Anthropometric, laboratory and image data used in this article belong to the third wave of the EpiFloripa Adult Cohort Study and the second wave of the EpiFloripa Aging Cohort Study. Similar methods and equipment were used in both studies. Anthropometric measures (waist circumference [WC], height and body mass) were measured twice according to the recommendations of Lohman et al.,¹⁶ and the mean of two measurements was considered for the study.

WC was measured to the nearest 1mm at the narrowest part of the trunk using an inextensible anthropometric tape, and height was measured to the nearest 1 mm using a stadiometer. Body mass was measured to the nearest 0,1 Kg using a digital scale (150 Kg capacity), calibrated before the study.

Blood pressure was measured in the sitting position, on the right arm using a pulse sphygmomanometer (Techline® digital read, São Paulo, Brazil), with the cuff at heart level. Blood pressure levels were measured twice, with a rest time of at least 15 minutes before and between measurements. When the difference between measurements was greater than 20 mmHg for systolic blood pressure or 10 mmHg for diastolic blood pressure, a third measurement was performed to replace the highest value. The mean of measurements was considered for analysis.

Blood samples were collected early in the morning after fasting for at least eight hours. Samples were stored and analyzed according to standard techniques of the Laboratory of Clinical Analysis (University Hospital, UFSC). Fasting blood glucose and lipid profile (triglycerides and high-density lipoproteins - HDL) were determined by colorimetric test, while CRP was determined by turbidimetry.

Carotid ultrasound was performed by a cardiologist using a portable ultrasound equipment Viamo™ (Toshiba Medical Systems, Tokyo, Japan) using a 5-11 MHz linear transducer. At least three IMT images of the common carotid artery were obtained on each side (right and left) and used the one with the best quality (quality index > 0.50). Carotid IMT measurements were analyzed using the M’Ath® software (version 3.1, METRIS Co., Argenteuil, France), which performs 100 automated measures per centimeter. The mean (in mm) of the left and right IMT was used in the analyses.¹⁷ The presence of atherosclerotic plaques (common carotid, carotid bulb, or ramifications in any side) was also identified during the imaging examination. Atherosclerotic plaque was defined as a focal structure encroaching at least 0.5 mm into the arterial lumen or having an arterial wall thickness ≥ 50% the surrounding IMT, with or without calcifications.^18,19

Three different outcomes were considered for this study: 1) carotid IMT (mean left and right IMT in millimeters, continuous symmetrical variable); 2) the presence of atherosclerotic plaque (binary variable, yes/no), and; 3) CRP levels (continuous variable analyzed as natural logarithm, ln, due to its asymmetry).²⁰

The joint criteria for definition of metabolic syndrome (MS) were used to establish the cutoff points of the different risk factors.²¹ Central obesity was defined as a WC > 80.0 cm in women and > 90.0 cm in men. High blood pressure was defined as a systolic blood pressure ≥ 130 mmHg and/or diastolic blood pressure ≥ 85 mmHg. Individuals who reported being diagnosed with hypertension by a physician and/or were taking antihypertensive drugs were also classified as hypertensive. Hyperglycemia was defined as a fasting blood glucose ≥ 100 mg/dl, self-report of a medical diagnosis of diabetes mellitus, and/or making use of antidiabetic medications.¹⁹ Participants were considered to have dyslipidemia when they had elevated triglycerides (≥ 150 mg/dL), low HDL (< 40 mg/dl in men and < 50 mg/dl in females), or reported use of antilipemic drugs.²¹

Each MS component was analyzed either as an independent exposure variable or as a cluster. Two different variables were generated to identify clusters. The first variable considered the number of positive cardiometabolic risk factors within the same individual (0, 1, 2, 3 or 4 MetS components). The second one combined the four independent variables into 16 possible combinations: negative for all risk factors, positive for only one factor (four possible combinations), two-factor cluster (six possible combinations), three-factor cluster (four possible combinations) or positive for all cardiometabolic risk factors.

