| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
|
|
|
|||||||
|
|||||||||
Correspondence: Address correspondence to Laurette Dubé, PhD, Desautels Faculty of Management, McGill University, 1001 Sherbrooke Street West, Montreal, Québec, Canada H3A 1G5. E-mail: Laurette.dube{at}mcgill.ca
 |
Abstract |
|---|
 TOP Abstract MethodsResultsDiscussionReferences |
|---|
Key Words: Interpersonal behavior • Interpersonal circumplex • Meal fellowship • Nutrition • Social facilitation
Nevertheless, in contrast to the dietetic aspects of nutritional care, research on the influence of social environment on food intake performed in institutional contexts has been limited. In particular, existing investigations have mainly focused on the impact of the presence or absence of other people during meals. For instance, Edwards and Hartwell (2004) found that the energy intake of patients consuming their meals with others around a table was superior to the intake of those eating alone either in or at their bed. Similarly, Reed and colleagues (Reed, Zimmerman, Sloane, Williams, & Boustani, 2005) found that the food intake of long-term-care residents was higher among those who had their meals in a public dining area than among those who did not. This social facilitation of intake is consistent with findings from studies performed in free-living individuals (Clendenen et al., 1994; de Castro & de Castro, 1989; Redd & de Castro, 1992).
Although the aforementioned findings provide initial evidence for the importance of mealtime social interaction, they offer limited guidance for the development of environmental interventions that could support adequate intake. Results of investigations performed among free-living individuals suggest that the social facilitation of food intake may go beyond the effect of the mere presence of others; it may vary with the number and nature of interactions taking place between dining partners. For instance, the social facilitation of intake has been shown to vary with the number of individuals present (de Castro & Brewer, 1992; de Castro & de Castro, 1989) and the nature of the relationship existing between the interacting individuals, with the effect being stronger for interactions with family members and friends than with coworkers and strangers (Clendenen et al., 1994; de Castro, 1994). In this article, we propose that social facilitation of intake among hospitalized elderly patients can also go beyond the effect of the presence of dining companions. Specifically, we build upon the interpersonal circumplex model of human interactions to evaluate how specific elements of the meal social environment contribute to the social facilitation of elderly patients' intake.
Interpersonal Circumplex Model
The interpersonal circumplex framework conceptualizes any type of human interaction along two fundamental dimensions, namely agency (one's strivings for mastery and power) and communion (individuals' efforts to promote intimacy and union with other members in the interaction; see Wagner, Kiesler, & Schmidt, 1995; Wiggins, 1991). The framework views behaviors performed by both parties in the exchange as influencing both processes and outcomes of the interaction. The model provides a comprehensive picture of interpersonal exchanges and their ebb and flow in terms of the frequency and nature of behaviors performed by different interacting agents, as well as the complementarity of their individual behaviors (Wagner et al.; Wiggins). Typically, interactions taking place during a given time period are observed or self-reported in terms of agentic and communal behaviors exchanged between the individuals interacting, from which different components of the interaction can be assessed.
For instance, one can capture the amount of interaction from the sum of all behaviors performed by individuals participating in the exchange. Moreover, one can assess the nature of the interaction by separately considering the frequency of agentic and communal behaviors expressed by interacting agents. In addition, the framework, by capturing behaviors individually performed by all interacting partners, also offers the advantage of distinguishing the role of behaviors expressed by individuals from the role of behaviors to which they are exposed during a given interaction. This attribute is particularly relevant given evidence that individuals' own interpersonal behaviors can have consequences on their own state and behavior. For instance, behaviors expressed by individuals can induce in them pleasant or unpleasant emotions, depending on whether they are in concordance or discordance with specific personality traits (Côté & Moskowitz, 1998).
Furthermore, the expression of agentic behaviors is akin to the notion of autonomy, which has been shown to play an adaptive role on performance, well-being, medical compliance, and learning (for a review, see Ryan & Deci, 2000). In the context of the social facilitation of intake, dissociating behaviors expressed from those "received" can help uncover their relative role in the effect. Finally, the interpersonal circumplex offers a basis on which to evaluate the degree to which behaviors exchanged between partners complement each other. In the interpersonal circumplex framework, interpersonal complementarity (i.e., the extent to which behaviors of interacting partners fit with each other in a prescribed way; see Tracey, 1994) is characterized by similarity for communal behaviors (e.g., agreeableness invites agreeableness but constrains quarrelsomeness, and vice versa) and by reciprocity for agentic behaviors (e.g., dominance invites submissiveness but constrains dominance, and vice versa; see Kiesler, 1983).
