HomeLarge Type Edition
HOME ARCHIVE SEARCH TABLE OF CONTENTS
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Services
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
PubMed
Right arrow PubMed Citation
The Journals of Gerontology Series A: Biological Sciences and Medical Sciences 57:M128-M133 (2002)
© 2002 The Gerontological Society of America

The Association of Eating Behavior With Risk for Morbidity in Older Women

Nicholas P. Haysa, Gaston P. Bathalona,b, Ronenn Roubenoffa, Ruth Lipmana and Susan B. Robertsa

a Jean Mayer USDA Human Nutrition Research Center on Aging, Tufts University, Boston, Massachusetts
b U.S. Army Research Institute of Environmental Medicine, Natick, Massachusetts

Susan B. Roberts, Energy Metabolism Laboratory, Jean Mayer USDA Human Nutrition Research Center on Aging at Tufts University, 711 Washington St., Boston, MA 02111 E-mail: sroberts{at}hnrc.tufts.edu.


   Abstract
Top
 Abstract
Methods
 Results
 Discussion
 References
 
Background. Although an influence of eating behavior on dietary intake and physiology has been documented in several studies, the extent to which eating behavior influences long-term health is uncertain.

Methods. Current dietary restraint, disinhibition, and hunger were assessed using the Eating Inventory in 1252 nonsmoking women aged 55 to 65 years. In addition, subjects reported the presence or absence of 22 specific morbidities, along with general demographic information. Logistic regression was used to examine associations between eating behavior scores and morbidity, adjusting for age, prior smoking status, hormone replacement therapy, education level, and body mass index (BMI).

Results. In adjusted models excluding BMI, higher disinhibition scores were associated with small increased risks for hypercholesterolemia (odds ratio [OR] 1.04, p = .045), leg cramps (OR 1.05, p = .044), indigestion (OR 1.06, p = .020), and cataract (OR 1.09, p = .036), and a decreased risk of eczema (OR 0.91, p = .008). In addition, higher hunger scores were associated with increased risk of eczema (OR 1.09, p = .026). However, after adjusting for confounding variables plus BMI, higher disinhibition scores were associated with increased risks for low back pain (OR 1.06, p = .031) and constipation (OR 1.10, p = .004), and associations of disinhibition and hunger with eczema were unchanged (OR 0.90, p = .008 and OR 1.09, p = .024, respectively). Dietary restraint was not associated with morbidity in any model.

Conclusions. Higher disinhibition and hunger scores were associated with small alterations in reported morbidity risk in a large population of nonsmoking older women. Although our cross-sectional study design makes the directionality of these relationships unclear, our results suggest at most a relatively minor independent influence of eating behavior constructs on long-term health.

"EATING behavior" is a general term that describes an individual's attitudes toward and relationships with eating and food. There is increasing recognition that eating behavior may have an important impact on both dietary intake and physiology. However, the extent to which eating behavior influences long-term health has received little attention.

Much of the work to date regarding eating behavior and health has focused on dietary restraint, which is one of three eating behavior constructs defined by the Eating Inventory (EI) of Stunkard and Messick (1). Dietary restraint refers to the tendency to consciously restrict food intake either to prevent weight gain or to promote weight loss by control over both energy intake and types of foods eaten (2). The EI also quantifies disinhibition, which refers to the tendency to overeat in the presence of palatable foods or other stimuli, such as emotional distress (3), and hunger, which refers to perceived body symptoms that signal the need for food (3). High levels of dietary restraint and disinhibition appear to be relatively common. For example, large surveys conducted in German and U.S. adult populations indicate a 20% to 38% prevalence of high dietary restraint scores (4)(5)(6) and a 46% prevalence of high disinhibition scores (7); however, to our knowledge, there is little information on the prevalence of high hunger scores as assessed by the EI.

