Introduction

Obesity is an established risk factor for multiple diseases [1–4]. The lag between exposure to obesity and the onset of disease should make prevention possible. But prevention will be difficult if overweight people who are at risk of disease are unaware of any developing health problem. Awareness of health deterioration only after becoming obese, and not at lower thresholds, may come too late to change behavior or avert the risk.

Health-related quality of life [HRQoL] measures an individual’s perception of their health in multiple dimension [5, 6]. It allows examination of whether people whose weight puts them at increased risk of disease perceive themselves to be in worse health. The evidence is mixed [7–16]. Associations between HRQoL and adiposity have been found to be nonlinear [7, 12, 16, 17], differ across dimensions of HRQoL [9, 14, 15], and vary with the measure of adiposity [14]. Associations also differ across populations defined by country [12, 15, 18], ethnicity [8], and sex [9, 12, 14, 16, 18–20].

Most of the evidence on the association, if any, between HRQoL and adiposity comes from high-income countries in North America and Europe [7, 9, 10, 21]. Evidence from East Asia also tends to be for high-income countries [12, 19–22]. Lack of evidence on the HRQoL-adiposity association in low- and middle-countries (LMIC) in Asia, and elsewhere, is limiting given the rising prevalence of obesity in these countries [23, 24]. Without this evidence, it is more difficult to design and target disease prevention programs that are critical to protecting resource-constrained health systems with underdeveloped chronic disease management in primary care. Evidence on the existence, nature, and strength of any HRQoL-adiposity association in Asian LMIC is potentially particularly valuable because disease risks may occur at lower adiposity thresholds in Asian populations [25, 26].

Most studies of the HRQoL-adiposity relationship have used the RAND 36-item Short Form Health Survey (SF-36) to measure HRQoL and the body mass index (BMI) to measure adiposity [7, 11, 12, 15, 17–19, 27–29]. Many found a negative relationship between physical health, measured by the SF-36 physical component score, and BMI [11, 12, 15, 17, 18–21]. Evidence is more mixed on the relationship between the mental health component score and BMI [7, 12, 14, 15, 17, 18, 20, 21]. Some studies found that overweight and moderately obese people reported better mental health [7, 14, 15, 17]. There is some evidence from Asia of this obesity paradox: class 1 obesity has been found to be associated with better mental health, physical functioning, role functioning, social functioning, mental health, vitality, and general health [15, 17], and mild abdominal obesity was found to be associated with better physical functioning and vitality among women [29]. People who are underweight have often been found to have the lowest HRQoL [9, 15], resulting in an inverse U-shaped relationship between some dimensions of HRQoL and BMI [7, 17].

The focus on BMI, with some exceptions [20], is limiting because central adiposity is a more relevant risk factor for some diseases [30–32]. Since measures of adiposity are not perfectly correlated [33–38] they do not necessarily all have the same association with HRQoL. For example, if HRQoL were to decrease as WHR increased but it was relatively unresponsive to BMI, perhaps because of greater self-awareness of high WHR, then it may be easier to convince people to adopt healthier habits when they are close to a WHR risk threshold than when they are close to a BMI risk threshold.

The aim of this study was to test for and estimate the magnitude of any association between HRQoL and adiposityin a low-income population. It aimed to explore heterogeneity in the association by sex, dimension of HRQoL, and the measure of adiposity. Further, we aimed to examine nonlinearity in the relationship across the distribution of each adiposity measure.

Methods

Results

Of 4126 randomly selected individuals, 37 were ineligible, 137 refused, and 156 did not complete the interview. The analysis sample of 3796 individuals consisted of 1298 (34.2%) males and 2498 (65.8%) females.

Table 1 shows the characteristics of this sample by sex. Mean ages were 52.6 and 51.2 years for males and female, respectively. Most participants (> 55%) had not graduated from high school and almost three quarters were located in rural areas. About 85% of males, but only 45% of females, were employed. The percentage of smokers was 58% for males and 12% for females. Only 21% of males and 23% of females reported being diagnosed with a chronic illness, but over 70% had a family member with such an illness.

Download:

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

Table 1. Participant characteristics by sex.

https://doi.org/10.1371/journal.pone.0275798.t001

In the sample, mean BMI was 22.1 for males and 23.7 for females. On the basis of BMI, 27.5% of males and 36.7% of females were at increased risk, while 8.6% and 17.3%, respectively, were at high risk. Sample mean WC was 84.8 cm for males and 86.5 cm for females. Only 5.2% of males, but 44.7% of females, were at high risk on the basis of WC. Mean WHR was 0.9 for both sexes. Only 7.6% of males, but 83.1% of females, were classified as high risk for WHR.

