SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Disclosure
  8. REFERENCES

Research is needed to better elucidate the relationship between obesity and depression, which has been most consistently demonstrated for women, but not for men. We examined exclusively a population-based sample of US women who participated in the 2005 or 2006 National Health and Nutritional Examination Survey. Current depression was defined as having a score of ≥10 (a conventional threshold for moderate symptoms of depression) or meeting the Diagnostic and Statistical Manual of Mental Disorders, 4th edition (DSM-IV) diagnostic criteria for major depression on the nine-item Patient Health Questionnaire. Weight and height were measured and BMI was calculated. Waist circumference, a clinical measure of abdominal obesity, was also measured. BMI was positively associated with the probability of moderate/severe depressive symptoms (r = 0.49, P = 0.03) and major depression (r = 0.72, P < 0.0001). The probability curves increased progressively, beginning at BMI of 30. Degree of obesity was an independent risk factor for depression even within the obese population, and women in obesity class 3 (BMI ≥40) were at particular risk (odds ratio (OR) = 4.91, 95% confidence interval (CI): 1.17–20.57), compared to those in obesity class 1 (BMI 30 to <35). Abdominal obesity was positively associated with depressive symptoms, but not major depression, independent of general obesity (BMI). In addition to severe obesity, compromised physical health status, young or middle-aged adulthood, low income, and relatively high education were also independently associated with greater odds of depressive symptoms among obese women. These characteristics may identify specific at-risk subgroups of obese women in which hypothesized causal pathways and effective preventive and therapeutic interventions can be profitably investigated.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Disclosure
  8. REFERENCES

The literature abounds with clear and consistent evidence of the epidemiology of obesity and its detrimental effects on physical health and metabolism (1,2,3). However, much less is known about obesity and risk for psychopathology, although research has been growing over the most recent decades. Depression is one of the most prevalent psychiatric disorders and, like obesity, contributes substantially to global morbidity and mortality (4,5,6,7). Not only do depression and obesity share common health complications such as cardiovascular disease and diabetes (2,6), there is evidence of synergistic negative effects on health and treatment response when the two disorders coexist (8,9,10,11).

Studies of depressive outcomes (i.e., depressive symptomatology or clinical depression) account for much of the growing literature on the relation between obesity and psychological morbidity. Earlier studies yielded largely discrepant findings (probably reflecting methodological shortcomings typical of early work in the field (12)), whereas some consistency has emerged from more recent studies of the relationship between obesity and depression (13,14). Nevertheless, the nature of the relationship between the two disorders and the underlying mechanism remain poorly understood. There is increasing recognition that obesity is strikingly heterogeneous with respect to depression psychopathology, as is the case for effects of excess weight on medical problems (12,15).

Gender is one of the few factors that has been consistently shown to modify the association between obesity and depression, with most studies demonstrating a positive relationship for women and a null or inverse relationship for men (12,15). This suggests that factors linking obesity and depression may be gender specific and that differential mechanisms may be involved. Further investigations may generate useful insights into the relationship between obesity and depression by focusing on women, separate from men, given their increased risk for the comorbidity.

This study examines data from a population-based sample of women in the National Health and Nutritional Examination Survey (NHANES) of 2005 and 2006. We test the hypotheses that the association between relative body weight (measured by BMI) and depression is dose-dependent and that abdominal obesity (measured by waist circumference) is associated with depression independent of general obesity (BMI). Further, we investigate which obese women are more likely to be depressed by evaluating an array of purported risk factors found in the published literature (14,16), including demographic characteristics, socioeconomic status, health status, and behavioral factors.

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Disclosure
  8. REFERENCES

Data source

The NHANES is conducted by the National Center for Health Statistics in the 50 states and the District of Columbia of the United States. It was changed from a periodic annual survey to a continuous annual survey in 1999, and the continuous NHANES survey data have been released in 2-year increments for public use. The NHANES uses a stratified, multistage probability sample design and weighting methodology that allows unbiased national estimates to be produced for the civilian, noninstitutionalized US population. NHANES sample weights adjust for unequal probabilities of selection, nonresponse, and planned oversampling (of young children, the elderly, persons with low-income, and ethnic minorities). Detailed information about this survey and public use data files can be found at http:www.cdc.govnchsnhanes.htm.

