.05）；与WHtR低受试者比较，WHtR中、高受试者新发胆石症的风险增加（HR=1.34，1.71，95%CI：1.12～1.61，1.43～2.04，P〈0.05）。（4）不同肥胖症评价指标联合应用对新发胆石症模型的拟合情况和预测价值的比较：男性受试者中建立多因素模型，将年龄、TC、TG、糖尿病、高血压病、吸烟、饮酒、体育锻炼等胆石症传统危险因素共同带入模型中，计算此模型的-2log L值和AIC值，分别为71 257和71 275。而后将BMI、腰围、WHtR、BMI＋腰围、BMI＋WHtR、腰围＋WHtR、BMI＋腰围＋WHtR分别带入此模型中，计算-2log L值和AIC值分别为71 156和71 178、71 170和71 192、71 197和71 219、71 134和71 160、71 132和71 162、71 170和71 196、71 132和71 162。其中AIC值最小模型为多因素模型＋BMI＋腰围组合，与多因素模型的-2log L值的差值为123，差异有统计学意义（x2=123.00，P〈0.05）。女性受试者中建立多因素模型，将年龄、TC、TG、糖尿病、高血压病、吸烟、饮酒、体育锻炼等胆石症传统危险因素共同带入模型中，计算此模型的-2log L值和AIC值，分别为19 612和19 630。而后将BMI、腰围、WHtR、BMI＋腰围、BMI＋WHtR、腰围＋WHtR、BMI＋腰围＋WHtR分别带入模型中，计算-2log L值和AIC值分别为19 568和19 590、19 575和19 597、19 574和19 596、19 558和19 584、19 557和19 583、19 571和19 597、19 556和19 586。其中AIC值最小模型为多因素模型＋BMI＋WHtR组合，与多因素模型的-2log L值的差值为55，差异有统计学意义（x2=55.00，P〈0.05）。结论：男性和女性人群中，高水平的BMI、腰围、WHtR均为胆石症发病的独立危险因素。在男性人群中联合BMI、腰围作为肥胖症评价指标可提高对新发胆石症的预测价值；而在女性人群中联合BMI、WHtR则是预测新发胆石症的最优组合。

Objective：To explore the predictive value of combined application of the different obesity measures on incident gallstone disease （GD） and find the optimal combination. Methods：The retrospective cohort study was conducted. The data of 88 947 participants who participated in health examination at the Kailuan General Hospital, Kailuan Linxi Hospital, Kailuan Zhaogezhuang Hospital, Kailuan Tangjiazhuang Hospital, Kailuan Fan′gezhuang Hospital, Kailuan Jinggezhuang Hospital, Kailuan Lyujiatuo Hospital, Kailuan Linnancang Hospital, Kailuan Qianjiaying Hospital, Kailuan Majiagou Hospital and Kailuan Branch Hospital from July 2006 to December 2015 were collected. All participants received the same-order health examinations by the fixed team of doctors in 2006, 2008, 2010, 2012 and 2014 at the same place. Epidemiological investigation, anthropometric parameters and biochemical indicators were collected. Observation indicators： （1） comparisons of general data between 2 genders; （2） incidence of GD; （3） risk factors analysis of the different obesity measures affecting incident GD; （4） comparisons of the fitting degree and predictive value of combined application of the different obesity measures on incident GD model. Measurement data with normal distribution were represented as ±s, and comparisons between groups were analyzed using the t test. Measurement data with skewed distribution were described as M （P25, P75）, and comparisons between groups were analyzed using the rank sum test. Comparisons of count data were analyzed using the chi-square test. The incidences of GD between 2 genders were calculated by person-year of follow-up. The hazard ratio （HR） and 95% confidence interval （CI） of the different obesity measures on incident GD were estimated by the COX proportional hazard model. The fitting degree of different combination of obesity measures on incident GD model was calculated by the likelihood ratio test and akaike information criterion （AIC）. Results：（1） Comparisons of general data between 2 genders： of 88 947 participants, age, body mass index （BMI）, waist circumference （WC）, systolic pressure, diastolic pressure, total cholesterol （TC）, triglyceride （TG）, fasting plasma glucose （FPG）, cases with diabetes, hypertension, smoking, drinking and physical exercise were respectively （51±12）years old, （25±3）kg/m2, （88±10）cm, （132±20）mmHg （1mmHg=0.133 kPa）, （84±12）mmHg, （4.95±1.16）mmol/L, 1.18 mmol/L （0.81 mmol/L, 1.74 mmol/L）, （5.5±1.6）mmol/L, 6 223, 31 816, 26 993, 15 779, 11 063 in male participants and （49±11）years