Professional Documents
Culture Documents
& 2010 Macmillan Publishers Limited All rights reserved 0954-3007/10 $32.00
www.nature.com/ejcn
REVIEW
The optimal cutoff values and their performance of
waist circumference and waist-to-hip ratio for
diagnosing type II diabetes
Q Qiao1,2 and R Nyamdorj1,2
1
Department of Public Health, University of Helsinki, Helsinki, Finland and 2Department of Chronic Disease Prevention, National
Institute for Health and Welfare, Helsinki, Finland
Studies on the theme of optimal cutoff values of waist circumference (WC) and waist-to-hip ratio (WHR) for assessing risk of type
II diabetes were reviewed. Twenty-eight studies of individuals aged 1874 years are eligible for inclusion. Four of these studies
are prospective and the rest are all cross-sectional. Tongans had the highest WC (103 cm for both men and women) cutoff value
(but not for WHR), followed by studies in the USA and U.K. The WC cutoff values were higher for all races in the USA and the UK
studies compared with their counterparts in their original countries. The optimal WC (WHR) cutoff values were 9799 cm (0.95)
for White men and 85 cm (0.830.85) for White women living outside the USA and the UK, whereas they were 85 cm (0.90) for
Asian men and 7580 cm (0.790.85) for Asian women; the values for other ethnic groups were between those for White and
Asians. Men had higher values than women in White, Chinese, Japanese, Indians and Bangladeshis, but not in Thai, Iranians,
Iraqi, Tunisians, Mexicans, Africans and Tongans. At these optimal cutoff points the sensitivities were around 6070%, which
was higher or equal to the specificity. There is no universal cutoff value that can be applied worldwide, and a country-specific
value should be considered taking into account the purposes and resources.
European Journal of Clinical Nutrition (2010) 64, 2329; doi:10.1038/ejcn.2009.92; published online 19 August 2009
Keywords: waist circumference; BMI; type II diabetes; cutoff values; sensitivity and specificity
Introduction and 7 years after the trial in the Finnish Diabetes Prevention
Study (Lindstrom et al., 2006) have also been reported, with
Obesity, as a major risk factor for type II diabetes, has been an effect size of 43% for both the trials.
very well documented. Clinical trials have unequivocally The most commonly used anthropometric measurements
shown that reducing weight and increasing physical activity for obesity are body mass index (BMI) indicating general
can prevent or at least delay the onset of type II diabetes in obesity, waist circumference (WC) and waist-to-hip ratio
individuals with impaired glucose tolerance in many ethnic (WHR) as markers for central obesity. Other indicators such
groups around the world including Swedish (Eriksson and as hip circumference (Lissner et al., 2001; Snijder et al., 2004),
Lindgarde, 1991), Chinese (Pan et al., 1997), Finnish waist-to-height ratio (Hsieh and Muto, 2005) and waist-to-
(Tuomilehto et al., 2001), American (Diabetes Prevention stature ratio (Ho et al., 2003) have also been suggested to be
Program Research, 2002), Indians (Ramachandran et al., useful markers of obesity. The concept of central obesity was
2006) and Japanese (Kosaka et al., 2005). The relative risk first introduced by Vague in the 1940s (Vague, 1947) and
reduction ranged from 28% in Indians to 67% in Japanese. later in 1956 he pointed out for the first time that central
A sustainable long-term effect of lifestyle intervention 20 obesity (android) was more important than peripheral
years after trial in the Chinese Daqing Study (Li et al., 2008) obesity (gynaecoid) in relation to diabetes, gout, athero-
sclerosis and urate calculus diseases (Vague, 1956). Since the
Correspondence: Dr Q Qiao, Department of Public Health, University of 1990s interest in WC has increased because it correlates more
Helsinki, PL41, Mannerheimintie 172, FIN-00014 Helsinki, Finland. closely with the abdominal visceral fat than either WHR or
E-mail: qing.qiao@ktl.fi
Received 3 February 2009; accepted 29 May 2009; published online 19
BMI (Pouliot et al., 1994; Han et al., 1995; Lean et al., 1995).
August 2009 The association of these obesity indicators with type II
Waist cutoff for diabetes
Q Qiao and R Nyamdorj
24
diabetes has been studied in different ethnic groups in recent cutoff values for the WC were higher for all races in the
years with regard to the optimal cutoff values for definition USA and in the UK studies compared with their counter-
of obesity. Attempts to identify optimal cutoff values to parts in their original countries (Stevens et al., 2001; Diaz
assess diabetes risk have come out with different findings. et al., 2007).
