EATING ATTITUDES AND BEHAVIORS IN PREGNANCY AND POSTPARTUM: GLOBAL STABILITY VERSUS SPECIFIC TRANSITIONS 1

Christina Wood Baker, M.S., Alice S. Carter, Ph.D., Lisa R. Cohen, Ph.D., and Kelly D. Brownell, Ph.D. Yale University

having infant feeding problems and worrying about infant size (7,8). Given the widespread discontent with shape and weight present among women in our society (9), it is important to understand how women respond and adjust to the multitude of changes in eating, weight, and shape that are associated with pregnancy and the period following delivery. There is a need for prospective research spanning pregnancy and the postpartum phase. The literature on eating disorders, attitudes, and behaviors in pregnancy and postpartum reflects inconsistent findings. There are some reports that eating problems worsen in pregnancy (10), while others have found that some women experience a reprieve from eating disorder symptoms which then return in the postnatal period (7,11). There is also evidence that some women make positive changes in eating during pregnancy which persist following delivery (7). In general, women are less likely to diet during pregnancy (2,12); however, there is some evidence that eating concerns decrease early in pregnancy but increase again as pregnancy progresses (1). Discrepant findings may be due in part to methodological limitations, such as reliance on case reports and retrospective studies (13), again emphasizing the need for prospective studies across pregnancy and the postpartum period. Although research on the postpartum period is sparse, the available evidence suggests that women are particularly vulnerable to weight concerns during this time. Hisner (14) found that 75% of women were concerned about their weight in the first few weeks postpartum. Lacey and Smith (7) reported that nearly half of their sample of 20 mothers with bulimia experienced postpartum increases in eating disorder severity over prepregnancy levels. Research on body image during this phase of life has also proved inconsistent (12,15-17), perhaps reflecting that the construct of body image has specific components which respond differentially to the perturbations of pregnancy and postpartum. For example, in a study of 63 pregnant women, Strang and Sullivan (17) found that women felt more positively about their bodies when no longer pregnant, in the early postpartum period, when compared to the third trimester of pregnancy. Davies and Wardle (12), in a sample of 76 pregnant women and 97 controls, found that pregnant women were less dissatisfied with their bodies than nonpregnant women, but the groups did not differ in choices of body image ideals. These latter results suggest that some women make specific positive cognitive and/or behavioral adjustments while pregnant, despite the presence of underlying negative eatingand weight-related cognitions. One way to understand the inconsistencies observed in the literature is to distinguish between global and specific attitudes and behaviors. Although global attitudes may be resistant to change, perhaps specific behaviors and attitudes shift more quickly in response to environmental demands (i.e. changes in food consumption necessary to nurture a developing fetus and relaxed external expectations about body weight and shape). A number of studies

ABSTRACT Global changes in eating attitudes were examined prospectively across pregnancy and 4 months postpartum in a sample of 90 women. In addition, specific changes in dieting behavior and weight~shape satisfaction were assessed at 4 months postpartum for concurrent and retrospective time points. Measures included the Eating Attitudes Test (EAT) and weight/shape satisfaction in pregnancy and at 4 months postpartum, as well as prepregnancy, pregnancy, and postpartum weight loss efforts. While global EAT scores were stable across time, dieting scores (Factor 1) increased between pregnancy and postpartum, Weight~shape satisfaction was higher in pregnancy, and satisfaction was related to EAT scores at 4 months postpartum but not during pregnancy. Prepregnancy dieters and nondieters were best discriminated by higher weights, elevated pregnancy dieting scores, and lower postpartum weight/ shape satisfaction. Results emphasize the importance of looking beyond changes in global eating attitudes and behaviors to more specific eating concerns or behaviors. Lastly, the results have implications for identifying women at risk for eating- and weightrelated concerns during this period of rapid physical change. (Ann Behav Med

1999, 21(2):143-148)

INTRODUCTION Over the past decade, there has been an increase in research on eating attitudes and behaviors in pregnancy, as it has been recognized that many women of childbearing age struggle with significant eating and weight concerns, including the presence of eating disorders (1,2). Research on eating attitudes and behaviors during pregnancy is important because it has implications for both maternal and fetal health and well-being. Maternal weight gain in pregnancy and infant birth weight are positively correlated (3,4), and low birth weight has been found to be associated with infant mortality and negative consequences for infant development (5). The postpartum period has received less attention; although this, like pregnancy, is a time when attitudes and behaviors related to eating and weight can be particularly salient and subject to change. A woman's body undergoes major transformations through the course of pregnancy. Following delivery, the body rarely returns immediately to its prepregnancy shape, and many women are not prepared for the extent of their physical changes (6). Again, infants may be affected by maternal eating attitudes in the postpartum period. Mothers with eating concerns may be at risk for

1 Preparation of this manuscript was supported in part by a small grant from NIMH (R03 MH 49684) awarded to Alice S. Carler, Ph.D. Reprint Address: C. W. Baker, M.S., P.O. Box 208205, Department of Psychology, Yale University, New Haven, CT 06520-8205. 9 1999 by The Society of Behavioral Medicine.

