 |
INTRODUCTION |
Food sharing, a nearly universal practice (Goldschmidt 1966
), refers to a complex of cultural rules and behaviors governing the distribution and redistribution of food among individuals. A common food-sharing practice may be termed shared-plate eating, wherein foods are presented to two or more individuals on a single bowl or plate and are shared among the group. Although a substantial amount of work has been done on food sharing both within and between households (Blurton Jones 1984, Engle 1990, Feinman 1979, Pelto 1984, Rogers and Schlossman 1990, Van Esterik 1985), limited data are available that examine dietary intake in a shared-plate situation or relate this common practice to nutritional risk.
The feeding of young children differs substantially among cultures (Dettwyler 1989); from studies of intrahousehold food allocation related to food sharing, we know that household behaviors can result in certain individuals receiving disproportionately more or less of the household food supply. Most commonly, females are reported to receive less food of "poorer quality" than their male counterparts (Basu et al. 1986, Carloni 1981, Chen et al. 1981, Gittelsohn 1989 and 1991). In Guatemala, male heads of households received proportionally more dietary protein, whereas female heads of household consumed more energy than other household members (Engle and Nieves 1993). Other researchers have found that although female adults may consume poorer quality foods, their female children were not discriminated against in this manner (Abdullah and Wheeler 1985, Basu et al. 1986, Gittelsohn 1991, Harriss 1990).
The specific cultural rules that govern intrahousehold food allocation patterns can also influence shared-plate eating. For example, the status of the food consumer within the household often determines the social interactions during mealtime (Panter-Brick 1995, Van Esterik 1985). If female children are generally served after older males, based on the social hierarchy of the family, a young girl may be less likely to reach for choice foodstuff (e.g., meat) first if eating with an older brother.
Much of the literature that does exist on shared-plate eating comes from ethnographic case studies. These have shown that general societal norms promoting beneficial exchange and cooperation are often echoed within the etiquette surrounding the sharing of food. For example, in Zimbabwe, children learn early who eats first and who eats with whom, based on their social standing (Farb and Armelogos 1980). In The Gambia, food-sharing etiquette requires that those sharing from the same bowl wait for the senior among them to invite them to start the meal and never encroach on their neighbor's portion (Beckerleg 1995). In a study in the hill regions of Western Nepal, Gittelsohn (1989) found that shared-plate eating occurred infrequently, possibly the result of beliefs surrounding the pollution of food that occurs by touching another's served food. In this setting, shared-plate eating also seemed to reflect economic status, with members of poorer households more likely to consume food from the same plate.
Other factors such as the characteristics and dynamics of a shared-plate episode may also directly influence an individual's consumption. Such characteristics include the number of people involved; the age, sex, identity (e.g., mother) and health of each participant; the type and amount of food shared; and the sequence in which the sharing occurs (i.e., did they begin eating together or did one or more come later?). Young children who compete with others in a shared-plate situation may be less able to procure food because of their smaller size. In households in which food is scarce, intense competition may place a child at an even greater disadvantage. On the other hand, other factors can ensure that a child will receive his/her due portion. For example, mothers or others may actively encourage small or sick children to eat more food (Engle and Zeitlin 1996). The effects of food sharing on the nutritional status of children or women, however, remains unknown.
The dearth of published literature on this subject is due in part to the methodological difficulty of examining the dynamics of shared-plate eating. The only way to capture actual behaviors is to observe specific mealtime interactions. One published study in The Gambia (Hudson 1995) estimated food intake (actually energy intake) from shared bowls of food on the basis of a distribution algorithm that used a nonproportional relationship between energy requirements and body weights. The validity of this algorithm was tested against total energy expenditure (as determined by measuring doubly-labeled water) and found to be within 80% of total energy expenditure. Although this may be suitable to estimate total energy intake within a shared-plate setting, it does not assess intakes of micronutrients, such as vitamin A.
