3.1 Descriptive
A summary of the characteristics for the participants is shown in Table 2. Slightly more women (54.4%) than men were included in the sample. The data shows that prevalence of DMT2 was higher among men (4.3%) compared to women (3.3%). The total prevalence for the sample was 3.7%. Men had a significantly higher average age, weight, height, BMI, HC, WC, WHR, and WHtR compared to women. Almost twice as many men (22.2%) were smokers compared to women (11.7%), with a total smoking prevalence of 16.5%.
Table 2 - Descriptive statistics according to sex
Characteristics
|
Men (n=22,830)
|
Women (n=27,212)
|
Total (n=50,042)
|
P-value
|
DMT2, n (%)
|
975 (4.3)
|
901 (3.3)
|
1,876 (3.7)
|
<0.001
|
Smoking, n (%)
|
5,072 (22.2)
|
3,195 (11.7)
|
8,267 (16.5)
|
<0.001
|
Age (years)
|
53.6±15.6
|
52.8±16.3
|
53.1±16.1
|
<0.001
|
Weight (kg)
|
86.9±13.3
|
72.9±13.7
|
79.2±15.2
|
<0.001
|
Height (cm)
|
177.8±6.7
|
164.6±6.4
|
170.6±9.3
|
<0.001
|
HC (cm)
|
103.6±6.6
|
103.8±9.3
|
103.7±8.2
|
0.024
|
WC (cm)
|
97.4±10.5
|
90.3±12.8
|
93.6±12.3
|
<0.001
|
BMI (kg/m2)
|
27.5±3.8
|
26.9±4.9
|
27.2±4.4
|
<0.001
|
<18.49, n (%)
|
74 (0.3)
|
238 (0.9)
|
312 (0.6)
|
|
18.5-24.99, n (%)
|
5,678 (25)
|
10,414 (45.9)
|
16,092 (32)
|
|
25-29.99, n (%)
|
11,955 (52.4)
|
10,346 (37.9)
|
22,301 (44.4)
|
|
>30, n (%)
|
5,123 (22.4)
|
6,398 (22.3)
|
11,521 (22.9)
|
|
WHR
|
0.94±0.07
|
0.87±0.07
|
0.90±0.08
|
<0.001
|
WHtR
|
0.55±0.06
|
0.55±0.08
|
0.55±0.07
|
0.04
|
Results are shown as means±SD unless stated otherwise.
|
The comparison of characteristics between men and women (unpaired Student’s t-test for numerical data, χ2 test for categorical data).
|
BMI, body mass index; DMT2, diabetes mellitus type 2; HC, hip circumference; WC, waist circumference; WHR, waist-to-hip ratio; WHtR, waist-to-height ratio.
|
Table 3 shows descriptive statistics of the diabetic and non-diabetic group (diabetic referring to DMT2). Average age, weight, HC, WC, BMI, WHR, and WHtR were all significantly higher in the diabetic group. Of the total sample, 67.3% were either overweight or obese according to universal cut-offs for BMI and 87.3% were either obese or overweight in the diabetic group. The group without diabetes had a higher prevalence of smoking (16.6%) compared to the diabetic group (13.8%).
Table 3 - Descriptive statistics according to diabetic. and non-diabetic groups
Characteristics
|
Diabetic (n=1,876)
|
Non-diabetic (n=48,166)
|
Total
(n=50,042)
|
P-value
|
Smoking, n (%)
|
260 (13.8)
|
8,007 (16.6)
|
8,267 (16.5)
|
<0.001
|
Age (years)
|
65.4±11.8
|
52.7±15
|
53.1±16.1
|
<0.001
|
Weight (kg)
|
86.5±16.6
|
78±15.1
|
79.2±15.2
|
<0.001
|
Height (cm)
|
169±9.4
|
170.6±9.2
|
170.6±9.3
|
<0.001
|
HC (cm)
|
106.7±9.3
|
103.6±8.1
|
103.7±8.2
|
<0.001
|
WC (cm)
|
103.4±12.6
|
93.2±12.1
|
93.6±12.3
|
<0.001
|
BMI (kg/m2)
|
30.2±5.0
|
27.1±4.3
|
27.2±4.4
|
<0.001
|
<18.49, n (%)
|
4 (0.2)
|
308 (0.6)
|
312 (0.6)
|
|
18.5-24.99, n (%)
|
235 (12.5)
|
15,857 (32.8)
|
16,092 (32)
|
|
25-29.99, n (%)
|
750 (40)
|
21,551 (44.6)
|
22,301 (44.4)
|
|
>30, n (%)
|
887 (47.3)
|
10,634 (22)
|
11,521 (22.9)
|
|
WHR
|
0.97±0.07
|
0.9±0.08
|
0.90±0.08
|
<0.001
|
WHtR
|
0.61±0.08
|
0.55±0.07
|
0.55±0.07
|
<0.001
|
Results are shown as means±SD unless stated otherwise.