Sociodemographic and lifestyle data were included as possible confounding factors. Most of these variables were collected during the third wave of the EpiFloripa Adult Cohort Study (2014/2015) and second wave of the EpiFloripa Aging Cohort Study (2013/2014): sex (male or female); age in years; education attainment (0-8, 9-11, ≥12 years of schooling); per capita family income (including all sources of income and divided by the number of family members) distributed in tertiles (1^sttertile = <R$ 900.00; 2^ndtertile = R$ 907.33 to R$ 2,122.00; 3^rdtertile =>R$ 2,125.00 ; 1 USD = R$ 3.14 in 2015). Smoking was analyzed as never smoked, former smoker or smoker, regardless of intensity and frequency. A validated questionnaire was used to evaluate physical activity in each cohort.^22,23 The adults were considered active when they reported ≥ 30 minutes of moderate-intensity physical activity on five or more days per week or vigorous-intensity activities for at least 20 minutes on three or more days per week.²⁴ Older adults were considered physically active when performed ≥ 150 minutes of physical activity/week. Data on energy intake from alcohol, saturated fat, sugar, fiber and sodium intake were obtained by two 24-hour food recalls, and information was extracted using procedures recommended in the literature, with adjustments for intra- and inter-individual variability.²⁵ Consumption of saturated fats and fiber were transformed into ln due to their asymmetric distribution.

Symmetric, continuous variables were described as mean and standard deviation, and asymmetric variables as median and interquartile range (p25-p75). Geometric mean was also estimated for CRP. Histograms and Q-Q plots were used to verify the normality of continuous variables. Categorical variables were presented as percentages (%). Depending on the nature of the investigated variables, the chi-square test, t-test for independent samples, or the Mann-Whitney test was used to identify possible differences between cohort members evaluated (or not) in the last wave, as compared to the baseline. Although the literature has reported an important relationship between increased years of life and cardiometabolic risk,²⁰ considering the outcomes of the present study (IMT, carotid plaque and CRP), interactions between all cardiometabolic risk variables (analyzed individually or as clusters) with age were tested. However, in the present study, a possible effect of age was detected only on CRP and dyslipidemia (p = 0.61).²⁶ Thus, to maintain statistical power to test the associations of interest in this study, the analyses were not stratified by age.

To evaluate the most prevalent clusters of cardiometabolic risk factors, the ratio between the observed and expected prevalence (O/E ratio) for each of the 16 possible combinations was estimated.²⁷ The expected prevalence was calculated by multiplying the observed probability of each risk factor, assuming that they occur independently in the population. An O/E ratio > 1.2 was used to identify highly prevalent clusters (i.e., higher than a random occurrence).²⁷

Linear regression models were used to test associations between the clusters of cardiometabolic risk factors (independent variables) and IMT and natural logarithm of CRP (lnCRP). Results were presented as regression coefficients (β) with the respective standard error (SE). For lnCRP, β was transformed into the exponential form (_EXPβ) and was interpreted as the percentage increase in serum CRP levels in relation to the geometric mean of those without any cardiometabolic risk factor. Assumptions of linearity for continuous variables, constant variance of standardized residuals, and goodness of fit of the model were assessed by plotting residuals against fitted values.

Logistic regression was used to analyze the presence of atherosclerotic plaque as an outcome, and results were expressed as odds ratio (OR) with their respective SE. All analyses were adjusted for all possible confounders (sociodemographic and lifestyle variables),²⁰ regardless of their level of statistical significance in the association with the outcomes.

Data analysis was conducted in the statistical software Stata 13.0 (StataCorp LP, College Station, USA), considering sampling weights, i.e., probability of selection at baseline and probability of location during follow-up, re-weighted to the estimated population of Florianópolis by sex and age group¹² and the survey design in each study. (A p value < 0.05 denoted statistical significance.

In the EpiFloripa Adult Cohort Study, 862 individuals were interviewed in 2014-15 (50.1% of the original cohort) Table 1). Family income and smoking status of participants were comparable with those of the baseline sample (2009). In the EpiFloripa Aging Cohort Study, 1,197 elderly were interviewed in 2013/14 (70.3%), who showed comparable data of sex, family income and smoking status to the baseline sample (2009).

Table 1

	EpiFloripa Adult Cohort Study			EpiFloripa Aging Cohort Study
	2009 (n=1,720)	2014/15 (n=862)		2009 (n=1,705)	2013/14 (n=1197)
Variables*	%	%	p value	%	%	p value
Sex: male	44.3	42.5	0.03‡	37.5	36.9	0.36‡
Age in years: mean ± SD	38.1 ± 11.6	39.9 ± 11.5	< 0.01‡	70.5 ± 7.9	69.7 ± 7.1	< 0.01‡
Per capita Family income (R$)†: Median [p25-p75]	900 [500-1750]	900 [500-1667]	0.55§	767 [380-1600]	800 [400-1667]	0.09§
Smoking
Never smoked	54.7	55.8	0.05//	59.6	60.2	0.53//
Former smoker	26.1	27.7		32.0	31.1
Smoker	19.2	16.5		8.4	8.7

*: Considering information obtained in 2009; ^†: 1 USD = R$ 1.70 in 2009; ^‡: T- test; ^§: Mann-Whitney test; ^//: Chi-square test.