In this paper we use highly specific event-sampled data from multiple day-long continuous observations to characterize shared-plate eating in a rural population of Nepali children among whom this practice is common. The nutritional risks of shared-plate eating are examined in children at high risk of vitamin A deficiency (having a known previous history of xerophthalmia) compared with lower risk children having no history of xerophthalmia.
| |
SUBJECTS AND METHODS |
This study was conducted in three village development committees (formally panchayats) with a population of ~15,000 in the Sarlahi District, located in the central lowland region of Nepal. In January 1992, we conducted a community-wide ocular survey of 3735 children (aged 1-6 y) from 2389 households. A detailed description of the methodology has been published elsewhere (Shankar 1995). Children with clinical signs of xerophthalmia or a current history of night blindness were identified by senior ophthalmic assistants. Eighty-six households had at least one xerophthalmic child between the ages of 1 and 6 y. Cases were treated with high-potency vitamin A according to the WHO guidelines (Sommer 1982) and referred to the local health post for follow-up. No nutritional education or counseling was provided to the parents or guardians regarding the cause or prevention of vitamin A deficiency at the time of diagnosis. Five families either refused to participate in the study or migrated from the district, resulting in 81 households enrolled in the study. Each clinically diagnosed child between the ages of 1 and 6 y served as the "case focus child" in the household.
Controls were selected from the pool of households whose surveyed children exhibited no xerophthalmia and for which there was no reported history of night blindness in any sibling or the child's mother. Control households were matched to case households based on the age of a control to a case focus child (±3 mo) and on age of the next youngest sibling (±4 mo), if present. Matching by younger sibling was done to adjust a priori for potential confounding influences in the feeding behavior of a focus child that could result from the presence of a younger sibling in the household and to examine feeding behaviors of younger siblings from high risk and lower risk households.
Following the survey and selection, the 162 households were monitored longitudinally for 15 mo using day-long direct continuous observations to assess dietary intake and related behaviors. Study households were visited seven times from March 1993 to June 1994, with each visit ~2 mo apart. Our observers followed the focus children for ~8 h during the day, covering the two major meals and much of the day-long snacking. Of the 162 households enrolled, 144 completed all seven visits, another 10 had six visits and the remaining 8 households had fewer than six. Ten local male Nepalese whom we trained for 3 mo in observational techniques and food weight estimation served as our observers. Observers were masked to their household's pretreatment status.
Each observer was responsible for collecting data for an average of nine case-control household pairs (18 total) over the 7 d of observation. The observers were required to prioritize their recording of behaviors with respect to individuals and activities. Food consumption by the focus child had the highest priority. The observers used event sampling (Bentley et al. 1994) to record over 110 different behaviors of interest. If there was no change in the child's behavior over an extended period of time (e.g., sleeping or playing), the observers recorded the behavior every 5 min regardless. The behavioral codes were broadly categorized into seven groups including hygiene-related (12 codes), food-related (36), child care (14), talking (14), work (14), general activities (12) and observer-related (5). To facilitate memorization by the observers, all codes were 5 letter mnemonic codes created from Nepali descriptions of the behavior of interest.
The dietary data included herein represent food consumed during the observational periods only. In the case of individual-plate eating, observers visually estimated the amount of food served and food lost to spillage or other means. The total food-specific portion size for that feeding episode was calculated based on the estimated amount served minus the estimated food lost. During shared-plate feeding episodes, observers would visually estimate the food-specific totals served on the plate and the total consumed by each person, except for the last person eating on the shared-plate. The portions for the last person in the shared-plate feeding episode were calculated by subtracting the amount consumed by every other person eating from the shared-plate from the total amount of food served on the plate. Each line of dietary data included the food name, unit of measurement (grams), whether it was served, consumed or lost, the identities of the server and recipient of the food, and the food quality or condition (e.g., watery pulses or meat with mostly bones). In the case of food mixtures from a single recipe, only the two most prominent foods were listed, in order of decreasing quantity. In the case of non-recipe food mixtures, such as rice, pulses and vegetables, each food (or two most prominent ingredients of that food) was recorded, in order; the total consumed was specified as a food mix and each food assigned a quantity. Each specific food was then categorized within a food group. The following food groups were used: grains (e.g., rice, corn or wheat); vegetables with low or no carotenoid content (e.g., white gourds, cauliflower or potatoes); carotenoid-rich vegetables (e.g., dark green leafy vegetables or pumpkin); pulses (e.g., lentils or chick peas); fruits with low or no carotenoid content (e.g., apples, pears or bananas); carotenoid-rich fruits (e.g., mango, papaya or jackfruit); meats or fish (e.g., chicken, goat or pond fish); and dairy products (e.g., milk or yogurt). Specific observational training procedures used in the study have been previously described (Gittelsohn et al. 1994).