Sample mean HRQoL was highest in the role functioning dimension (Males: 97.5; Females: 97.4) and lowest in the health perception dimension (Males: 67.5; Females: 66.7) for both sexes. In most dimensions, mean HRQoL was slightly lower for female participants.

Fig 1 shows, for each sex, the estimated conditional mean of each dimension of HRQoL as a function of BMI (top), WC (middle), and WHR (bottom). Shading indicates 95% CI. From the left, the first column shows physical functioning, which was higher at higher BMI through the thresholds that define increased risk (BMI>22.9) and high risk (BMI>27.4). Above the obesity threshold (BMI>29.9), physical functioning was lower for individuals with higher BMI. Physical functioning was lower for women with higher WC only once WC was above the high risk threshold. For males, physical function was lower for individuals with higher WC even before the high risk threshold, although the 95% CI is wide in this region. For both sexes, physical functioning was lower for individuals with higher WHR, including through the thresholds for borderline and high risk. There was an upturn in physical functioning at very high WHR for males, but the data are very sparse in this region.

Download:

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

Fig 1. Means of health-related quality of life dimensions by body mass index (BMI), waist circumference (WC), waist-to-hip ratio (WHR), and by sex.

Notes. Each graph shows a conditional mean function estimated by local linear bivariate regression using the Epanechnikov kernel function and the plugin estimator of the asymptotically optimal constant bandwidth. Shading shows 95% confidence intervals. Dashed vertical lines indicate increased and high risk for BMI, and borderline and high for WC and WHR.

https://doi.org/10.1371/journal.pone.0275798.g001

For role functioning the patterns are similar. This dimension of HRQoL was also not lower, and for males it was higher, at higher values of both BMI and WC through the thresholds that define increased/borderline risk and high risk. Again, the relationship with WHR was negative.

For social functioning, the relationships differ by sex. For the most part and through the relevant thresholds, males with higher BMI and WC had better social functioning. This dimension of HRQoL was lower for females with higher values of WC and WHR.

Mental health, particularly of males, was better at higher BMI and WC. The (positive) health perception of males peaked before BMI and WC reached the respective thresholds that define high risk. Only above these thresholds did females with higher BMI and WC have perceptions of worse health. There was no clear relationship between pain and BMI. Particularly for males, HRQoL in the domain of pain was higher (less pain) for individuals with higher WC and WHR.

Table 2 gives the mean of each dimension of HRQoL by categories of BMI, WC, and WHR, and by sex. Males in categories of higher BMI had significantly better physical functioning (P < 0.0001). But males in categories of higher WHR had significantly worse physical functioning (P < 0.05). Male role functioning, social functioning, and mental health were all significantly better in categories of higher BMI (P < 0.05, P < 0.1, and P < 0.05, respectively). Mental health of males was also better in categories of higher WC (P < 0.05). The health perception of males peaked at the increased risk category of BMI and the borderline category of WC.

Download:

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

Table 2. Means of health-related quality of life dimensions by categories of body mass index (BMI), waist circumference (WC), and waist-to-hip ratio (WHR), and sex.

https://doi.org/10.1371/journal.pone.0275798.t002

For females, there were fewer significant differences. Females in higher categories of WC and WHR had significantly worse social functioning (P < 0.05 and P < 0.001, respectively). Female mental health was significantly better in categories of higher BMI (P < 0.05). These results are robust to replacing the Asian population BMI categories with the international standard categories, with the exception that obese males and females have lower physical functioning (S2 Table in S1 File).

Fig 2 shows multivariable regression estimates of differences in the conditional mean of each dimension of HRQoL across the categories of BMI, WC, and WHR (S3 Table in S1 File for estimates in table format). After controlling for covariates, underweight males continued to have significantly lower physical functioning (-4.93, P < 0.01), while men at increased and high risk according to their BMI did not have significantly different physical functioning from those with BMI in the acceptable risk interval. The physical functioning of men with borderline (-1.63) and high (-3.93) WHR was significantly worse than the functioning of those with low WHR (P < 0.05). Role functioning was significantly higher for males at increased risk by BMI (1.40, P < 0.1) and borderline WC (1.58, P < 0.05) compared with those at acceptable BMI risk and low WC, respectively. After adding controls, role functioning continued to be worse for males with borderline and high WHR, although the differences were not significant. Male social functioning and mental health remained higher in categories of higher BMI and WC, while health perceptions still peaked at increased risk by BMI and borderline WC, although not all of these differences were significant (P < 0.1) after adding controls.

Download:

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

Fig 2. Adjusted differences in means of health-related quality of life dimensions between categories of body mass index (BMI), waist circumference (WC), and waist-to-hip ratio (WHR) by sex.