Previous work examining eating behavior and morbidity has been confined mainly to analyses of short-term outcome measures in young and middle-aged subjects. For example, studies using the EI have found higher dietary restraint scores to be associated with higher urinary cortisol excretion (8), lower insulin and norepinephrine concentrations (9), and higher triacylglycerol values (10). Higher disinhibition and hunger scores, along with lower restraint scores, have also been associated with a poor health-related quality of life (11)(12). Although few studies have examined eating behavior and morbidity using longer-term measures, associations of higher dietary restraint scores with subclinical ovulatory disturbances (13)(14) and lower bone mineral content (15) have also been reported in young women. In contrast, prior work from our laboratory (16) did not find a relationship between dietary restraint scores and detrimental effects in a range of short- and long-term physiological and metabolic characteristics in postmenopausal women. However, there are concerns with a number of these studies that make their conclusions uncertain. In particular, the relatively small sample sizes of many of the studies, including our own, limits their interpretation. Furthermore, the impact of eating behavior patterns on specific long-term indicators of health may not be apparent in younger subjects, nor is it clear that eating behavior has an effect on health independent of its reported association with body mass index (BMI) and obesity (2)(17)(18).

We therefore conducted a cross-sectional, descriptive study in a large population of older women to test the hypothesis that higher dietary restraint, disinhibition, and hunger scores are associated with a greater likelihood of self-reported morbidity.


   Methods
Top
 Abstract
 Methods
Results
 Discussion
 References
 
Subject Recruitment and Data Collection
As described previously (17), subjects aged 55 to 65 years were asked to participate in a study described as examining nonspecified eating patterns and health, and those willing to participate were mailed an informed consent form and two questionnaires, the EI and a health history questionnaire (HHQ). A total of 2042 women (representing a response rate of approximately 60%) completed both questionnaires. Details of the study population are given in Table 1 .


View this table:
[in this window]
[in a new window]
  		Table 1. Subject Characteristics (n = 1252)

 
The EI consists of 36 true/false and 15 multiple-choice questions, with three separate groups of questions used to calculate the three constructs' (restraint, disinhibition, and hunger) scores. Higher scores reflect a greater tendency to exhibit that particular eating behavior characteristic. Questionnaires were scored according to published guidelines (1); however, to compensate for missing answers on some returned questionnaires, proportional scores were calculated for each eating behavior scale as described elsewhere (17). Only subjects who had completed >=80% of each EI subscale were included in the analyses.

Information obtained using the HHQ included height, weight, general demographic and lifestyle parameters, and the presence or absence of 22 specific health conditions. The morbidity prevalence in our population was generally similar to the morbidity prevalence in a recent national survey of women aged 45 to 64 years for whom comparable data were available (19).

BMI (kg/m2) was calculated from reported height and weight. In addition, associations of eating behavior with subject-reported blood pressure and total cholesterol values were examined in a subset of respondents who supplied this information (n = 584 for blood pressure; n = 494 for cholesterol). Reported height, weight, blood pressure, and total cholesterol values were validated in approximately 5% of the sample as part of a separate study (16), and differences between reported and measured values were not significant (all p > .18 by paired t test).

For the main analyses, 137 subjects were excluded because they were current smokers, and 34 subjects were excluded because of egregiously incomplete EI questionnaires. In addition, 546 subjects were excluded because of missing health or demographic information, and another 73 subjects were excluded because of reported bulimia, anorexia nervosa, or binge eating patterns. The final sample size was 1252 and was predominantly (95%) white. Ethical approval was given by the New England Medical Center/Tufts University Human Investigations Review Committee.

Statistics
Statistical analyses were performed using SPSS 10.0.7 for Windows and SYSTAT 9.0.1 (SPSS, Chicago, IL). Logistic regression was used to examine relationships between eating behavior and each specific morbidity. In general, regression models considered the influence of dietary restraint, disinhibition, and hunger scores, the interactions between these eating behavior variables, as well as several potential confounding variables (listed below). Examination of the continuous variables in each analysis using normal probability plots failed to reveal any serious departures from normality. Outliers were investigated using the SPSS Explore procedure but were not excluded because their influence on analytic outcome was minimal. The goodness-of-fit of each logistic regression model was checked using Hosmer and Lemeshow's test (p > .05 for all models presented).