Assessment of the shared-plate eating event.
Shared-plate eating events were recorded in two ways. Once it was clear that two or more individuals were eating from a single plate, a specific activity code was used to signify that the actor and recipient were eating together. This code was assigned to all members of the shared-plate episode and also to individuals who joined a meal in progress. The second method for recording food eaten in a shared-plate setting was to use a special food condition code for every food that was eaten in a shared-plate setting. This double-coding system enabled us to readily identify shared-plate episodes.
A feeding episode was defined to be all foods consumed by the focus child, as part of a meal or a snack, not more than 30 min apart. We created an operational definition of a meal, based on the assumption that, within the Nepali context, meals tend to involve several individuals consuming a number of different foods together. A meal was defined as any feeding episode that involved three or more individuals and involved consumption of five or more distinct foods. On the basis of this definition, 74% of the feeding episodes (n = 4277) observed were classified as meals, whereas the remaining 26% were classified as snacks.
Shared-plate feeding episodes were categorized by age and sex of at least one sharer, excluding the focus child. Because there may be more than one food sharer in any shared-plate feeding episode, these are nonexclusive groups. We also noted the extent of interplate food sharing, which is defined as food already served to an individual in one plate that is given to another recipient having another plate.
Statistical analyses.
Observations were quantified at the level of the feeding episode (meals and snacks) for various outcomes including the estimated quantity in grams consumed for each of eight food groups or the presence or absence of each food group. Because each focus child was observed on multiple days and during multiple meals per day, these outcome measures are expected to be positively correlated within children.
To adjust for within-child correlation, we used one of two approaches. First, for binary responses, including the presence or absence of each food group, we computed robust variances based on the generalized estimating equations for marginal models (Zeger et al. 1988), assuming equivalent correlations. Second, for normally distributed response variables, such as the estimated quantities consumed for each food group, we assumed a random coefficients model, allowing a separate slope for each child, a repeated structure for observation days and an autoregressive correlation between meals within a day. Statistical differences were calculated on the basis of the difference between the least-square means. A Scheffé's adjustment was used to account for multiple comparisons. These regressions were run using PROC MIXED (SAS Institute 1995).
Odds ratios (OR) were adjusted for the sex of the focus child and socioeconomic status (SES) of the household, sequence of the observation day and feeding episode within the day. The SES score was based on an additive scale comprised of material possessions and house quality. The final scale included the following eight items (in currently usable condition): flashlights (1+), watch (1+), cycle (1+), armoire (1+), beds (2+), number of rooms in the house (2+), a roof made with more than just straw and irrigation pumpset. The 
coefficient for this scale was 0.78, the mean score value was 3. Households were categorized as having relatively higher SES on the basis of an SES scale score >3.
Verbal informed consent from head of households and mothers was obtained before their participation in the survey and again after selection in the case-control study. This study was approved by the Nepal Health Research Council, Kathmandu, Nepal and the Committee on Human Research, Johns Hopkins School of Hygiene and Public Health, Baltimore, MD.
| |
RESULTS |
Classification of child feeding episodes.
