| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
ORIGINAL ARTICLES |
From the Departments of Psychology, University of Massachusetts Boston (A.S.C.), Boston, MA, and Yale University (C.W.B., K.D.B.) New Haven, CT.
Address reprint requests to: Alice S. Carter, Department of Psychology, University of Massachusetts Boston, 100 Morrissey Blvd., Boston, MA. Email: alice.carter{at}umb.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
METHODS: As part of a larger study, women were recruited during pregnancy and followed prospectively to 14 months postpartum. Measures included self-reported prepregnancy and 4-month postpartum BMI as well as pregnancy, 4-month, and 14-month postpartum eating attitudes (EAT), depressive symptoms (CES-D), and anxiety symptoms (STAI).
RESULTS: During pregnancy, symptoms of depression or anxiety were not significantly correlated with concurrent eating attitudes or measures of BMI. However, at 14 months postpartum, measures of eating attitudes and both depression and anxiety symptoms were associated. Measures of BMI were associated with depressive and anxiety symptoms at both 4 and 14 months postpartum. Four-month eating attitudes and BMI predicted 14-month postpartum depressive symptoms, beyond pregnancy, and 4-month postpartum measures of affective symptoms. Results suggested that overweight women were at risk for elevated anxiety at 4 months and depressive symptoms at both 4 and 14 months postpartum.
CONCLUSIONS: These results provide evidence for a significant, albeit moderate, relationship between BMI, eating attitudes, and symptoms of depression and anxiety in the postpartum period that are not present during pregnancy.
Key Words: BMI • eating • depression • anxiety • pregnancy • postpartum
Abbreviations: BMI = body mass index; EAT = Eating Attitudes Test; CES-D = Center for Epidemiologic Studies-Depressed Mood Scale; STAI = State-Trait Anxiety Inventory.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Research has consistently described associations between eating psychopathology and both depression and anxiety (9–11), although causality remains a subject of debate. The association between weight itself and both mood and anxiety is less clear (12, 13). Some studies have found a direct relationship between obesity and depression, whereas others have found no association or an inverse relationship (13, 14). There is a need for research examining these associations during pregnancy and the postpartum period. A recent study (15) provides support for this area of research, reporting that during the third trimester of pregnancy, higher deviations from medically ideal weight predicted increased dysphoria in European American women.
Pregnancy is often accompanied by positive behavioral and attitudinal changes with regard to eating and weight, but it seems that underlying concerns about eating and weight persist (16) and may reemerge as pregnancy progresses (17). The postpartum period is a vulnerable time for weight concerns. In the early postpartum period, a majority of women are carrying more weight than they did prepregnancy and, in contrast to pregnancy, may no longer attribute the weight gain to positive aspects of providing for a developing infant. Hisner (18) reported that 75% of women were concerned about their weight in the first few weeks postpartum. Baker and colleagues (16) found that 70% of women were trying to lose weight at 4 months postpartum compared with 8% during pregnancy; moreover, 57% of women who were not trying to lose weight before pregnancy were engaged in efforts to lose weight at 4 months postpartum. Thus, efforts to lose weight in the early postpartum period are normative. Stein and Fairburn (19) found increases, between late pregnancy and 3 months postpartum, in eating disorder symptoms, including concerns about weight and shape. Concern about weight continued to increase to 6 months postpartum, to above the level reported for preconception. Follow-up studies on eating disorder psychopathology in pregnancy have suggested that the positive behavioral changes that women make in pregnancy are usually not sustained after birth and may actually get worse compared with prepregnancy (4).
Investigating associations between eating and weight concerns and depression and anxiety may provide important information about affective or eating and weight-related vulnerabilities in the postpartum period. Eating and weight problems during the postpartum period may be exacerbated by elevated levels of depression or anxiety. Alternatively, increased concerns about eating and weight, in addition to dieting behavior itself, may contribute to anxiety or depression.
