Associations among gender, overweight and obesity, medical comorbidity, and health-related quality of life (HRQoL) were examined in a general population sample of 4,181 women and men aged 18–65 years. Anthropometric measurements and medical comorbidity were assessed as part of a computer-assisted physician interview. HRQoL was assessed with the Physical and Mental Component Summary scales of the Medical Outcomes Study Short Form (SF-36 PCS, MCS). General linear models were used to examine the associations among gender, weight status, medical comorbidity, and HRQoL. Controlling for age, social status, the occurrence of specific medical conditions, and the total number of medical conditions, mild obesity was associated with impairment in physical health functioning, as measured by the PCS, among women, whereas impairment in men's physical health was apparent only for moderate obesity. There was no association between weight status and psycho-social functioning, as measured by the MCS, in women, whereas overweight was associated with better perceived psycho-social functioning in men. The findings are consistent with the hypothesis that women suffer a disproportionately large share of the disease burden of overweight and obesity that is not due solely to differences in medical comorbidity. The possibility that aspects of emotional well-being may mediate the association between obesity and physical health functioning warrants further attention in this regard. The findings also indicate the need to stratify data by gender and to include more sensitive measures of psycho-social functioning in future studies.
It is well known that obesity is associated with a range of chronic medical conditions among both women and men in the community (1,2,3,4). Further, both the number of comorbid conditions and prevalence of specific conditions have been found to increase proportionate to the degree of overweight (2,3,4). In view of this medical comorbidity, it is not surprising that obesity is associated with marked impairment in physical health functioning, as assessed by those items of generic measures of health-related quality of life (HRQoL) tapping physical health status (5,6,7,8,9).
There is also some evidence that the association between obesity and impairment in physical health may be moderated by gender. Specifically, findings from a number of studies suggest that obesity may be associated with greater perceived impairment in physical health functioning in women than in men (5,8,10,11). These findings likely reflect, at least in part, the higher prevalence in overweight women of those conditions most directly associated with limitations in everyday activities, namely, respiratory symptoms and chronic neck, back, and joint pain (3,4,12). Evidence bearing on these issues is limited, however, because in those few studies that have included assessment of both medical illness and HRQoL, in addition to BMI, gender has been included as a covariate rather than as an independent variable (6,13).
In contrast to the clear associations between obesity and impairment in physical functioning observed in epidemiological studies, population-based studies of the association between obesity and impairment in psycho-social functioning have yielded weak and inconsistent findings (5,6,9,11,13,14,15). This inconsistency likely reflects, in part, the different ways in which the term “impairment in psycho-social functioning” has been operationalized (in the present article, we use the term “impairment in psycho-social functioning” to designate impairment in role functioning associated with poor mental health, as measured by the Mental Component Summary scale of the Medical Outcomes Study Short Form (SF-36) and similar measures), although studies of the association between obesity and specific psychopathology have also yielded inconsistent findings (16,17,18,19,20,21,22). The evidence does suggest, however, that psycho-social functioning is impaired in certain subgroups of obese individuals, namely, the severely obese (17), those with chronic medical conditions and/or pain (23), and those with high levels of body image dissatisfaction and/or disordered eating (24). Because the prevalence of body image dissatisfaction and eating disorder psychopathology is higher in women than in men, irrespective of body weight (e.g., ref. 25), it is not surprising that associations between obesity and impairment in psycho-social functioning and other aspects of emotional well-being, where these have been observed, have been found to be more pronounced in women (8,11,16,22,26). In fact, obesity has been found to be associated with better perceived mental health among men in several studies (16,27,28,29,30,31).
The presence of chronic medical conditions appears to be a particularly important mediator of the relationship between obesity and impairment in psycho-social functioning. Thus, Doll and colleagues (6), in a large, general population sample of men and women aged 18–64 years, found that obesity was associated with poorer perceived psycho-social functioning only in the subgroup of obese participants who reported three or more chronic medical conditions. Further, among participants with chronic medical illness, the degree of impairment in psycho-social functioning did not differ by weight status. In a community sample of women and men in Australia (26), obesity was associated with lower levels of general psychological distress after (but not before) controlling for physical health status and this was the case in both women and men. However, medical comorbidity per se was not assessed in this study.
The aim of the current study was to expand current knowledge in this field by examining associations among overweight, medical comorbidity, and HRQoL in a community sample of women and men. Based on a review of the literature, we hypothesized, first, that obesity would be associated with marked impairment in physical health functioning, but little or no impairment in psycho-social functioning; and second, that impairment in both physical health and psycho-social functioning associated with overweight, where this was observed, would be more pronounced for women than for men. We were particularly interested to consider whether the nature of the associations among overweight, medical comorbidity, and HRQoL might differ for women and men, given the paucity of evidence concerning gender differences in this regard.
