DOI: https://doi.org/10.21203/rs.3.rs-582331/v1
Purpose We aimed to determine the incidence and prediction nomogram for new-onset metabolic syndrome (MetS) in a middle-aged Vietnamese population.
Methods A population-based prospective study was conducted in 1150 participants aged 40-64 years without MetS at baseline and followed-up for 5 years. Data on lifestyle factors, socioeconomic status, family diabetes history, and anthropometric measures were collected. MetS incidence was estimated in general population and subgroup of age, gender, and MetS components. A Cox proportional hazards regression was used to estimate hazard ratios (HRs) with 95% confidence intervals (CI) for MetS. A prediction nomogram was developed and checked for discrimination and calibration.
Results During median follow-up of 5.14 years, the accumulate MetS incidence rate was 23.4% (95% CI: 22.2-24.7). The annual incidence rate (95%CI) was 52.9 (46.7-60.1) per 1000 person-years in general population and higher in women [56.6 (48.7-65.9)] than men [46.5 (36.9-59.3)]. The HRs (95% CI) for developing MetS were gender [females vs males: 2.04 (1.26-3.29)], advanced age [1.02 (1.01-1.04) per one year], waist circumference [1.08 (1.06-1.10) per one cm] and other obesity-related traits, and systolic blood pressure [1.02 (1.01-1.03) per one mmHg]. The prediction nomogram for MetS had a good discrimination (C-statistics = 0.742) and fit calibration (mean absolute error = 0.009) with a positive net benefit in the predicted probability thresholds between 0.13 and 0.70.
Conclusions The study is the first to indicate an alarmingly high incidence of MetS in a middle-aged population in Vietnam. The nomogram with simply applicable variables would be useful to qualify individual risk of developing MetS.
Metabolic syndrome (MetS), a cluster composed of obesity, hypertension, insulin resistance, disturbed glucose and dyslipidemia [1], is an important risk factor for diabetes [2], cardiovascular diseases [3] and cancers [4] which are leading causes of deaths [5]. MetS is estimated to affect about 20−25% global [6] and 34.7% US adults [7]. In the Asia-Pacific region, the prevalence of MetS is about 20%, and continues to increase [8]. Fortunately, MetS can be prevented or delayed by lifestyle modifications such as healthy eating [9] and intensity of leisure time physical activity [10]. Thus, early identification of those who are at high risk of MetS may help to establish effective strategies for MetS prevention.
Developing the prediction model for incident MetS with appropriate factors are important and useful clinical practice for individual and community. In recent years, some models have been constructed using both non-invasive and invasive factors in China [11] and France [12]. However, these models are not usually suitable for low-resource settings. Furthermore, prediction models developed in one population may not be applied in other populations. Therefore, it is necessary to develop population-specific models for MetS prediction, especially non-invasive models which are more suitable for low-resource settings.
In Vietnam, there was a significant upward trend in the prevalence of MetS, from 10% in 2003 to 18.1% in 2011 [13–15]. Although the MetS prevalence has been investigated in several regions including both rural and urban settings, there has been no report on the MetS incidence. Moreover, Vietnam is experiencing a rapid change in both living environment and lifestyle which may lead to high risk of non-communicable diseases [16]. Precise estimation of MetS incidence can help high-risk individuals take interventions timely to reduce MetS-related morbidity and mortality. In this context, it is crutial to have a tool for predicting new-onset MetS that also promote an effective planning prevention and early treatment of MetS. Therefore, this five-year cohort study was conducted to estimate the incidence of MetS, and to develop a nomogram for MetS prediction among middle-aged Vietnamese population.
This study was an important part of the DiaMetS-VN population-based prospective study designed to identify the epidemiological patterns and genetics of MetS and type 2 diabetes in the middle-aged population in Vietnam. The participants of this study were Kinh Vietnamese aged 40–64 years, lived in Ha Nam, a typical rural province in the Red River Delta region. Further details of this population were previously reported [15]. Of 2042 participants without MetS at the baseline, 871 did not take part in the follow-up and 21 were missed glycemic and lipid profile. As a result, 1150 subjects were entered this analysis.