143

144

ANNALS OF BEHAVIORAL MEDICINE TABLE 1 Maternal Demographics

Mean age Mean gestational age (# weeks) % Married % Primiparas Race Caucasian African-American Hispanic Asian Household income .70 (21) (see Table 2). Factor II appeared to be reliable at 4 months postpartum (0.87), but not in pregnancy (0.56). The evidence for reliability of Factor III was poor overall (pregnancy: 0.37; postpartum: 0.56). Given indications of a lack of reliability in Factors II and III, analyses were performed using only total and Factor I scores.

A repeated measures ANOVA was conducted to investigate whether satisfaction with weight/shape changed between pregnancy and postpartum. There was a significant effect of ratings of weight/shape satisfaction across time, F(1, 87) = 35.29, p < .01. Subjects were more satisfied during pregnancy (M = 3.26, SD = 1.24, n = 88) than at 4 months postpartum (M = 2.34, SD = 1.22, n = 90). Only 8% (n = 7) of the women were trying to lose weight during pregnancy, while 53% (n = 48) reported that they had been trying to lose weight prior to pregnancy, and 70% (n = 63) were trying to lose weight at 4 months postpartum. Fifty-seven percent of the women who were not dieting prepregnancy reported that they were trying to lose weight at 4 months postpartum.

Demographics No associations were found between demographic variables and EAT total or Factor I scores. Of note, pregnancy EAT scores were not associated with the point in pregnancy (i.e. number of

Eating Attitudes Repeated measures ANOVA was also conducted for EAT total and Factor I (Dieting) scores to investigate whether eating attitudes differed between pregnancy and 4 months postpartum. There was no significant effect of EAT total across time, F(1, 89) = 2.98, p = .09; however, there was a significant effect of Dieting, F(1, 89) = 5.04, p < .05. Means are presented in Table 2. As hypothesized,

146

B a k e r et al.

ANNALS OF BEHAVIORAL MEDICINE

TABLE 3 Pearson rs of EAT and Weight/Shape Satisfaction Covarying for Prepregnancy BMI (n = 85)

TABLE 4 Means (SD) and Univariate ANOVAs of Dependent Variables by Prepregnaney Dieting Status

Weight/Shape Satisfaction Pregnancy Pregnancy EAT Total score Dieting subscale Postpartum EAT Total score Dieting subscale

Postpartum

-.05 -.16

-.29"* -.36**

-.09 - . 14

-.43"* - .45" *

** p < .01. global scores were stable across time, while women had lower scores during pregnancy on the dieting and shape concern subscale. Of interest, EAT total and Factor I scores were not associated at either time point with BMI. Weight/Shape Satisfaction and Eating Attitudes To investigate the association between weight/shape satisfaction and eating attitudes, correlations between weight/shape sarisfaction ratings and EAT total and Factor I scores were examined (Table 3). In addition, to assess differences in the associations between pregnancy and 4 months postpartum, correlations were compared using Steiger's test for significance of difference between two dependent rs without a test in common (22). Due to the association between BMI and weight/shape satisfaction, prepregnancy BMI was included as a covariate in the analyses. Prepregnancy BMI was used because 4-month postpartum BMI is more variable due to dependence on factors such as pregnancy weight gain and rate of weight loss following pregnancy. The same pattern of significant results as described below was found when 4-month postpartum BMI was used as the covariate. It is important to note that the pregnancy weight/shape satisfaction measure was collected retrospectively at 4 months postpartum. Results indicated that the relationship between global eating attitudes and weight/shape satisfaction was significantly different in pregnancy from 4 months postpartum (z = 3.95, p < .01). Pregnancy weight/shape satisfaction was not associated with pregnancy EAT total score (pr = - . 0 5 ) , while 4-month postpartum weight/shape satisfaction was significantly associated with 4-month postpartum EAT total (pr = - 0 . 4 3 , p < .01); women with higher global EAT scores felt more negatively about their bodies. Correlations between EAT Factor I (Dieting) and weight/ shape satisfaction revealed a similar pattern (z = 2.99, p < .01). During pregnancy, the two variables were not significantly correlated (pr = -0.16), while the association was significant at 4 months postpartum (pr = - 0 . 4 5 , p < .01). These results support the hypothesis that eating attitudes may be influenced by different factors in pregnancy and postpartum. Finally, pregnancy EAT total and Factor I (Dieting) scores were significantly correlated with postpartum weight/shape sarisfaction (prs = - . 2 9 , - . 3 6 , respectively; ps < .01), such that women with higher EAT scores in pregnancy reported being less satisfied with their bodies at 4 months postpartum. Dieters versus Nondieters Due to previous findings that prepregnancy dieters and nondieters differ in attitudinal and behavioral responses to being pregnant (2), one-way MANOVA, between-groups design, and discriminant function analysis were used to investigate which

