Intrahousehold allocation of energy intake among children ... - Nature

2 downloads 0 Views 102KB Size Report
02115, USA; 2Tufts University School of Nutrition, 126 Curtis Street, Medford, MA, ... that, when daily energy intake is adjusted for energy expenditure, no age or ...
European Journal of Clinical Nutrition (1997) 51, 750±756 ß 1997 Stockton Press. All rights reserved 0954±3007/97 $12.00

Intrahousehold allocation of energy intake among children under ®ve years and their parents in rural Bangladesh EM Kramer1 , KE Peterson1 , BL Rogers2 and MD Hughes3 1

Harvard School of Public Health, Department of Maternal Child Health/Department of Nutrition, 677 Huntington Avenue, Boston, MA 02115, USA; 2Tufts University School of Nutrition, 126 Curtis Street, Medford, MA, USA; 3Harvard School of Public Health, Department of Biostatistics, 677 Huntington Avenue, Boston, MA, USA Objective: This study assesses intrahousehold allocation of energy in rural Bangladesh and tests the hypothesis that, when daily energy intake is adjusted for energy expenditure, no age or gender bias will be apparent in intrahousehold energy allocation. Design: Data were collected at two-month intervals over a one year study. Setting: Four villages in Matlab Thana, rural Bangladesh. Subjects: Two hundred and seven children up to 5 y of age and their 145 mothers and 123 fathers. Interventions: Data included six measurements of observed 24 h dietary energy intake and physical activity recorded from waking to sleeping. Total daily energy expenditure was derived using the factorial method. Results: Women's energy intake ranged from 75±88% of the FAO/WHO recommended energy intake over the six periods of data collection, signi®cantly less (P < 0.0001) than the men's (range 89±114%). Although the women had moderate levels of physical activity, frequent pregnancies and long lactation periods increased their energy needs. Among children no longer breast fed, energy consumption, unadjusted for energy expenditure, provided 86±108% of the FAO/WHO recommended energy intake by weight. Conclusions: Women consistently received less of their energy requirements than either their children or their husbands. Sponsorship: This work was supported in part by the Stare Nutrition Education Fund at the Harvard School of Public Health. Descriptors: intrahousehold allocation; energy expenditure; energy intake; Bangladesh

Introduction Intrahousehold allocation is the ®nal step in the chain of events that determines food availability to individuals. While household food supply re¯ects various factors (for example market availability, government food policies and programs, income of household members, agricultural production), intrahousehold allocation mediates between the household food supply and that of individual members, ultimately in¯uencing the health and nutritional status of the individual (Van Esterik, 1984). Because systematic gender bias may in¯uence the food an individual receives (Van Esterik, 1984; Behrman, 1992), knowing the amount of food available to an entire household is not necessarily suf®cient to understand the amount of food (or nutrients) available to any given family member (Haaga & Mason, 1987). Food policies and programs are frequently targeted to infants, small children, and pregnant and lactating women. Better understanding of food distribution within impoverished families in poor countries has great promise for strengthening policies and programs, when allocation of household resources does not protect the nutritional status of women and children (Haddad, 1994a; Alderman et al, 1994; Rogers, 1990). The effects of gender and age on intrahousehold allocation of energy vary by region (Behrman, 1992). South Asia Correspondence: Dr EM Kramer. Received 25 March 1997; revised 26 June 1997; accepted 3 July 1997