The trained surveyors collected data through face-to-face interview with questionnaires. The participants in the follow-up 2016 survey were interviewed and measured with the same protocol in the baseline 2011 survey as described previously[17]. Briefly, all participants were asked fasting overnight for venous blood collection next morning. Blood sample was centrifuged immediately, then aliquots of plasma were stored at 2-8oC and transported to laboratory for analyzing biochemical index including glucose, total cholesterol, high-density lipoprotein (HDL-C), and triglycerides (TG). Demographic data included current age, household income, education level, occupation, marital status, family history of diabetes, and history of using medicine for hypertension or dyslipidemia. A family history of diabetes was defined as having at least one parent or sibling with diabetes. In addition, occupation was categorized as heavy occupation (farmer and manual worker) and non-heavy occupation (office clerks, teacher, retired, and housework).
Anthropometric measurements including height, weight, waist circumference, hip circumference were done twice for each individual and the average of two values was used for data analysis. Waist circumference (WC) was measured at the midway between the lower rib margin and the iliac crest, whereas hip circumference was measured at the level of the largest circumference around the buttocks. From, body mass index (BMI), Waist-height ratio (WHtR) and waist–hip ratio (WHR) were calculated. In addition, blood pressure was manually also taken twice in a sitting position after at least 5 minutes of rest using a mercury sphygmomanometer and the mean of value was considered as participant’s blood pressure.
To assess exposure, data on smoking habits, night sleeping duration, siesta, leisure time, alcohol consumption were obtained by interview. Smoking was defined in 3 groups (none smoker, ex-smoker and current smoker). Sleep duration was total of hours spend for sleeping at night daily. Participants were asked about the daily total of leisure-time including time for watching TV and sitting without physical activity. Alcohol intake was categorized in 5 groups: (none, < 1 drink/mo, ≥ 1 drink/mo to < 1 drink/wk, 1 drink/wk to ≤ 1 drink/d, and ≥ 2 drink/d, one drink was defined as a 50–ml cup of rice wine at about 30%).
MetS was defined according to the criteria of the US National Cholesterol Education Adult Treatment Panel III with adjusting waist circumference cutoff in Asian population [1, 18]. An individual was diagnosed MetS as the presence of three or more of the following: (1) waist circumference ≥ 90 cm for men and ≥ 80 cm for women; 2) fasting plasma glucose ≥ 100 mg/dL (5.6 mmol/L) or used of drug treatment of elevated glucose; 3) systolic blood pressure ≥ 130 mmHg or diastolic blood pressure ≥ 85 mmHg or history of hypertension; 4) HDL-C < 40 mg/dL (1.04 mmol/L) for men and < 50 mg/dL (1.29 mmol/L) for women; 5) Triglycerides ≥ 150 mg/ dL (1.7 mmol/L) or taking a lipid-lowering medication.
The MetS incidence was estimated using the weight based on the study design, the probability of sampling, finite population correction, and none–response rate. Incidence rates with 95% confidence interval (CI) were calculated by dividing the number of events by person-year at risk. Time at risk was estimated as the mid-time between of the two surveys in individuals who developed MetS and as the interval between the first and the last observation dates in participants without MetS. The age and sex- adjusted incidences were estimated using direct standardization method based on the 2019 Vietnam Population and Housing Census [19]. Baseline characteristics between two groups were compared using T-test or Mann-Whitney-U test or Chi-square test or Fisher’s exact test when appropriate. The Cox proportional hazard regression model was used to estimate the hazard ratio (HR) for MetS.
The nomogram for estimating new-onset MetS was constructed according to the variables in prediction models resulting from multivariable logistic regression and Bayesian Model Average methods. The discrimination of the nomogram was evaluated by using the area under a receiver operating characteristic curve (AUC). Moreover, C-statistic was applied to quantify the discrimination ability of the nomogram and implementing internal validation of this model. In general, C-index > 0.7 is considered to present a good discrimination. Calibration of the nomogram was assessed by plotting the actual probability of MetS compared to the probabilities predicted by prediction model. The decision curve analysis of the model was built for using in the clinical utility [20]. In this study, we used bootstrap analysis with 500 subsamples for internal validation and with 1000 resamples for calibration. The statistical analyses were performed using SPSS version 16 (SPSS, Chicago, USA), Stata version 14 (Stata Corporation, College Station, TX, USA) and and R version 3.5.3 for Windows. All of statistical tests were two-tailed and P < 0.05 was considered as significant.