Both main effects and interactions between eating behavior variables were examined; however, interaction terms were not significant in any model examined and were not included in the final analyses. Analyses were performed with covariates of current age (year), prior smoking status (never/ever), current hormone replacement therapy (no/yes), postsecondary education level (low = none, vocational school, two-year junior college/high = four-year college, graduate, or professional school), and both with and without current BMI. These specific covariates were chosen because of their likely association with both eating behavior and morbidity in this population.

Multiple linear regression was used to analyze relationships between eating behavior variables and reported blood pressure and cholesterol values. Again, both main effects and interactions between eating behavior variables were examined, and analyses were performed both with and without current BMI, in addition to the covariates listed above. For these analyses, a backward stepwise procedure was used to determine the best set of predictors for each dependent variable.


   Results
Top
 Abstract
 Methods
 Results
Discussion
 References
 
Two sets of analyses were performed, with one set adjusting for age, prior smoking status, current hormone replacement therapy, and education level, and a second set adjusting for these variables plus BMI. This procedure was adopted to examine associations of eating behavior both with and without adjusting for BMI, which has been shown to be independently associated with eating behavior (17)(20)(21). In the present study, higher disinhibition scores were associated with higher current BMI (partial r = .33, p < .001), and dietary restraint, although not independently significant (partial r = .04, p = .130), attenuated the relationship between disinhibition and BMI (interaction term partial r = -.13, p < .001). Detailed results of the analysis of eating behavior and BMI are described elsewhere (17).

Significant models unadjusted for BMI are shown in Table 2 . Disinhibition was associated with an increased likelihood of hypercholesterolemia, leg cramps or pain, indigestion/hiatal hernia/heartburn, and cataract, whereas hunger was associated with an increased likelihood of eczema. Disinhibition was also associated with a decreased likelihood of eczema. Dietary restraint was not associated with any reported morbidity (all p > .055).


View this table:
[in this window]
[in a new window]
  		Table 2. Multivariate Logistic Regression Models of Factors Associated With Specific Morbidity in Women Aged 55–65 y, Adjusting for Age, Past Smoking Status, Hormone Replacement Therapy, and Education Level

 
Fig. 1 illustrates the significant associations between eating behavior variables and reported morbidity, after adjusting for the confounding variables listed above (except BMI). In this figure, each OR has been standardized to reflect a 1-standard-deviation (SD) increase in the corresponding eating behavior subscale (for example, each disinhibition OR reflects a 4.1-point increase in this 16-point subscale). The use of standardized ORs allows a more direct comparison of the relative impact on morbidity of independent variables that are measured using different scales.



View larger version (14K):
[in this window]
[in a new window]
  		Figure 1. Odds ratios (ORs) with 95% confidence intervals for significant independent associations of dietary disinhibition and hunger scores (1) with specific reported morbidities. Logistic regression models from which ORs were calculated were adjusted for current age, prior smoking status, hormone replacement therapy, and education level, but unadjusted for body mass index. To facilitate comparison among variables, each OR was standardized to reflect a 1-standard-deviation increase in the respective eating behavior subscale (i.e., 4.1 and 3.4 units for disinhibition and hunger, respectively).

 
Regression models in which BMI was considered as an additional possible confounding variable are shown in Table 3 . For several diseases, eating behavior variables were no longer significant once BMI was added to the model, suggesting that the influence of eating behavior on these health conditions is mediated by its effect on body weight. In other cases, eating behavior variables that merely approached significance in models that did not include BMI reached significance once BMI was added (for example, the OR for disinhibition in a model examining low back pain was 1.04 [ p = .098] without BMI as a covariate and was 1.06 [ p = .031] with BMI). In models adjusted for BMI, disinhibition was associated with an increased likelihood of low back pain and constipation, and hunger remained associated with an increased likelihood of eczema. Disinhibition also remained associated with a decreased likelihood of eczema, whereas associations of restraint with morbidity were again insignificant (all p > .09).