It is plausible to hypothesize that associations between weight concerns and affective symptoms of depression and anxiety may shift between pregnancy and the postpartum period. Self-directed cognitions about eating and weight seem to be less negative in pregnant women, thus there may be a reduced association between weight, eating attitudes, depression, and anxiety compared with the postpartum period when shape and weight concerns seem to be particularly salient. The Reflected Self-Appraisal theory described by Ross (20) poses that perceptions of being overweight have a greater effect on depression when being overweight is perceived to be less common, normal, and/or acceptable. Inversely, as acceptance of overweight increases, its effect on depression should decrease. This theory has not received empirical support; however, it is useful in comparing pre- and postnatal feelings about weight. In pregnancy, when weight gain is more "accepted" and seen as externally driven (ie, attributed to caring for the developing fetus), eating and weight-related concerns may not be tied to depression and anxiety. In contrast, in the postpartum period when increased weight is no longer accepted and weight loss is expected, weight and eating concerns may be associated with mood and/or anxiety.
Utilizing an existing data set, the aim of this study was to prospectively investigate relationships between maternal body mass index, eating attitudes, and symptoms of anxiety and depression in pregnancy and at 4 and 14 months postpartum. Specifically, we examined three hypotheses: 1) that eating attitudes and BMI would not be associated with symptoms of depression and anxiety in pregnancy, but would be associated at the two postpartum time points, 2) that women who were overweight at 4 months postpartum would be more likely to have elevated symptoms of eating concerns, depression, and anxiety during the postpartum phase, and 3) that BMI and eating attitudes would be significant predictors of postpartum depression and/or anxiety.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
As part of a larger study focused on the impact of maternal depression on early parent-child interactions and child behavior (22), women were recruited from an Obstetric and Gynecologic clinic in an ethnically and socioeconomically diverse health maintenance organization in an urban setting. Pregnant women were approached by a research assistant and invited to participate in a study about how mothers feel during pregnancy and how these feelings relate to mothers’ and babies’ adjustment after birth. Consent was obtained at each assessment point. Women who agreed to participate completed the pregnancy questionnaires between the fifth and eighth months of pregnancy. Previous studies suggest that eating concerns are more salient in the later months of pregnancy (17). Mothers were asked to participate in two subsequent study visits when their infants were 4 and 14 months old. The 4- and 14-month assessments included filling out additional questionnaires and participating in videotaped laboratory visits.
Measures
Sociodemographic Variables.
Information was gathered about age, race, family income, maternal education level, and parity.
BMI.
At 4 months postpartum, subjects reported current height and weight as well as prepregnancy weight. Weights, reported in pounds, were converted into standardized BMI units (kg/m2). Although self-reported weight and retrospective reports of prepregnancy weight are limitations of the present study, research documents that self-report and objective measures of weight are highly correlated (23), even among pregnant women (15) who, like postpartum women, are experiencing shifts in shape and weight.
Depressive Symptoms.
Participants completed the CES-D (24) during pregnancy and at 4 and 14 months postpartum. The CES-D is a 20-item self-report measure, typically used to assess depressive symptoms in the general population. The measure has good internal consistency (
of approximately 0.85 for the general population and 0.90 for a psychiatric population), excellent concurrent validity, and good known groups validity (25). Because of potential confounding with normative elevations in somatic symptoms in the prenatal and early postnatal periods, the CES-D was modified by removing three somatic items (Item 2. I did not feel like eating; my appetite was poor; Item 11. My sleep was restless; and Item 20. I could not get going). The pattern of results was similar when the data were analyzed using the 20-item CES-D.
Anxiety Symptoms.
Participants completed the state subtest of the STAI (26) in pregnancy and at 4 and 14 months postpartum. The STAI is a widely administered self-report questionnaire with well-documented psychometric properties.
Eating Attitudes.