Table 1 shows the baseline characteristics of participants according to new-onset MetS at follow up. The subjects with MetS had higher baseline values of age, blood pressures, weight, BMI, WC, WHR, WHtR, and body fat percentage than those without MetS. No significant differences between the two groups were found in socioeconomic status (residence, marital status, education, occupation, and income level), sedentary time (watching TV and sitting), sleeping time, fasting plasma glucose, and lipid profile.
|
Variable |
MetS (n = 280) |
Non-MetS (n = 870) |
P-value |
|
|---|---|---|---|---|
|
Age (median, quartile) years |
51.1 (46.0−56.0) |
50. (45.6−55.3) |
< 0.0001a |
|
|
Gender (n, %) |
0.015c |
|||
|
Male |
84 (30.0) |
331 (38.0) |
||
|
Female |
196 (70.0) |
539 (62.0) |
||
|
Body mass index (mean ± SD) kg/m2 |
22.1 ± 2.6 |
20.8 ± 2.2 |
< 0.0001b |
|
|
Body fat (mean ± SD) % |
29.2 ± 5.6 |
21.8 (22.5−30.2) |
< 0.0001a |
|
|
Wait circumference (mean ± SD) cm |
76.9 ± 7.3 |
72.0 (67.5−77.0) |
< 0.0001a |
|
|
Systolic blood pressure (median, quartile) mmHg |
110.0 (100.0−130.0) |
110.0 (100.0−120.0) |
< 0.0001a |
|
|
Diastolic blood pressure (median, quartile) mmHg) |
70.0 (70.0−80.0) |
70.0 (60.0−80.0) |
< 0.0001a |
|
|
Triglycerides (median, quartile) mmol/L |
1.29 (1.00−2.00) |
1.29 (0.97−1.91) |
0.555a |
|
|
HDL-cholesterol (median, quartile) mmol/L |
1.30 (1.05−1.66) |
1.27 (1.00−1.66) |
0.356a |
|
|
Fasting glucose (median, quartile) mmol/L |
4.60 (4.00−5.09) |
4.50 (4.00−5.00) |
0.185a |
|
|
Region (n, %) |
||||
|
Rural |
265 (94.6) |
844 (97.0) |
0.063c |
|
|
Urban |
15 (5.4) |
26 (3.0) |
||
|
Education level (n, %) |
0.828c |
|||
|
Elementary |
24 (8.6) |
72 (8.3) |
||
|
Intermediate |
178 (63.6) |
542 (62.3) |
||
|
Secondary |
31 (11.1) |
115 (13.2) |
||
|
Post–secondary |
47 (16.8) |
141 (16.2) |
||
|
Income level (n, %) |
0.506c |
|||
|
< 25 percentiles |
68 (24.3) |
227 (26.1) |
||
|
< 25 percentiles |
66 (23.6) |
228 (26.2) |
||
|
50– < 75 percentiles |
67 (23.9) |
206 (23.7) |
||
|
≥ 75 percentiles |
79 (28.2) |
209 (24.0) |
||
|
Smoking (n, %) |
< 0.0001c |
|||
|
None |
220 (78.6) |
609 (70.0) |
||
|
Ex-smokers |
34 (12.1) |
92 (10.6) |
||
|
Current smokers |
26 (9.3) |
169 (19.4) |
||
|
Leisure time (median, quartile) hour/day |
5.0 (3.3−7.0) |
4.8 (3.0−7.0) |
0.489a |
|
|
Sleeping time (median, quartile) hour/day |
7.0 (6.0−7.0) |
7.0 (6.0−7.0) |
0.260a |
|
|
Family history of diabetes (n, %) |
||||
|
Yes |
17 (6.1) |
51 (5.9) |
0.897c |
|
|
No |
263 (93.9) |
819 (94.1) |
||
| MetS metabolic syndrome | ||||
| a Mann-Whitney U test | ||||
| b T-test | ||||
| c Chi-Square test | ||||
The comparison of baseline characteristics between participants and non-participants in the follow-up survey is shown in Additional Table. There were no significant differences between two groups in terms of anthropometrics, socioeconomic status, and lifestyles.