In this article, we adopt the interpersonal circumplex framework to investigate the nature of the association between the food intake of elderly patients and the social interactions taking place among the elderly patients during meals. Specifically, we assess, over repeated episodes, elderly patients' food intake and its relationship with the behaviors they engage in and those to which they are exposed during the meal, in terms of the total amount, nature (agentic or communal behavior frequency), and complementarity of interpersonal behaviors observed.
|
Methods |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
In order to assess the impact of variations in the meal environment on intake, the design of the study consisted of sampling meal episodes within individuals. This methodology is part of a set of empirical methods called experience sampling methods or ecological momentary assessment techniques, which researchers commonly use to assess the role of situations in individual behavior in ecologically valid settings (Schwartz & Stone, 1998) while controlling for individual-level sources of variance. Observations were made for a given participant for three meals a day, every other day until discharge, or for a maximum of 6 weeks. A total of 1,477 meal episodes were observed, for an average of 46.2 (± 28.6) meal episodes per participant.
Participants
Participants were selected among patients admitted to a geriatric facility, where they stayed 4 weeks on average. Because perceptual and emotional reports were collected as part of the broader study, individuals with cognitive impairments and depression, as respectively assessed by the Mini-Mental Status Examination (score below 23 excluded; see Folstein, Folstein, & McHugh, 1975) and the Geriatric Depression Scale-15 (scores above 9 excluded; see Yesavage et al., 1983), were not considered. Overall, 167 eligible patients were identified among the 355 patients admitted to the unit over the 18-month study period. An observation quota of two participants on any given day limited the number of invitations to participate to 83, among which 46 were declined. The most frequent reasons for declining participation were concurrent enrollment in other research projects, extended length of the observation period, unwillingness to eat meals in the dining room, and reluctance toward the collection of three blood samples as part of the broader study. The participation rate was 45% (37 of 83). Among the 37 patients who accepted, 32 completed the study. The sample characteristics are reported in Table 1. Eligible individuals who declined to participate did not differ from those who participated in terms of gender and diagnostic value (t test p values >.75). The protocol was approved by the hospital's Research Ethics Committee. Participants signed an informed consent form and received $50 in compensation.
|
We used three equivalent versions of the recording forms sampling different behavioral items in a counterbalanced order over the three meals of the day. Observations were made during a sequence of 2-minute observation intervals during which coders observed behaviors performed by or directed toward a given participant. This period was followed by a second 2-minute interval during which coders went over all behavioral items on their list and checked those that had just been observed. The same schedule was followed whether one or two participants were under observation during a given meal. For each participant, between three and five 2-minute time intervals (4.4 of them, on average) were observed within each meal episode, which lasted an average of 32 minutes.
All 12 coders received 30 hours of training, which entailed a detailed discussion of all behavioral items and the subsequent coding of videos of interpersonal exchanges. We assessed intercoder reliability immediately following training and on 36 occasions over the course of the study. We measured intercoder reliability as the percentage of agreement by taking the ratio of the number of concordant individual items (i.e., both coded as observed or nonobserved) over the total number of items observed for a given meal (16 items x number of observation periods for that meal). Consistent with past research (Moskowitz, 1990, 1994), reliability was above.85.
We followed a previously validated methodology (Moskowitz & Côté, 1995) to compute behavioral meal-level scores for participants and other patients. We first compiled the number of times each behavioral item was checked for a given meal episode and aggregated items falling under the same interpersonal domain so as to obtain four domain-specific meal-level behavioral scores. We aggregated other patients' behavior that was directed toward participants across different patients when participants interacted with multiple patients. We assessed the interitem reliability for all four domains by measuring Cronbach's alpha for the four behavioral items falling under the same behavioral domain; we found it to be satisfactory for all domains, for both participants (dominance, 
= 0.87; submissiveness, = 0.66; agreeableness, = 0.94; quarrelsomeness, = 0.60) and other patients (dominance, = 0.81; submissiveness, = 0.79; agreeableness, = 0.92; quarrelsomeness, = 0.60). Total and relative frequencies for each interpersonal domain are reported in Table 2. Reported frequencies indicated that agreeableness was, for both participants and other patients, the most frequently expressed behavior, followed by dominance, submissiveness, and quarrelsomeness.
