Of the 8 000 potential participants who were mailed invitation letters, 442 were found to be ineligible (had since died, moved away, or disenrolled from the health plan). Of the remaining 7 558 eligible women, 865 could not be reached by telephone and 2 033 declined to participate, leaving 4 660 participants (62% of those eligible). Nineteen participants had missing data on one or more key variables, so 4 641 were included in analyses presented here. Participation varied significantly across sampling strata (63% among those who reported BMI of 30 kg/m2 or more on the breast cancer screening questionnaire vs. 59% among those reporting BMI less than 30 kg/m2 vs. 34% among those declining to participate in breast cancer screening, (p <0.001).
Among the 4 641 respondents included in the analyses, the mean age was 52 years; 66% were married; 58% were college graduates; and 9.7% were current smokers (). A total of 3.7% reported Hispanic ethnic origin; and for race categories, 2.5% reported American Indian/Alaskan Native, 8.3% Asian/Pacific Islander (7.4% Asian and 0.9% Native Hawaiian or other Pacific Islander), 7.0% African American, and reported 82.3% White race. The average BMI was 28.5 kg/m2, with 27% of women classified as overweight (BMI 25.0-29.9 kg/m2) and 34% classified as obese (BMI ≥30 kg/m2). Most of the women (64%) had no significant depressive symptoms, while 23% had mild depressive symptoms (PHQ score 5.0-9.9) and 12.9% had moderate-to-severe depressive symptoms (PHQ ≥10). Consistent with the usual demographics of this health plan’s members, 80% of the women in this sample were currently employed full time or part time. Another 2.3% were unemployed, and 3.0% were disabled or on sick leave. Among employed women, nearly one-quarter (23%) of the sample had been absent from work during the past month, and 9% reported work limitations occurring all or most of the time they were at work. Finally, the prevalence of self-reported disability varied by domain, with a low of 7% reporting any limitations in self-care activities and a high of 46% reporting any limitations in household activities (see ).
Population characteristics – a representative sample of 4641 female health plan enrollees aged 40-65 years
In unadjusted regression models, higher BMI and greater depressive symptoms were both strongly associated with greater self-reported disability in every domain of the WHODAS (P<0.001 for cognitive, mobility, self-care, social, role, household, and work limitation domains; results not shown). After adjusting for age, race, college education, current smoking status, and Rx Risk score (), women with PHQ scores indicating mild (5.0-9.9) and moderate-to-severe depressives symptoms (significantly greater odds of reporting any cognitive, mobility, self-care, social, role, household, and work limitation domains compared to women with no depressive symptoms (PHQ <5). The magnitude of the association between PHQ scores and disability outcomes was only slightly reduced after adjusted for BMI categories (1st and 3rd bars). After adjusting for age, race, college education, current smoker, and Rx Risk score (), women with a BMI in the overweight range did not have significantly greater odds of reporting disability in any of the domains we examined compared to women with a normal BMI; however, women with a BMI in the obese range did have significantly greater odds of mobility, role, household, and work limitations than women with a normal BMI (P<0.05; ). In this case however, the magnitude of the association between BMI and disability outcomes was significantly reduced after adjusted for PHQ scores (1st and 3rd bars). Finally, we observed no statistically significant interactions between depressive symptoms and BMI categories in any of our analyses of disability outcomes (P>0.05 for all comparisons).
Multivariable Adjusted* Odds Ratios of Having Any Limitations in Seven Domains of Disability across Levels of Depressive Symptoms
Multivariable Adjusted* Odds Ratios of Having Any Limitations in Seven Domains of Disability across Categories of Body Mass Index
When examining the relationships between obesity and depression with disability, there were notable differences in the patterns of the association. Obese women reported relatively large limitations in mobility, role, household, and work functioning (). On the other hand, women with severe depression reported relatively large limitations in cognitive and social functioning in addition to large impacts on household and work functioning (). Notably, the magnitude of the associations between severe depression and disability (OR range: 3.7-12.0) were much greater than the associations between obesity and disability (OR range: 1.1-2.2).