|
The comparison of characteristics between diabetic and non-diabetic (unpaired Student’s t-test for numerical data, χ2 test for categorical data).
|
BMI, body mass index; HC, hip circumference; WC, waist circumference; WHR, waist-to-hip ratio; WHtR, waist-to-height ratio.
|
3.2 Associations between DMT2 and anthropometric measures
In the regression analysis, the lowest quartile was taken as the reference in the analysis. In addition to unadjusted analysis, two models where applied. The first one adjusting only for sex and the second one for sex, age and smoking. Each of the four anthropometric indices showed a statistically significant increase in OR when comparing the quartiles to reference values, with the largest difference found in the highest quartiles in all models. The increase in OR across the quartiles indicates that the likelihood of having diabetes type 2 increases for higher values of all the four anthropometric indices. In the unadjusted model, OR was largest for WHtR: 12.34 (9.98 – 15.24) followed by the highest quartile in WHR: 11.16 (9.14-13.63), when comparing confidents intervals. The adjusted models showed a statistically significant difference in OR compared to the unadjusted model only in three cases: The OR for the highest quartile for WHR in model 1, 14.64 (11.89-18.02) was larger than in Model 2, 8.92 (7.22-11.02). For WHtR, the OR for the second to highest quartile was larger in model 2, 3.22 (2.56-4.03) compared to the unadjusted OR, 5.22 (4.18-6.51) in the same quartile. Lastly, the highest quartile for WHtR had a lower OR in model 2 compared to model 1 and the unadjusted OR, in the same quartile (Table 4).
Table 4 - OR and 95% CI of the presence of DMT2 according to quartiles of anthropometric index
Variable
|
N
|
%
|
Unadjusted
|
|
Model 1*
|
|
Model 2**
|
|
BMI
|
|
|
Odds Ratio
|
95% CI
|
Odds Ratio
|
95% CI
|
Odds Ratio
|
95% CI
|
< 24.1
|
12,235
|
24.5%
|
reference
|
|
reference
|
|
reference
|
|
24.1 - 26.6
|
12,597
|
25.2%
|
1.83
|
1.51 - 2.23
|
1.76
|
1.45 - 2.15
|
1.53
|
1.25 - 1.86
|
26.7 - 29.5
|
12,455
|
24.9%
|
3.10
|
2.58 - 3.72
|
2.95
|
2.46 - 3.55
|
2.45
|
2.03 - 2.94
|
29.6+
|
12,755
|
25.5%
|
6.23
|
5.25 - 7.39
|
6.06
|
5.10 - 7.19
|
5.18
|
4.35 - 6.16
|
WC
|
|
|
|
|
|
|
|
|
< 85
|
11,752
|
23.5%
|
reference
|
|
reference
|
|
reference
|
|
85 - 93
|
11,866
|
23.8%
|
2.01
|
1.61 - 2.51
|
2.06
|
1.65 - 2.57
|
1.67
|
1.33 - 2.08
|
93 - 101
|
12,918
|
25.8%
|
3.38
|
2.75 - 4.14
|
3.5
|
2.84 - 4.30
|
2.53
|
2.08 - 3.13
|
101.1+
|
13,476
|
26.9%
|
8.57
|
7.08 - 10.37
|
8.91
|
7.33 - 10.82
|
6.19
|
5.08 - 7.54
|
WHR
|
|
|
|
|
|
|
|
|
< .85
|
12,533
|
25.0%
|
reference
|
|
reference
|
|
reference
|
|