These findings are reviewed and summarized in this article, (2) When the two studies from the USA and UK were not
with the aim to answer questions of (1) What is the optimal counted in, there were fewer variations for both the WC
WC (WHR) cutoff point for assessing the risk of type II and the WHR in White men and women, and in Asian
diabetes? (2) How effective is this cutoff point? men as well; the optimal cutoff values for the WC was
9799 cm for White men, 85 cm for White women,
whereas they were 85 cm for Asian men and 7580 cm
Materials and methods for Asian women; the optimal cutoff values for the WC
for other ethnic groups were between those of White and
Inclusion criteria Asians.
All publications in English that reported optimal cutoff (3) If not including studies from the USA and UK, the
values for the WC or the WHR in adults, derived at the point optimal WHR cutoff values were 0.95 for White men and
that maximizes the sum of sensitivity and specificity or 0.90 for Asian men, 0.830.85 for White women and
applying other valid methods, were included. 0.790.85 for Asian women; the values for the other
ethnic groups were between the two.
(4) The optimal cutoff values for the WC and the WHR
Data sources and limitations were higher in men than in women in White,
Published articles related to these topics have been Chinese, Japanese, Indians and Bangladesh; but the
searched in PubMed from 1975 onwards or knowing from WC was similar to or even lower in men than in
conferences and colleagues etc., have been reviewed by two women in Pakistani living in the USA and the UK, Thai,
independent researchers (RN, QQ). Most of the studies Iranians, Iraqi, Tunisians, Mexicans, Africans and
include individuals aged 1874 years, except for one Chinese Tongans.
study (Woo et al., 2002). Participants in this Chinese study (5) For men the optimal cutoff values were about 10 cm
were older than 70 years and selected from a list of recipients higher in White than in Asians for the WC and 0.020.03
of Old Age and Disability Allowance. Compared with higher for the WHR; but the difference for women was
other Chinese studies, the WC was obviously higher whereas not as striking as that for men.
the BMI was lower in this elderly Chinese population. (6) At the optimal cutoff values the sensitivity was higher
In different studies waist has been measured in different than or equal to the specificity, with the lowest
anatomic locations, and diabetes defined by either a sensitivity of 47% for the WC and 57% for the WHR in
previous history of diabetes or based on FPG or FPG plus Mexican men; for most of the studies the sensitivities
2hPGT levels. Most of the studies are population-based with were around 6070%.
random sampling approaches and a few are hospital-based
with participants coming for the health check-up as
indicated in Table 1. Twenty-eight studies have tried to Conclusions
identify optimal cutoff values by selecting the point that
maximizes the sum of sensitivity and specificity. Four of the In the literature almost all studies have applied the
28 studies were prospective (Stevens et al., 2001; Sargeant sensitivity and specificity approach to determine the
et al., 2002; Woo et al., 2002; Katzmarzyk et al., 2007), and 24 optimal cutoff values for assessing type II diabetes risk. The
studies cross-sectional. cutoff values selected using such an approach are all
arbitrary, because they are based purely on the analyzing
trade-off between sensitivity and specificity. To increase the
Results sensitivity, it will reduce the specificity and vice versa. We
may hope to have a high sensitivity for the WC measure-
To easily compare the similarity and difference between ment in the health promotion to increase public awareness
ethnic groups, findings of different studies have been re- of obesity and diabetes, but a high specificity for an
arranged by putting the results of the same ethnic group individual diagnosis is expected in clinical practice. How-
side-by-side (Table 2). Studies with more than one ethnicity ever, whether the WC measurement can be applied, as a first-
were split according to ethnicity, and those with mixed races step diagnostic tool for assessing an individuals diabetes risk