Study population

A total of 10,348 individuals of all ages were included in NHANES 2005–2006. The overall response rates were 80.5% for the household interviews and 77.3% for the examinations. In this study, we focused on the 2,489 women ≥20 years who had been randomly selected for the depression screening interview. A total of 287 respondents (11.5%) were excluded due to missing data on the depression screener questionnaire (n = 268) or missing data on height or weight (n = 46). We also excluded 38 people who reported taking antidepressants or other psychotherapeutic agents and 307 pregnant women. The final sample size for this study was 1,857. Compared with study subjects, a larger percentage of persons excluded from the study had a less than high school education (24% vs. 16%; χ2 test: P < 0.001) whereas a smaller percentage were white (63% vs. 73%; P < 0.001). In addition, those excluded were younger (mean (s.d.): 44.8 (1.2) vs. 47.8 (0.8); Student's t-test: P < 0.001). There were no differences by BMI category, marital status, and income.

Measurements

Depression outcomes. In NHANES 2005–2006, depression was assessed using the nine-item Patient Health Questionnaire (PHQ-9). The questionnaire was administered by trained interviewers using Computer-Assisted Personal Interview technology during Mobile Examination Center visits. The PHQ-9 refers to the previous 2-week interval and consists of nine items of depression symptoms and one follow-up question on functional impairment (17). The PHQ-9 is both a measure of depressive symptomatology and a diagnostic instrument for the Diagnostic and Statistical Manual of Mental Disorders, 4th edition (DSM-IV) depressive disorders (18). The symptom score is calculated as the total of all nine items (possible range 0–27), and a score of ≥10 represents a moderate level of depressive symptoms. Consistent with the DSM-IV criteria, a diagnosis of major depression requires a positive response to one of the two core symptoms (depressed mood or anhedonia) and a total of five or more positive symptoms. Individuals with major depression represent a subgroup of those with PHQ-9 scores ≥10.

Relative body weight. During the Mobile Examination Center physical examination, height and weight were measured by trained technicians using standardized protocols and calibrated equipment. BMI was calculated as weight in kilograms divided by the square of height in meters and was rounded to the nearest tenth. Continuous BMI values ranged from 13.4 to 76.1. Analyses of BMI as a continuous variable investigated associations with depression over the entire spectrum of relative body weight. BMI categories were defined using widely accepted cutpoints, i.e., BMI <18.5 for underweight, BMI 18.5–24.9 for normal weight, 25.0–29.9 for overweight, and BMI ≥30.0 for obesity (19). Also, the obese category was subdivided into three levels of severity: obese class 1 (BMI 30.0–34.9), obese class 2 (35.0–39.9), and obese class 3 (≥40.0).

Waist circumference. Waist circumference, an accepted clinical indicator of abdominal obesity, was measured in a horizontal plane around the abdomen at the level of the uppermost lateral border of the right ilium. Measurements were recorded to the nearest 0.1 cm.

Potential risk factors. We examined factors in four domains: demographics (age, race, marital status), socioeconomic status (education and income), health status (self-rated health and chronic conditions), and behavioral factors (physical activity, cigarette smoking, and alcohol use).

Demographic and socioeconomic characteristics: We classified age into three categories: 20–44, 45–64, and ≥65 and race/ethnicity into four categories: non-Hispanic white, non-Hispanic black, Mexican American, and other race. This latter category included participants reporting multiple racial/ethnic identities. We categorized participants by marital status (never married, married or live with partner, or divorced/widowed/separated), annual family income (<20,000, 20,000–44,999, 45,000–74,999, or ≥75,000), and highest achieved education level (<high school, high school/general equivalency diploma, or >high school).

Health status: Self-rated physical health was classified as excellent/very good, good, or fair/poor. Chronic conditions referred to the total number of physical health conditions that a respondent had ever been told by a doctor or other health profession that they had. The list of conditions included asthma, diabetes, arthritis, coronary heart disease, angina, myocardial infarction, congestive heart failure, stroke, emphysema, chronic bronchitis, any kind of cancer, any thyroid problem, or any kind of liver condition. The number of chronic conditions was categorized as 0, 1–2, and 3+ conditions.

Behavioral factors: The physical activity was recorded with the accelerometer ActiGraph AM-7164 (Manufacturing Technology, Fort Walton Beach, Florida) for 7 days. We restricted the analyses to those respondents with valid and reliable accelerometer data, i.e., subjects who had at least 4 days of complete monitoring. A day was considered complete if the wearing time was at least 10 h a day. No wear time was defined as continuous zero counts for ≥20 min. Intensity readings were summed over each 1-min epoch. The activity levels were based on number of counts per minute: moderate-intensity (1,952–5,724) and vigorous-intensity physical activity (≥5,725) (ref. 20). Time associated with light activity and sleep was excluded. We further computed metabolic equivalent score minutes per week using 4.0 metabolic equivalents/min for moderate-intensity physical activity and 8.0 metabolic equivalents/min for vigorous-intensity physical activity (21). Tertiles of total metabolic equivalent scores were created. We categorized smoking status as nonsmoker, current smoker, or past smoker; alcohol drinking as never drinker, former, moderator drinker (≤1 drink/day), or heavy drinker (>1 drink/day).

