Internal and External Antecedents of Binge EatingEpisodes in a Group of Women with

Bulimia Nervosa

Anne Waters,1* Andrew Hill,2 and Glenn Waller1

1 Department of Psychology, University of Southampton,Southampton, United Kingdom

2 Division of Psychiatry and Behavioural Sciences, University of Leeds,Leeds, United Kingdom

Accepted 10 March 1999

Abstract: Objective: This research examined the internal and external factors that predictwhether a food craving develops into a binge in bulimia nervosa. Method: Fifteen bulimicwomen kept diary records of food cravings over the course of a week, noting their internalstate, environmental triggers, and whether the craving developed into a binge. Eighty cravingswere examined, of which approximately one half developed into binges. Results: At thebivariate level, binge eating resulted from cravings more often if the women were alone, inthe morning, and following eating the craved food. These factors interacted, so that bingeswere more likely if the women avoided a craved food in the morning, but not later in the day.Conclusions: These findings support a model of binge eating for which both internal states(craving, eating the craved food) and external factors (time of day, social circumstances) needto be taken into account. Relevant models of craving are discussed, along with their impli-cations for clinical practice. © 2000 by John Wiley & Sons, Inc. Int J Eat Disord 29: 17–22,2001.

Key words: food craving; environmental triggers; binges

INTRODUCTION

The binge-purge cycle of bulimic patients is largely understood within models thatacknowledge a combination of general predisposing factors and immediate specific an-tecedent triggers (McManus & Waller, 1995). Research into the triggers of these bingeeating episodes has predominantly focused on internal factors such as cognitions, hunger,and cravings for specific foods (Fairburn & Cooper, 1989). Dietary restraint and foodcravings have been seen as the most powerful triggers of the urge to overeat (Striegel-

*Correspondence to: Anne Waters, Department of Psychology, University of Southampton, Highfield,Southampton SO17 1BJ, United Kingdom.

© 2000 by John Wiley & Sons, Inc.

Prod. #1562

Moore, Silberstein, & Rodin, 1986). However, there exists a body of research that dem-onstrates that this model is insufficient to explain bulimic behaviors and that other inter-nal and external triggers should be taken into account. For example, other internal factorsthat trigger binge eating include negative affect (Meyer, Waller, & Waters, 1998). Externaltriggers include time of day, being alone, and unstructured time (Davis, Freeman, &Solyom, 1985; Grilo, Shiffman, & Carter-Campbell, 1994; Woell, Fichter, Pirke, & Wolfram,1989).

It is clear that internal and external cues to binge eating cannot usefully be consideredin isolation from each other (Schlundt, Virts, Sbrocco, & Pope-Cordle, 1993; Weingarten &Elston, 1991). In particular, it remains to be determined which internal and externaltriggers result in binge eating episodes when the bulimic individual experiences a foodcraving. This study explores a range of environmental, situational, and food-related fac-tors, in order to identify critical internal and external antecedents of binge eating episodesin bulimic women when they experience cravings. As well as considering the role ofindividual factors, this study explores the interaction of internal and external cues inproducing binge eating.

METHOD

Design

This was a naturalistic study of food craving and binge eating behavior among a clinicalpopulation of bulimia nervosa patients. Participants completed a self-monitoring ques-tionnaire each time they experienced a craving for a particular food and recorded relevantantecedents (environmental and situational factors). Data were then divided into cravingsthat were and were not followed by a binge eating episode. Thus, the craving was thesubject of the analysis, rather than the participant.

Participants

A clinical sample of 15 women with bulimia nervosa (American Psychiatric Association,1994) were recruited from the waiting lists of two clinical psychology outpatient depart-ments. The mean age of the women was 24.8 years (SE = 0.7) and their mean body massindex (BMI; kg/m2) score was 22.9 (SE = 1.3). All participants were assured of confiden-tiality.

Measures and Procedures

Participants completed two measures relevant to this study, a food intake diary and acraving record. The food intake diary was used continuously over 7 days to assess foodintake. This included self-identified binge eating episodes (defined as an episode of eatingassociated with a sense of loss of control). Over the same time period, the craving recordwas used each time that a craving (defined as a strong urge or desire for a particular food)was experienced, regardless of whether a binge occurred. This measure provided detailedinformation about environmental precipitants of participants’ food cravings. Checklistswithin the craving record identified situational aspects and specific triggers of the foodcraving (seeing the craved food, seeing other food, thinking of craved food, thinking of

18 Waters, Hill, and Waller

other food, eating the craved food, eating other food, being in company, time of day,location).

Data Analysis

Data provided by the craving record were divided into two groups, that is, cravingsthat were followed by a binge eating episode and those that were not. Chi-square analyseswere used to determine whether associations existed between the dependent variables(situational and environmental factors) and the occurrence of a binge. Next, hierarchicallog-linear modeling was used to test for interactions involved between the factors that hadindependent effects.