0.85 - 0.89
|
12,630
|
25.2%
|
2.15
|
1.71 - 2.71
|
2.38
|
1.89 - 2.99
|
1.94
|
1.53 - 2.44
|
0.90 -0 .94
|
12,316
|
24.6%
|
4.27
|
3.45 - 5.29
|
5.16
|
4.15 - 6.40
|
3.66
|
2.94 - 4.55
|
0.95+
|
12,563
|
25.1%
|
11.16
|
9.14 - 13.63
|
14.64
|
11.89 - 18.02
|
8.92
|
7.22 - 11.02
|
WHtR
|
|
|
|
|
|
|
|
|
< 0.50
|
12,603
|
25.2%
|
reference
|
|
reference
|
|
reference
|
|
0.50 - 0.53
|
12,520
|
25.0%
|
2.57
|
2.02 - 3.27
|
2.45
|
1.93 - 3.11
|
1.85
|
1.46 - 2.36
|
0.54 - 0.58
|
12,510
|
25.0%
|
5.22
|
4.18 - 6.51
|
4.96
|
3.97 - 6.19
|
3.22
|
2.56 - 4.03
|
0.59+
|
12,429
|
24.8%
|
12.34
|
9.98 - 15.24
|
12.30
|
9.96 - 15.2
|
7.19
|
5.80 - 8.92
|
Dependent variable: DMT2
*adjusted for sex
**adjusted for sex, age and smoking
BMI, body mass index (kg/m2); CI, confidence interval; DMT2, diabetes mellitus type 2; OR, odds ratio; WC, waist circumference (centimeters); WHR, waist-to-hip ratio; WHtR, waist-to-height ratio.
|
3.2 Predictable values
WHR was the strongest predictor for DMT2 followed by WHtR, WC, and BMI. The predictable value for BMI was significantly weaker than for all the other indices. The predictable value for WC was significantly weaker compared only to WHtR. The order of indices did not differ between men and women but the predictable potential was statistically higher for women than men according to the AUC. Table 5 shows the area under the ROC-curve, stratified for sex.
Table 5 – Area under the curve with 95% CI and Optimal cut-off points
Area Under the Curve (men)
|
|
Test Result Variables
|
AUC
|
95% CI
|
|
|
BMI
|
0.666
|
0.648 -
|
0.684
|
|
|
WHtR
|
0.723
|
0.707 -
|
0.739
|
|
|
WHR
|
0.734
|
0.718 -
|
0.750
|
|
|
WC
|
0.698
|
0.681 -
|
0.715
|
|
|
|
|
|
|
|
|
Area Under the Curve (women)
|
|
Test Result Variables
|
AUC
|
95% CI
|
|
|
BMI
|
0.713
|
0.696 -
|
0.729
|
|
|
WHtR
|
0.764
|
0.749 -
|
0.779
|
|
|
WHR
|
0.770
|
0.755 -
|
0.786
|
|
|
WC
|
0.742
|
0.726 -
|
0.757
|
|
|
|
|
|
|
|
|
|
|
BMI, body mass index (kg/m2); WC, waist circumference (centimeters); WHR, waist-to-hip ratio; WHtR, waist-to-height ratio; AUC, area under the curve; CI, confidence interval
|
Cut-offs according to Yoden’s index indicate a trend toward lower cut-offs for women than men, for WHR and WC. Other indices show small differences between groups (Table 6).
Table 6 - Optimal cut-off points of BMI, WHtR, WHR and WC according to Youden´s index
Optimal cut-offs*
|
Total
|
Men
|
Women
|
|
BMI
|
28.35
|
28.25
|
28.35
|
|
WHtR
|
0.57
|
0.58
|
0.57
|
|
WHR
|
0.93
|
0.96
|
0.90
|
|
WC
|
97.25
|
98.70
|
91.00
|
|
*According to Youden´s index
|
BMI, body mass index (kg/m2); WC, waist circumference (centimeters); WHR, waist-to-hip ratio; WHtR, waist-to-height ratio.