During a median follow-up of 5.14 years (quartile: 5.05−5.19 years), 280 (23.4%) subjects developed MetS. As shown in Table 2, the estimated incidences of MetS increased with age in general population. After 55 years of age, about a quarter of men and one third of women suffered from MetS during 5-year period. The MetS incidence was higher in groups: urban, non-heavy occupation, and overweight.
|
Baseline characteristics |
No at risk |
No new-onset MetS |
Person- year |
5 years cumulative incidence rate |
Incidence density/1000 person-years |
|||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
|
Total |
Men |
Women |
Total |
Men |
Women |
|||||||
|
Total |
1150 |
280 |
5224.2 |
23.4 (22.2−24.7) |
20.1 (18.0−22.3) |
25.3 (23.7−27.0) |
52.9 (46.7−60.1) |
46.5 (36.9−59.3) |
56.6 (48.7−65.9) |
|||
|
Age group (years) |
||||||||||||
|
40 − 44 |
220 |
42 |
1030.4 |
17.7 (15.3−20.4) |
18.8 (13.9−24.9) |
17.4 (14.7−20.4) |
38.2 (28.0−53.1) |
41.4 (21.9−85.6) |
37.2 (26.1−54.6) |
|||
|
45 − 49 |
294 |
59 |
1368.1 |
18.7 (16.5−21.2) |
15.4 (12.0−19.5) |
20.5 (17.6−23.6) |
42.7 (32.7 −56.5) |
36.0 (21.8−63.2) |
46.0 (33.6−64.0) |
|||
|
50 − 54 |
282 |
63 |
1295.6 |
21.5 (19.1−24.2) |
16.9 (13.8−20.6) |
24.6 (21.2−28.3) |
48.4 (37.2−64.0) |
37.8 (24.5−60.7) |
55.5 (39.9−79.0) |
|||
|
55 − 59 |
214 |
71 |
922.0 |
33.3 (29.8−36.9) |
26.5 (21.3−32.4) |
38.6 (34.0−43.4) |
74.1 (57.5−96.4) |
56.5 (35.4−95.1) |
89.5 (66.3−122.0) |
|||
|
60 − 64 |
140 |
45 |
608.1 |
30.1 (26.3−34.3) |
24.5 (19.1−30.9) |
34.2 (29.1−39.7) |
75.2 (55.1−104.5) |
68.8 (39.4−128.8) |
80.0 (55.4−118.2) |
|||
|
Region |
||||||||||||
|
Rural |
1109 |
265 |
5051.4 |
22.6 (21.4−23.9) |
19.1 (17.1−21.3) |
24.6 (23.0−26.2) |
51.8 (45.6−58.9) |
43.8 (34.9−55.8) |
56.3 (48.5−65.7) |
|||
|
Urban |
41 |
15 |
172.8 |
37.2 (29.9−45.0) |
38.7 (25.9−53.3) |
36.4 (27.9−46.0) |
68.8 (37.6−133.7) |
84.9 (30.2−292.0) |
60.1 (28.2−139.5) |
|||
|
Occupation |
||||||||||||
|
Heavy |
915 |
214 |
4179.5 |
22.7 (21.3−24.2) |
17.5 (15.1−20.2) |
25.3 (23.5−27.1) |
50.3 (43.6−58.2) |
38.6 (28.6−53.2) |
56.2 (47.9−66.2) |
|||
|
None heavy |
235 |
66 |
1044.7 |
26.2 (23.5−29.2) |
26.8 (23.0−31.1) |
25.6 (21.8−29.9) |
63.0 (48.3−83.1) |
68.0 (47.5−99.9) |
58.5 (39.7−88.5) |
|||
|
Nutrition status |
||||||||||||
|
Underweight |
160 |
23 |
768.1 |
12.9 (10.5−15.8) |
9.6 (6.56−13.72) |
15.3 (11.9−19.5) |
28.9 (18.9−46.1) |