Notes. Multivariable ordinary least squares regression estimates. For each sex and HRQoL dimension, three regressions were estimated. One included BMI categories, another included WC categories, and the third included WHR categories. References categories are indicated by a dot at 0 with no whiskers that elsewhere indicate 95% CI. All regressions included the covariates in Table 1, with age entered as indicators of 5-year intervals and participant and family history of chronic condition excluded. Estimates in table format in S3 Table in S1 File.

https://doi.org/10.1371/journal.pone.0275798.g002

After controls were added, female social functioning remained significantly worse in categories of higher WC (-3.61, P<0.05) and WHR (-5.48, P < 0.05). The multivariable regression results were robust to adding participant and family history of chronic conditions to the controls and to estimating a restricted model that only controlled for age (S1 Fig and S4 and S5 Tables in S1 File).

Discussion

In a predominantly low-income, rural province of the Philippines, we did not find that people who were at disease risk according to measures of adiposity had lower health-related quality of life, providing further evidence on the nuanced association between HRQoL and adiposity in Asian population [25, 26]. There was some variation in the association between HRQoL and adiposity by sex, adiposity measure, and dimension of HRQoL. For both sexes, we found no evidence that people with a BMI that put them at increased or high risk had lower HRQoL in any dimension. In fact, particularly for males, there was some evidence that people at high BMI risk had better social functioning, consistent with findings in the USA and East Asia [7, 14, 15, 17]. Similar to findings in the USA and China, underweight males had lower physical functioning and health perception [9, 15].

Risks defined by WC and WHR were associated with lower HRQoL more often than was observed for BMI risks. Physical functioning was lower for people at high risk by WC and for males at high risk by WHR. Males at borderline risk and high risk by WHR also had lower role functioning. Females at high risk by WC and WHR had lower social functioning. At least in the sample, males at high risk by WC and WHR had lower health perceptions. Otherwise, there was no evidence of higher adiposity risk being associated with lower HRQoL. In fact, males at high risk by BMI had better social functioning, and males at high risk by WC had better social functioning and mental health,which are similar to previous findings [7, 14, 15, 17].

These findings reveal that Filipinos whose body weight places them at higher disease risk are generally unaware of any health deterioration across a number of dimensions of related quality of life. Consequently, it may be difficult to persuade these people that their health is truly at risk. If they are continuing to function as well as, or even better than, people who are not overweight, then they must accept an abstract argument that they are exposed to higher risk of succumbing to disease in the future without direct experience of health problems giving them reason to believe that the risk will indeed materialize.

It may be important to take this discordance between risk to future health and experience of current health into account in the design of effective prevention programs. For example, information on disease risks associated with overweight could be accompanied by a message that currently feeling in good health does not guarantee the absence of weight-related health problems in the future. Such messages may need to be stronger for less educated individuals who may have less capacity to interpret risks and a greater tendency to behave in response to experiences rather than information.

We found that risks defined by WC and WHR were more often associated with lower HRQoL than were BMI risks. This is consistent with some evidence that high WC and WHR represent greater disease risks than high BMI [30–32]. If the experience of health problems is more closely associated with the types of adiposity that give exposure to the highest objective risks, then there is greater scope to convince people to take steps to reduce their waist size.

We should not, however, exaggerate the heterogeneity by type of adiposity. High WC and WHR was not strongly associated with lower HRQoL in most dimensions. Even people with large waists will need to be convinced that the fact that they are not experiencing worse health than slimmer people does not mean that they can be relaxed with respect to their future health prospects. Further men with high WC experienced better social functioning and mental health, and so it may be particularly difficult to convince males of the risks associated with central obesity. Their body size and shape could well reflect their enjoyment of life’s pleasures. With little or no experience of offsetting health problems, there will be less motivation to abstain from excessive eating and drinking. On the other hand, women with high WHR had worse social functioning, which could potentially motivate them to lose weight. Finally, while disease risks associated with high BMI may be lower than those for high WC and high WHR, high BMI is certainly not without any risk to health. Prevention programs need to work at changing any perceptions that only abdominal obesity is a risk to health.

The main limitation of the study is that it collected data from only one province of the Philippines. The sample was representative of that province. But we cannot claim that the results are representative of the whole country. However, the study setting was reasonably typical of the socioeconomic conditions and rural setting in which much of the lower-income Filipino population lives. Studies of such populations that measure both adiposity and HRQoL are relatively rare.

The evidence we provided of a general lack of association between HRQoL and measures of adiposity is relatively rare in a low-income setting. It can help in motivating and designing health promotion programs that aim to prevent the disease burden that threatens to accumulate from the rising prevalence of obesity in LMIC. Our central finding deserves attention precisely because it is negative. Overweight people at high risk of future disease who are not currently experiencing health problems have less reason to change their behavior. They will need greater external encouragement to do so.

Supporting information