Study design and subject recruitment

The Cork and Kerry Diabetes and Heart Disease Study (Phase II) was a single centre, cross-sectional study conducted between 2010 and 2011 [22]. In brief, a population representative random sample was recruited from a large primary care centre in Mitchelstown, County Cork, Ireland (Mitchelstown cohort). The Livinghealth Clinic includes 8 general practitioners and serves a catchment area of approximately 20,000 with a mix of urban and rural residents. Participants were randomly selected from all registered attending patients in the 50-69 year age group. In total 3,807 potential participants were selected from the practice list. Following exclusion of duplicates, deaths and ineligibles, 3,043 were invited to participate in the study and of these 2,047 (49.2% male) completed the questionnaire and physical examination components of the assessment (response rate 67%). Ethics committee approval conforming to the Declaration of Helsinki was obtained from the Clinical Research Ethics Committee of University College Cork. All participants provided written informed consent. Following exclusion of missing BMI data the remaining 2,040 participants were analysed.

Clinical data

Participants attended the clinic in the morning after an overnight fast (minimum 8h). Fasting blood samples were taken on arrival. Participants completed a General Health Questionnaire (GHQ), a Food Frequency Questionnaire (FFQ) and the International Physical Activity Questionnaire (IPAQ). Data on age, gender, family history, medication use (diabetes, dyslipidaemia and hypertension) and medical history (diabetes, hypertension and cardiovascular disease) was gathered through a self-completed GHQ. Blood pressure was measured according to the European Society of Hypertension Guidelines using an Omron M7 Digital BP monitor on the right arm, after a 5 minute rest in the seated position. The average of the second and third measurements was used for analyses.

Anthropometric data

Anthropometric measurements were recorded with calibrated instruments according to a standardised protocol. Body weight was measured in kilograms without shoes, to the nearest 100g, using a Tanita WB100MA weighing scales (Tanita Corporation, IL, USA). Height was measured in centimetres to 1 decimal place using a Seca, Leicester height gauge (Seca, Birmingham, UK). Waist circumference (defined as mid-way between lowest rib and iliac crest) was measured in centimetres to 1 decimal place using a Seca 200 measuring tape (Seca, Birmingham, UK). The average of two measures were used for analyses. Body mass index (BMI) was calculated. Body fat percentage (BF%) was estimated using a recently described body adiposity estimator [23]. Individuals with a BMI ≥30kg/m² were defined as obese.

Dietary and lifestyle data

Diet was assessed using a modified version of the self-completed EPIC FFQ used in the Cork and Kerry Phase 1 study [24] which has been validated for use in the Irish population. A dietary score (the DASH (Dietary Approaches to Stop Hypertension) score) was calculated using the FFQ responses. It was a composite score derived from standard food groups within the FFQ as described by Fung et al. [25]. For each food group, consumption was divided into quintiles and participants were classified according to their intake ranking. Consumption of healthy food components were rated on a scale of 1-5, the higher the score the more frequent the consumption of that food. Less healthy dietary constituents, where low consumption is desired, were scored on a reverse scale with lower consumption receiving the higher scores. Component scores were summed and an overall DASH score for each person was calculated. A lower score indicated a poorer dietary quality. The DASH score was also categorised by median. Daily servings from each food pyramid shelf and compliance with the food pyramid were also determined.

Physical activity levels were assessed using the short form IPAQ [26] which provided information on frequency, duration and intensity of physical activity. Physical activity was categorized into 3 groups; low, moderate and high, based on a combination of activity frequency, duration of each activity bout and metabolic equivalent (MET) minutes per week in all activity types. Subjects were additionally classified according to current Irish physical activity guidelines of at least 30 minutes of moderate exercise 5 days a week or 150 minutes per week. Smoking status was defined as never, former and current smokers. Alcohol consumption included questions based on weekly intake to define never, moderate and heavy drinkers [27]. The Alcohol Use Disorders Identification Test (AUDIT-C) was used to identify subjects who are hazardous drinkers or have active alcohol use disorders.

Biological analyses

Plasma and serum were prepared from fasting blood samples from each subject. Fasting plasma glucose (FPG), serum total, high density lipoprotein (HDL) cholesterol, low density lipoprotein (LDL) cholesterol and triglycerides levels were measured by Cork University Hospital Biochemistry Laboratory using fresh blood samples. FPG concentrations were determined using a glucose hexokinase assay and serum lipids were analysed using enzymatic colorimetric tests (Olympus Life and Material Science Europa Ltd., Lismeehan, Co. Clare, Ireland) on an Olympus 5400 automatic analyser (Olympus Diagnostica Gmbh, Hamburg, Germany). Serum insulin and C reactive protein (CRP) were determined using a biochip array system (Evidence Investigator; Randox Laboratories, UK). Homeostasis model assessment of insulin resistance (HOMA-IR) was derived from fasting glucose and insulin concentrations as [(fasting plasma glucose x fasting serum insulin)/ 22.5] [28].