Methods and Procedures
The study utilized data from the German Health Interview and Examination Survey, a national, population-based study of the prevalence and correlates of physical and mental health impairment in individuals aged 18–79 years conducted between 1997 and 1999 (32). The study design and methods have been detailed in several previous publications (21,32,33,34). In brief, participants were recruited by means of a multi-stage, stratified sampling procedure which drew upon 113 communities throughout Germany. All participants (n = 7,124) completed a core assessment, in which anthropometric data and medical conditions were assessed, and a number of supplemental surveys. BMI (kg/m2) was calculated based on measured height and weight. The response rate of the core assessment was 61.4% and there was no evidence that respondents differed systematically from nonrespondents (32,34).
The core assessment included a brief screening questionnaire for mental disorders comprising 11 questions (32). All participants aged 18–65 who responded positively to at least one of these questions (n = 3,474), along with a random sample of ∼50% of the remaining participants (n = 1,301), were approached to participate in the Mental Health Supplement of the survey, which included the assessment of HRQoL. Findings from the Mental Health Supplement have been outlined previously (33). Participants in the present study were the 4,181 individuals aged 18–65 years (87.6% of those approached) who completed both the core assessment and the Mental Health Supplement. Of these, 49.7% (n = 2,079) were women. As has also been detailed previously (32,34), this two-phase selection procedure resulted in a sample that was representative of the total population of (noninstitutionalized) individuals aged 18–65 years in Germany possessing language skills sufficient to complete the interview assessment.
The mean (s.d.) age of participants was 43.5 (11.6) years. Their mean (s.d.) BMI was 26.3 (4.6) kg/m2. The mean BMI of male participants was higher than that of female participants (women: mean = 25.9, s.d. = 5.2; men: mean = 26.7, s.d. = 3.9; F = 37.25, P < 0.02). Approximately two-thirds of participants (64.1%) were married and approximately half (49.8%) were employed full-time. An additional 9.7% were retired, 7.2% listed home duties as their main activity, 5.7% were full-time students, and 6.5% were unemployed. Most participants (89.0%) had completed ≥9 years of formal education and 17.8% had completed ≥12 years. The majority (57.6%) were classified as being middle-class, whereas 19.1 and 23.3% were classified as being of lower and upper social class, respectively, according to an index combining educational level, employment status, and income (32).
Assessment of weight status
For the purposes of the present study, four weight categories were distinguished, namely: (i) 18.5 ≤ BMI < 25.0 (normal weight); (ii) 25.0 ≤ BMI < 30.0 (overweight); (iii) 30.0 ≤ BMI < 35.0 (mild obesity); and (iv) 35.0 ≤ BMI < 40.0 (moderate obesity) (3). Participants classified as underweight (BMI < 18.5) were excluded because we were interested in the associations between overweight and quality of life. Severely obese participants (BMI ≥40) were excluded because the low prevalence of severe obesity in men precluded analysis of these data by gender (Table 1).
Table 1. Weight status in the total sample (n = 4,181) by gender: percentage (%) of female and male participants in each of six weight (BMI) categories
Assessment of medical conditions
The core assessment included a self-report questionnaire, a standardized, computer-assisted medical interview, anthropometric and blood pressure measurements, and the collection of blood and urine samples. The self-report questionnaire evaluated the subjects' current and past somatic symptoms and complaints, health care utilization, and impairments and disabilities. The computer-assisted interview, which was conducted by a physician, was designed to supplement and refine the self-report data and thereby determine the lifetime and 12-month prevalence of each of the 42 medical conditions assessed. Final diagnoses were made by the physician on the basis of the medical examination, the interview assessment and, where appropriate, results of the laboratory tests. Only 12-month diagnoses were considered in the present analysis.
In order to further simplify the present analysis, medical conditions were grouped into six categories as follows: (i) allergies, including eczema, contact allergy, food allergy, nettle rash, and hay fever; (ii) endocrine and metabolic diseases, including thyroid disease, diabetes, and hyperlipidemia; (iii) cardiovascular disorders, including hypertension, coronary heart disease, and stroke; (iv) gastrointestinal diseases, including gastritis, ulcer, and liver diseases; (v) neurological disorders, comprising Parkinson's disease, multiple sclerosis, epilepsy, encephalitis, and migraine; and (vi) respiratory diseases, including asthma and chronic obstructive pulmonary disease.
