Lifestyle and weight predictors of a healthy overweight profile over a 20-year follow-up

Authors


  • Funding agencies: This work was funded by a research grant from the Canadian Institute of Health Research (CIHR - #131594) to JL Kuk.

  • Disclosure: The authors declare no conflict of interest.

Abstract

Objective

To examine whether changes in modifiable risk factors [physical activity, cardiorespiratory fitness (CRF), body weight, and diet composition] are associated with the transition to metabolically healthy overweight/obese (MHOW) versus metabolically abnormal overweight/obese.

Methods

Analysis included 1,358 adults [aged 25.0 (3.5) years] from the CARDIA study who were healthy at baseline and had overweight/obesity at follow-up. Participants with zero or one of the following six risk factors were classified as MHOW: elevated triglycerides, LDL, blood pressure, fasting glucose, and HOMA-insulin resistance and low HDL.

Results

Over the 20-year follow-up, the sample gained weight (BMI 24.5 to 31.1 kg/m2), and the prevalence of MHOW was 47% at follow-up. After adjusting for changes in CRF, diet, and weight change, physical activity and macronutrient intake were not independently associated with MHOW (P > 0.05), while changes in CRF [fit–unfit: RR (95%) = 0.58, 0.52-0.66; unfit–unfit: RR = 0.67, 0.58-0.76, versus fit–fit] and weight [gain: RR (95%) = 0.54, 0.43-0.67; cycle: RR = 0.74, 0.57-0.94, versus stable] were independently associated with MHOW.

Conclusions

Focusing on high CRF and strategies to limit weight gain may be important for individuals with overweight and obesity in early to mid-adulthood to maintain a metabolically healthy profile.

Introduction

It is well established that obesity is positively associated with several cardiovascular disease (CVD) risk factors [1, 2]. However, there is a subgroup of obese individuals who appear to be “protected” from the common ill effects of obesity such as insulin resistance, hyperglycemia, hypertension, and dyslipidemia [3]. These “metabolically healthy obese” individuals display a metabolic profile that is nearly indistinguishable from that of healthy normal-weight individuals (3–5). Despite the clinical awareness of metabolically healthy obese individuals, factors underlying the apparently protective profile are not well established.

As the global prevalence of overweight and obesity is rising in both adult and youth populations [6, 7], it is important to examine what factors distinguish those who will become metabolically healthy overweight/obese (MHOW) versus the more common forms of overweight and obesity that are associated with metabolic abnormalities (metabolically abnormal overweight/obese; MAOW). While physical activity, cardiorespiratory fitness (CRF), weight, and diet composition have been indicated to be associated with CVD risk [8, 9] and metabolically healthy obesity [5, 10], research has yet to demonstrate whether changes in these factors can predict the maintenance or development of metabolically healthy obesity. Therefore, the purpose of the current investigation is to track individuals who were metabolically healthy at baseline and were overweight/obese after a 20-year follow-up and to examine whether changes in physical activity, fitness, body weight, and diet composition predict the maintenance or transition from metabolically healthy to MHOW versus MAOW.

Methods

Study population and database

The CARDIA study is a multicenter, longitudinal study designed to investigate the development and progression of CVD risk factors in young adults. In 1985-1986, a cohort of 5,115 participants from Birmingham, AL; Chicago, IL; Minneapolis, MN; and Oakland, CA, were randomly sampled and balanced for age (18-24 years and 25-30 years old), sex, ethnicity (African-American and White), and educational status (high school graduate or less and more than high school). Further details of the CARDIA eligibility criteria, recruitment process, and baseline demographic characteristics have been published elsewhere [11, 12]. This current study was reviewed and approved by the York University Research Ethics Board. Limited data access was obtained through the National Heart, Lung, and Blood Institute (NHLBI) of the National Institutes of Health (NIH). This manuscript was prepared using research materials obtained from the NHLBI Data Repository Information Coordinating Center and does not necessarily reflect the opinions or views of the CARDIA investigators or the NHLBI.

Participants

Of the 5,115 individuals in the original sample at year 0, approximately 31% (n = 1,569) were lost to follow-up by year 20 and were not included in the current analysis. Participants were excluded if they did not attend at least one other examination (year 5, year 10, and/or year 15) (n = 92) to determine weight change. Participants were also excluded if they reported exceedingly high (men ≥8,000; women ≥6,000 kcal) or low (men ≤800; women ≤600 kcal) energy intakes (n = 104), had 2 or more CVD risk factors at baseline (n = 627), were pregnant at baseline (n = 35), had a body mass index (BMI) of <25 kg/m2 at follow-up (n = 1,029), or had missing data for CRF (n = 707), physical activity (n = 21), BMI (n = 31), dietary data (carbohydrate-, fat- or protein intake) (n = 407) or smoking status (n = 54) at baseline or follow-up.