has been identi®ed as a region with gender bias in intrahousehold food allocation (Haaga & Mason, 1987; Lipton, 1987). But not all studies within the region found evidence for pro-male biases in food intake (Haddad et al, 1994b). Early studies in Bangladesh concluded that men received more calories than women (Carloni, 1981), and that preschool children, particularly girls, received proportionately fewer calories than others in the household (Abdullah, 1989; Pitt et al, 1990; Brown et al, 1982). Physical activity and energy expenditure dataÐimportant, but often overlooked, determinants of energy requirementsÐare essential to rigorous investigation of the pattern of sex bias against women in South Asia in the distribution of energy (Harris, 1990). The separation between men's and women's work roles in rural Bangladesh is pronounced (Barkat-E-Khuda, 1980). Men are responsible for farming and agricultural labor outside the home and for marketing; women's primary tasks are cooking, child care, and postharvest grain processing. Seclusion of women limits their work participation outside the home (Rahman, 1986). While cultural roles that limit total energy expenditure of women may increase energy adequacy, frequent pregnancy and lactation increase their energy requirements. Few studies of intrahousehold allocation have controlled for variation in energy expenditure when calculating energy requirements (Haddad et al, 1994b). Several studies approximated energy expenditure using estimates of physical activity (Abdullah, 1989; Pitt et al, 1990; Chen et al, 1981; James & Scho®eld, 1990; Ahmed, 1983): extrapolation from a question about occupation (Pitt et al, 1990);

Intrahousehold allocation in Bangladesh EM Kramer et al

estimation of the activity level of the male head of household based on the seasonal demands of agricultural production (Abdullah, 1989); and crude labor force participation data (Chen et al, 1981). In the one study that found allocation of energy intake favored adult women, estimation of energy expenditure was based on the `activity reported by the individuals' (Kumar & Bhattarai, 1992). The validity of these measures is limited, nor do they account for the energy requirements of pregnancy and lactation. This study describes energy intake among children up to ®ve years of age and their parents. It tests the hypothesis that when a measure of energy expenditure is included in estimating adequacy of daily energy intake, energy needs will be met and no age group or gender bias will be apparent.

Methods The data were derived from a ®eld study conducted between June, 1977 and August, 1978 at Matlab Thana, the experimental ®eld station of the International Centre for Diarrhoeal Disease Research, Bangladesh (ICDDR,B). The study year began in June, the monsoon season, which lasts through mid August; hot humid weather persists through October. There are three harvests, with the main one in October. In Matlab Thana the period of greatest scarcity is August through October, when rice prices are highest, household food stocks are lowest and agricultural labor demand is weakest (Chowdhury et al, 1981). Sample selection The 135 households in the study came from four villages in the Matlab Surveillance area. They do not represent a strictly random sample of households with children under ®ve years of age but were chosen to re¯ect a wide range of socio-economic status in the study population. Forty-four families owned less than 0.5 acres of land; 51 owned between 0.5 and 2.0 acres; and 40 owned more than 2.0 acres. Subjects included 145 mothers with a mean age of 29 y (range 17±47 y), and 123 fathers with a mean age of 40 y (range 20±62 y). Their 207 young children were 2±58 months of age at the start of the study. There were 94 girls (mean age 28.6 months) and 113 boys (mean age 29.2 months). The study was conducted according to ethical guidelines of the International Center for Diarrhoeal Disease Control of Bangladesh. Dietary data and anthropometry The survey team consisted of two female dietary workers with high school education and four assistants. Visits were conducted by one dietary worker and one assistant. For each subject, observed 24 h dietary intake data was measured six times over the study year. At the beginning of the study ®eld workers measured all large cooking vessels and serving utensils in the household. On the day of the home visit, the ®eld workers arrived before the ®rst meal was prepared. After calibrating their scales, the workers weighed and measured foods before and after cooking. Field workers provided families with pre-measured spoons and cups and estimated individual consumption by counting the number of pre-measured spoonfuls or cupfuls