who could not be separated were not included in the Table 2. is still unknown and need to be further investigated based
The main findings are summarized below: on the well designed prospective studies with incidence of
diabetes as an outcome. As the majority of the studies in the
(1) Except for Tongans who had highest cutoff values for the literature were cross-sectional, the evidence obtained based
BMI and the WC (but not for the WHR), the optimal on these studies is less confirmative because they might
Men Women
Mixed Canadian 25.6 85.5 0.85 25.6 85.5 0.85 Random household, n 1543, 1869 years;
(Canada) 66 and 66 67 and 67 63 and 63 66 and 66 67 and 67 63 and 63 incidence of diagnosed DM , follow-up
16 years; lowest floating rib
Katzmarzyk et al. (2007)
White (majority) X28.0 X103 X0.97 X27.8 X94.0 X0.87 Primary care-based, n 5377, X18 years;
(Germany) 64 and 64 65 and 64 58 and 60 68 and 68 68 and 67 58 and 63 physicians diagnosis of DM; midway between
the lowest rib and pelvis
Schneider 2007 (DETECT) (Schneider et al., 2007)
Chinese (Taiwan) 24.5 84.5 0.88 23.4 74.5 0.79 From four nationwide health screening centers,
66 and 66 97 and 70 71 and 71 75 and 75 78 and 77 79 and 79 n 55 563, mean 37 years; FPG X7.0 mmol/l;
midway between the lowest rib and the
crest of the ilium; Lin et al. (2002)
Chinese 24.3 84.0 0.91 24.3 78.4 0.83 Occupational study n 1513; FPG X7.8/2hPG
(Hong Kong- 67 and 66 67 and 67 76 and 76 67 and 66 70 and 70 71 and 70 X11.1 mmol/l; at minimum circumference
working age) between the umbilicus and xiphoid process;
Ko et al. (1999)
Chinese 22.3 88.2 0.97 18.4 85.3 0.83 From recipients of old age and disability allowance,
(Hong Kong) 89 and 56 78 and 67 39 and 85 100 and 15 58 and 55 96 and 18 n 2032, X70 years; incidence of diagnosed
DM, Fu 3 years; Between umbilicus and
xiphoid process; Woo et al. (2002)
Chinese 24.4 83.9 0.92 23.3 78.0 0.83 Random sample n 2895, 2574 years;
(Hong Kong) 71 and 56 76 and 58 64 and 76 81 and 52 75 and 69 78 and 73 FPG X7.0/2hPGX11.1/self-reported DM;
midway between xiphisternum and umbilicus;
Ho et al. (2003)
Men Women
Japanese (Japan) 23.6 85 0.92 23.1 73 0.81 Check-up, n 2728, 2079 years;
59 and 59 62 and 62 71 and 71 67 and 67 70 and 70 78 and 78 FPGX7.0/HbA1c X6.5%/treatment; NA;
Ito et al. (2003)
Tongan (Kingdom 31.7 102.9 0.93 35 102.8 0.86 Random sample, n 767; 2hPG X11.1 mmol/l
of Tonga) 66 and 68 63 and 64 69 and 71 62 and 61 65 and 63 69 and 71 with elevated HbA1c/2hPG X11.1 alone;
at midpoint between the ribs and the iliac crest;
Craig et al. (2007)
Turkish (Turkey) X95 X91 Random sample, n 1682 women and n 1638
70 and 53 75 and 55 men, 2879 years; incidence DM, Fu 7.0 years,
FPGX7.0/2 h PG X11.1 mmol/l/self-report; At
midway between the lower rib margin and the
iliac crest; Onat et al. (2007a, b) (TARFS)
Jamaican 24.8 88.0 0.87 29.3 84.5 0.90 Random sample, n 728, 2574 years,
(African ancestry) 71 and 71 71 and 79 Similar 62 and 65 62 and 65 Similar incidence of DM, Fu 4 years, FPGX7.0/2 h PG
(Jamaica) X11.1 mmol/l; at smallest horizontal
circumference between the ribs and
the iliac crest; Sargeant et al. (2002)
Mexican (Mexico) 25.3 90 0.90 25.4 85 0.86 Hospital workers (check-up), n 8365, X20
57 both 47 both 57 both 55 Both 53 both 62 both years (mean 39); FPGX7.0/2 h PG X11.1 mmol/l;
Berber et al. (2001)
Men Women
Guadeloupean (Guadeloupe) (African descent) Health center check-ups, n 5149; 1874 years
(mean 40); FPGX7.0/drug use; above the iliac
crests and below the lowest rib margin;
Foucan et al. (2002)
1839 years 26.0 85.0
83 and 69 84 and 78
4074 years 27.0 88.0
62 and 52 70 and 60
1874 years 27.0 89.0
70 and 54 70 and 60
have been confounded by other concurrent condi- universal optimal cutoff value that can be applied world-
tions such as hypertension and dyslipidemia, etc. Never- wide. A country- or region-specific cutoff value should
theless, the literature review shows that the optimal cutoff be considered taking into account the purposes and the
values vary across different ethnicities, and there is no resources.
Cutoff Sen Spe Cutoff Sen Spe Cutoff Sen Spe Cutoff Sen Spe
Iranian (Iran)
1834 years 86 0.88 82 0.82
3554 years 91 0.94 93 0.87
5574 years 92 0.96 95 0.91