Participant health demographics

Age, sex, ethnicity, socioeconomic status, cigarette smoking status, alcohol consumption, family history, and medication use were assessed by interviewer-administered questionnaires [11] Measures for triglycerides, HDL, LDL, blood pressure, fasting glucose, and homeostasis model assessment of insulin resistance (HOMA-IR) were collected according to the standardized CARDIA protocol and processed at central laboratories [11]. Triglycerides and cholesterol were measured enzymatically and LDL was estimated using the Friedewald equation [13]. Blood pressure was measured three times at 1 min intervals after 5 min of seated rest and the mean of the final two measurements was used [13]. Blood was separated via centrifuge and frozen until laboratory analysis for glucose and insulin concentration. HOMA-IR was calculated as: [glucose (mmol/L) × insulin (mU/L)]/22.5 [14]. Participants were asked to arrive at the examination having fasted for at least 12 h, abstained from strenuous physical activity, and avoided tobacco use for at least 2 h. Height and weight was measured under the supervision of lab technicians on a calibrated scale and was used to calculate BMI (kg/m2) and weight change.

Definition of metabolic phenotypes

At baseline individuals were all metabolically healthy with zero or one of the six following subclinical CVD risk factors: triglycerides ≥1.7 mmol/L or taking cholesterol medication; HDL cholesterol <1.04 mmol/L or taking cholesterol medication for men or HDL <1.29 mmol/L or taking cholesterol medication for women; LDL cholesterol ≥4.1 mmol/L or taking cholesterol medication; systolic blood pressure ≥130 mm Hg or diastolic blood pressure ≥ 85 mm Hg or taking hypertension medication; fasting glucose ≥5.6 mmol/L or taking blood glucose lowering medication; and (HOMA-IR) ≥2.5. At follow-up, individuals were classified as MAOW if they had a BMI ≥25 kg/m2 and had two or more of the six aforementioned subclinical CVD risk factors at follow-up and were classified as MHOW if they had a BMI ≥25 kg/m2 and had zero or one of the six aforementioned CVD risk factors.

Assessment of physical activity and CRF

Physical activity was assessed using the CARDIA Physical Activity Questionnaire, which is an interviewer-based self-report questionnaire of frequency of participation in moderate [3-6 metabolic equivalent of tasks (METs)] and vigorous intensity activities (>6 METs). Separate moderate- and vigorous-intensity activity scores (expressed in exercise units; EUs) were computed as the product of the frequency and intensity of the activity [15]. Physical activity was subsequently categorized into sex-specific tertiles (inactive, moderately active, and active). For these analyses, the moderately active and active categories (males ≥364 EU, females ≥192 EU) were collapsed as previous studies have demonstrated that the greatest improvement in cardiometabolic risk factors is found between inactive and moderately active individuals [16, 17].

CRF was measured using a graded maximal exercise treadmill test [11]. CRF was subsequently categorized into sex-specific tertiles (unfit, moderately-fit and fit) at baseline and follow-up. The cutoffs at baseline for males are: unfit <689 s, moderately fit 689-780 s, and fit ≥ 781 s; and females: unfit <435 s, moderately fit 435-540 s, and fit ≥541 s. The cutoffs at follow-up for males are: unfit <466 s, moderately fit 466-570 s, and fit ≥571 s; and females: unfit <254 s, moderately fit 254-361 s and fit ≥362 s. Moderately fit and fit categories were collapsed for this analysis as previous studies have demonstrated that the greatest protective health benefits are found between unfit and moderately fit individuals [18, 19].

Weight changes

Weight changes between examinations (years 0, 5, 10, and 20) were divided into stable weight, weight loss, and weight gain categories. Weight change of <5% of body weight between all available time points with an overall weight change of <15% from baseline was classified as stable weight. Weight losers were defined as losing weight without gaining weight between any of the time points. Due to a low number of weight losers (n = 23), the stable weight and weight loser groups were collapsed and classified as stable weight. Weight gainers were defined as gaining ≥5% body weight without losing ≥5% body weight between any of the time points and having ≥15% overall increase in body weight from baseline to follow-up. Participants were classified as weight cyclers if they experienced both weight gain (>5%) and weight loss (>5%) between examinations.