Statistical analysis

All analyses were conducted in SAS Enterprise Guide 4.1 (SAS Institute, Cary, NC) and took account of the complex sample design and sample weights of the NHANES. The prevalence estimates for moderate or severe depressive symptoms (PHQ-9 ≥10) and DSM-IV defined major depression were age standardized to the 2000 US population (22). The age-adjusted odds ratios (ORs) were obtained to test associations of depressive symptoms and major depression with standardized continuous BMI, BMI classification, standardized waist circumference, and waist circumference quartiles separately. BMI-independent associations of depressive symptoms and major depression with waist circumference were also examined. A multivariable logistic regression analysis was performed to detect independent associations of depressive symptoms with the severity of obesity, abdominal obesity, and each potential risk factor among obese women (BMI ≥30). The number of obese women with meeting DSM-IV major depression criteria was too small for meaningful multivariable analyses.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Disclosure
  8. REFERENCES

The study sample included 1,857 women whose average age was 47.8 ± 0.8 years. The demographic and socioeconomic characteristics are shown in Table 1. In addition, the sample was 35.7% normal weight, 2.6% underweight, 25.8% overweight, and 35.9% obese (17.3% in obese class 1, 10.7% in obese class 2, and 7.9% in obese class 3).

Table 1.  Demographic and socioeconomic characteristics of the study population (n = 1,857), National Health and Nutrition Examination Survey, 2005–2006
inline image

The relationship between BMI and the probability of current depression is shown in Figure 1. BMI was positively associated with the probability of moderate/severe depressive symptoms (a score ≥10) (r = 0.49, P = 0.03) and major depression (r = 0.72, P < 0.0001) on the PHQ-9. The probability curves represented a continuum that began to increase progressively for both depression outcomes at BMI of 30.

image

Figure 1. The probability of current depression by BMI. Current depression is defined as having a score of ≥10 (a conventional threshold for moderate symptoms of depression) or meeting the Diagnostic and Statistical Manual of Mental Disorders, 4th edition (DSM-IV) diagnostic criteria for major depression on the nine-item Patient Health Questionnaire (PHQ-9). The full lines are the smoothed-out probabilities, the fragmented lines are the raw prevalence numbers, and the dotted lines are ± s.d.

Download figure to PowerPoint

The relationship between waist circumference and the probability of current depression is shown in Figure 2. Waist circumference was positively associated with the probability of having at least moderate depressive symptoms (r = 0.5, P = 0.004); however, its association with the probability of major depression did not reach statistical significance (r = 0.35, P = 0.1).

image

Figure 2. The probability of current depression by waist circumference. Current depression is defined as having a score of ≥10 (a conventional threshold for moderate symptoms of depression) or meeting the Diagnostic and Statistical Manual of Mental Disorders, 4th edition (DSM-IV) diagnostic criteria for major depression on the nine-item Patient Health Questionnaire (PHQ-9). The full lines are the smoothed-out probabilities, the fragmented lines are the raw prevalence numbers, and the dotted lines are ± s.d.

Download figure to PowerPoint

The age-standardized prevalence of moderate/severe depressive symptoms and major depression according to continuous and categorical measures of BMI and associated ORs are shown in Table 2. The age-adjusted OR was 1.01 (95% confidence interval (CI): 1.00–1.02) per s.d. increase in BMI for moderate/severe depression symptoms and 1.02 (1.01–1.03) for major depression. Compared to normal weight, obesity (BMI ≥30) was associated with 2.18 (95% CI: 1.30–3.68) times the odds of satisfying the DSM-IV major depression diagnosis and with modestly increased odds of having moderate/severe depressive symptoms (OR = 1.45; 95% CI: 0.87–2.40). The prevalence of both depression outcomes was the highest in women with class 3 obesity—7.0% for having major depression and an additional 7.5% for having moderate/severe depressive symptoms (but not meeting the diagnostic criteria for major depression). These prevalence rates reflected a relative OR of 5.25 (95% CI: 2.30–11.97) for major depression and 3.44 (95% CI: 1.56–7.56) for moderate/severe depressive symptoms, compared to normal weight women.