Determination of metabolic health status

Subjects were classified according to their BMI as obese (BMI ≥30kg/m²) and non-obese (BMI <30kg/m²). Metabolic health status was defined using a selection of five existing MH definitions based on a range of cardiometabolic abnormalities [9,29–32] (Table 1). In secondary analyses we also examined the combined overweight and obese (BMI ≥25kg/m²) subjects and the normal weight only (BMI <25kg/m²) subjects generating the following groups: metabolically healthy overweight and obese (MHOWO), metabolically unhealthy overweight and obese (MUOWO), metabolically healthy normal weight (MHN) and metabolically unhealthy normal weight (MUN).

Statistical analysis

Statistical analysis was conducted using PASW Statistics version 18.0 for Windows (SPSS Inc, Chicago, IL). Continuous variables are expressed as means ± SEM and categorical variables as percentages. Biochemical variables were assessed for normality of distribution, and skewed variables were normalised by log₁₀ or square root transformation as appropriate. Differences between groups were analysed by independent t-tests for continuous variables and Chi-Square test for categorical variables. Logistic regression was used to determine associations between MHO and MHNO, demographic, dietary and lifestyle factors. Multivariate logistic regression analysis was performed including age, gender, dietary quality, food pyramid compliance, physical activity, smoking status and alcohol consumption as confounding factors. Pair-wise comparison of agreement between metabolic health definitions was assessed by Cohen’s kappa where values >0.75, 0.4-0.75 and <0.4 indicate excellent, fair to good and poor agreement, respectively [33]. A P-value of < 0.05 was considered significant.

Prevalence of metabolically healthy and unhealthy obese and non-obese phenotypes

The proportion of individuals in the Mitchelstown cohort defined as being metabolically healthy by BMI categories varied considerably between definitions (Figure 1A). Among all individuals, 2.2% to 11.9%, were obese yet metabolically healthy (MHO), 20.6% to 30.1% were obese and metabolically unhealthy (MUO), 14.7% to 59% were non-obese yet metabolically unhealthy (MUNO), and 8.8% to 52.7% were non-obese and metabolically healthy (MHNO). Among the obese participants, the proportion of metabolically healthy subjects ranged from 6.8% (Aguilar-Salinas), 14.2% (Karelis), 23.7% (Wildman), 30.2% (Meigs (A)) to 36.6% (Meigs B). Among the non-obese participants, the proportion of metabolically unhealthy subjects ranged from 21.8% (Meigs B), 32.4% (Meigs (A)), 41.7% (Wildman), 75.2% (Karelis) to 87% (Aguilar-Salinas). Examination of metabolic health obese phenotypes according to gender revealed higher prevalence of metabolic health among both obese and non-obese females, whereas the prevalence of the metabolically unhealthy phenotype was generally greater among obese and non-obese males (Figure 1B and 1C). MHO prevalence increased with age, except for MHO defined by Aguilar-Salinas and Karelis (Figure 2), whereas the prevalence of MHNO decreased with age according to each definition. Secondary analysis of the prevalence estimates of MHOWO, MUOWO, MHN and MUN followed the same definition dependent trends identified in the primary analysis (Table S1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Prevalence of metabolically healthy and unhealthy obese and non-obese phenotypes in the Mitchelstown cohort according to different metabolic health criteria among all subjects and stratified by gender.

Results are expressed as the percentage of all subjects (A), female (B) and male (C) participants within each metabolic health definition. The metabolically healthy non-obese (MHNO), metabolically unhealthy non-obese (MUNO), metabolically healthy obese (MHO) and metabolically unhealthy obese (MUO) groups are depicted as white, dark grey, black bars and light grey bars, respectively.

https://doi.org/10.1371/journal.pone.0076188.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Prevalence of metabolically healthy and unhealthy obese and non-obese phenotypes according to different criteria stratified by age group.

The prevalence of metabolically healthy non-obese subjects (A) decreased with age according to each definition. Conversely estimates of metabolically unhealthy non-obese subjects (B) increased with age according to each definition. Metabolically healthy obesity (C) prevalence differed according to which definition was used. The Aguilar-Salinas et al., and Karelis et al., definitions identified the highest prevalence among the 45-54 yr olds whereas the Wildman et al., and Meigs et al., definitions found the greatest prevalence of MHO among the 65-74 yr olds. Prevalence of metabolically unhealthy obesity (D) was consistently highest among the 55-64 yr olds, with the exception of the Aguilar-Salinas et al., definition. Meigs A is defined using metabolic syndrome variables and Meigs B is defined using HOMA. Results are expressed as percentage of all subjects within each age category. The 45-54, 55-64 and 65-74 yr old age groups are depicted as white, grey and black bars, respectively.

https://doi.org/10.1371/journal.pone.0076188.g002

Agreement between MHO classifications

We examined whether the same individuals were classified as MHO according to the different definitions. Comparison of the four definitions which led to the greatest prevalence (Karelis et al., Meigs et al., and Wildman et al.) revealed that only 20 subjects were simultaneously classified as MHO (Figure 3A). Examination of the Meigs et al., definitions revealed some degree of concordance. These scores identified a total of 436 MHO subjects: n=202 (Meigs A) and n=234 (Meigs B). Of the 234 participants classified as MHO according to Meigs B, 44.4% were simultaneously classified as MHO by Meigs A (Cohen’s kappa 0.23, p < 0.001) (Figure 3B). Comparison of the three MH definitions which included HOMA (Karelis et al., Meigs et al., (B) and Wildman et al.) identified only 25 subjects being similarly classified as MHO (Figure 3C).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. Agreement between MHO definitions.