given to each family member. Field workers measured individual plate waste by using these same utensils. The sum of the distributed food had to equal the volume recorded for the cooking pot. Any discrepancies were reconciled in the ®eld. Meals eaten outside the household and snacks were assessed by recall conducted twice daily, using the pre-measured utensils described above to help subjects estimate quantity of food consumed. About half of the subjects (46% of females and 59% of males) reported eating snacks on the day of the recall. Total 24 h consumption was computed using standard food composition tables from India (Gopalan, 1978). The ®eld worker obtained and recorded the age of all subjects and information on whether the child was currently breast fed. Because the energy contribution of breast milk to dietary energy could not be quanti®ed, breast fed children were excluded from analysis of energy intake data. Children's energy intake adequacy was assessed using the FAO/WHO/UNU (1985) standards for age and weight. Trained ®eld workers measured height and weight each month among preschool children and their mothers. Heights and weights were assessed every three months among fathers who were less likely to undergo rapid changes in weight. Weights were obtained by balance beam scales for adults and Salter 25 kg scales for children. Standardized height sticks and horizontal length boards (for children under two years of age) were used for body length measurements. (Jelliffe, 1966). Anthropometric ®ndings are reported elsewhere in a paper submitted for review. Adequacy of adult's energy intake was assessed using the FAO/WHO/UNU (1985) recommendation based on age, sex, basal metabolic rate (BMR), levels of physical activity, and pregnancy and lactation determined by maternal self report. BMR was calculated using the FAO/WHO/ UNU (1985) formula based on weight, age and sex. Because energy intake and physical activity was collected for all household members every two months, the average of weights measured in each two-month period was used to calculate BMR for mothers and children. For fathers the weight measured closest to the time of observing the 24 h dietary intake was used. The levels of physical activity were derived from observation of daily activity patterns.

Activity At the home visit ®eld workers observed and recorded daily activity of father, mother and the youngest child in 15 min blocks during waking hours using one of 63 possible activities. Based on these observations we estimated energy expenditure using the FAO/WHO/UNU (1985) factorial method to enhance comparability to other data. We assigned a MET value or metabolic constant to each of the 63 activities using published tables (FAO/WHO/UNU, 1985; Ainsworth, 1993). For many activities, a range of MET values was available and preference was given to the lower end of the value range to account for breaks and rest periods. We used Torun's (1990a) recommendations for interpreting observations of activity of children under 5 y. We calculated total energy expenditure for the 24 h day using the following equation: (metabolic constant for each activity) 6 BMR 6 (minutes spent on activity). Sleeping was assigned a MET value of 1.00. For a more complete discussion of how MET values were assigned see Kramer, 1996.

751

Intrahousehold allocation in Bangladesh EM Kramer et al

752

Analysis Summary measures, for example mean, standard deviation and range, of energy intake and expenditure were calculated for each data collection round and checked for normal distribution in mothers, fathers, and children using SAS (SAS, 1988). Three methods were used to analyze adequacy of energy intake. The ®rst two methods evaluated the mean energy intake of mothers, fathers and children in each of the six rounds of data collection. The ®rst method compared the observed intakes of mothers and fathers with the FAO/WHO/UNU (1985) recommendations for energy intake based on age, sex, BMR, level of physical activity and pregnancy and lactation status. The FAO/WHO/UNU (1985) recommendations for children's energy intake can be calculated based on either weight or age; both were calculated. Linear regression was used to examine differences in mean energy intake and expenditure between men and women and between adults and children. The FAO/UNU/WHO (1985) recommendations for energy intake for children, used in the ®rst method, did not include a recommendation based on children's energy expenditure. Our second method of assessing energy adequacy was based on the ratio of observed energy intake to total daily energy expenditure calculated from observations of physical activity. About three quarters of the women were lactating and 19% were pregnant in each round. Mothers' total daily energy expenditure included 500 calories during lactation and 285 calories in pregnancy as recommended by the FAO/WHO/UNU (1985). Mean energy intake as a percent of total energy expenditure for each group (mothers, fathers and children) was calculated for each round. The third method evaluated household distribution of energy intake. Calculation of an overall household energy