|
Food Intake and Meal Duration
Using the Comstock scale (Comstock, St-Pierre, & Mackierman, 1981), we performed an assessment of participants' food intake through a visual estimation of the serving's proportion that remained as plate leftover for each meal component. The Comstock scale is a 5-point scale (100%, 75%, 50%, 25%, and 0% of portion served). This method has been extensively validated against other techniques (Berrut et al., 2002; Comstock et al.; Dubois, 1990; Williamson et al., 2003). Raters received an exact duplicate of participants' trays to facilitate their estimation of leftovers. We derived the amount of food consumed for each meal component on the basis of the estimated proportion consumed and according to standardized portions and recipes in use in foodservice operations.
To reduce measurement error, we performed rigorous and systematic monitoring of portion sizes and yield from standardized recipes during the study period (Paquet, St-Arnaud-McKenzie, Ferland, & Dubé, 2003). We obtained energy consumption information (in kilocalories) for each meal by using the NutriWatch Nutrient Analysis program (version 6.1.5F Delphi, E. Warwick, Cornwall, PEI, 1997). We assessed the intercoder reliability of the intake measures provided by all coders before and halfway through the study, and by pairs of coders on 17 occasions over the course of the data collection. Reliability measured through the intraclass correlation was found satisfactory in all cases (average reliability above.98, with a minimum of.84). Finally, coders also recorded the time that participants started and finished eating, from which we derived the meal duration.
Statistical Analyses
The sample consisted of 1,477 meal episodes nested within 32 participants. In order to account for the clustering of observations within participants, we performed multilevel regression analyses, which offer a flexible method of modeling repeated measurements by accounting for both within-participant (i.e., from one meal to another for a given participant) and between-participant (i.e., from one participant to another) variability (Goldstein, 1986). The utility of multilevel approaches is increasingly recognized in a variety of research domains, including research in gerontology (Pekkarinen, Sinervo, Perälä, & Elovainio, 2004; Sikorska-Simmons, 2006; Tornatore & Grant, 2002). In the present study, we estimated multilevel regression models by using the MLwiN software (version 1.10.0006, Multilevel Models Project Institute of Education, Bristol, UK). We considered meal-level interpersonal variables and relevant covariates as predictors of energy intake (we performed similar analyses with protein intake, and we found similar results. For the sake of conciseness, we present only the results for energy intake). In addition, we performed analyses separately for participants' and other patients' behavior, because of the strength of the correlations between such measures on a given episode (see Table 3).
|
|
Results |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
|
Underlying Mechanisms
The frequency of interactions is intrinsically tied to the duration of the meal. Because longer meals are also more likely to be tied to larger intake, our results could be explained by the relationship between meal duration and the number of interactions. In order to disentangle the role of interpersonal behavior and meal duration in determining patients' intake, we assessed the mediating role of meal duration in the relationships that emerged between interpersonal variables and intake. We defined meal duration as the interval between the time participants started and finished eating.
Following standard mediation analysis procedures (Baron & Kenny, 1986), we first verified that the interpersonal variables that were previously found to be related to intake were also significant predictors of meal duration. To do so, we performed multilevel regression analyses with interpersonal variables as predictors of meal duration. Results reported in Table 5 show that duration was indeed positively related to all such variables.
|
|
|
Discussion |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
We also found evidence that the relationship varied across different dimensions of interpersonal behavior. Specifically, we found that participants' and other patients' communal behaviors were related to energy intake, whereas we found no such relationship for agentic behaviors. The communion effect on energy intake echoes previous results on the greater consumption of dessert in the presence of friends than in the presence of strangers (Clendenen et al., 1994). When considering the complementarity of such behaviors, we found that both agentic and communal dimensions emerged as significant predictors of intake. The aforementioned results highlight the fact that interaction per se may not be enough to explain the impact of the social environment on intake and that the specific nature of these individual behaviors and their complementarity may play an important role in the effect. This conclusion does not contradict our findings concerning the total amount of interaction because the majority of behaviors observed were at the positive end of both communal and agentic dimensions. Finally, this conclusion also corroborates existing evidence that the social facilitation of intake in free-living individuals varies with the nature of the relationship that exists between the individuals who are interacting, with the effect being stronger for interactions with family members and friends than with coworkers and strangers (Clendenen et al.; de Castro, 1994).