View this table:
[in this window]
[in a new window]
  		Table 3. Multivariate Logistic Regression Models of Factors Associated With Specific Morbidity in Women Aged 55–65 y, Adjusting for Age, Past Smoking Status, Hormone Replacement Therapy, Education Level, and BMI

 
Significant associations of eating behavior with health conditions, after adjusting for confounding variables including BMI, are shown in Fig. 2. Again, ORs have been standardized to reflect a 1-SD increase in each EI subscale.



View larger version (12K):
[in this window]
[in a new window]
  		Figure 2. Odds ratios (ORs) with 95% confidence intervals for significant independent associations of dietary disinhibition and hunger scores (1) with specific reported morbidities. Logistic regression models from which ORs were calculated were adjusted for current age, prior smoking status, hormone replacement therapy, education level, and body mass index. To facilitate comparison among variables, each OR was standardized to reflect a 1-standard-deviation increase in the respective eating behavior subscale (i.e., 4.1 and 3.4 units for disinhibition and hunger, respectively).

 
In multiple linear regression models predicting reported blood pressure, disinhibition was associated with diastolic blood pressure (partial r = .097, p = .019) in a model unadjusted for BMI. However, disinhibition was no longer significant once BMI was added as a covariate; only age and BMI were significant in the fully adjusted model (partial r = .120, p = .004; partial r = .230, p < .001, respectively). In a model predicting reported cholesterol values, education level was negatively associated with cholesterol value (partial r = -.105, p = .019), but no eating behavior variable was significant (all p > .065).


   Discussion
Top
 Abstract
 Methods
 Results
 Discussion
References
 
In this study, higher dietary disinhibition and hunger scores were associated with small alterations in reported morbidity risk in a large population of nonsmoking older women. In general, higher eating behavior scores were associated with increased risks of morbidity. Whereas the magnitude of risk associated with higher disinhibition and hunger scores was generally small, differences in risk between individuals with the lowest disinhibition or hunger scores and those with the highest scores may be more pronounced (for example, OR for constipation is 4.7 when comparing lowest with highest disinhibition score).

Associations of disinhibition and increased morbidity have been reported previously. In a large population of obese patients undergoing weight-reduction therapy, patients electing surgical intervention had significantly higher BMI, higher disinhibition and hunger scores, lower restraint scores, lower self-reported health status, and greater depression and psychosocial dysfunction at baseline compared with obese controls (12); however, the association of eating behavior with health status independent of body weight was not examined. A similar result was observed in a study that examined eating behavior in relation to mental and physical health-related quality of life in overweight and obese subjects (11). Numerous other studies have also reported associations between dietary restraint and physiological alterations related to bone mineral density and reproductive and endocrine function (8)(9)(13)(14)(15)(22). However, to our knowledge, no previous study has examined the association of eating behavior with specific long-term, self-reported health and illness parameters in older women.

In this study, we examined the relationships of eating behavior and morbidity in regression models both adjusted and unadjusted for BMI. Higher disinhibition scores were associated with an altered risk for six specific morbidities (including diastolic blood pressure) in models unadjusted for BMI. However, disinhibition is also associated with BMI (17), and once our models were adjusted for BMI, the number of these significant associations dropped to three. These analyses, combined with the known morbidity risk of overweight and obesity (23), suggest that in this study population, the influences of eating behavior on morbidity act primarily via mechanisms mediated by body weight.

Several limitations to our study deserve mention. One limitation is our inability to identify or exclude subjects with diagnosed or self-reported depression, a potential confounding variable. Another limitation is the use of self-reported morbidity, anthropometric measures, and blood pressure and cholesterol values. Additionally, the direction of association in our models is unclear, and eating behavior may either influence disease risk, or the presence of disease may influence eating behavior. However, limited evidence suggests that higher disinhibition scores may be associated with dietary patterns that promote the development of disease, such as increased consumption of sweets and butter/margarine (24), decreased consumption of vegetables (Megan A. McCrory, unpublished data, 2001), and increased reported intake of dietary fat (25)(26)(27)(28). Thus, highly disinhibited eaters may have a greater intake of highly refined foods, which may put them at increased risk for constipation and disc disease/low back pain (via decreased dietary fiber and calcium intakes, respectively). Further research is needed to examine potential mechanistic relationships between eating behavior and eczema.