Table 2.  Association of BMI with current depression, National Health and Nutrition Examination Survey, 2005–2006
inline image

The age-standardized prevalence of moderate/severe depressive symptoms and major depression by waist circumference and associated ORs are shown in Table 3. Greater waist circumference was associated higher likelihood of having moderate/severe depressive symptoms (OR per s.d. increase in waist circumference = 1.02; 95% CI: 1.01–1.03) and major depression (OR per s.d. increase in waist circumference = 1.02; 95% CI: 1.01–1.04), although these associations were attenuated after adjusting for BMI. In addition, we observed a trend of higher prevalence and higher ORs for both depression outcomes with increasing quartiles of waist circumference. Most notably, among women in the fourth quartile of waist circumference, the prevalence of major depression was 3.5% and an additional 4.9% had moderate/severe symptoms of depression. These women had 2–3 times the odds of either depression outcome, compared to women in the lowest quartile. However, the OR for major depression was no longer statistically significant after adjusting for BMI.

Table 3.  Association of waist circumference with current depression, National Health and Nutrition Examination Survey, 2005–2006
inline image

The results of the multivariable logistic regression analysis of association with moderate/severe depressive symptoms on PHQ-9 are shown in Table 4 for women with BMI ≥30. After controlling for all measured potential risk factors, the effect of obesity class 3 persisted but the effect of abdominal obesity was no longer statistically significant. Increased odds of having moderate/severe depressive symptoms were associated with younger age, lower income, more chronic conditions, and fair/poor self-rated health status. In addition, compared to obese women with less than high school education, the odds of moderate/severe depressive symptoms were greater in those with a high school education but not in those with at least some college education. The ORs were not statistically significant for race, marital status, cigarette smoking, and alcohol use.

Table 4.  Independent associations of moderate/severe depressive symptoms (PHQ-9 ≥10) with severity of obesity (by BMI), abdominal obesity (by waist circumference), and potential risk factors, National Health and Nutrition Examination Survey, 2005–2006
inline image

Physical activity was not included because 35% of the study sample did not have accelerometer data or had data that were deemed unreliable or invalid. However, we conducted a sensitivity analysis by performing the same multivariable logistic regression analysis in the subset of women with reliable and valid accelerometer data. Physical activity was not significantly associated with the odds of moderate/severe depressive symptoms in obese women (P = 0.4 for tertile 1 vs. tertile 3 and P = 0.74 for tertile 2 vs. tertile 3). The results for the other variables were consistent with those found in the full sample without adjusting for physical activity and are not reported.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Disclosure
  8. REFERENCES

In their review of the literature on psychological correlates of obesity, Friedman and Brownell (12) concluded that there are probably important but inconsistent relations between obesity and psychopathology, which was consistent with the existing literature demonstrating striking heterogeneity of obesity with respect to its etiology, effects on medical problems, and response to various treatments (15). They also proposed that, to truly understand the psychological consequences of obesity, the field would need to move beyond comparing obese and nonobese individuals (“first generation” of research) and instead focus on identifying factors likely to place an obese individual at risk for psychological problems—a risk factor approach (“second generation” of research). Subsequently, investigations of the underlying mechanisms and effective preventive and therapeutic interventions (“third generation” of research) would be most fruitful by targeting the at-risk populations.

Female gender is the only factor in studies from the United States that has been consistently associated with an increased risk of depression among obese individuals (12,13,14). With rare exceptions (e.g., the study by Simon et al. (23)), past studies used samples consisting of both men and women, even though some studies analyzed them separately (24,25,26,27). Research is needed to better elucidate the relationship between obesity and depression, which has been most consistently demonstrated for women, but not for men. In this study, we focused exclusively on a nationally representative sample of women in the United States.

Confirming our hypothesis that the association between obesity and depression among women would be dose-dependent, we found that the probability of moderate/severe depressive symptoms and major depression, assessed with the PHQ-9, increased progressively, beginning at BMI of 30. With regard to our second hypothesis, we found that abdominal obesity was not associated with major depression but was positively associated with moderate/severe depressive symptoms, independent of BMI; the latter became null when assessed in obese women only and adjusted for other covariates. Past studies of the association between abdominal obesity and depression have yielded mixed results. Some studies suggested that abdominal obesity did not confer additional risk for depression over and above that of general obesity (28,29), whereas others reported that depression might be associated with abdominal obesity after adjusting for BMI (30). Additionally, it has been suggested that the association between depression and abdominal fat deposition may be mediated by hypercortisolemia (31,32). Insights are to be gained from the emerging research on depression in individuals with metabolic syndrome.