EAT total score pregnancy EAT total score postpartum EAT Dieting subscale pregnancy EAT Dieting subscale postpartum Weight/shape satisfaction pregnancy Weight/shape satisfaction postpartum BMI prepregnancy BMI 4 months postpartum BMI ideal

Dieters (n = 43)

Nondieters (n = 42)

F(1, 83)

7.0 (7.5) 8.2 (8.5) 5.1 (6.1) 5.8 (5.7)

5.0 (3.5) 5.6 (5.1) 2.7 (2.4) 3.8 (4.1)

2.48 3.01 5.86* 3.60

3.2 (1.2)

3.4 (1.3)

0.81

2.0 (1.1) 26.4 (5.4) 27.5 (5.0) 22.4 (2.1)

2.8 (1.2) 23.3 (5.0) 23.9 (3.9) 21.1 (1.9)

9.94*** 7.43** 13.57"** 8.75***

Note: Bonferroni type adjustment requires p < .006. *p < .05. **p < .01. ***p < .006. variables best differentiated dieters (n = 43) from nondieters (n = 42). Dependent variables included EAT total, EAT Factor I and weight/shape satisfaction at both time points, as well as BMI (actual and ideal). Whereas Fairburn and Welch (2) grouped women based on lifetime history of dieting, the grouping variable in this study was whether a woman reported retrospectively that she was trying to lose weight in the period just prior to pregnancy. Fifty-three percent of women identified themselves as prepregnancy dieters. Analysis revealed a significant difference between the groups when compared simultaneously on the outcome measures, Wilks' Lambda: F(9, 75) = 2.86, p < .01. Structure coefficients of the discriminant function suggested that the variables that best discriminated dieters from nondieters were measures of BMI (prepregnancy = 0.57, 4-month = 0.74, ideal = 0.61), weight/shape satisfaction at 4 months postpartum ( - 0 . 6 5 ) , and EAT Factor I (Dieting) scores during pregnancy (0.51). Coefficients less than .50 were not interpreted. Means, as well as univariate comparisons, are presented in Table 4. Dieters had higher postpartum and ideal BMIs, as well as lower ratings of weight/shape satisfaction in the postpartum period. The discriminating value of Factor I scores suggests that women who reported trying to lose weight prior to pregnancy may be at greater risk for dieting or preoccupation with shape during pregnancy. Indeed, 100% of the women who reported weight loss efforts during pregnancy had been trying to lose weight prior to becoming pregnant (Fisher's exact test yielded a p < .05). DISCUSSION The results supported the primary hypothesis that global concerns regarding eating and weight remain stable through pregnancy and the postpartum period, while specific attitudes and behaviors vary. While EAT total scores did not differ across time, women reported being more satisfied with their weight and shape during pregnancy and had lower scores on the EAT subscale which assessed dieting behaviors and attitudes towards shape. Furthermore, fewer women reported trying to lose weight in pregnancy (8%) when compared to both prepregnancy (53%) and postpartum (70%). Other studies have found similar (2) or lower (12) rates of dieting among pregnant women; Davies and Wardle (12) reported that 1.3% of a pregnant group of women were dieting compared with 44.3% of a nonpregnant group. The rates in the present study, 8% versus 53%, are comparable but suggest a slightly higher prevalence of dieting during pregnancy. Global eating attitudes, as