In our investigation of the mechanism underlying the aforementioned relationships, we found that the effect of the social environment on energy intake was only partially explained by meal duration. This mediation by meal duration is consistent with results of previous investigations performed among free-living individuals (de Castro, 1990; Feunekes, de Graaf, & van Staveren, 1995) or in laboratory settings (Pliner, Bell, Hirsch, & Kinchla, 2006). The partial mediation of the effect on energy intake indicates that the social environment may not impact intake only by prolonging meal duration. The results could also potentially be explained by a more positive perception of food taste in response to mealtime interactions (de Castro, 1994; Feunekes et al.). However, post hoc analyses in which we looked at the impact of the same interpersonal variables on measures of perceived palatability of the food (which were also collected as part of the broader study) indicated no impact of interpersonal variables on such measures. Future investigations should explore other potential underlying mechanisms, such as change in emotional states in response to mealtime interactions (de Castro, 1994; Feunekes et al.).
The results of the present study should be interpreted in light of the study's limitations. First, it could be argued that our study, by neglecting to consider interactions with care providers, fails to comprehensively account for the social environment of our participants. However, we deliberately decided to focus on interactions among patients to gain insights into the effect of the presence of dining partners reported in the literature. Second, participation in this study was 45%; although patients who declined did not differ from participants in terms of gender and diagnostic value, we cannot rule out the possibility that participants differed in terms of health status or expressed different interpersonal behaviors than did nonparticipants. Future research should therefore examine the generalizability of the results to other institutionalized elderly populations. Finally, we acknowledge that the number of participants in the study is small compared with sample sizes obtained in cross-sectional studies. However, the objective of repeated-sampling designs like the one used herein is to focus the analyses on fluctuations in the environment, while controlling for individual differences, and not on individuals per se, which minimizes the consequences of a limited number of participants. Finally, future research could also assess potential changes in the social interaction effect over the length of the stay, as the relationship among dining companions and potentially the nature of their interactions evolve.
Conclusions
Our findings highlight the importance of meal fellowship in determining elderly individuals' intake, which has already been identified as a potential factor affecting appetite in the elderly population (Wikby & Fägerskiöld, 2004). Specifically, our findings provide insights into the specific components of the meal social environment that contribute to shape food intake. Such results can inspire new strategies to ensure adequate intake in institutions.
For instance, staff members could encourage meal fellowship by rearranging the meal environment in ways that are more conducive to interaction, either by asking patients to eat at a common table or by minimizing the number of mealtime distractions. Similar changes in layout have already been shown to affect eating behaviors among elderly patients. For instance, Melin and Götestam (1981) showed that by rearranging furniture and changing mealtime routines in a way that is more conducive to interactions, staff members found that psychogeriatric patients were communicating more among themselves in addition to showing better eating behavior during coffee breaks. Other potential interventions include the assessment of the complementarity of the behaviors of patients before seating arrangements are made and increasing patients' awareness of the benefits of interacting with their fellow patients. Finally, we acknowledge that decisions to have patients eating in a common dining room may depend on many factors (e.g., number of staff available, patients' needs for assistance or supervision, patients' own preferences) and that these factors should not be overseen. However, on the basis of our results, we argue that the need for social interactions should be among these considerations.
|
Footnotes |
|---|
1 Department of Social and Preventative Medicine, Université de Montréal, Québec, Canada.
2 Research Center on Aging, Graduate Institute of Geriatrics, Université de Sherbrooke, Québec, Canada.
3 School of Business and Information Technology, University of Ontario Institute of Technology, Oshawa.
4 Graduate Institute of Geriatrics, Université de Montréal Research Center, Québec, Canada.
5 Desautels Faculty of Management, McGill University, Montréal, Québec, Canada.
Decision Editor: William J. McAuley, PhD
Received for publication July 14, 2007. Accepted for publication September 21, 2007.
|
References |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| All GSA journals | Journals of Gerontology Series A: Biological Sciences and Medical Sciences | Journals of Gerontology Series B: Psychological Sciences and Social Sciences |