Consistent with previous studies (33), we observed that, compared to normal weight women, those with severe obesity had a significantly greater OR for the risk of either depression outcome than those with lesser degrees of obesity. Moreover, we detected a strong association between severe obesity and moderate/severe depressive symptoms among obese women after adjusting for covariates. This finding suggests that degree of obesity is an independent risk factor for depression even within the obese population. Severely obese women may represent a distinct subgroup of the obese population in which mechanisms linking obesity to depression can be profitably investigated. The association between severe obesity and depression has clinical importance in that women seeking treatment for severe obesity may need to be routinely screened for depression and appropriate treatment should be given if diagnosed.

Nevertheless, excess weight occurs on a continuum. Obesity, not just severe obesity, is associated with an increased risk for a variety of chronic diseases (1,2), most of which are associated with depression (6), and depression can precipitate chronic disease due to diminished treatment adherence and/or response (10,11). Our results suggest that the rates of moderate/severe depressive symptoms are higher among obese women with increasing number of chronic conditions and poorer self-rated health status, which is consistent with the results of a recent study using data from the 2006 Behavioral Risk Factor Surveillance System (27). In addition, as noted above, there may be distinct but overlapping associations of general obesity and abdominal obesity with depression. Similar to obesity, depression may manifest with varying severity and in different forms. Depressed mood, without reaching the level of major depression, are associated with impaired physical and social functioning (34,35), and chronic disease can exacerbate symptoms of depression (6). Also, the association of depression with obesity may vary by subtype; for example, major depression with atypical symptoms in women may be significantly more likely to be associated with obesity than major depression with typical symptoms (36). These data suggest that not only obesity and depression are each polygenic, heterogeneous conditions but the obesity–depression association is likely complex and multiple covariations may exist, as posited by Faith et al. (37). Further investigation of specific subtypes of depression in obesity and, vice versa, specific subtypes of obesity in depression is greatly needed, and it will undoubtedly shed light on possible pathophysiologies of the obesity–depression comorbidity.

We found that obese women under age 65, with low income, or with a high school education were more likely to experience moderate/severe depressive symptoms than their respective counterparts who were older, richer, or had less education. Heo et al., using data from the 2001 Behavioral Risk Factor Surveillance System, previously reported that young obese women, particularly Hispanics, are at greater risk for depressive mood than older obese women after controlling for income, education, employment status, and marital status (26). In the 2001 Behavioral Risk Factor Surveillance System, BMI was calculated from self-report height and weight and depression was measured using a single question about how many days in the past 30 days that the participant reportedly felt sad, blue, or depressed. These notable measurement differences preclude direct comparisons with this study. Nevertheless, findings from these two studies indicate that the strength of association with depression varies in subgroups of obese women. Indeed, systematic reviews of the literature (14,16) suggest that in addition to gender, age and socioeconomic status may also moderate the relation between obesity and depression. Together with degree of obesity and physical health status, these population characteristics help explain the heterogeneity in risk for depression in the obese.

The present findings should be interpreted in light of the limitations of this study. First, the cross-sectional design clearly precludes us from establishing the direction or causality of the relationship between obesity and depression or the underlying mechanism. These questions have to be addressed in longitudinal and intervention studies, which represent the third generation of research proposed by Friedman and Brownell (12). Second, depression was assessed with a score-based symptom scale rather than the gold standard of a structured clinical interview. However, it is prohibitively expensive to conduct structured clinical interviews to ascertain diagnoses in large community surveys. The PHQ-9 has been validated against structured diagnostic interviews by clinicians in the general population (38), as well as in general medical outpatient (17,39) and in-patient samples (40). Third, our analyses do not consider other variables that have been suggested as possible risk factors such as repeated dieting, lifetime depression, and anxiety (14,16) but that were not measured in the NHANES 2005–2006.

In conclusion, the risk of depression begins to increase noticeably among US women with BMI of ≥30. Abdominal obesity may confer additional risk for depression above and beyond that of BMI. Among obese women, severe obesity, compromised physical health status, young or middle-aged adulthood, low income, and relatively high education represent potentially important risk factors for depression. These characteristics identify specific at-risk subgroups in which future investigations may need to prioritize to elucidate hypothesized causal pathways and evaluate treatment effectiveness.

Disclosure

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Disclosure
  8. REFERENCES

The authors declared no conflict of interest.

REFERENCES

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Disclosure
  8. REFERENCES