At all three time points, participants completed the Eating Attitudes Test (EAT-26) (27). The EAT is a 26-item self-report measure that can be used to identify eating disturbances in a nonclinical population (28). The EAT has good reliability ( = 0.90 for an anorexic group) and acceptable criterion validity (eating disorders vs. controls; Ref. 27). Three subscales of the EAT have been identified (27). There is evidence to suggest that, although Factor I scores are reliable in pregnancy and the postpartum period, EAT Factors II and III may not be reliable for use with a pregnant population (16). Consequently, only total and Factor I scores were used in this study. Factor I, "dieting," assesses preoccupations with shape and pathological avoidance of fattening foods.
Analytic Plan
Before testing, all variables were examined for normality. To rule out influences due to demographic differences, associations between demographic information and all relevant variables were explored. Initial descriptive analyses examined changes in BMI, eating attitudes, depressive symptoms, and anxiety across the three time points. To investigate the hypothesis that eating attitudes and BMI were not associated with symptoms of depression and anxiety in pregnancy, but were associated at the two postpartum time points, analyses investigated differences across time in the associations among these continuous variables. To examine clinical significance of overweight, we examined BMI as a categorical variable. Using multivariate analysis of variance (MANOVA) and discriminant function analysis, we tested the hypothesis that women who were overweight were more likely to have elevated symptoms of eating concerns, depression, and anxiety. Multiple regression was used to test whether BMI and eating attitudes were significant predictors of postpartum depression and/or anxiety.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
$40K per year; mean = 40.4 and 32.6, respectively). Depressive symptoms at 4 months postpartum were associated with maternal age (r = 0.27, p < .05); older women had higher CES-D scores. Finally, maternal education was associated with pregnancy CES-D, such that pregnant women with less education reported more depressive symptoms (F(1,59) = 6.22, p < .05). No associations were found between demographic variables and BMI, EAT total, or EAT Factor I scores. Parity was not related to either prepregnancy or 4-month postpartum BMI (r values = 0.02 and 0.10). Criteria for demographic covariates consisted of association with outcome variables at least two time points; no variables met this criteria.
Descriptive Statistics
Mean BMI for prepregnancy, reported retrospectively at 4 months postpartum, was 24.7 kg/m2 (SD = 5.5); 22.0% of women had BMIs >27 kg/m2 and 12.5% had BMIs >30 kg/m2. Mean BMI at 4 months postpartum was 25.6 kg/m2 (SD = 5.1) with 26.6% of women >27 kg/m2, and 17.2% >30 kg/m2. Change in BMI across time was analyzed using an ANOVA, repeated measures design. Maternal BMI at 4 months after birth was significantly higher than prepregnancy BMI (F(1,63) = 10.92, p < .01). Pre- and postpregnancy BMI were highly correlated (r = 0.91, p < .01). Means of the EAT, CES-D, and STAI, as well as changes in these variables across time, are presented in Table 1.
|
|
EAT and Symptoms of Depression and Anxiety.
As hypothesized, there were no significant relations during pregnancy between eating attitudes (EAT total), dieting attitudes and shape concerns (Factor I), and either the CES-D or STAI. Although the predicted associations were found between eating attitudes and affective symptoms at 14 months postpartum, there were no significant associations between the EAT and the STAI or CES-D at 4 months postpartum. Comparisons of correlations involving the EAT and CES-D did not reveal significant differences in associations at different times; however, the correlation between EAT total score and the STAI was significantly higher at 14 months postpartum than in pregnancy.
Because of the relatively small sample size, we decided to control for Type I error for the correlational analyses post hoc by comparing the ratio of significant to nonsignificant results. Using this approach, risk of Type II error could be minimized while maintaining the ability to assess the probability that significant results were due to chance. Of sixteen correlations (Table 2) that were predicted to be statistically significant, 12 (75%) were consistent with the hypothesis; as predicted, none of the eight pregnancy correlations were statistically significant.