Assessment of health-related quality of life
HRQoL was assessed using the (36-item) SF-36 (35), a widely used self-report measure that provides for the assessment of impairment in role functioning associated with physical and mental health problems in eight domains (general health, mental health, physical functioning, vitality, pain, physical role functioning, emotional role functioning, and social functioning), as well as scores on each of two summary scales reflecting overall physical health functioning (Physical Component Summary scale; PCS) and psycho-social functioning (Mental Component Summary scale; MCS). Only the summary scale scores were considered in the present study. Scores on each scale range from 0 to 100, with higher scores indicating better perceived functioning. The SF-36 has robust psychometric properties (35) and it has been translated into several languages including German (32).
Data were stratified by gender in all analysis. Differences in the prevalence of specific medical conditions, and in the total number of conditions, by weight status were examined by means of χ2-tests. Logistic regression was employed to generate odds ratios for the presence of each condition by weight status. Inspection of the relationships between BMI and scores on the PCS and MCS, using scatterplots with 50% Lowess smoothing, did not suggest any departures from linearity. Analysis of covariance (ANCOVA) was used to compare scores on the PCS and MCS between BMI groups and between participants with and without each medical condition. Multiple linear regression analysis with simultaneous variable entry was used to determine the independent contributions of weight status and medical comorbidity to variance in scores on the PCS and MCS. Dummy variables indicating overweight, mild obesity, and moderate obesity were created for this purpose. Age and social status were controlled in all multivariable analysis. To account for the oversampling of screen positives and for differential nonresponse, data were weighted by selection probabilities and demographic characteristics (32). Analyses were conducted using SPSS version 15.0 (SPSS, Chicago, IL) and Stata release 7.0 (StataCorp, College Station, TX). In view of the multiple comparisons required, a conservative significance level of 0.01 was employed for all tests.
The unadjusted prevalence of medical conditions by weight status and gender is shown in Table 2, whereas (adjusted) odds ratios for the occurrence of each condition are given in Table 3. It can be seen that, for both men and women, the prevalence of cardiovascular and endocrine/metabolic disorders increased with increasing BMI, although the prevalence of allergic conditions tended to decrease with increasing BMI in men (but not women). After adjusting for age and social status, however, only the associations with cardiovascular disease (both men and women) and endocrine/metabolic disorders (men, but not women) remained significant.
Table 2. Unadjusted (12-month) prevalence (%) of specific medical conditions by weight category and gender
Table 3. Odds ratios for the occurrence of specific medical conditions by weight category and gender
In women, the proportion of participants with at least one medical condition increased from 58.0% for normal weight to 77.4% for moderate obesity (χ = 32.8, P < 0.01), whereas the proportion of women with ≥2 medical conditions increased from 28.3 to 42.2% (χ = 23.5, P < 0.01). Similarly, in men, the proportion of participants with at least one medical condition increased from 43.5% for normal weight to 85.7% for moderate obesity (χ = 55.6, P < 0.01), whereas the proportion of men with ≥2 medical conditions increased from 16.2 to 50.0% (χ = 45.2, P < 0.01). In both women and men, scores on the PCS, and to a lesser extent on the MCS, tended to decrease (indicating poorer perceived HRQoL) as the total number of medical conditions increased (Figure 1). Details of the bivariate associations between specific medical conditions and PCS and MCS scores are available from the first author upon request.
Associations between weight status and HRQoL are shown in Table 4. It can be seen that scores on the PCS were lower for mildly obese women than for normal- and overweight women, whereas scores on the MCS did not differ by weight status in women. In men, scores on the PCS were lower for moderately obese participants than for normal-weight, overweight, and mildly obese participants, whereas scores on the MCS were higher (indicating better perceived psycho-social functioning) for both overweight and mildly obese participants than for normal-weight participants.
Table 4. Mean (s.e.) scores on the SF-36 Physical (PCS) and Mental (MCS) Component Summary scales by weight category and gender
Results of the regression analyses are given in Table 5. For women, overweight and mild obesity were significantly associated with poorer perceived physical health (as measured by the PCS), after controlling for age, social status, and medical comorbidity. For men, overweight and moderate obesity tended to be associated with poorer physical health (P = 0.05, P = 0.02, respectively), whereas there was no association between mild obesity and scores on the PCS. For both women and men, the number of medical conditions was a better predictor of impairment in physical health functioning than the occurrence of any specific condition.
