PCD logo

Discriminatory Capacity of Anthropometric Indices for Cardiovascular Disease in Adults: A Systematic Review and Meta-Analysis

PEER REVIEWED

We identified 2,457 articles for possible analysis through database searching (PubMed, n = 771; Web of Science, n = 575; Cochrane, n = 439; Scopus, n = 478; Proquest, n = 194) and identified 23 from other sources (Google Scholar, n = 9; Magiran, n = 3; SID, n = 1; reference checking, n = 10). We removed duplicate publications (n = 892) and were left with 1,588 articles. We reviewed the titles and abstracts of those 1,588 articles, and excluded 1,356 that were found not to be relevant. We then had 232 full-text articles for possible inclusion. Of those, we excluded 194 articles for the following reasons: 146 had no relevant outcome data; 9 were studies on a specific subpopulation; 10 were not available in English; 5 were systematic reviews and/or meta-analyses; 19 did not report cut-offs, AUC, or sensitivity and specificity; 2 were conference abstracts; and 3 did not study an adult population. After screening and exclusion, we had 38 articles for qualitative and quantitative syntheses.


Figure 1.

PRISMA flowchart of the study selection process. Abbreviations: AUC, area under the receiver operating characteristic curve; PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses.

Return to Article


Figure 2.

Random-effects pooled area under the ROC curve (AUC) and test of heterogeneity for body mass index with cardiovascular disease or cardiometabolic disease for men and women. The dashed lines indicate the null model. Weighted percentages determined by using random effects analysis. Abbreviation: ROC, receiver operating characteristic.

Random-effects pooled area under the ROC curve (AUC) and test of heterogeneity for body mass index with cardiovascular disease or cardiometabolic disease for men and women. The dashed lines indicate the null model. Weighted percentages determined by using random effects analysis. Abbreviation: ROC, receiver operating characteristic.
Study, Year AUC (95% Confidence Interval) Weighted %
Men
Al-Lawati, 2008 0.60 (0.56-0.64) 7.54
Ptanga, 2005 0.64 (0.57-0.71) 5.84
Zeng, 2014 0.71 (0.71-0.71) 8.48
Samsen, 2012 0.67 (0.66-0.68) 8.38
Haun, 2009 0.64 (0.57-0.71) 5.84
Kim, 2016 0.54 (0.53-0.55) 8.42
Lin, 2002 0.69 (0.68-0.70) 8.46
Wakabayashi, 2012 0.71 (0.70-0.72) 8.46
Hadaegh, 2009 0.57 (0.53-0.61) 7.35
Zabetian, 2009 0.57 (0.53-0.61) 7.35
Ho, 2003 0.76 (0.73-0.79) 7.84
Mirmiran, 2004 0.71 (0.68-0.74) 7.84
Katulanda, 2011 0.71 (0.69-0.73) 8.19
Overall (I2= 99.1%,P< .001) 0.66 (0.63-0.69) 100.0
Women
Al-Lawati, 2008 0.64 (0.60-0.67) 5.79
Foucan, 2002 0.71 (0.69-0.73) 6.18
Wildman, 2004 0.51 (0.48-0.54) 6.08
Ptanga, 2005 0.59 (0.53-0.65) 4.98
Pua, 2004 0.69 (0.64-0.74) 5.33
Zeng, 2014 0.75 (0.74-0.75) 6.37
Samsen, 2012 0.64 (0.63-0.64) 6.34
Haun, 2009 0.59 (0.53-0.65) 4.98
Kim, 2016 0.66 (0.64-0.67) 6.25
Satoh, 2010 0.78 (0.77-0.78) 6.34
Lin, 2002 0.72 (0.71-0.73) 6.34
Wakabayashi, 2012 0.75 (0.74-0.76) 6.33
Hadaegh, 2009 0.53 (0.48-0.58) 5.30
Zabetian, 2009 0.53 (0.48-0.58) 5.30
Ho, 2003 0.72 (0.69-0.73) 5.96
Mirmiran, 2004 0.65 (0.62-0.68) 5.96
Katulanda, 2011 0.71 (0.69-0.73) 6.18
Overall (I2= 98.8%,P< .001) 0.66 (0.63-0.69) 100.0

Return to Article


Figure 3.

Random-effects pooled area under the ROC curves (AUC) and test of heterogeneity for waist circumference with cardiovascular disease or cardiometabolic disease for men and women. The dashed lines indicate the null model. Weighted percentages determined by using random effects analysis. Abbreviation: ROC, receiver operating characteristic.