Comparison of Overweight and Nonoverweight Women
To investigate whether being overweight imparts unique risk for eating concerns, depression, or anxiety in the postpartum period, one-way MANOVA, between-groups design, and discriminant function analysis were used to examine differences between women with 4-month postpartum BMIs 27 (N = 17) and women with BMIs < 27 (N = 47). Women with BMIs 27 were classified as overweight (31). BMI at 4 months postpartum was used to determine overweight status; it was considered to be more reliable because it was not reported retrospectively. Analysis revealed a significant difference between the two groups on EAT total, CES-D, and STAI scores from both postpartum time points (Wilks 
: F(6,57) = 5.33, p < .01). Structure coefficients of the discriminant function suggested that the variables that best differentiated the two groups were 4-month postpartum STAI (0.65) and CES-D (0.57), as well as 14-month postpartum CES-D (0.89). Coefficients less than 0.50 were not interpreted. Means and univariate tests for differences are presented in Table 3.
|
16 has been used as a cutoff for "depressed" vs. a score of 
15 for "nondepressed" (14). At both 4 and 14 months postpartum, overweight women had mean CES-D scores above 16.
Predicting Affective Symptoms at 14 Months Postpartum
Hierarchical regression was used to assess whether eating attitudes and/or BMI predicted symptoms of anxiety or depression at 14 months postpartum. To avoid multicollinearity, one measure of BMI (4 months postpartum) and the EAT total score, which provides a broader measure of eating disturbance than Factor I, were included as predictors.
Predicting 14-Month CES-D.
The first step of the model included previous reports of depressive and anxiety symptoms (see Table 4). These variables were included because we were interested in the predictive value of eating and weight-related concerns above and beyond other significant predictors, which included prior symptoms of depression and anxiety. Without BMI in the model, EAT scores predicted depressive symptoms at 14 months postpartum, over and above previous symptoms of anxiety and depression. The final model including both BMI and eating attitudes was significant, as was the 
R2 (F(1,56) = 10.18, p < .01) from step 2. The two variables significant in the final model were 4-month postpartum eating attitudes and BMI. Early postpartum eating attitudes and BMI predicted later depressive symptomatology, even after controlling for all of the other variables in the model.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
One consistent pattern that supported our hypothesis was that measures of anxiety and depression were not significantly correlated with eating concerns or with BMI during pregnancy. This supports previous research findings concerning changes in eating attitudes and behaviors during pregnancy. Negative self-directed cognitions concerning eating, weight, and shape seem to be less pervasive during pregnancy (16) and consequently may be less apt to impact mood and/or anxiety. Women may feel more justified to eat during pregnancy to care for the infant. The meaning of weight may be altered if changes in weight are perceived to reflect growth of the developing fetus and competence in the nurturing role of motherhood. Thus, priorities and role identification may shift such that eating concerns still exist but are not as important and thus less likely to influence mood or anxiety. Women may also experience weight gain and shape change during pregnancy as socially sanctioned, and, consequently, being overweight in pregnancy may feel less stigmatizing. The Self-Reflected Appraisal theory (20) proposes that when increased weight is accepted, as perhaps it is perceived to be in pregnancy, higher weight will be less likely to influence depression.
The pattern of significant correlations with affective symptoms at 4 and 14 months postpartum was different for BMI and for eating attitudes. Consistent with predictions, BMI was associated with anxiety and depressive symptoms in both the early and later postpartum periods, and many of these correlations were significantly different from those in pregnancy. Predictions for eating-related concerns were partially supported. Eating attitudes were not associated with affective symptoms during pregnancy, but were related at 14 months postpartum. The expected associations at 4 months postpartum were not found.
One of the questions raised by the results is why BMI, but not eating attitudes, was associated with depressive and anxiety symptoms at 4 months postpartum, whereas both were related to affective symptoms in the later postpartum period. BMI reflects the state of one’s body, whereas the eating attitudes measure contains more information about one’s actual eating and dieting behavior. As already proposed, in the early postpartum period, the state of a woman’s body may be related to affective symptoms because of distress associated with pregnancy weight gain and the desire to return to prepregnancy shape and weight. In contrast, women’s eating behaviors may not be associated with negative affect because the women may still be hopeful that their behaviors will lead to weight loss or shape changes.