Dietary intake

Dietary data were obtained using the CARDIA Diet History Questionnaire [20], which is an interviewer-administered questionnaire of food intake in the past month. Carbohydrate, fat, and protein consumed (g/day) were converted into caloric values and percentage of total caloric intake. Macronutrient intake was subsequently divided into two categories: high intake (carbohydrates ≥65%, fats ≥35%, or protein ≥15%) and low intake (carbohydrates <65%, fats <35%, and protein <15%) [21]. Dietary intake levels for saturated fat, fiber, calcium, cholesterol, folic acid, magnesium, polyunsaturated fatty acid, potassium, sodium, vitamin D, vitamin C, and vitamin A were divided into two categories using a median split (high intake versus low intake). Changes in nutrient intake from baseline to follow-up were then analyzed from high intake to low intake, high intake to high intake, low intake to high intake, and low intake to low intake.

Statistical analysis

Differences in baseline and follow-up characteristics stratified by metabolic status (MHOW and MAOW) were assessed using repeated measures analysis of variance (ANOVA) for the continuous variables and chi-square tests for the categorical variables. Log-binomial regression was performed to estimate the longitudinal associations of physical activity, CRF, weight cycling, and diet composition (carbohydrate-, fat- and protein-intake) on the relative risk (RR) of becoming MHOW at the 20-year follow-up adjusting for age, sex, ethnicity, education, smoking status, alcohol consumption and baseline BMI (Model 1). The mutually adjusted model was further adjusted for changes in physical activity, CRF, weight, and diet composition where applicable (Model 2). All statistical analyses were performed using SAS v9.4. Statistical significance was set at an alpha of 0.05.

Results

Table 1 presents the baseline and follow-up characteristics of participants stratified by metabolic status at follow-up. Participants had a mean BMI of 24.5 kg/m2 at baseline and 31.1 kg/m2 at follow-up. Of the 1,358 adults who became overweight or obese, 47% (642) were MHOW. Individuals who were MHOW at follow-up had a lower baseline BMI and caloric intake and higher baseline protein intake than those who were MAOW at follow-up (P < 0.05). At follow-up, those who were MHOW also had a lower BMI, higher levels of physical activity and CRF, lower carbohydrate intake, and a smaller increase in BMI from baseline to follow-up as compared to those who were MAOW at follow-up (P < 0.05).

Table 1. Baseline and follow-up characteristics in 1,358 adults who were overweight/obese at follow-up, stratified by follow-up metabolic status
 BaselineFollow-up
Total sample (n = 1,358)MHOW at follow-upMAOW at follow-upMHOW at follow-upMAOW at follow-up
  1. MHOW: metabolically healthy overweight/obese; MAOW: metabolically abnormal overweight/obese; BMI: body mass index.

  2. Data are presented as mean (SD) unless otherwise indicated.

  3. *Significantly different from MHOW at follow-up (P < 0.05).

n (%)642 (47%)716 (53%)  
Sex (% male)43.050.8--
Age (years)25.0 ± 0.124.9 ± 0.1--
BMI (kg/m2)24.2 ± 0.124.8 ± 0.1*29.6 ± 0.232.4 ± 0.2*
Active (%)434 (68%)477 (67%)365 (57%)352 (49%)*
High CRF (%)460 (72%)458 (64%)*510 (79%)400 (56%)*
Caloric intake (kcal)2652 ± 502807 ± 47*2293 ± 412345 ± 38
Dietary carbohydrates (%)46.3 ± 0.346.7 ± 0.346.2 ± 0.447.4 ± 0.4*
Dietary fat (%)37.2 ± 0.237.1 ± 0.236.8 ± 0.336.4 ± 0.3
Dietary protein (%)14.9 ± 0.114.6 ± 0.1*15.5 ± 0.115.6 ± 0.1
BMI change (kg/m2)--5.4 ± 0.27.6 ± 0.1*

After adjustment for age, sex, ethnicity, education, smoking status, alcohol consumption, and baseline BMI (Model 1), only individuals who were active at baseline but became inactive were less likely to be MHOW at follow-up [active–inactive: RR (95%) = 0.85, 0.75-0.95] when compared with individuals who remained active. After accounting for changes in CRF, diet, and body weight in the mutually adjusted model (Model 2), physical activity was no longer associated with becoming MHOW (Figure 1A, P > 0.05). In both models, remaining inactive or becoming active at follow-up was not associated with MHOW at follow-up [P > 0.05). Regardless of baseline CRF, individuals who were unfit at follow-up [fit–unfit: RR (95%) = 0.58, 0.52-0.66; unfit–unfit: RR (95%) = 0.67, 0.58-0.76] were less likely to be MHOW at follow-up compared with individuals who remained fit. Similar results for CRF were observed in Model 2 (Figure 1B). Individuals who gained weight [RR (95%) = 0.54, 0.43-0.67] or cycled their weight [RR = (95%) 0.74, 0.57-0.94] from baseline to follow-up were less likely to be MHOW at follow-up compared with individuals who maintained a stable weight or lost weight. Similar results were observed in Model 2 (Figure 1C). Changes in macronutrient intake were not associated with becoming MHOW in either model (Table 2). Furthermore, changes in dietary intake levels of saturated fat, fiber, calcium, cholesterol, folic acid, magnesium, polyunsaturated fatty acid, potassium, sodium, vitamin D, vitamin C, and vitamin A were also not associated with becoming MHOW in either model.