Only 20 subjects were simultaneously classified as MHO according to the four definitions which led to the greatest prevalence (Figure 3A). Less than half (44.4%) of the MHO subjects identified by Meigs B were simultaneously classified by Meigs A (Cohen’s kappa 0.23, p < 0.001) (Figure 3B). Comparison of the three MH definitions including HOMA identified only 25 subjects being similarly classified as MHO (Figure 3C).

https://doi.org/10.1371/journal.pone.0076188.g003

Anthropometric measures and clinical characteristics

Clinical and anthropometric characteristics of the study population according to both metabolic health status and BMI category are presented in Table 2. Compared to their metabolically unhealthy counterparts both the MHNO and MHO subjects had smaller waist circumference, lower BMI, TAG, FPG and higher HDL-C concentrations, reduced insulin resistance and were less hypertensive, with the exception of the MHO subjects defined by Aguilar-Salinas. In the secondary analyses comparison of the MHOWO and MUOWO subjects mirrored these findings, whereas some definition dependent differences were noted between the MHN and MUN subjects (Table S2).

Dietary composition, quality and food pyramid shelf servings

No differences in calorie consumption, macronutrient intake and dietary quality were observed between MHO and MUO subjects (Table 3), with the exception of MHO subjects defined by Meigs A who had higher fat and lower carbohydrate intakes. Some definition dependent differences were noted between daily servings of fruit and vegetables, dairy, meats and fats between MHO and MUO subjects. Compliance with the food pyramid recommendations was greater among the MHO subjects (Meigs A and Wildman). Among the non-obese subjects (Table 4), calorie intake and dietary macronutrient composition were similar, with the exception of MHNO defined by Aguilar-Salinas. Dietary quality was higher among the MHNO subjects defined by Karelis and Meigs B. Daily number of servings of fruit and vegetables and dairy were higher among the MHNO defined by Meigs A, Meigs B and Wildman. Micronutrient intakes were examined. MHO subjects had lower retinol intake (Aguilar-Salinas and Meigs B) and higher iodine intake (Meigs A and Wildman) compared to the MUO subjects (p < 0.05, data not shown). The secondary analyses did not reveal any significant differences regards energy intake, macronutrient composition and dietary quality between the MHOWO and MUOWO subjects. Comparison of the MHN and MUN subjects produced similar findings to comparison between MHNO and MUNO subjects (data not shown).

Lifestyle behaviours: physical activity, smoking status and alcohol consumption

Examination of lifestyle behaviours among the obese subjects (Table 3) did not reveal any differences in level of physical activity intensity, daily total time of physical activity and whether subjects met current physical activity guidelines. Differences in smoking status and alcohol consumption were noted for MHO defined by Wildman, wherein fewer MHO subjects were current smokers and engaged in hazardous drinking. Among the non-obese subjects (Table 4), duration of physical activity was higher among the MHNO subjects defined by Meigs A. Physical activity level, smoking status and alcohol consumption were not different between the MHNO and MUNO subjects. In the secondary analyses examination of physical activity measures, smoking behaviour and alcohol consumption did not reveal any significant differences between the metabolically healthy and unhealthy subjects, regardless of BMI (data not shown).

Determinants of the metabolically healthy phenotype

Among both the obese and non-obese subjects, females were more approximately 2-4 times more likely to be metabolically healthy relative to males (Table 5). Associations with age were conflicting. Among the obese subjects, greater compliance with food pyramid recommendations was positively associated with MH defined by Meigs A (OR 1.49, 95% CI 1.07-2.08, p = 0.018), Meigs B (OR 1.45, 95% CI 1.05-2.00, p = 0.025) and Wildman (OR 1.53, 95% CI 1.07-2.19, p = 0.021) in the unadjusted analyses only. In the adjusted analyses a moderate to high level of physical activity was positively associated with MHO defined by insulin resistance, particularly among those with a high level of activity (OR 2.35, 95% CI 1.28-4.34, p = 0.006) relative to those with a low level of physical activity. Odds of presenting with MHO defined by Karelis were greater among alcohol drinkers.

Among the non-obese subjects, again in the unadjusted analyses higher dietary quality was positively associated with MH defined by Karelis (OR 1.39, 95% CI 1.04-1.84, p = 0.024) and Meigs B (OR 1.37, 95% CI 1.00-1.87, p = 0.048). These differences did not remain significant in the multivariate adjusted analysis. No relationship was identified with tobacco smoking or alcohol consumption. While physical activity was positively associated with metabolic health defined by insulin resistance and by Wildman, these did not reach statistical significance. In the secondary adjusted analysis comparison of the MHOWO and MUOWO subjects mirrored these findings, particularly with regard to the positive associations between being female (for most definitions) and having a moderate to high level of physical activity with metabolic health defined by insulin resistance among the overweight and obese subjects (Table S3).