Next, we examined the relationships between BMI categories, depressive symptoms, current employment status and absenteeism (). Once again, in unadjusted regression models, higher BMI and greater depressive symptoms were both strongly associated with a higher prevalence of disabled work status (P<0.05). Similarly, among employed women, higher BMI and depressive symptoms were associated with the number of days absent from work (P<0.05). Overall, 22% of overweight women and 31% of obese women reported any absenteeism in the past month, compared with 17% of normal weight women. Furthermore, 32% of women with mild depressive symptoms and 47% of women with moderate-to-severe depressive symptoms reported any absenteeism in the past month, compared to 16% of women with no depressive symptoms. After adjusting for BMI, age, race, college education, current smoker, and Rx Risk score (data not shown), women with PHQ scores in the mild or moderate-to-severe depressive symptom range had significantly greater odds of reporting any disability leave and absenteeism compared to women with no depressive symptoms (PHQ <5). After adjusting for PHQ, age, race, college education, current smoker, and Rx Risk score (data not shown), overweight and obese women had significantly greater odds of reporting any days absent from work compared to normal weight women; however, overweight and obese status were no longer significantly associated with unemployment/disability/sick leave.
Trends in unadjusted work-related unemployment, sick leave, disability, absenteeism, and work limitations among employed women, by BMI and PHQ categories*
Population-attributable calculations for absenteeism
We also estimated the population-level impact of excess body weight and depressive symptoms among employed women in this cohort (). After adjusting for PHQ, age, and race, the 2 166 overweight and obese employed women in this cohort averaged 0.48 more days of absenteeism per month than normal weight employed women. Thus, excess body weight was associated with 5 760 days of absenteeism per 1 000 person-years. Women with mild, moderate, and severe depressive symptoms averaged 1.52 more days of absenteeism per month than non-depressed women. Thus, depressive symptoms were associated with 18 240 days of absenteeism per 1 000 person-years. Our cohort of 4 641 women reported a total of 6 990 days absent from work per month, which annualizes to approximately 83 000 absentee days per year. Based on our calculations, roughly 35 000 (42%) of days absent from work may be attributed to excess weight and depressive symptoms.
Population-level impact of excess body weight and depressive symptoms on absenteeism in our study sample
Finally, to further illustrate the impact of impact of increasing BMI and depressive symptoms on presenteeism (work performance) among employed women, presents the associations between BMI categories, PHQ categories, and the proportions of employed women reporting mild, moderate, and severe difficulty with day-to-day work activities. Once again, the larger magnitude of the relationship between increasing depressive symptoms and disability is clearly illustrated.
Proportions of employed women (80% of total sample) reporting mild, moderate, and severe difficulty carrying out their day-to-day work, by BMI category and PHQ score*
Obesity and depression are both very common health conditions among middle age women. In this large, representative cohort from a single health plan in Washington State, we found that obesity was a significant predictor of greater disability in the areas of mobility, role functioning, household functioning, and work, but not the cognition, self-care, and social domains. Notably, women who were overweight, but not obese, were no more likely to report disability than normal weight women after adjusting for depression and other covariates. We also found that depressive symptoms were a strong independent predictor of worse disability in all 7 domains that were examined, and that both mild and moderate levels of depression were associated with worse levels of disability. The associations between obesity or depression and disability persisted after adjusting for age, race, education, smoking status, and a pharmacy-based risk adjuster. The strong independent relationships between these health conditions and disability suggest that obesity and depression should be routinely screened and treated in this population.
Our analyses also indicate that obesity and depression have differential effects on disability. Obesity appears to have its strongest effects on mobility limitations, which may also be reflected in measures of household and work disability. For example, the area of household limitations that was most commonly endorsed was “getting household work done quickly as needed,” an area in which mobility limitations may reduce the speed at which work can be performed. On the other hand, depressive symptoms were most strongly associated with cognitive and social limitations and least strongly associated with mobility problems. It is notable, however, that the magnitude of the effect of depressive symptoms on mobility appeared to be at least as strong as the effect of obesity on mobility. Women reporting depressive symptoms were much more likely, however, to endorse “difficulties in concentrating,” “remembering to do important things,” and “starting or maintaining conversations.”