As a prerequisite to determining differences in feeding patterns between case and control children, it was necessary to develop a generalized classification paradigm for feeding episodes. Table 1 categorizes feeding episodes by type of food sharing, eating location and meal type (snack or meal) for both case and control households. No food sharing occurred in 38% of all feeding episodes (meals) and 22% of all snacks. Shared-plate eating was the predominant type of food sharing (26% of all feeding episodes) compared with interplate food sharing (14%). Shared-plate eating in study households occurred nearly exclusively at meals within the home (24% of all feeding episodes).
View this table:
[in this window]
[in a new window]
|
Table 1.
Feeding patterns for all study children includes type of food sharing, whether a meal or snack, and location
of feeding episode1
|
|
The majority of shared-plate feeding episodes (88%) can be described as multiple simultaneous sharing in which all individuals are served at the same time. The food was not divided into sections when initially served; individuals tended to separate, mix and consume their own portion and, in general, had equal access to the food provided. At any time during the meal, any member of the meal group could request more food. Each member leaves the shared plate as they finished. Post-served sharing (7%) occurs when an individual begins eating and another individual sits down later to eat, often separating a portion of the mixed food for him/herself. The remaining 5% of shared plate feeding episodes were classified as multiple post-served sharing. In this case, the newest addition to the shared plate must acquire food from either individuals currently eating off the plate or ask for more food. There were no differences in the frequency of the above-described types of food sharing by case status.
Shared-plate vs. individual plate eating.
We first examined the effect of shared-plate eating on the variety and quality of food consumption in both groups of focus children. For these analyses, individual-plate eating includes both feeding episodes in which no food sharing or only interplate food sharing occurred. In general, a child eating from a shared plate was more than two times as likely to consume any of the food groups examined compared with children eating alone (Table 2). Specifically, this included grains, vegetables, carotenoid-rich vegetables, pulses, meats and fish and dairy products.
View this table:
[in this window]
[in a new window]
|
Table 2.
Odds of consuming different food groups during shared-plate relative to individual-plate feeding episodes
for all study children1
|
|
Comparing food portion sizes between individual and shared-plate feeding episodes, for most of food groups examined, we found, that children also consumed larger portion sizes if eating from a shared plate (Table 3). Portion sizes for grains were more than 1.5 times larger for children in shared-plate situations than if they ate individually. In the case of non-carotenoid-rich vegetables, pulses, non-carotenoid-rich fruits and dairy products, eating from a shared plate increased the portion size by nearly one third. There were no significant differences in portion sizes between cases and controls for any food group examined.
View this table:
[in this window]
[in a new window]
|
Table 3.
Average total food portion size consumed per feeding episode of study children, by food group,
plate-eating status and case status1,2
|
|
Feeding patterns in case and control households.
General feeding behaviors of focus children were nearly identical between case and control households (data not shown). The average number of feeding episodes per child per observation day was 3.9. At least one shared-plate eating episode was observed on 65% of all observation days. If eating from the same plate, on average, 2.6 people ate together. Of all feeding episodes, the focus child ate in a shared plate with another child (19% of all feeding episodes), with an adult (12%) and with both an adult and another child (7%) (these are non-exclusive groups). Because the demographic composition of the family can modify food-sharing behaviors, we examined but found no differences between case or control households with respect to the number of adults (>15 y of either sex, median = 2) or children (<15 y, median = 4) of either sex living in the household.
There were significant differences in the types of foods consumed during feeding episodes of case and control focus children (Table 4). Comparing individual-plate feeding episodes between case and control children, cases were more likely to consume non-carotenoid-rich vegetables (OR = 1.3, CI = 1.1-1.6) than control children. This significant difference was also found when all feeding episodes were compared. Cases were more likely to consume carotenoid-rich vegetables during individual-plate eating compared with controls; however, this difference was not seen for all feeding episodes.
View this table:
[in this window]
[in a new window]
|
Table 4.