Figure 1.

Relative risk for becoming MHOW with changes in (A) physical activity, (B) fitness, and (C) weight after adjusting for age, sex, ethnicity, education, smoking status, alcohol consumption, baseline BMI, and changes in physical activity, fitness, weight, and diet composition in 1,358 overweight/obese adults. MHOW: metabolically healthy overweight/obese; RR: relative risk. *Significantly different than reference group (P < 0.05). A: Active-Active: Active (males ≥ 364 EU, females ≥ 192 EU) at both baseline and follow-up. Inactive-Inactive: Inactive (males < 364 EU, females < 192 EU) at both baseline and follow-up. Active-Inactive: Active at baseline and inactive at follow-up. Inactive-Active: Inactive at baseline and active at follow-up. B: Fit-Fit: Fit at both baseline (males ≥ 689 s, females ≥ 435 s) and follow-up (males ≥ 466 s, females ≥ 254). Unfit-Unfit: Unfit at both baseline (males < 689 s, females < 435 s) and follow-up (males < 466 s, females < 254). Fit-Unfit: Fit at baseline and unfit at follow-up. Unfit-Fit: Unfit at baseline and fit at follow-up. C: Stable/Loss: stable weight or weight loss at follow-up. Gainer: ≥5% weight gain without ≥5% body weight lost between any time points and ≥15% overall increase in body weight from baseline to follow-up. Cycler: Participants with both weight gain (>5%) and weight loss (>5%) between examinations.

Table 2. Relative risk for becoming MHOW with changes in diet composition (fat, carbohydrate, and protein intake)
 Model 1Model 2
  1. Model 1: simple model—adjusted for age, sex, ethnicity, smoking status, education, alcohol consumption, and baseline BMI.

  2. Model 2: mutually adjusted model—adjusted for age, sex, ethnicity, smoking status, education, alcohol consumption, baseline BMI, physical activity, CRF, weight change, and diet where applicable.

  3. Low–low: Low intake at both baseline and follow-up. Low–high: Low intake at baseline and high intake at follow-up. High–low: High intake at baseline and low intake at follow-up. High–high: High intake at both baseline and follow-up.

Fat
Low–low1.01 (0.88-1.14)1.01 (0.89-1.14)
Low–high1.05 (0.90-1.23)1.01 (0.89-1.14)
High–low0.99 (0.88-1.11)0.99 (0.88-1.10)
High–high11
Carbohydrates
Low–low11
Low–high0.95 (0.81-1.10)0.96 (0.85-1.09)
High–low1.02 (0.86-1.19)1.04 (0.90-1.19)
High–high0.92 (0.79-1.05)0.97 (0.85-1.11)
Protein
Low–low0.95 (0.83-1.10)0.96 (0.86-1.08)
Low–high0.97 (0.84-1.12)0.98 (0.88-1.09)
High–low1.03 (0.88-1.21)0.99 (0.87-1.12)
High–high11

Discussion

To our knowledge this is the first longitudinal study to demonstrate that CRF and weight changes are both independently associated with the maintenance of a metabolically healthy profile and that physical activity and diet are not associated with MHOW status. This may suggest that focusing on maintaining a high level of CRF and limiting weight gain in overweight and obese individuals in early adulthood may be more important factors to consider when it comes to maintaining metabolic health in obesity.

Changes in physical activity were associated with the maintenance of a metabolically healthy profile in overweight or obese individuals. Previous research offers conflicting results regarding physical activity and metabolically healthy obesity [22, 23]. However, these studies only examined physical activity at one time point and did not track changes in physical activity over time. In the current study, only the individuals who were active at baseline but became inactive, and not those who were consistently inactive, were less likely to be MHOW at follow-up when compared to those who remained active. These findings are analogous to observations in the Harvard Alumni Study which report that former university athletes who subsequently decreased sports play were at elevated morality risk compared to those who chronically participated in no sports [24]. However, these associations were no longer significant after adjusting for changes in CRF, BMI, and macronutrient intake. Thus, measuring physical activity at one time point may be insufficient in optimally predicting the development of a MHOW profile and furthermore, self-reported physical activity appears to be an inferior predictor for MHOW when compared to CRF.