Notably, we found that there were no significant interactions between obesity and depression in any of the disability domains examined in this cohort. This finding suggests that the effects of obesity and depression on disability are independent and do not require the presence of the other health condition to be manifested. In other words, the independent effects of obesity and depression on disability are likely to be additive, rather than multiplicative, in this population. However, we also observed that much of the effect of obesity on disability was explained by co-morbid depression (the odds ratios were significantly reduced in when we adjusted for PHQ), but that obesity explained very little of the effect of depression on disability (the odds ratios were not impacted much in when we adjusted for BMI). The main clinical implication of this finding is that each of these conditions needs to be treated and reduced in this population to most effectively reduce the prevalence of disability, but that depression has the larger independent effect.
Another important finding in our study was that obesity and depression were each associated with significant absenteeism. The overweight and obese women in this study population were absent one-half day more per month than normal weight women, and women with mild-or-greater depressive symptoms were absent one-and-a-half days more per month than women without depressive symptoms. Thus, among middle-aged women, excess body weight is associated with 5 760 days of absenteeism per 1 000 person-years, and depression is associated with 18 240 days of absenteeism per 1 000 person-years. To further illustrate the effect, in this modest-sized cohort of 4 641 women, these two health conditions were associated with over 35 000 days of absenteeism annually, representing roughly 42% of all absentee days experienced each year. Absenteeism of this magnitude is likely to have significant economic consequences for employers.(26
) Indeed, prior research by Cawley and colleagues estimated the annual cost of obesity-related absenteeism at $4.3 billion annually in the United States (2 004 dollars).(29
) Given that both obesity and depression are potentially treatable (and reversible) health conditions, we believe these findings should be a call to action among employers to place greater emphasis on prevention, screening, and treatment of obesity and depression in their wellness programs.
Several limitations of this study should be noted. Height and weight were assessed by self-report. Previous methodological research suggests that self-reported height and weight are highly correlated with direct physical measurements, but self-report tends to consistently underestimate weight and overestimate height. In a sub sample of participants in this study, weight was underreported by less than 1.5 kg on average, and correlation between self-reported and measured weight was 0.98. We would not expect that a general tendency to underreport weight would affect the association between obesity and depression. Previous studies demonstrating an association between obesity and depression have included both self-report and physical measurements of height and weight. Second, depression was assessed by non-clinicians using a structured research interview rather than by a full clinical assessment. Compared to the gold standard of a clinician assessment, the validity of the PHQ is as good as or better than alternative measures used in previous studies.(22
) This cross-sectional study cannot establish the direction of the causal relationship between obesity, depression, and disability or the mechanisms underlying those relationships. The link between obesity, depression, and disability is likely mediated through numerous mechanisms, including the related chronic health condition such as low back pain, osteoarthritis, cardiovascular disease, and diabetes.(32
) One possible limitation of this study is the potential for overlap on a few of the individual items on the PHQ and WHODAS (e.g., items related to difficulty concentrating and moving slowly); however, the overwhelming majority of items do not overlap and, overall, we believe that only a very small amount of the association between disability and depression seen in this study is likely to be attributable to overlap in symptoms. Another limitation is our use of disability measures that were drawn from a longer self-report survey, the WHODAS II. While the WHODAS II has been previously validated in the full version, our modified (shorter) version may misclassify women with disabilities in this modified format. Finally our sample only included middle-age women, so we are unable to assess these relationships among other age ranges and among men.
In conclusion, in this large, population-based sample of middle aged women, we find strong evidence for independent effects of obesity and depression on disability. Disability has been linked to greater work absenteeism, and in our multidimensional assessment of disability, we find that the primarily physical impacts of obesity are associated with less absenteeism than the more globally disabling aspects of depression. However, the greater prevalence of overweight and obesity translate into significant population-level impacts. We believe that these findings have important implications for employers and health insurance organizations as they consider the optimal design of employee wellness programs.