比較三種肥胖指標預測高密度膽固醇不足的最適切點

Optimal Cutoff Value of Three Obesity Indexes for Predicting the High-density Lipoprotein Cholesterol Less than 46mg/dL

李揚琦(Young-Chi Lee)

指導教授 ： 廖勇柏

中山醫學大學/健康管理學院/公共衛生學系碩士班/碩士(2012年)

https://doi.org/10.6834/CSMU.2012.00055

摘要

研究目的：探討常見的肥胖指標對於高密度膽固醇的預測能力。

研究方法：資料來自於國民營養健康狀況變遷調查(Nutrition and Health Survey in Taiwan, NAHSIT, 1993-1996，1999-2000，2005-2008)。比較三種肥胖指標–身體質量指數 (body mass index,BMI)、腰臀比 (waist-hip ratio,WHR)及新指標–身體質量指數加權腰臀比 (BMI_WHR) 與HDL-C的相關性。利用邏輯式迴歸分析(Logistic Regression)與接受者操作型曲線下面積 (Receiver Operation Characteristic,ROC curve)分析探討高密度膽固醇＜46mg/dL最適切點。

結果：利用邏輯式迴歸分析，將BMI、WHR和BMI_WHR分三層納入三種邏輯式迴歸模式中，分別計算對高密度膽固醇＜46mg/dL之危險性對比值。BMI、WHR和BMI_WHR比值上升，無論任何性別或年齡層，高密度膽固醇＜46mg/dL之危險性也顯著增加，且達劑量效應關係 (test for trend：p=<0.0001)。而接受者操作型曲線下面積 (Receiver Operation Characteristic，ROC curve)分析肥胖指標對高密度膽固醇＜46mg/dL的預測能力，對高密度膽固醇＜46mg/dL診斷預測性之最適切點，可以發現WHR的最適切點無論那個年齡層均隨著年代的增加而增加，男性19-44歲組由1993-1996的0.83 (Sen=0.68；Spe=0.55)增加到2005-2008的0.87 (Sen=0.73；Spe=0.72)，45-64歲組由0.88 (Sen=0.62；Spe=0.57)增加到0.92 (Sen=0.58；Spe=0.59)；女性WHR切點則分別由19-44歲組的0.76 (Sen=0.67；Spe=0.66)增加到0.81 (Sen=0.85；Spe=0.77)，45-64歲組由0.80 ( Sen=0.72；Spe=0.59)到0.87 (Sen=0.54；Spe=0.73)。BMI預測HDL-C的最適切點在男性19-44歲組隨著年代增加而增加，23.56 (Sen=0.61；Spe=0.68)增加到25.29 (Sen=0.55；Spe=0.78)，45-64歲組則隨年代增加而下降,25.15 (Sen=0.49；Spe=0.72)下降為24.85 (Sen=0.59；Spe=0.64)；而女性19-44歲組的最適切點隨著年代增加而下降24.27 (Sen=0.56；Spe=0.75)下降為23.47 (Sen=0.73；Spe=0.74)，45-64歲組卻增加25.12 (Sen=0.61；Spe=0.58)增加到25.88 (Sen=0.56；Spe=0.71)。BMI_WHR的最適切點男性19-44歲組由1993-1996的20.26 (Sen=0.57；Spe=0.73)增加到2005-2008的20.84 (Sen=0.71；Spe=0.65)，45-64歲組由21.67 (Sen=0.59；Spe=0.61)增加到22.36 (Sen=0.65；Spe=0.61)；女性BMI_WHR切點則分別由19-44歲組的17.52 (Sen=0.63；Spe=0.67)增加到18.79 (Sen=0.79；Spe=0.72)，45-64歲組由20.48 (Sen=0.64；Spe=0.59)到21.40 (Sen=0.73；Spe=0.63)。三個年代65歲以上的老人在WHR和BMI的最適切點皆隨著年代增加而增加。男性WHR切點由0.88 (Sen=0.63；Spe=0.63) (1993-1996)增加到0.92 (Sen=0.59；Spe=0.65) (1999-2000)到0.94 (Sen=0.62；Spe=0.60) (2005-2008)；女性WHR切點在三個年代分別為0.83 (Sen=0.63；Spe=0.54)增加到0.86 (Sen=0.69；Spe=0.58)到0.93 (Sen=0.55；Spe=0.68)。在男性，BMI最適切點分別為23.66 (Sen=0.57；Spe=0.71)增加到23.70 (Sen=0.62；Spe=0.61)到23.99 (Sen=0.64；Spe=0.58)；女性則為23.41 (Sen=0.68；Spe=0.44)增加到 24.40 (Sen=0.62；Spe=0.55)到25.73 (Sen=0.60；Spe=0.61)。男性BMI_WHR的最適切點三個年代分別為19.93 (Sen=0.72；Spe=0.60)增加到21.51 (Sen=0.66；Spe=0.63)到22.66 (Sen=0.65；Spe=0.59)。女性BMI_WHR的最適切點三個年代分別為20.20 (Sen=0.62；Spe=0.49)增加到22.12 (Sen=0.57；Spe=0.65)到24.32 (Sen=0.56；Spe=0.68)。