In conclusion, our results suggest that higher disinhibition and hunger scores are associated with small alterations in reported morbidity risk in a large population of nonsmoking older women; no association of dietary restraint and morbidity risk was observed. Although our cross-sectional study design makes the directionality of these relationships unclear, our results indicate at most a relatively minor independent association of eating behavior with morbidity.


   Acknowledgments
 
This research was supported in part by NIH Grants T32AG00209 (NPH) and AG12829 and DK46124 (SBR), and the U.S. Department of Agriculture, Agriculture Research Service under cooperative agreement 58-1950-9-001.

Received May 22, 2001

Accepted June 7, 2001


   References
Top
 Abstract
 Methods
 Results
 Discussion
 References
 

  1. Stunkard AJ, Messick S, 1985. The three-factor eating questionnaire to measure dietary restraint, disinhibition, and hunger. J Psychosom Res. 29:71-83. [Medline]
  2. Pirke KM, Laessle RG, 1993. Restrained eating. Stunkard AJ, Wadden TA, , ed.Obesity: Theory and Therapy 2nd ed. 151-162. Raven Press, New York.
  3. Lowe MR, Maycock B, 1988. Restraint, disinhibition, hunger and negative affect eating. Addict Behav. 13:369-377. [Medline]
  4. Tuschl RJ, Platte P, Laessle RG, Stichler W, Pirke K, 1990. Energy expenditure and everyday eating behavior in healthy young women. Am J Clin Nutr. 52:81-86. [Abstract/Free Full Text]
  5. Rand CS, Kuldau JM, 1991. Restrained eating (weight concerns) in the general population and among students. Int J Eat Disord. 10:699-708.
  6. Westenhoefer J, 2001. Prevalence of eating disorders and weight control practices in Germany in 1990 and 1997. Int J Eat Disord. 29:477-481. [Medline]
  7. Westenhoefer J, 1991. Dietary restraint and disinhibition: is restraint a homogeneous construct?. Appetite. 16:45-55. [Medline]
  8. McLean JA, Barr SI, Prior JC, 2001. Cognitive dietary restraint is associated with higher urinary cortisol excretion in healthy premenopausal women. Am J Clin Nutr. 73:7-12. [Abstract/Free Full Text]
  9. Pirke KM, Tuschl RJ, Spyra B, et al. 1990. Endocrine findings in restrained eaters. Physiol Behav. 47:903-906. [Medline]
  10. Laessle RG, Tuschl RJ, Kotthaus BC, Pirke KM, 1989. Behavioral and biological correlates of dietary restraint in normal life. Appetite. 12:83-94. [Medline]
  11. Marchesini G, Solaroli E, Baraldi L, et al. 2000. Health-related quality of life in obesity: the role of eating behavior. Diabetes Nutr Metab. 13:156-164. [Medline]
  12. Karlsson J, Sjostrom L, Sullivan M, 1998. Swedish obese subjects (SOS)—an intervention study of obesity. Two-year follow-up of health-related quality of life (HRQL) and eating behavior after gastric surgery for severe obesity. Int J Obes Relat Metab Disord. 22:113-126. [Medline]
  13. Barr SI, Janell KC, Prior JC, 1994. Vegetarian vs nonvegetarian diets, dietary restraint, and subclinical ovulatory disturbances: prospective 6-mo study. Am J Clin Nutr. 60:887-894. [Abstract/Free Full Text]
  14. Barr SI, Prior JC, Vigna YM, 1994. Restrained eating and ovulatory disturbances: possible implications for bone health. Am J Clin Nutr. 59:92-97. [Abstract/Free Full Text]
  15. van Loan MD, Keim NL, 2000. Influence of cognitive eating restraint on total-body measurements of bone mineral density and bone mineral content in premenopausal women age 18–45 y: a cross-sectional study. Am J Clin Nutr. 72:837-843. [Abstract/Free Full Text]