Eating Attitudes in Pregnancy well as attitudes specific to dieting and shape (EAT Factor I scores), were not associated with weight/shape satisfaction in pregnancy but were significantly associated at 4 months postpartum. A number of different factors could influence or motivate behavioral and/or attitudinal change during pregnancy. Moreover, maintenance of positive changes may depend in part on the nature of the motivational factor. Women may respond differently to weight gain in pregnancy due to the internalization of a new role (i.e. becoming a mother), the perception of bodily changes as reflections of infant growth and maternal nurturance, or because cultural expectations are relaxed and weight gain is accepted in pregnant women (7,12). Despite the persistence of underlying concerns with weight or shape, the maternal role could allow some women to move appearance lower on their list of priorities. This may explain in part why some women are able to make permanent, positive changes in their eating during pregnancy or the postpartum period. In the Lemberg and Philips study of eating disorders and pregnancy, the 23% of their sample who remained free of eating disorder symptoms during the first year after childbirth reported that "the meaning of their life had changed since having a child, and many reported that a factor in their improvement was to remain healthy in order to care for the baby" (11, p. 290). Another alternative was described by Lacey and Smith (7). In their study, some women reported making a "positive and strong attempt to control their eating disorder because they feared harming their babies" (7, p. 780). Women motivated by this fear may experience a regression to previous attitudes and behaviors following pregnancy, because their impetus for change is directly tied to pregnancy. It may be particularly difficult when added weight from pregnancy is present and the woman can no longer justify the weight gain. Future investigations are needed to investigate motivations for behavioral and attitudinal changes. Understanding factors that motivate women to make changes during pregnancy could inform intervention efforts that encourage the maintenance of positive changes following delivery. Given that so many women are motivated to change their behavior in pregnancy, the 8% of women who reported efforts to lose weight while pregnant may be a group at particularly high risk for eating problems. Due to the small number in this group, we did not perform analyses comparing these women to the rest of the sample. An investigation of differences between those who diet while pregnant and those who do not should be a priority for future research. The potential for intervention in pregnancy received support from the findings that prepregnancy weight loss efforts may identify women at future risk. Women who self-identify as dieting prior to pregnancy may be more likely to diet during pregnancy and experience dissatisfaction with their bodies in the postpartum period. Consistent with this, pregnancy EAT scores, both total and Factor I, predicted less satisfaction with weight and shape in the postpartum period. These two findings, taken together, suggest the value of screening women for eating concerns and dieting behavior early in pregnancy. Finally, that dieters and nondieters were best discriminated by BMI raises the possibility that weight may be an important risk factor in itself and may be a useful screening measure during prenatal care. The results of this study indicate that the early postpartum phase may be a particularly vulnerable time. A majority of women carry more weight than they did prepregnancy (M: prepregnancy BMI = 24.8, 4-month postpartum BMI = 25.7) and can no longer attribute the weight gain to positive aspects of providing for a

VOLUME 21, NUMBER 2, 1999

147

developing infant. Fifty-seven percent of the women in this study who were not trying to lose weight prior to pregnancy reported that they were trying to lose weight at 4 months postpartum. The decrease in weight/shape satisfaction seen at 4 months postpartum is consistent with the risks posed by the postpartum period. It is very important to note that the EAT was not uniformly reliable in pregnancy, which raises concerns about the use of some of the subscales of this measure with pregnant women. Specifi-. cally, Factors II (Oral Control) and III (Bulimia and Food Preoccupation) were not reliable in this sample. As such, exploration of the hypothesis concerning global versus specific changes in eating attitudes and behaviors was limited. We were not able to investigate two of the specific components (Factors II and III) that we were hoping to examine. Perhaps due to the very different meaning of eating and restraint during this developmental period, alternative assessment tools need to be developed to fully explore eating attitudes in pregnancy. Methodologically, this raises interesting questions for longitudinal research spanning pregnancy and postpartum. Assessment tools are needed that are reliable and valid both in pregnancy and postpartum. The issue of assessment reliability needs to be considered for future studies, but also for interpretation of previous studies in the area of eating and weight concerns during pregnancy. The primary purpose of this report is to present reliability information on the EAT in pregnancy and in the early postpartum period and to examine relations between global and specific eating attitudes and behaviors at these two time periods. Weaknesses of this study include the fact that prepregnancy weight and pregnancy weight/shape satisfaction were reported retrospectively and that full assessment of eating disturbances by structured interview was not possible. However, the retrospective reporting of weight/shape satisfaction had the advantage that all women could consider their entire pregnancy; had this questionnaire been administered during pregnancy, women at different gestational points would have been making judgements about their experience at that specific time in pregnancy. The demographic heterogeneity of the present sample is a strength, in that it increases the generalizability of the findings. However, the relatively small sample size of the present study precluded more fine-grained analyses of potentially very important demographic variables such as ethnicity, age, and socioeconomic status, all of which need further investigation. Beyond demographics, there are other factors that may contribute to eating and weight-related attitudes and behaviors during pregnancy and postpartum (e.g. breastfeeding, work status, and lifestyle changes). These variables should be addressed in future research. Lastly, the 50% response rate and attrition between the two assessments may compromise the generalizability of these findings. Despite the limitations described above, these data do provide a partial prospective view of global and specific eating and weight-related attitudes and behaviors during pregnancy and the postpartum period. The results highlight the persistence of underlying global eating concerns in spite of apparent changes in specific eating behaviors and attitudes. Moreover, this study demonstrates that different factors contribute to eating concerns in pregnancy and postpartum, with weight/shape satisfaction playing a more important role in the postpartum period. These findings have implications for professionals working with pregnant women, in that they suggest that women with eating and weight concerns who make positive changes in attitudes or behaviors during pregnancy may still be quite vulnerable and may experience increased concerns and associated unhealthy behaviors in the postpartum period.