old, （25±4）kg/m2, （83±11）cm, （124±21）mmHg, （79±11）mmHg, （4.98±1.08）mmol/L, 1.30 mmol/L （0.92 mmol/L, 2.00 mmol/L）, （5.3±1.6）mmol/L, 1 409, 5 866, 248, 87, 2 450 in female participants, with statistically significant differences [t=587.20, 894.27, 1 064.97, 813.49, 986.22, 630.97, H（x2）=642.39, t=452.87, x2=35.10, 1 205.40, 9 619.42, 4 901.75, 84.82, P〈0.05]. （2） Incidence of GD： 88 947 participants were followed up for 713 345 person-year, 4 291 participants had incident GD, with a total person-year incidence of 6.02 thousand person / year. The total follow-up time, cases with incident GD and person-year incidence were respectively 562 821 person-year, 3 268, 5.81 thousand person / year in male participants and 150 524 person-year , 1 023, 6.80 thousand person / year in female participants. （3） Risk factors analysis of the different obesity measures affecting incident GD： the results of COX proportional hazard model： in male participants, adjusted for age, TC, TG, diabetes, hypertension, smoking, drinking and physical exercise, BMI was associated with increased risk of incident GD （HR=1.35, 1.63, 95%CI： 1.24-1.46,1.48-1.80, P〈0.05）; WC was associated with increased risk of incident GD （HR=1.27, 1.53, 95%CI： 1.15-1.40, 1.39-1.67, P〈0.05）; waist-to-height ratio （WHtR） was associated with increased risk of incident GD （HR=1.20, 1.44, 95%CI： 1.09-1.32, 1.31-1.58, P〈0.05）. In female participants, BMI was associated with increased risk of incident GD （HR=1.35, 1.77, 95%CI： 1.16-1.56, 1.49-2.10, P〈0.05）; WC was associated with increased risk of incident GD （HR=1.38, 1.72, 95%CI： 1.15-1.66, 1.44-2.07, P〈0.05）; WHtR was associated with increased risk of incident GD （HR=1.34, 1.71, 95%CI： 1.12-1.61, 1.43-2.04, P〈0.05）. （4） Comparisons of the fitting degree and predictive value of combined application of the different obesity measures on incident gallstone diseases model： multi-factor model of male participants was constructed after adding risk factors of age, TC, TG, diabetes, hypertension, smoking, drinking and physical exercise, and -2log L and AIC were 71 257 and 71 275. Then BMI, WC, WHtR, BMI＋WC, BMI＋WHtR, WC＋WHtR and BMI＋WC＋WHtR were respectively added into the multi-factor model, and -2log L and AIC were respectively 71 156 and 71 178, 71 170 and 71 192, 71 197 and 71 219, 71 134 and 71 160, 71 132 and 71 162, 71 170 and 71 196, 71 132 and 71 162. The minimal mode of AIC was multi-factor model＋BMI＋WC, with a difference of 123 compared with multi-factor model of -2log L, showing a statistically significant difference （x2=123.00, P〈0.05）. The multi-factor model of female participants was constructed after adding risk factors of age, TC, TG, diabetes, hypertension, smoking, drinking and physical exercise, and -2log L and AIC were 19 612 and 19 630. Then BMI, WC, WHtR, BMI＋WC, BMI＋WHtR, WC＋WHtR and BMI＋WC＋WHtR were respectively added into the multi-factor model, and -2log L and AIC were respectively 19 568 and 19 590, 19 575 and 19 597, 19 574 and 19 596, 19 558 and 19 584, 19 557 and 19 583, 19 571 and 19 597, 19 556 and 19 586. The minimal mode of AIC was multi-factor model＋BMI＋WHtR, with a difference of 55 compared with multi-factor model of -2log L, showing a statistically significant difference （x2=55.00, P〈0.05）. Conclusions：The increased BMI, WC and WHtR are independent risk factors for incident GD, no matter the gender. In males, the combination of BMI and WC can improved the predictive value of the incident GD, while in females, BMI and WHtR are the best combination for predicting incident GD.