Correlational findings related to BMI were supported by analyses comparing women with BMIs 27 to women with BMIs <27. Women in the overweight group had higher ratings of depressive symptoms at 14 months postpartum and anxiety symptoms at 4 months postpartum. The group differences in depressive symptoms may be clinically significant, in that overweight women had CES-D scores in a "depressed" range (14). Thus, being overweight at 4 months postpartum may be associated with continued risk for depression in later postpartum, suggesting that intervention for women who are overweight and report elevated depressive symptoms is warranted. Of interest, eating attitudes showed the expected associations with depression, anxiety, and BMI despite the fact that the observed EAT scores in this sample, which included eight women with a lifetime history of an eating disorder, were low relative to previously studied college student and eating disordered populations (28). More research with pregnant women who have clinically elevated EAT scores is needed to fully understand the clinical implications of these findings for eating disturbances. In addition, norms on measures of eating attitudes and behaviors need to be established for noncollege-aged populations.
The partial prospective nature of the data allowed us to look at potential predictors of postpartum affective symptoms. This information could be valuable for identifying women in pregnancy or early postpartum who might be at risk for continuing or exacerbated anxiety and/or depressive symptoms due to weight status or concerns about eating and weight. Testing for factors that predict depressive symptoms past 6 months postpartum is particularly important in light of evidence that protracted maternal depression seems to confer risk in mother-infant interactions (32). Although eating and weight-related factors did not seem to be important predictors of anxiety symptoms at 14 months postpartum, both early postpartum eating attitudes and weight status were relevant for predicting depressive symptoms. Our results suggest that there may be value to identifying women who have significant eating-related concerns in the early postpartum period. But the more valuable marker for risk of late postpartum depressive symptoms may simply be a woman’s BMI, with increased BMI imparting potential risk. Although BMI at 4 months postpartum was included in the regressions, prepregnancy BMI could also be considered a potential marker for risk given its high correlation with postpartum BMI.
There were limitations to the study that need to be addressed in future investigations. Ideally, body weight would be assessed through objective means in addition to self-report. Although self-reported weights of both nonpregnant individuals and pregnant women seem to be highly correlated with observed weights (15, 23), it is possible that there are biases in self-reported weights among individuals who have experienced in the past, or are currently experiencing, significant eating concerns or symptoms of depression or anxiety. It was unfortunate that we did not have information on women’s weights at 14 months postpartum. In addition, a comparison group of nonpregnant women would allow us to rule out threats to internal validity and assess whether the changes in relationships we observed were unique to the process of pregnancy and childbirth rather than simply the passage of time. The demographic profile of the sample limits the generalizability of the findings.
In summary, this study suggests the presence of moderate relationships between BMI and eating concerns and both depressive and anxiety symptomatology in the postpartum period. BMI, as well as eating-related concerns in the early postpartum period, may be potentially useful predictors of depressive symptomatology in the later postpartum period. BMI itself seems to be a particularly important marker of risk for increases in maternal depressive symptoms in the first year after delivery. Although correlations with the EAT suggested the expected pattern of association, ‘comparisons between all but one time point did not reveal significant differences. Research involving larger and more diverse samples is necessary to further explore associations between eating attitudes and depression and anxiety during pregnancy and the postpartum period.
Received for publication November 9, 1998.
Revision received August 24, 1999.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  E. J. Corwin, L. E. Murray-Kolb, and J. L. Beard Low Hemoglobin Level Is a Risk Factor for Postpartum Depression J. Nutr., December 1, 2003; 133(12): 4139 - 4142. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. B. Dixon, M. E. Dixon, and P. E. O'Brien Depression in Association With Severe Obesity: Changes With Weight Loss Arch Intern Med, September 22, 2003; 163(17): 2058 - 2065. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