Odds of consuming different food groups, by type of feeding episode for case relative to control focus children1
|
|
On the other hand, case children were significantly less likely to have access to key vitamin A-rich foods including meats and fish (OR = 0.6, CI = 0.5-0.9) and dairy products (OR = 0.7, CI = 0.5-1.0), compared with control children over all feeding episodes and especially during shared-plate feeding. No significant differences were noted in the consumption of carotenoid-rich vegetables or fruits.
The age and sex of an individual food sharer may influence consumption patterns of children eating with them. We examined whether case and control children differed with respect to the identity of their food sharers (Table 5). Case children were significantly more likely to eat in a shared-plate situation with a male adult (age > 15 y) (OR = 1.7, CI = 1.0-2.8), whereas case children were significantly less likely to eat with a female adult (OR = 0.6, CI = 0.4-0.9) or a female child (OR = 0.6, CI = 0.4-1.0) compared with non-xerophthalmic control children. No significant differences were noted between case and control children sharing a plate with a male child.
View this table:
[in this window]
[in a new window]
|
Table 5.
Odds of participating in a shared-plate feeding episodes with an adult or child, by sex and age of at least one food sharer, for case relative to control focus children1
|
|
Table 6 presents the average total portion size (in grams) for food groups consumed in a feeding episode by the focus child on the basis of their case status and the age and sex of at least one food sharer. No statistical differences in food portion sizes could be discerned between any age-sex food-sharer group or between case and control children. Nonetheless, some trends can be observed. The total average portion sizes for vegetables and meats or fish consumed by case children appear to be larger than that of control children. On the other hand, portion sizes for carotenoid-rich vegetables and dairy products appear to larger for control compared with case children.
View this table:
[in this window]
[in a new window]
|
Table 6.
Average total food portion size consumed during shared-plate feeding episodes of study children, by food group, case status and by age and sex of at least one food sharer1,2
|
|
| |
DISCUSSION |
To our knowledge, this is the first study that has systematically examined shared-plate eating and has been able to compare directly shared and nonshared-plate eating of the same children over multiple observations. The fact that shared-plate eating occurs regularly throughout the year (65% of days observed) and infrequently within the day (25% of feeding episodes) provided us with a unique opportunity to explore the relative advantages and disadvantages of both feeding patterns in this high risk population.
Assessments of dietary intake are very difficult in regions in which shared-plate eating is common. We were able to explore in detail the dynamics of shared-plate eating with the use of continuous monitoring observations. This data collection technique permitted us to track the temporal sequence of events contributing to the shared-plate episode and to better characterize different types of shared-plate eating. In our sample, we found the predominant form of shared-plate eating consisted of two or more individuals being served simultaneously. The other identified patterns, such as post-served and multiple post-served sharing may be important because the time at which an individual comes to share food during a meal may affect the total portion size that the individual receives. We were not able to discern such differences between sharing types because of the infrequency of the last-mentioned types of sharing in our sample.
There are a number of aspects of shared-plate eating that influence food consumption for children. Taken as a whole, we found shared-plate eating to be a positive behavior in that more and different types of foods were consumed. It is clear from our sample that shared-plate eating occurs mainly during meals within the household and very infrequently during snacks or meals that were consumed outside the home. As a result, it is not surprising that we found shared-plate feeding episodes were more likely to include a greater variety of food groups compared with other feeding episodes. In general, children eating on a shared plate were twice as likely to receive a specific food group than if they ate alone.
Food portion sizes consumed by young children during a shared-plate meal were also larger. Again, this may be partially due to the fact that individual-plate feeding episodes included snacks. However, the very large differences in portion size would suggest that, on average, one consumes more if eating with others than by oneself. This may be because the child has increased access to foods as it may be obtained or requested by a food sharer more senior than the child or simply a result of more than one individual requesting food. It should be noted, however, that despite the larger portion sizes obtained during a shared-plate feeding episode, the average portion sizes, especially for vegetables and meats, were still small.