Data of the outcomes of the present study were available in 1,301 participants in both studies combined, and results are described below. The mean IMT was 0.64 mm (± 0.15) and the median CRP was 1.34 mg/L (p25-p75 0.61, 3.48; geometric mean 2.64 mg/L). Also, 27.7% of participants had carotid plaques.

The prevalence of central obesity and elevated blood pressure levels was 56.8% and 71.5%, respectively; 17.7% of participants had dyslipidemia and 22.4% hyperglycemia (Table 2). Combined, 6.1% of the sample was positive for all four factors, while 21.8% had no MS component.

Table 2

	%	IMT (mm)	Carotid plaque	lnCRP
		β (SE)‡	OR (SE)	EXPβ (SE)§
Individual risk factors*
Central obesity (% yes)	56.8	0.6 (0.8)	0.69 (0.14)	1.86 (0.15)¶
Elevated blood pressure (% yes)	71.5	2.4 (0.7)¶	2.12 (0.51)¶	1.09 (0.12)
Dyslipidemia (% yes)	17.7	1.9 (1.1)	1.67 (0.43)¶	1.12 (0.11)
Hyperglycemia (% yes)	22.4	2.9 (0.9)¶	1.21 (0.24)	1.22 (0.10)¶
Number of positive risk factors†
None//	21.8	59.1±8.8 #	27.7% #	0.88±1.5 #
1	27.6	1.6 (0.8)	1.02 (0.35)	1.37 (0.20)
2	27.1	3.0 (1.0)	1.53 (0.64)	2.07 (0.28)
3	17.4	5.5 (1.1)	1.88 (0.78)	2.76 (0.41)
4	6.1	7.5 (2.3)	2.01 (0.96)	2.17 (0.33)

IMT: carotid intima-media thickness; lnCRP: natural logarithm of C-reactive protein; SE: standard error; mm: millimetres. * – Results adjusted for sex, age, family income, educational level, smoking, nutritional variables (intake of saturated fat, sugar, alcohol, fibre, and sodium), physical activity level, and mutual adjustment between individual risk factors. † Results adjusted for sex, age, family income, educational level, smoking, nutritional variables (intake of saturated fat, sugar, alcohol, fibre, and sodium), and physical activity level ‡ – Results presented as power (10^-2). § – Results interpreted as increment percentage of geometric mean. // – Mean ± SD or prevalence in the reference category (“none”). ¶– p-value <0.05 in comparison with the reference category. # - p-value for trend <0.05.

Considering each MS component as an independent variable, high blood pressure and hyperglycaemia were associated with a higher IMT. High blood pressure and dyslipidemia were associated with the presence of atherosclerotic plaque, whereas central obesity and dyslipidemia were related to higher lnCRP levels.

Table 2 also shows there was a direct trend between the number of cardiometabolic risk factors present in the same individual and the three outcomes (p-value for trend < 0.05 in all cases) (Table 2).

The observed prevalence for the coexistence of the four risk factors in the same individual was 6.1%, 410% higher (O/E ratio = 5.1) than it would be expected at random. For the simultaneity of three risk factors, all combinations with central obesity presented an O/E ratio > 1.2. The simultaneous presence of central obesity and elevated blood pressure (20.5%) and the isolated presence of elevated blood pressure (19.5%) were the most frequent combinations of two and one risk factors, respectively. However, even among these combinations, the O/E ratio was < 1.0. Finally, the observed prevalence of patients with none of the cardiometabolic risk factors was 110% higher than the expected prevalence. (Table 3).