RESULTS

Over the course of 7 days, the 15 women experienced food cravings on 80 occasions(mean = 5.33 cravings per person per week). They binged on 43 of those occasions (mean= 2.87 binges per person per week). Table 1 shows the associations between individualpotential environmental factors and whether or not the women binged. Chi-square tests

Table 1. Environmental factors associated with binge eating among women following theexperience of cravings

Internal/External Factors

Behavioral Response to Craving

x2 pDid not Binge Binged

See craved foodNo 24 33 1.37 NSYes 13 10

See other foodNo 36 41 0.21 NSYes 1 2

Think of craved foodNo 19 26 0.67 NSYes 18 17

Think of other foodNo 33 35 0.95 NSYes 4 8

Eat craved foodNo 36 32 8.17 .004Yes 1 11

Eat other foodNo 33 41 1.09 NSYes 4 2

In companyNo 29 21 7.41 .007Yes 8 29

Time of dayMorning 4 14 7.04 .03Afternoon 20 22Evening 13 7

LocationHome 21 22 2.95 2.95Work 8 16Other 8 5

Antecedents of Binge Eating 19

were used to determine the reliability of those associations. When expected frequencieswere below 5 (“see other food,” “eat other food”), Fisher’s Exact Test was used, yieldingidentical results to the chi-square analysis in both cases.

There were no links between binge eating behavior and a number of the potentialenvironmental cues (seeing the craved food, seeing another food, thinking of the cravedfood, thinking of another food, eating another food, location). However, binging was asignificantly more common response to craving when the women ate the craved food,when they were alone at the time of the craving, and when the craving occurred in themorning. When the craving occurred in the evening, they were less likely to binge.

These findings suggest that there are three specific environmental correlates that influ-ence whether or not craving is likely to develop into binging: eating the craved food, beingalone, and time of day. However, it is unlikely that these factors are independent of eachother. It is, therefore, valuable to determine whether they interact to enhance or reduce thelikelihood that cravings will result in a binge. Table 2 shows these three factors in acontingency table, together with whether or not the women binge in response to cravings.It also shows the results of a hierarchical log-linear model (backward eliminationmethod), testing the interactions involved between the factors. As is conventional in suchanalyses, the best model is described by the set of the highest-order interactions thatinvolve all of the factors.

The best model consisted of three interactions. The first of these effects (the presence ofother people × whether or not a binge occurred) has already been described (Table 1). Itshowed that the women were more likely to binge when alone. The second interaction(eating the craved food × the presence of other people × time of day) shows that thewomen were less likely to eat the craved food when alone in the evening (none of 10

Table 2. Association of binging with time of day, presence of other people, and whether or notthe craved food had been eaten

Time of Day

In Company

Ate Craved Food

Morning Afternoon Evening

No Yes No Yes No Yes

No Yes No Yes No Yes No Yes No Yes No Yes

Binged No 1 1 2 0 2 0 18 0 4 0 9 0Yes 5 2 7 0 4 5 10 3 6 0 0 1

Effect X2 p

Binged (B) 0.45 NSAte craved food (E) 43.3 .001Company (C) 5.05 .025Time of day (T) 12.5 .002B × E 5.34 .021B × C 7.83 .006B × T 7.12 .029C × E 3.57 NSC × T 7.44 .024E × T 3.22 NSB × E × C 0.01 NSB × E × T 7.72 .022B × C × T 4.09 NSE × C × T 7.64 .021B × E × C × T 0.32 NS

20 Waters, Hill, and Waller

occasions) than when alone in the afternoon (5 of 11 occasions) or in the morning (threeof nine occasions). The third effect (eating the craved food × the time of day × whether ornot a binge occurred) shows that the women were more likely to binge when they had noteaten the craved food in the mornings (binging on 12 of 15 occasions). However, theywere less likely to binge if they had not eaten the craved food in the afternoon (bingingon 14 of 34 occasions) or in the evening (binging on 6 of 19 occasions).

To summarize, binging in response to cravings was influenced by three individualenvironmental factors: time of day, whether the craved food was eaten, and whether thewomen were alone. The women were more likely to binge when alone. However, therewere two higher-order effects to take into account. First, eating the craved food whilealone was more likely in the morning or afternoon than in the evening. Second, not eatingthe craved food is not always protective, because such an avoidance was more likely to beassociated with a binge in the morning than later in the day.