Table 5. Multiple regression analysis of variables predictive of impairment in physical health functioning, as measured by the SF-36 Physical Component Summary (PCS) and Mental Component Summary (MCS) scales, by gender
For women, overweight tended to be associated with better perceived psycho-social functioning (as measured by the MCS), after controlling for age, social status, and medical comorbidity (P = 0.04), whereas both overweight and mild obesity were associated with (better) psycho-social functioning in men (Table 5). There was no association between mild obesity and scores on the MCS in women, and no association between moderate obesity and scores on the MCS in either women or men. For women, neither the total number of medical conditions nor the occurrence of any specific condition was significantly associated with scores on the MCS, whereas in men, gastrointestinal and neurological conditions were significantly associated with lower MCS scores.
Summary of main findings
We examined the associations among obesity, medical illness, and HRQoL in a community sample of women and men. The main findings were consistent with those of previous studies and with our study hypotheses. First, in both women and men, obesity was associated with increased risk of medical comorbidity, in particular, cardiovascular and endocrine disorders. Second, and also in both women and men, obesity was associated with significant impairment in physical health functioning, as measured by the SF-36 PCS, but little or no impairment in psycho-social functioning, as measured by the SF-36 MCS. Third, gender differences were observed in the associations between weight status and both physical health and psycho-social functioning. Whereas impairment in physical health functioning among women was most pronounced for mild obesity, impairment in men's physical health was most pronounced for moderate obesity; and whereas there was no association between weight status and psycho-social functioning in women, obesity was associated with better perceived psycho-social functioning in men.
Strengths and limitations
Several limitations of the present study should be noted. First, only summary scales of the SF-36 were considered. Although analysis of associations involving individual subscales would have provided a richer data set, it also would have added considerably to the complexity of the analysis and detracted from the intended focus of the study on gender differences in the associations among bodyweight, medical comorbidity, and HRQoL. Second, only categories of medical illness, as opposed to specific medical conditions, were considered. Further, some relevant medical conditions were not assessed. The omission of osteo-arthritic conditions in particular is significant, because these conditions are prevalent in the adult population and known to have a pronounced impact on HRQoL (3,4,12). There was also no assessment of several other variables known to impact HRQoL and likely to be related to both BMI and gender, including physical activity (36), chronic pain (23), body dissatisfaction (24), and psychiatric comorbidity (34). As suggested below, the role of general psychological distress associated with negative body image in women may be a particularly important direction for future research.
Due to the small number of men with severe obesity, analysis of the associations between weight status and HRQoL was confined to overweight, mild obesity, and moderate obesity. However, there is good evidence that severe obesity is associated with marked impairment in both physical and psycho-social functioning in both women and men (2,6,7). Finally, this was a cross-sectional study, so that any inferences concerning the direction of the observed associations are speculative. The primary strengths of the study were the use of a large, representative general population sample, comprehensive assessment of medical morbidity, the use of a standardized measure of HRQoL, determination of weight status on the basis of measured height and weight, and stratification of the analysis by gender. To our knowledge, this is the first study of its kind in the German population and one of only a handful of studies to consider gender differences in the associations among bodyweight, medical comorbidity, and HRQoL. The findings indicate the need to stratify data by gender in future research of this kind.
Obesity and medical comorbidity
The observed associations between overweight and the occurrence of medical comorbidity were generally consistent with findings from previous research, in that the most pronounced effects were for heart disease and endocrine/metabolic disorders, including diabetes, and the prevalence of medical comorbidity was found to increase with increasing levels of overweight (2,3,4). The finding that moderate obesity conveyed a higher risk for cardiovascular disease in men than in women (odds ratio = 8.7 vs. 4.7) is at odds with findings from some previous studies (3,4), although it is apparent that both the direction and magnitude of these associations vary according to the demographic features of the population sampled. The finding that overweight tended to be associated with a lower prevalence of allergic conditions, particularly in men, may reflect less exposure to allergens among obese individuals by virtue of their more restricted activities.
Medical illness and HRQoL
Bivariate analysis of the associations between medical illness and HRQoL indicated that the impact of medical illness on quality of life was primarily in terms of its effects on physical health functioning, as would be expected. Thus, for both women and men and for all conditions assessed, physical health functioning, as measured by the PCS, was poorer among individuals with medical illness than those who did not have medical illness. Nevertheless, psycho-social functioning, as measured by the MCS, was found to be poorer for participants (both women and men) with allergic, gastrointestinal, and neurological conditions, than for those who did not have these conditions, and poorer for women with respiratory conditions than those who did not have such conditions. Further, for both women and men, scores on both the PCS and MCS tended to decrease as the total number of medical conditions increased. These findings indicate that the adverse effects of medical illness on HRQoL, although most pronounced for the physical health domain, extend to impairment in individuals' psycho-social functioning.