  16. Bathalon GP, Hays NP, Meydani SN, et al. 2001. Metabolic, psychological, and health correlates of dietary restraint in healthy postmenopausal women. J Gerontol Med Sci. 56A:M206-M211. [Abstract/Free Full Text]
  17. Hays NP, Bathalon GP, McCrory MA, Roubenoff R, Lipman R, Roberts SB. Eating behavior correlates of adult weight gain and obesity at age 55–65 years in healthy women. Am J Clin Nutr. In press.
  18. Foster GD, Wadden TA, Swain RM, Stunkard AJ, Platte P, Vogt RA, 1998. The Eating Inventory in obese women: clinical correlates and relationship to weight loss. Int J Obes Relat Metab Disord. 22:778-785. [Medline]
  19. Adams PF, Hendershot GE, Marano MA. Current estimates from the National Health Interview Survey, 1996. Vital and Health Statistics. Hyattsville, MD: National Center for Health Statistics; 1999:83–84.
  20. Westenhoefer J, Pudel V, Maus N, 1990. Some restrictions on dietary restraint. Appetite. 14:137-141. [Medline]
  21. Williamson DA, Lawson OJ, Brooks ER, et al. 1995. Association of body mass with dietary restraint and disinhibition. Appetite. 25:31-41. [Medline]
  22. Schweiger U, Tuschl RJ, Platte P, Broocks A, Laessle RG, Pirke KM, 1992. Everyday eating behavior and menstrual function in young women. Fertil Steril. 57:771-775. [Medline]
  23. National Task Force on the Prevention and Treatment of Obesity2000. Overweight, obesity, and health risk. Arch Intern Med. 160:898-904. [Abstract/Free Full Text]
  24. Lahteenmaki L, Tuorila H, 1995. Three-factor eating questionnaire and the use and liking of sweet and fat among dieters. Physiol Behav. 57:81-88. [Medline]
  25. Lawson OJ, Williamson DA, Champagne CM, et al. 1995. The association of body weight, dietary intake, and energy expenditure with dietary restraint and disinhibition. Obes Res. 3:153-161. [Medline]
  26. Fogelholm M, Kukkonen-Harjula K, Oja P, 1999. Eating control and physical activity as determinants of short-term weight maintenance after a very-low-calorie diet among obese women. Int J Obes Relat Metab Disord. 23:203-210. [Medline]
  27. Lindroos AK, Lissner L, Mathiassen ME, et al. 1997. Dietary intake in relation to restrained eating, disinhibition, and hunger in obese and nonobese Swedish women. Obes Res. 5:175-182. [Medline]
  28. d'Amore A, Massignan C, Montera P, Moles A, De Lorenzo A, Scucchi S, 2001. Relationship between dietary restraint, binge eating, and leptin in obese women. Int J Obes Relat Metab Disord. 25:373-377. [Medline]



This article has been cited by other articles:


Home page
 Journals of Gerontology Series A: Biological Sciences and Medical SciencesHome page
C. A. Rideout, W. Linden, and S. I. Barr
High cognitive dietary restraint is associated with increased cortisol excretion in postmenopausal women.
J. Gerontol. A Biol. Sci. Med. Sci., June 1, 2006; 61(6): 628 - 633.
[Abstract] [Full Text] [PDF]

Home page
 Journals of Gerontology Series A: Biological Sciences and Medical SciencesHome page
J. E. Morley
Anorexia and Weight Loss in Older Persons
J. Gerontol. A Biol. Sci. Med. Sci., February 1, 2003; 58(2): M131 - 137.
[Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Services
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
PubMed
Right arrow PubMed Citation

HOME ARCHIVE SEARCH TABLE OF CONTENTS