22.6 (10.4 −57.8) |
33.3 (19.9−59.0) |
|||
|
Normal |
749 |
154 |
3475.7 |
20.1 (18.7−21.7) |
14.9 (12.7−17.4) |
22.7 (20.8−24.7) |
43.2 (36.6−51.4) |
32.2 (22.8−47.0) |
48.9 (40.4−59.5) |
|||
|
Overweight |
162 |
64 |
672.7 |
38.7 (34.8−42.6) |
37.1 (31.0−43.6) |
39.6 (34.7−44.7) |
97.3 (75.0−128) |
107 (70.2−169) |
92.4 (66.5−131) |
|||
|
Obesity |
77 |
37 |
302.6 |
47.6 (41.3−53.8) |
45.0 (36.1−54.3) |
50.8 (42.4−59.2) |
123 (88.9−172) |
109 (67.6−183) |
140 (90.7−222) |
|||
|
Severe obesity |
2 |
2 |
5.2 |
N/A |
N/A |
N/A |
N/A |
N/A |
N/A |
|||
| MetS metabolic syndrome | ||||||||||||
| Data are weighted by the study design, the probability of sampling, finite population correction, and none-response rate. Data are shown as % (95% CI) for cumulative incidence rate and cases (95% CI) for incident density/1000 person-years. N/A, not applicable due to 2 participats with BMI ≥ 30 kg/m2. Nutrition status was classified as underweight (BMI < 18.5 kg/m2), normal (18.5 ≤ BMI < 23 kg/m2), overweight (23 ≤ BMI < 25 kg/m2), obesity (25 ≤ BMI < 30 kg/m2), and severe obesity (BMI ≥ 30 kg/m2). Occupation was categorized as heavy occupation (farmer and manual worker) or none heavy occupation (office clerks, teacher, retired worker, and houseworker). | ||||||||||||
The sex- and age- standardized MetS incidences were 24.5% (95% CI: 24.3−24.7) in general population and significantly higher in women [27.3 (95% CI: 27.1−27.5)] than in men [20.4 (95% CI: 20.1−20.6)]. The corresponding new-onset MetS cases per 1000 person-years were 55.6 (95% CI: 54.7−56.5) in general population and much higher in women [62.2 (95% CI: 60.9−63.5)] compared to men [47.5 (95% CI: 46.4−48.7)].
Table 3 presents the development of MetS in participants with single component and pairwise combination of MetS components at the baseline. The more number of MetS component they had at baseline, the more MetS incidence rate they suffered at follow-up. In addition, among participants with one component at baseline, subjects with central obesity had the highest MetS incidence, while people with elevated blood glucose had the lowest incidence. Moreover, among 349 people with 2 MetS components, the highest incidence was seen in those with central obesity and raised blood pressure combination, while the lowest incidence was found in those with low HDL-C and elevated blood glucose combination.