148

ANNALS OF BEHAVIORAL MEDICINE REFERENCES

(1) Fairburn CG, Stein A, Jones R: Eating habits and eating disorders during pregnancy.Psychosomatic Medicine. 1992, 54:665-672. (2) Fairburn CG, Welch SL: The impact of pregnancy on eating habits and attitudes to shape and weight. International Journal of Eating Disorders. 1990, 9:153-160. (3) Abrams BE Laros RK: Prepregnancy weight, weight gain, and birth weight. American Journal of Obstetrics and Gynecology. 1986, 154:503-509. (4) Gormican A, ValentineJ, Satter E: Relationships of maternal weight gain, prepregnancy weight, and infant birthweight: Interaction of weight factors in pregnancy. Journal of the American Dietetic Association. 1980, 77:662-667. (5) Palmer JL, Jennings GE, Massey L: Development of an assessment form: Attitude toward weight gain during pregnancy. Journal of the American Dietetic Association. 1985, 85:946-949. (6) Stein A, Fairburn CG: Eating habits and attitudes in the postpartum period. Psychosomatic Medicine. 1996, 58:321-325. (7) Lacey JH, Smith G: Bulimia nervosa: The impact of pregnancy on mother and baby. British Journal of Psychiatry. 1987, 150:777-781. (8) Stein A, Fairburn CG: Children of mothers with bulimia nervosa. British Medical Journal. 1989, 299:777-778. (9) Rodin J, Silberstein LR, Striegel-Moore RH: Women and weight: A normative discontent. In Sonderegger TB (ed), Nebraska Symposium on Motivation: Vol. 32. Psychology and Gender. Lincoln, NE: University of Nebraska Press, 1985, 267-307. (10) StewartDE, Raskin J, Garfinkel PE, MacDonald OL, Robinson GE: Anorexia nervosa, bulimia, and pregnancy. American Journal of Obstetrics and Gynecology. 1987, 157:1194-1198.

Baker et al. (11) Lemberg R, Philips J: The impact of pregnancy on anorexia nervosa and bulimia. International Journal of Eating Disorders. 1989, 8:285-295. (12) Davies K, Wardle J: Body image and dieting in pregnancy, Journal of Psychosomatic Research. 1994, 38:787-799. (13) Franko DL, Walton BE: Pregnancy and eating disorders: A review and clinical implications. International Journal of Eating Disorders. 1993, 13:41--47. (14) Hiser PL: Concerns of multiparas during the second postpartum week. Journal of Obstetric, Gynecologic, and Neonatal Nursing. 1987, 16."195-203. (15) McConneU OL, Daston PG: Body image changes in pregnancy. Journal of Projective Techniques. 1961, 25:451-456. (16) Slade PD: Awareness of body dimensions during pregnancy: An analogue study. Psychological Medicine. 1977, 7:245-252. (17) Strang VR, Sullivan PL: Body image attitudes during pregnancy and the postpartum period. Journal of Obstetric, Gynecologic, and Neonatal Nursing. 1985, 7:332-337. (18) Garner DM, GarfinkelPE: The Eating Attitudes Test: An index of the symptoms of anorexia nervosa. Psychological Medicine. 1979, 9:273-279. (19) Garner DM, Olmsted MP, Bohr Y, Garfinkel PE: The Eating Attitudes Test: Psychometricfeatures and clinical correlates. Psychological Medicine. 1982, 12:871-878. (20) Cronbach LJ: Coefficient alpha and the internal structure of tests. Psychometrika. 1951, 16:297-334. (21) Nunnally JC: Psychometric Theory. New York: McGraw-Hill, 1978. (22) Steiger JH: Tests for comparing elements of a correlation matrix. Psychological Bulletin. 1980, 87:245-251.