Previous research reports that low CRF is associated with metabolic syndrome [25] and that women with metabolically healthy obesity have significantly higher CRF than women with obesity and metabolic syndrome [26]. In the current study, CRF was also a significant predictor of the maintenance of a metabolically healthy profile, even after adjusting for changes in physical activity, weight and diet. More specifically, individuals that had low CRF at follow-up were less likely to be MHOW compared to individuals that maintained or achieved a high level of CRF at follow-up. These results suggest that having a high level of CRF may be more important than high levels of physical activity for metabolic health in overweight and obese individuals. However, a high CRF is generally attained through engaging physical activity [25, 26], and thus this may highlight the limitations of using self-report physical activity. Alternatively, individuals simply may not have engaged in the types of activities that are associated with increases in CRF. Studies have demonstrated that individual response to a given training intervention can range from no change to a doubling in CRF (VO2max) [27, 28]. These differences are generally attributed to genetic variability, but suggest that individuals with the greatest improvement in CRF with physical activity may also have the greatest health benefits and are more likely to maintain a healthy profile with overweight or obesity. However, it is important to note that we defined high CRF as being in the upper two tertiles and the relatively low levels of CRF necessary to attain health benefits aligns with previous studies which demonstrate the largest gains in health lies between those who are in the lowest fitness quintile and the upper 80% [29].

Individuals that gained or cycled weight were less likely to be MHOW independent of changes in physical activity, CRF and diet when compared to those who maintained a stable weight or lost weight. These results extend previous findings demonstrating that both weight gain and weight cycling are associated with metabolic syndrome [29, 30]. The negative effects of weight cycling may be in part due to the reported detrimental effects on body composition [31]. Weight cyclers experience greater reductions in lean mass during weight loss and a greater proportion of fat mass gained during weight regain compared to nonweight cyclers [31]. As only 5-10% of individuals maintain clinically significant weight loss for more than 1 year [32], our findings suggests that it may be more beneficial for individuals to focus on limiting weight gain as opposed to repeatedly failing at weight loss.

Changes in macronutrient composition were not associated with the maintenance of a metabolically healthy profile. Although previous studies report that diet composition is associated with several CVD risk factors [33, 34], the relationship of macronutrient intake with the clustering of these CVD risk factors is still unclear [35]. For instance, there are studies that indicate high carbohydrate intake is a risk factor for having metabolic syndrome [36], while others do not [35]. Similarly, in studies examining metabolically healthy obesity specifically, there are conflicting results regarding macronutrient intake differences between metabolically healthy obese and metabolically abnormal obese individuals [5, 22, 37, 38]. Our analysis of micronutrient intake also found that changes in dietary intake levels of saturated fat, fiber, calcium, cholesterol, folic acid, magnesium, polyunsaturated fatty acid, potassium, sodium, vitamin D, vitamin C, and vitamin A were not associated with becoming MHOW. This area may require further study as the association between diet composition and health risk factors may require more holistic examination of dietary factors such as glycemic load and sodium content in addition to macronutrient and micronutrient content. Alternatively, the general lack of consensus regarding dietary factors and the metabolically healthy obese phenotype may suggest that there is no single dietary factor that is responsible for this unique healthy profile.

Several limitations of the present study warrant mention. There are several definitions of metabolically healthy obesity used in previous studies and thus comparisons between results must be made cautiously [39, 40]. Information as to why participants were lost to follow-up was unavailable. Excluded individuals may have had the highest health risk (e.g., low physical activity, low CRF, greater disease, etc.), which may have minimized the observed associations. Physical activity and dietary data was obtained using self-report. Another shortcoming was that we were unable to distinguish intentional from unintentional weight loss. Also, all variable changes that occurred in between assessments were not accounted for. Finally, there was insufficient sample size to analyze metabolically healthy obese separate from metabolically healthy overweight. However, a major strength of this study is the tracking of these variable changes over a 20-year period of time with direct assessment of weight and health risk factors.

In summary, CRF and changes in weight are independently associated with the maintenance of a MHOW profile whereas physical activity and macronutrient intake were not. Thus, it may be more important for individuals who develop overweight or obesity in early-mid adulthood to focus on maintaining a high CRF and limiting weight gain to maintain a metabolically healthy status.

Ancillary