結論：
一、不同年代、不同年齡層、不同性別，三種指標BMI, WHR, BMI_WHR值愈大，HDL-C不足(<46 mg/dL)的危險性愈高且達劑量效應關係。
二、三個年代65歲以上的老人在WHR、BMI和BMI_WHR預測HDL-C不足的最適切點皆隨著年代增加而增加，且男性在BMI_WHR相較於WHR, BMI有較高的敏感度。
三、在19-44, 45-64歲組， WHR和BMI_WHR的最適切點亦隨著年代增加而增加，但BMI則否。在敏感度方面早期資料(1993-1996)顯示WHR有較佳的敏感度，而2005-2008的資料顯示，19-44歲組WHR有較高的敏感度，45-64歲組則以BMI_WHR有較高的敏感度。
四、不同年齡層有不同的體位指標最適切點，1993-1996年中，19-44歲男性：BMI為23.56；WHR為0.83；BMI_WHR為20.26。19-44歲女性：BMI為24.27；WHR為0.76；BMI_WHR為17.52。45-64歲男性：BMI為25.15；WHR為0.88；BMI_WHR為21.67。45-64歲女性：BMI為25.12；WHR為0.80；BMI_WHR為20.48。而2005-2008年中，19-44歲男性：BMI為25.29；WHR為0.87；BMI_WHR為20.84。19-44歲女性：BMI為23.47；WHR為0.81；BMI_WHR為18.79。45-64歲男性：BMI為24.85；WHR為0.92；BMI_WHR為22.36。45-64歲女性：BMI為25.88；WHR為0.87；BMI_WHR為21.40。

關鍵字

高密度膽固醇 ； 身體質量指數 ； 腰臀比 ； 身體質量指數_腰臀比 ； 流行病學 ； 接受者操作型曲線分析

並列摘要

Objective: This study is relating to the obesity index which are able to predict HDL is under 46mg/dL.

Methods: The data were collected from Nutrition and Health Survey in Taiwan,NAHSIT,1993-1996，1999-2000，2005-2008. Using three index, including the body mass index (BMI), waist-hip ratio (WHR), and BMI weighted WHR correlate to HDL. The cut-point of the obesity index was able to predict HDL is under 46mg/dL via logistic regression and ROC curve.