These data suggest that shared-plate eating is a beneficial feeding behavior for young children, but the role it plays with respect to vitamin A deficiency is still unclear. Our study populations (cases and controls) were surprisingly similar in their basic feeding patterns. From earlier ethnographic work, we found that adults usually ate two meals (morning and evening) during the day and one snack (usually late afternoon). The children in our study appear to have eaten more frequently, having, on average, three meals and one snack each day. The frequency of shared-plate eating did not differ substantially between cases and controls.
The primary difference between the two groups was that, on a per meal basis, case children were significantly less likely to have any dairy products or meat and/or fish than control children. On the contrary, they were significantly more likely to consume non-carotenoid-rich vegetables, presumably as substitution for other foods. These data are consistent with feeding patterns identified using weekly food frequencies (Shankar et al. 1996) and infant diet histories (Gittelsohn et al. 1997) in the same population.
The identity of the person sharing food with the child appears to play a key role. Children having a history of xerophthalmia were significantly more likely to be found eating with a male adult, whereas similar aged non-xerophthalmic children were more likely to eat with a female adult or a female child (
15 y). On the basis of the literature on intrahousehold food distribution, this would suggest that non-xerophthalmic children are benefiting from the increased child attentiveness provided by the female adult (Rogers and Schlossman 1990). There is evidence that women, especially mothers, are more likely to encourage a child to eat than other caregivers (Engle and Zeitlin 1996) and are more attentive during feeding when a child is ill (Bentley 1988, Bentley et al. 1991). Even as young children, girls are found to offer and ask for food more than boys and are more likely to share their food with others (Dyson-Hudson and Van Dusen 1972).
Studies examining the effectiveness of male interactions with respect to child rearing indicate that although men can parent, they tend not to be as responsive as women to the child's needs (Mackey 1983). Therefore in a shared-plate setting with a male adult, the case child may be neglected or may otherwise procure less food. This is balanced by the fact that, in this setting, male adults, because of their high social status, may be the recipients of "higher quality" foods (Gittelsohn, 1991), which may increase a young child's access to these foods.
In our examination of food portion sizes, we found little evidence that the sex and age of the food sharer significantly influenced the total portion size of consumed foods. This may be due, in part, to the variability of the portion size estimates. These identified patterns in shared-plate eating may reflect the overall poorer care experienced by children with a history of xerophthalmia. In earlier analyses, we found that case children were less likely to be groomed (A. V. Shankar, unpublished manuscript), more likely to experience negative social behaviors, such as hitting and scolding, and less likely to have positive health behaviors, such as face cleaning or using soap than control children (Gittelsohn, et al. 1998). It is likely that eating with an adult male is a risk factor for vitamin A deficiency only in that it is characteristic of households in which poor child caretaking exists.
This study has examined the very common, but poorly researched phenomenon of shared-plate eating. With the use of continuous monitoring observation, we were able to identify the temporal sequence, actors and recipients of dietary behaviors as well as approximate the total amount of food consumed by each child and posit a relationship between these feeding patterns and nutritional status. The shared-plate eating patterns we have identified are slightly different than those reported in the literature. Shared-plate eating is not the primary method of food consumption in this population and children often eat only with other young children, not with adults. It is likely that this type of food sharing occurs in other areas of southern Nepal and northern India, although it may be evident in other regions of South Asia as well.
There are still many gaps in our knowledge with respect to food sharing. For example, how does illness affect total portion size consumed especially because a child's appetite plays a large role in food intake (Engle and Zeitlin 1996)? What are the specific behavioral mechanisms that occur in a shared-plate setting that assist the child to eat more? Can these behaviors be taught to other household members? Because food sharing from a single bowl or plate is so common in developing countries, shared-plate dynamics must be evaluated in different settings to identify their potential effect on dietary outcomes.
![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
Abstract
Introduction
Abstract
the Anthropology of eating. Houghton Mifflin Company, Boston, MA.