Weight status and psycho-social functioning
Also consistent with findings from a number of earlier studies (5,6,9,13,14), overweight was not significantly associated with impairment in psycho-social functioning in either women or men in the present study. The finding that mild to moderate obesity is associated with little or no impairment in psycho-social functioning in unselected samples is well-established and likely reflects the influence of a number of factors, including the fact that generic measures of HRQoL, including the SF-36, may not be sensitive to the types of impairment in emotional well-being likely to be associated with overweight in women, namely, body image dissatisfaction and the effects of obesity-related stigma (8,30,37). One way to address this hypothesis in future research of this kind would be to include an eating-disorders- (e.g., ref. 38) and/or obesity- (e.g., ref. 39) specific measure of HRQoL in addition to a generic measure such as the SF-36.
The finding that obesity in men was associated with higher scores on the MCS, indicating better perceived psycho-social functioning, is also consistent with findings from several previous studies, including the original statement of the “jolly fat” hypothesis (27) and more recent studies employing larger general population samples (16,28,29,30,31). The robustness of this finding may have been obscured somewhat by the failure to stratify data by gender in much of the relevant research (6,13,14,23), although it should be noted that associations between obesity and increased risk of depression and other psychiatric disorders have been observed in both women and men in at least one recent study (20). In the present study, both overweight and mildly obese men reported significantly better psycho-social functioning than normal weight men and this was the case whether or not medical comorbidity was statistically controlled. Although several factors might contribute to an association between overweight and better mental health (27,29), the mechanisms involved remain poorly understood (16,31).
Weight status and physical health functioning
Gender differences were also apparent in the associations between bodyweight and physical health status, as measured by the PCS. In women, both overweight and mild obesity were associated with poorer perceived physical health, relative to normal weight, whereas significant impairment in men's physical health was apparent only with moderate obesity. Given that the prevalence of overweight and mild obesity is far higher than that of moderate and severe obesity, in both women and men, these findings are consistent with the hypothesis that women suffer a disproportionately large share of the disease burden of overweight and obesity (40). Because perceived impairment in role functioning is strongly associated with the use of health services (e.g., ref. 41), this burden has both individual and societal components.
As has been noted, the greater adverse effects of overweight and mild obesity on physical health functioning in women likely reflects, in part, the fact that the prevalence of certain medical conditions is higher in overweight women than in overweight men (3,4,12). Interestingly, however, in the present study, overweight and mild obesity both maintained a strong, independent association with scores on the PCS in women when medical comorbidity was statistically controlled. Hence, other factors appear to be involved. In particular, it has been suggested that higher levels of perceived impairment in physical health associated with overweight in women might reflect, in part, an effect of psychological distress associated with negative body image and/or weight-related stigma on perceptions of general health (4,10,42). This possibility warrants further investigation. Whereas the role of poor physical health in mediating the association between obesity and impairment in psycho-social functioning has received considerable attention (e.g., refs. 6,23,26), very little attention has been given to the converse hypothesis, namely, that impairment in psycho-social functioning may influence the association between obesity and physical health (10,42).
In summary, gender differences were apparent in the associations between overweight and HRQoL. Whereas mild obesity was associated with marked impairment in physical health functioning among women, significant impairment in men's physical health was apparent only for moderate obesity; and whereas there was no association between weight status and psycho-social functioning in women, overweight was associated with better perceived psycho-social functioning in men. The findings are consistent with the hypothesis that women suffer a disproportionately large share of the disease burden of overweight and obesity, due to both higher levels of medical comorbidity and a more pronounced effect of obesity-related conditions on HRQoL. In addition, the findings indicate the need to stratify data by gender, and to include more sensitive measures of emotional well-being, in future studies. The possibility that poor mental health may mediate the association between obesity and aspects of physical health functioning, particularly in women, warrants greater attention.
The National Health Survey (BGS98) was supported by grant 01EH970/8 (German Federal Ministry of Research, Education and Science; BMBF). Dr Mond is supported by an NHMRC Sidney Sax Fellowship. Thanks to Hans-Ulrich Wittchen, Heribert Stolzenberg, PhD, and Bärbel-Maria Kurth for their assistance with the BGS Public use databases and to Bryan Rodgers for helpful comments on an earlier version of this article.