DISCUSSION

This study examined the critical internal and external antecedents of food cravings thatdevelop into binge eating episodes in bulimic patients. When examining individual pre-dictors, a number of significant environmental and food-related factors were identified.Considering external factors, both time of day and social context were identified as criticaldeterminants of whether food cravings develop into binge eating. A binge was more likelyto occur if the craving occurred in the morning, but was less likely if it occurred in theevening. A binge was also more likely if the woman was alone. Of the food-related factors,the act of eating the craved food was identified as a highly specific predictor of whetherthe craving experience resulted in binge eating (on any food). There were also two inter-action effects that involved these environmental and food-related predictors. First, eatingthe craved food while alone was more likely to occur in the morning or afternoon than inthe evening (although there was no relationship with the likelihood of binging). Second,not eating the craved food was more likely to be associated with a binge in the morningthan later in the day.

The current findings are partly compatible with previous studies that show that bingeeating episodes are more likely to occur when the bulimic patient is alone (Davis et al.,1985; Grilo et al., 1994). Binge eating is well established as being associated with height-ened negative affect and may play a particular role in the reduction of awareness ofloneliness (Meyer et al., 1998). However, the finding that food cravings were more likelyto result in the food being eaten if they occurred in the morning or afternoon goes beyondthe existing literature on these links (Schlundt et al., 1993), which has tended to focus onsimpler, bivariate associations.

The most important (and novel) finding is that binges are more likely to occur in thecontext of a combination of environmental and food-related variables (i.e., not eating thecraved food in the morning). This finding could be explained if one assumed that there aretwo craving mechanisms. First, a craving in the morning might reflect a homeostaticappetitive mechanism in response to not having eaten for a considerable time. In suchcircumstances, ignoring a craving for a food in the morning would result in an extremelevel of hunger (due to the overnight fast), and this hunger would drive a binge. Second,the cravings that are experienced later in the day might be driven by a process more akinto Marlatt’s (1987) model. For example, a craving serves the function of offering expec-tation of pleasurable relief from an aversive experience. This second process means that

Antecedents of Binge Eating 21

some cravings are more likely in response to negative emotional states, rather than fooddeprivation.

Current starve-binge models of bulimia nervosa (Fairburn & Cooper, 1989) suggest thatthe bulimic cycle is underpinned by restraint and dichotomous thinking, and do notaddress specific critical antecedents. Affective state models move us toward an under-standing of bulimia nervosa that emphasizes intrapsychic antecedents of binge eating.Recent work (McManus & Waller, 1995; Meyer et al., 1998) emphasizes the need for anintegration of these two models, in order to increase our psychological understanding ofthe processes maintaining bulimia nervosa. The findings of this study support that inte-grative position, stressing the role of both internal and environmental factors as imme-diate antecedents to binge eating episodes. Further empirical investigation of these criticalantecedents is essential for an integrated model of bulimia nervosa that generates im-proved treatment protocols. Clinical work to reduce the likelihood of binge eating mightinclude components that address the full range of antecedents, including internal factors(changing the availability of specific favored foods) and external factors (restructuring oftime, attention to social conditions).

REFERENCES

American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed.). Wash-ington, DC: Author.

Davis, R., Freeman, R., & Solyom, L. (1985). Mood and food: An analysis of bulimic episodes. Journal ofPsychiatric Research, 19, 331–335.

Fairburn, C.G., & Cooper, P. (1989). Eating disorders. In K. Hawton, P.M. Salkovskis, J. Kirk, & D.M. Clark (Eds.),Cognitive behaviour therapy for psychiatric problems (pp. 227–314). New York: Oxford University Press.

Grilo, C.M., Shiffman, S., & Carter-Campbell, J.T. (1994). Binge eating antecedents in normal-weight non-purging females: Is there consistency? International Journal of Eating Disorders, 16, 239–249.

Marlatt, G.A. (1987). Craving notes. British Journal of Addictions, 82, 42–43.McManus, F., & Waller, G. (1995). A functional analysis of binge-eating. Clinical Psychology Review, 15,

845–863.Meyer, C., Waller, G., & Waters, A. (1998). Emotional states and bulimic psychopathology. In H. Hoek, M.

Katzman, & J. Treasure (Eds.), The neurobiological basis of eating disorders. (pp. 271–289), Chichester: Wiley.Schlundt, D.G., Virts, K.L., Sbrocco, T., & Pope-Cordle, J. (1993). A sequential behavioural analysis of craving

sweets in obese women. Addictive Behaviors, 18, 67–80.Striegel-Moore, R.H., Silberstein, L.R., & Rodin, J. (1986). Toward an understanding of risk factors for bulimia.

American Psychologist, 41, 246–263.Weingarten, H., & Elston, D. (1991). Food cravings in a college population. Appetite, 17, 167–175.Woell, C., Fichter, M.M., Pirke, K.-M., & Wolfram, G. (1989). Eating behavior of patients with bulimia nervosa.

International Journal of Eating Disorders, 8, 557–568.

22 Waters, Hill, and Waller