Results: The three odd ratios received from BMI, WHR and BMI_WHR were calculated through the logistic regression. The risk of HDL＜46 mg/dL would be increased due to the higher odd ratios. This positive proportion also conform the dose response relationship. The cut-point of the obesity index was able to predict HDL is under 46mg/dL via ROC curve. The cut point of WHR is increasing by period no matter which ages they are. For males,ages between 19 to 44 years old in 1993 to 1996 is 0.83(Sen=0.68；Spe=0.55) and then go to 0.87(Sen=0.73；Spe=0.72) in 2005-2008. Same positive trend from 0.88(Sen=0.62；Spe=0.57) to 0.92(Sen=0.58；Spe=0.59) for ages between 45 to 64 years old. For females, ages between 19 to 44 years old in 1993 to 1996 is 0.76(Sen=0.67；Spe=0.66) and then go to 0.81(Sen=0.85；Spe=0.77) in 2005-2008. Same positive trend from 0.80( Sen=0.72；Spe=0.59) to 0.87(Sen=0.54；Spe=0.73) for ages between 45 to 64 years old.The cut-point of BMI for males is ages between 19 to 44 years old in 1993 to 1996 is 23.56(Sen=0.61；Spe=0.68) and then go to 25.29(Sen=0.55；Spe=0.78) in 2005-2008. Same positive trend from 25.15(Sen=0.49；Spe=0.72) to 24.85(Sen=0.59；Spe=0.64) for ages between 45 to 64 years old.For females, ages between 19 to 44 years old in 1993 to 1996 is 24.27(Sen=0.56；Spe=0.75) and then go to 23.47 (Sen=0.73；Spe=0.74)in 2005-2008. Same positive trend from 25.12(Sen=0.61；Spe=0.58) to 25.88 (Sen=0.56；Spe=0.71)for ages between 45 to 64 years old.The BMI is rising up by period as the age is 19-44, but decays as the age is between 45 and 64 for males. Nevertheless, this trend is tend to opposite for females. The cut-point of BMI_WHR for males is ages between 19 to 44 years old in 1993 to 1996 is 20.26(Sen=0.57；Spe=0.73) and then go to 20.84(Sen=0.71；Spe=0.65) in 2005-2008. Same positive trend from 21.67 (Sen=0.59；Spe=0.61)to 22.36(Sen=0.65；Spe=0.61) for ages between 45 to 64 years old.For females, ages between 19 to 44 years old in 1993 to 1996 is 17.52(Sen=0.63；Spe=0.67) and then go to 18.79(Sen=0.79；Spe=0.72) in 2005-2008. Same positive trend from 20.48(Sen=0.64Spe=0.59) to 21.40(Sen=0.73；Spe=0.63) for ages between 45 to 64 years old.
The cut-point of WHR for males is increase from 0.88(Sen=0.63；Spe=0.63) to 0.92 (Sen=0.59；Spe=0.65) to 0.94(Sen=0.62；Spe=0.60) and 0.83(Sen=0.63；Spe=0.54) to 0.86(Sen=0.69；Spe=0.58) to 0.93(Sen=0.55；Spe=0.68) for females when the age is up to 65 years old, which is also increased by period. The cut-point of BMI for males is increase from23.66(Sen=0.57；Spe=0.71) to 23.70(Sen=0.62；Spe=0.61) -23.99(Sen=0.64；Spe=0.58) and 23.41(Sen=0.68；Spe=0.44) to 24.40(Sen=0.62；Spe=0.55)-25.73 (Sen=0.60；Spe=0.61)for females. The cut-point of BMI_WHR for males is increase from 19.93(Sen=0.72；Spe=0.60) to 21.51(Sen=0.66；Spe=0.63)- 22.66(Sen=0.65；Spe=0.59) and 20.20(Sen=0.62；Spe=0.49) to 22.12(Sen=0.57；Spe=0.65)- 24.32(Sen=0.56；Spe=0.68) for females.
Conclusion: 1.Different period,different ages,different gender,three indicators,larger BMI,WHR,BMI_WHR value HDL-C less than 46mg/dL,the higher the risk and up to the dose-response relationship.2. The cup point of HDL_C<46mg/dL predicted from WHR, BMI and BMI_WHR for people who ages is up to 65 years old in these three periods is increased by periods. Also, BMI_WHR of males is more sensitive than WHR and BMI.
3.In the 19-44 and 45-64 age group,the WHR and BMI_WHR the optimal cut point with the age increase,but BMI did not.In the early data(1993-1996)WHR have better sensitivity,and 2005-2008 information,19-44 years old group WHR have higher sensitivity,45-64 years old groupBMI_ WHR have higher sensitivity.
4. Different ages have different optimal cutoff value of obesity indexes：To1993-1996：For males ages between 19 to 44 years old, the cut-point of BMI is 23.56；WHR is 0.83；BMI_WHR is 20.26。.For females ages between 19 to 44 years old, the cut-point of BMI is 24.27；WHR is 0.76；BMI_WHR is 17.52。For males ages between 45 to 64 years old, the cut-point of BMI is 25.15；WHR is 0.88；BMI_WHR is 21.67。For females ages between 45 to 64 years old, the cut-point of BMI is 25.12；WHR is 0.80；BMI_WHR is 20.48。To 2005-2008：For males ages between 19 to 44 years old, the cut-point of BMI is 25.29；WHR is 0.87；BMI_WHR is 20.84。.For females ages between 19 to 44 years old, the cut-point of BMI is 23.47；WHR is 0.81；BMI_WHR is 18.79。For males ages between 45 to 64 years old, the cut-point of BMI is 24.85；WHR is 0.92；BMI_WHR is 22.36。For females ages between 45 to 64 years old, the cut-point of BMI is 25.88；WHR is 0.87；BMI_WHR is 21.40。

並列關鍵字

High-density lipoproteins ； Body mass index ； Waist-hip ratio ； BMI weighted WHR ； Epidemiology ； ROC curve

參考文獻

31. Mellati AA, Mousavinasab SN, Sokhanvar S, Ali Naghi Kazemi S, Esmailli MH, Dinmohamadi H. Correlation of anthropometric indices with common cardiovascular risk factors in an urban adult population of Iran: data from Zanjan Healthy Heart Study. Asia Pac J Clin Nutr. 2009;18(2):217.

1. 行政院衛生署國民健康局2007台灣地區高血糖、高血壓、高血脂之追蹤調查研究.

2. Meilahn EN, Kuller LH, Matthews KA, Wing RR, Caggiula AW, Stein EA. Potential for increasing high-density lipoprotein cholesterol, subfractions HDL2-C and HDL3-C, and apoprotein AI among middle-age women. Preventive medicine. Jul 1991;20(4):462-473.

5. Mineo C, Deguchi H, Griffin JH, Shaul PW. Endothelial and antithrombotic actions of HDL. Circulation research. 2006;98(11):1352-1364.

6. Hausenloy DJ, Yellon DM. Targeting residual cardiovascular risk: raising high-density lipoprotein cholesterol levels. Heart. 2008;94(6):706-714.