Random-effects pooled area under the ROC curves (AUC) and test of heterogeneity for waist circumference with cardiovascular disease or cardiometabolic disease for men and women. The dashed lines indicate the null model. Weighted percentages determined by using random effects analysis. Abbreviation: ROC, receiver operating characteristic.
Study AUC (95% Confidence Interval) Weighted %
Men
Al-Lawati, 2008 0.71 (0.68-0.74) 5.06
Narisawa, 2008 0.69 (0.65-0.73) 4.26
Talaei, 2012 0.59 (0.55-0.63) 4.45
Wildman, 2004 0.52 (0.46-0.58) 2.68
Ptanga, 2005 0.73 (0.67-0.79) 2.88
Zeng, 2014 0.71 (0.71-0.71) 7.89
Samsen, 2012 0.68 (0.67-0.69) 7.47
Haun, 2009 0.73 (0.67-0.79) 2.88
Kim, 2016 0.62 (0.61-0.64) 7.09
Lin, 2002 0.69 (0.69-0.70) 7.77
Siren, 2012 0.86 (0.81-0.91) 3.42
Wakabayashi, 2012 0.72 (0.71-0.72) 7.81
Cheong, 2013 0.69 (0.69-0.70) 7.61
Hadaegh, 2009 0.60 (0.56-0.64) 4.35
Yoshida, 2009 0.61 (0.55-0.67) 2.72
Zabetian, 2009 0.58 (0.51-0.64) 2.64
Ho, 2003 0.78 (0.75-0.81) 5.50
Mirmiran, 2004 0.70 (0.66-0.74) 4.45
Lee, 2010 0.67 (0.62-0.72) 3.57
Katulanda, 2011 0.71 (0.68-0.74) 5.50
Overall (I2= 94.7%,P< .001) 0.69 (0.67-0.70) 100.0
Women
Al-Lawati, 2008 0.73 (0.69-0.76) 4.58
Narisawa, 2008 0.76 (0.70-0.82) 4.33
Talaei, 2012 0.59 (0.55-0.63) 4.53
Foucan, 2002 0.76 (0.74-0.78) 4.64
Wildman, 2004 0.51 (0.51-0.51) 4.68
Ptanga, 2005 0.66 (0.61-0.71) 4.48
Pua, 2004 0.72 (0.68-0.76) 4.53
Zeng, 2014 0.76 (0.76-0.76) 4.68
Samsen, 2012 0.67 (0.67-0.68) 4.67
Haun, 2009 0.66 (0.61-0.71) 4.46
Kim, 2016 0.72 (0.71-0.73) 4.66
Lin, 2002 0.74 (0.73-0.74) 4.67
Wakabayashi, 2012 0.75 (0.74-0.75) 4.67
Cheong, 2013 0.71 (0.71-0.72) 4.67
Almeida de, 2009 0.70 (0.63-0.77) 4.26
Hadaegh, 2009 0.60 (0.55-0.65) 4.48
Yoshida, 2009 0.66 (0.61-0.71) 4.49
Zabetian, 2009 0.57 (0.49-0.64) 4.21
Ho, 2003 0.74 (0.71-0.77) 4.60
Mirmiran, 2004 0.68 (0.66-0.70) 4.64
Lee, 2010 0.77 (0.72-0.82) 4.46
Katulanda, 2011 0.72 (0.70-0.74) 4.64
Overall (I2= 99.7%,P< .001) 0.69 (0.64-0.74) 100.0

Return to Article


Figure 4.

Random-effects pooled area under the ROC curves (AUC) and test of heterogeneity for waist-to-hip ratio with cardiovascular disease or cardiometabolic disease for men and women. The dashed lines indicate the null model. Weighted percentages determined by using random effects analysis. Abbreviation: ROC, receiver operating characteristic.

Random-effects pooled area under the ROC curves (AUC) and test of heterogeneity for waist-to-hip ratio with cardiovascular disease or cardiometabolic disease for men and women. The dashed lines indicate the null model. Weighted percentages determined by using random effects analysis. Abbreviation: ROC, receiver operating characteristic.
Study AUC (95% Confidence Interval) Weighted %
Men
Al-Lawati, 2008 0.77 (0.74-0.80) 10.27
Ptanga, 2005 0.76 (0.70-0.82) 8.65
Haun, 2009 0.76 (0.70-0.82) 8.65
Lin, 2002 0.68 (0.67-0.69) 11.68
Hadaegh, 2009 0.62 (0.58-0.66) 10.06
Li, 2014 0.61 (0.58-0.63) 11.10
Zabetian, 2009 0.57 (0.50-0.64) 8.12
Ho, 2003 0.77 (0.73-0.81) 10.13
Mirmiran, 2004 0.70 (0.66-0.74) 10.13
Katulanda, 2011 0.68 (0.65-0.71) 10.78
Overall (I2= 92.6%,P< .001) 0.69 (0.66-0.73) 100.0
Women
Al-Lawati, 2008 0.80 (0.77-0.83) 9.30
Ptanga, 2005 0.75 (0.70-0.80) 7.56
Pua, 2004 0.69 (0.64-0.74) 7.56
Haun, 2009 0.75 (0.70-0.80) 7.56
Lin, 2002 0.72 (0.72-0.73) 10.43
Almeida de, 2009 0.74 (0.67-0.81) 5.96
Hadaegh, 2009 0.66 (0.61-0.70) 7.90
Li, 2014 0.67 (0.64-0.69) 9.72
Zabetian, 2009 0.58 (0.50-0.65) 5.66
Ho, 2003 0.74 (0.71-0.77) 9.23
Mirmiran, 2004 0.66 (0.66-0.70) 9.23
Katulanda, 2011 0.68 (0.66-0.70) 9.91
Overall (I2= 89.8%,P< .001) 0.71 (0.68-0.73) 100.0

Return to Article

Top

Error processing SSI file

The opinions expressed by authors contributing to this journal do not necessarily reflect the opinions of the U.S. Department of Health and Human Services, the Public Health Service, the Centers for Disease Control and Prevention, or the authors’ affiliated institutions.