Table 3

Risk Factor	Central obesity	Elevated blood pressure	Dyslipidemia	Hyperglycemia	n	Prevalence
						Observed % (95% CI)	Expected %	O/E
4	+	+	+	+	112	6.1 (4.9-7.6)	1.2	5.1
3	+	+	+	-	68	5.9 (4.6-7.5)	4.3	1.4
	+	+	-	+	209	10.2 (8.1-12.6)	5.7	1.8
	+	-	+	+	09	0.7 (0.3-1.6)	0.5	1.4
	-	+	+	+	10	0.7 (0.3-1.5)	1.6	0.4
2	+	+	-	-	309	20.5 (17.5-23.8)	19.7	1.0
	+	-	+	-	10	0.8 (0.3-1.8)	1.7	0.5
	+	-	-	+	17	1.1 (0.6-2.1)	2.3	0.5
	-	+	+	-	19	2.4 (1.4-4.4)	5.6	0.4
	-	+	-	+	37	1.8 (1.1-2.9)	7.5	0.2
	-	-	+	+	05	0.4 (0.1-1.0)	0.6	0.7
1	-	-	-	+	17	1.5 (0.8-2.7)	3.0	0.5
	-	-	+	-	11	0.8 (0.4-1.4)	2.2	0.4
	-	+	-	-	202	19.5 (16.2-23.2)	25.9	0.8
	+	-	-	-	70	5.8 (4.4-7.6)	7.8	0.7
0	-	-	-	-	196	21.8 (18.7-25.1)	10.3	2.1

CI: confidence interval; + presence of risk factor; - absence of risk factor; O: observed prevalence; E: expected prevalence; O/E: ratio between observed and expected prevalence.

Table 4 shows the association between the 16 combinations of cardiometabolic risk factors and the investigated outcomes. All clusters including central obesity and high blood pressure (i.e. clusters of two, three, or four factors) showed higher IMT and lnCRP than individuals without any risk factor. On the other hand, central obesity was a common factor in the determination of systemic inflammation, since a higher lnCRP was observed in all combinations including that risk factor. Except for central obesity, the isolated presence of one risk factor was not associated with any of the investigated outcomes. Other clusters showed a low prevalence (< 1%) to allow robust conclusions. Conversely, although the highest O/E was for the coexistence of the four risk factors, the associations with the investigated outcomes were not stronger than the three-component clusters. None of the clusters was associated with a higher frequency of carotid plaque.

Table 4

		%	IMT (mm)	Carotid plaque	lnCRP
	n		β (SE)‡	OR (SE)	EXPβ(SE)§
All negative	196	21.8	59.9±8.8//	11.6% //	0.86±1.5//
Positive for one risk factor only
Central obesity	70	5.8	0.4 (1.3)	0.24 (0.13)	2.01 (0.38)†
Elevated blood pressure	202	19.5	1.7 (0.9)	1.48 (0.53)	1.19 (0.18)
Dyslipidemia	11	0.8	2.7 (1.9)	0.92 (0.85)	1.92 (0.76)
Hyperglycaemia	17	1.5	3.7 (2.3)	0.71 (0.66)	1.21 (0.37)
Two-components clusters
Central obesity + elevated blood pressure	309	20.5	3.2 (0.9)†	1.31 (0.54)	2.18 (0.32)†
Central obesity + Dyslipidemia	10	0.8	-1.4 (1.9)	1.08 (1.04)	3.63 (0.90)†
Central obesity + Hyperglycemia	17	1.1	-1.7 (4.6)	0.69 (0.59)	2.24 (0.88)†
Elevated blood pressure + Dyslipidemia	19	2.5	5.3 (4.2)	3.14 (2.44)	1.03 (0.27)
Elevated blood pressure + Hyperglycemia	37	1.8	0.8 (3.4)	1.80 (0.88)	2.23 (0.74)†
Dyslipidemia + Hyperglycemia	05	0.4	9.4 (4.2)†	-	4.20 (1.31)†
Three-components clusters
Central obesity + Elevated blood pressure + Dyslipidemia	68	5.9	3.9 (1.5)†	2.15 (1.07)	2.77 (0.48)†
Central obesity + Elevated blood pressure + Hyperglycemia	209	10.2	6.1 (1.4)†	1.58 (0.70)	2.83 (0.42)†
Central obesity + Dyslipidemia + Hyperglycemia	09	0.7	5.3 (7.3)	1.36 (1.81)	4.06 (1.44)†
Blood pressure + Dyslipidemia + Hyperglycemia	10	0.6	11.1 (4.6)†	3.01 (2.78)	1.88 (0.84)
Positives for the four risk factors	112	6.1	7.4 (2.3)†	1.92 (0.91)	2.21 (0.34)†

IMT: carotid intima-media thickness; CRP: natural logarithm of C-reactive protein; SE: standard error; mm: millimeters. *Results adjusted for sex, age, family income, educational level, smoking, nutritional variables (intake of saturated fat, sugar, fibre, and sodium) and physical activity level. † – p-value< 0.05 in comparison with the reference category, indicating a higher outcome in that group. ‡ – results presented as power (10-2). § – Results should be interpreted as increment percentage of geometric mean. // – Mean ± SD or prevalence in the reference category (“all negative”).