|
At risk (n) |
Person- years |
Case (n) |
5 years cumulative incidence rate (95% CI) (%) |
Incidence/ 1000 person-years (95% CI) |
|
|---|---|---|---|---|---|
|
The number of components |
|||||
|
0 |
243 |
1160.3 |
37 |
15.23 (11.25−20.28) |
31.89 (23.22−43.64) |
|
1 |
558 |
2605.1 |
109 |
19.53 (16.46−23.03) |
41.84 (34.80−50.23) |
|
2 |
349 |
1458.8 |
134 |
38.4 (33.45−43.60) |
91.86 (78.09−107.8) |
|
One of component |
|||||
|
CO |
16 |
61.9 |
8 |
50.00 (28.00−72.00) |
129.24 (66.79−234.8) |
|
RBP |
74 |
311.8 |
27 |
36.49 (26.44−47.87) |
86.59 (60.19−123.1) |
|
HTG |
183 |
860.7 |
34 |
18.58 (13.61−24.84) |
39.50 (28.40−54.69) |
|
LHC |
258 |
1237.9 |
37 |
14.34 (10.59−19.14) |
29.89 (21.76−40.93) |
|
EBG |
27 |
132.7 |
3 |
11.11 (3.58−28.06) |
22.61 (7.72−64.63) |
|
Pairwise combination of components |
|||||
|
CO-RBP |
18 |
59.3 |
13 |
72.22 (49.1−87.5) |
219.2 (132.8−339.8) |
|
LHC-CO |
24 |
85.1 |
15 |
62.50 (42.71−78.84) |
176.3 (109.8−270.7) |
|
LHC-RBP |
76 |
289.3 |
40 |
52.63 (41.55−63.46) |
138.3 (103.2−182.8) |
|
HTG-RBP |
60 |
242.7 |
26 |
43.33 (31.57−55.90) |
107.1 (74.16−152.3) |
|
HTG-CO |
35 |
142.7 |
15 |
42.86 (27.98−59.14) |
105.1 (64.74−166.2) |
|
RBP-EBG |
11 |
49.3 |
3 |
27.27 (9.75−56.56) |
60.85 (20.91−164.3) |
|
HTG-EBG |
19 |
85.8 |
5 |
26.32 (11.81−48.79) |
58.28 (25.15−129.3) |
|
LHC-HTG |
91 |
434.4 |
14 |
15.38 (9.39−24.18) |
32.23 (19.29−53.63) |
|
LHC-EBG |
14 |
67.6 |
2 |
14.29 (4.01−39.94) |
29.59 (8.15−101.6) |
|
CO-EBG |
1 |
2.5 |
1 |
100.0 (20.65−100.0) |
400.0 (74.5−846.6) |
| HR hazard ratio, CI confidence interval, CO central obesity, RBP raised blood pressure, | |||||
| HTG high triglycerides, LHC low HDL-Cholesterol, EBG elevated blood glucose | |||||
Table 4 shows HR values of MetS according to candidate risk factors in multivariable analysis. Gender, age, blood pressures, fasting plasma glucose, and obesity-related measurements (WC, HC, WHtR, WHR, BMI, and body fat percentage) were significant risk factors for MetS. The HRs (95% CI) of MetS for gender (females vs males), advanced age, systolic blood pressure, fasting plasma glucose, and WC were 2.04 (1.26−3.29), 1.02 (1.01−1.04), 1.02 (1.01–1.03), 1.12 (1.02–1.40), and 1.08 (1.06−1.10), respectively. There was no significant association between incident MetS with levels of some lifestyle factors, socioeconomic conditions and family history of diabetes.
|
Risk factors |
Hazard ratio (95% CI) |
P-value |
|---|---|---|
|
Gender (female vs male) |
2.04 (1.26−3.29) |
0.004 |
|
Age (years) |
1.02 (1.01−1.04) |
0.021 |
|
Fasting plasma glucose (mmol/L) |
1.12 (1.02−1.40) |
0.025 |
|
Systolic blood pressure (mmol/L)a |
1.02 (1.01−1.03) |
< 0.0001 |
|
Diastolic blood pressure (mmol/L)b |
1.03 (1.02−1.04) |
< 0.0001 |
|
Triglycerides (mmol/L) |
0.96 (0.85−1.07) |
0.429 |
|
High density lipoprotein-cholesterol (mmol/L) |
0.92 (0.73−1.16) |
0.481 |
|
Each of the following obesity-related traits: |
||
|
Waist circumference (cm)c |
1.08 (1.06−1.10) |
< 0.0001 |
|
Waist height ratio (per 0.03)d |
1.46 (1.31−1.64) |
< 0.0001 |
|
Waist hip ratio (per 0.07)e |
1.32 (1.19−1.46) |
< 0.0001 |
|
Body mass index (kg/m2)f |
1.22 (1.15−1.27) |
< 0.0001 |
|
Body fat (%)g |
1.11 (1.08−1.15) |
< 0.0001 |
|
Hip circumference (cm)h |
1.10 (1.07−1.13) |
< 0.0001 |
|
Family history of diabetes |
||
|
No |
1.0 |
|
|
Yes |
0.88 (0.53−1.44) |
0.605 |
|
Residence |
||
|
Rural |
1.0 |
|
|
Urban |
1.26 (0.73−2.18) |
0.413 |
|
Education level |
||
|
Elementary |
1.0 |
|
|
Intermediate |
1.10 (0.70−1.71) |
0.687 |
|
Secondary |
0.99 (0.55−1.76) |
0.964 |
|
Post–secondary |
0.77 (0.43−1.39) |
0.385 |
|
Marital status |
||
|
Married |
1.0 |
|
|
Never |
0.84 (0.30−2.36) |
0.739 |
|
Widowed |
1.15 (0.72−1.85) |
0.553 |
|
Others |
1.16 (0.47−2.84) |
0.746 |
|
Occupation |
||
|
Heavy |
1.0 |
|
|
None heavy |
1.25 (0.88−1.78) |
0.210 |
|
Income level |
||
|
< 25 percentiles |
1.0 |
|
|
25– < 50 percentiles |
1.06 (0.75−1.51) |
0.727 |
|
50– < 75 percentiles |
1.17 (0.82−1.67) |
0.374 |
|
≥ 75 percentiles |
1.09 (0.76−1.56) |
0.644 |
|
Alcohol consumption |
||
|
None |
1.0 |
|
|
< 1 drink/mo |
0.76 (0.44−1.29) |
0.308 |
|
≥ 1 drink/mo to < 1 drink/wk |
0.94 (0.55−1.60) |
0.819 |
|
1 drink/wk to ≤ 1 drink/d |
0.75 (0.43−1.30) |
0.303 |
|
≥ 2 drink/d |
1.01 (0.57−1.77) |
0.987 |
|
Smoking status |
||
|
None |
1.0 |
|
|
Ex-smoker |
0.72 (0.41−1.27) |
0.259 |
|
Current smoker |
1.12(0.66−1.93) |
0.672 |
|
Leisure time/day (hour) |
0.99 (0.94−1.03) |
0.549 |
|
Sleeping time/day (hour) |
0.96 (0.88−1.05) |
0.419 |
| CI confidence interval | ||
| b replace a in multivariable analysis | ||
| d, e, f, g, h step by step replace c in multivariable analysis | ||
A multivariable logistic regression analysis with backward stepwise and Bayesian Model Average approach were used to search for the most predictive MetS model with the highest AUC value and parsimonious variables. As a result, gender, age, WC, and SBP were considered in the final model (Table 5). The prediction nomogram for estimating the individual risk of MetS was constructed with above factors in the final model.
|
Predictor |
Unit |
β-coefficient (se) |
P-value a |
|---|---|---|---|
|
Gender |
(female = 1, male = 0) |
1.132 (0.178) |
< 0.0001 |
|
Age |
year |
0.060 (0.019) |
0.002 |
|
Waist circumference |
cm |
0.022 (0.005) |
< 0.001 |
|
Systolic blood pressure |
mmHg |
||
|
Intercept |
-14.099 (1.191) |
< 0.0001 |
|
| a Using multivariable logistic regression analysis | |||
| CI Confidence Interval. | |||
Figure 1 shows the nomogram for predicting new-onset MetS. The C-index before and after bootstrap was 0.742 and 0.7367 respectively, indicating a good discrimination of the nomogram. The calibration curve presented possibility of the nomogram-predicted probabilities versus the actual observation. As shown in Fig. 2a, both these lines were very close with the ideal line and the mean absolute error was 0.009. Thus, the prediction nomogram performed a good calibration. Figure 2b demonstrated that for the predicted probability thresholds between 0.13 and 0.70, the prediction nomogram showed a positive net benefit than strategies for all and none participant to treat.
To the best of our knowledge, the study is the first to report the high MetS incidence in a middle-aged Vietnamese population with the annual incidence rate of 52.9 (95% CI: 46.7−60.1) per 1000 person-years. With almost similar range of age and equal criteria for diagnosed MetS, our finding was higher than that in Taiwan [21], Korea [22], and Thailand [23]. The current MetS incidence was relatively higher compared to a study in Japan with higher cut-off of WC criteria in both men and women [24]. Conversely, the incidence of MetS in current study was lower in Iran [25].
Regarding to sex difference in the MetS incidence, our study showed that the MetS incidence in women was higher than that in men, in line with previous studies [21, 25]. One study in Iran reported the reverse finding that the MetS incidence in men was higher than in women [27]. Another study in Portugal showed the similar MetS incidence in both gender [28]. A possible explanation for this discrepancy might result from the difference in socioeconomic status, lifestyle patterns, applied definition of MetS, and genetic background.
After adjustment for lifestyle factors, socioeconomic status, family history of diabetes and individual MetS components, the present study found that aging was an independent predictor of the development of MetS. The highest MetS incidence was found in 60−64 years men and 55−59 years women. This sex-difference should be interpreted with caution because 55−59 years is the postmenopausal period in women. A study in India showed that menopause was an independent risk factor of MetS and MetS incidence in postmenopausal women was higher about 2.5 times than in premenopausal women [29].
MetS might not often cause death directly, but MetS has been an important predictor to developing cardiovascular diseases and type 2 diabetes [30], which are the top four of all global deaths [5]. Recently, several prognostic models have been developed for predicting MetS in Asian populations [11, 32, 33] and Whites [34, 35]. Most of these models include both non-invasive and invasive factors such as age, measures of obesity (e.g. BMI, WC), blood pressures (SBP, DBP), lifestyle (e.g. smoking, dietary), biochemical indies (e.g. HDL-C, TG, FPG) and genetic background. The AUC values of these models vary from 0.67 to 0.94. There is only a nomogram for predicting the risk MetS among these models and this nomogram was constructed with six non-invasive factors [33]. In our study, the nomogram for the prediction of MetS was constructed base on only four non-invasive predictors including gender, age, WC and SBP with AUC in range of previous studies. Additionally, all of variables in our model are easy and cheap to collect, thus this model might be conveniently applied to qualify the risk of developing MetS for each middle-age person in entire population. Furthermore, the decision curve analysis of this nomogram presents a high application in clinical practice. When the probability thresholds of new-onset MetS were from 0.13 to 0.7, the net benefit of this nomogram is better than those of examination all participant or none participant strategies.
This study had some limitations. First, 43.1% of the subjects were lost to follow-up. Given that the significant difference between participants and non-participants was not found in terms of anthropometrics, socioeconomic status, and lifestyles, the possible participation bias was minimal. Second, physical activity and dietary intake were not included in analysis of risk factors for MetS. Lastly, as the study cohort included Kinh people living in rural areas of Vietnam, the findings cannot be generalised to either urban areas or other ethnic groups.
In summary, the study indicates an alarming high incidence of MetS among a middle-aged Vietnamese population. The non-invasive nomogram should be applied to estimate the personalized risk of new-onset MetS to help high-risk individuals take interventions timely to reduce MetS-related morbidity and mortality.
The authors would like to thank all the participants and staffs of the Ha Nam Center for Preventive Medicine for their cooperation and assistance.
This study was supported by Vietnam's National Foundation for Science and Technology Development (NAFOSTED), grant no. 106.09–2010.29 and grant no. 106-YS.01-2015.10 from the Ministry of Science and Technology, Vietnam.
The authors declare that they have no competing interests
The datasets used and analyzed in this study are available from the corresponding author on reasonable request.
TQT contributed to research design, data analysis, manuscript preparation, discussion and editing of the final draft for publication. TQB contributed to conceptualization of the study, data collection, data analysis, and the intellectual revision of the manuscript. DHD, BTTN, NAN, DTL, PTP, and BTN conceptualized the study, collected and entered data, interpreted and discussed the final draft for publication.
The study was approved by the Ethics council of National Institute of Hygiene and Epidemiology, Vietnam (IRB-VN01057-34/2016).
All of participants gave written informed consent with signatures before entering the study.
Not applicable