adequacy index was not feasible due to the lack of complete energy intake data for breast fed children and missing data for fathers due to the seasonal demands for agricultural labor. Subjects contributing data to each round of data collection are noted in the tables. To assess the in¯uence of missing data on conclusions about household allocation of energy intake, we undertook the third method, a matched analysis of parent-parent and parent-child pairs, using the FAO/WHO energy standards and including only pairs of subjects who were at home to share household meals. The the difference in intake adequacy between children and their parents and between men and women who shared the household meal was examined with paired t-tests. Results Energy intake and expenditure Among the 123 fathers, average energy intake was lowest from August to November, the rainy season preceding the major harvest. Mean energy intake of the fathers met or exceeded the FAO/WHO/UNU recommendations for eight months of the year, December to July (Table 1). Average energy expenditure exceeded 10500 J (2500 calories) in 10 of 12 months and was lowest in the months of August and September. The highest total energy expenditure occurred in February and March. Among the 145 mothers, mean energy intake was lowest from August to November (Table 1). On average, they consumed  88% of recommended energy intake in each two month period. Average total energy expenditure of mothers remained fairly constant over the year at about 8400 J (2000 calories). Percent of energy adequacy and energy expenditure were signi®cantly lower (P  0.0001) in mothers than fathers.

Table 1 Total daily energy expenditure and energy intake among adults in rural Bangladesh Period of data collection Months Men June/July Aug/Sept Oct/Nov Dec/Jan Feb/March April/May Period of data collection Months Women June/July Aug/Sept Oct/Nov Dec/Jan Feb/March April/May

Total energy expenditure n intakeb (expenditure) 97 (73) 101 (95) 108 (101) 101 (94) 110 (98) 105 (98)

Joules mean [s.d.] 10,812 8,803 10,664 11,025 11,634 11,025

[2440] [2318] [2432] [2209] [1877] [2150]

Kcal (mean) (2579) (2096) (2539) (2625) (2770) (2625)

Joules mean [s.d.] 12,495 11,084 10,017 11,214 11,508 11,609

Total energy expenditure * n intakeb (expenditure) 137 134 135 128 129 132

(109) (132) (137) (131) (132) (129)

Joules mean [s.d.] 8354 8295 8551 8396 8644 8744

[1113] [903] [1130] [1407] [1247] [1378]

Percent of recommended energy intakea

Energy intake

[3759] [3121] [3263] [3599] [3914] [2864]

Kcal (mean)

% < 85% recommended intake

Mean %

(2975) (2401) (2385) (2670) (2740) (2764)

21 39 42 32 27 20

114 92 89 101 106 104

Percent of recommended energy intakea

Energy intake

Kcal (mean)

Joules mean [s.d.]

Kcal (mean)

% < 85% recommended intake

Mean % *

(1989) (1975) (2036) (1999) (2058) (2082)

9148 7867 8337 8606 9261 8904

(2178) (1873) (1985) (2049) (2205) (2120)

56 69 63 66 50 61

86 75 79 81 88 84

[2646] [2352] [2184] [2264] [2524] [2386]

*P=0.0001 (Beta coef®cient signi®cantly different between men and women in linear regression model adjusted for round of data collection). World Health Organization. Energy and Protein Requirements. Report of a Joint FAO/UNU/WHO Expert Consultation. Geneva: World Health Organization, 1985. Technical Report Series 724. b 123 men contributed information on energy intake, 107 on energy expenditure and 106 on recommended energy intake in one or more rounds of data collection. c 145 women contributed information on energy intake, 142 on energy expenditure and 141 on recommended energy intake in one or more rounds of data collection. a

Intrahousehold allocation in Bangladesh EM Kramer et al

In the ®rst round of data collection 82 children (43%) were not breast fed and provided complete dietary energy intake information. The boys averaged 43.6 months of age and girls averaged 44.6 months. Over the study year more children completed breastfeeding, and by the ®nal round complete energy intake data were available for 115 children (59%). The average energy intake of fully weaned children was close to 4200 J (1000 calories) (Table 2). Although average energy intake of boys exceeded that of girls in 5 of the 6 rounds, the difference was not signi®cant. Children's intake was compared to the FAO/WHO recommendations for weight and for age (Table 2). Weight for age Z scores of children in this study averaged 72.79 (s.d.  0.82) over the study year. There was no signi®cant difference (P ˆ 0.68) in energy adequacy between boy and girls. When the energy intake of children was determined as a percent of the recommendation based on weight in each round of data collection, children followed the parental pattern, with the lowest percent of recommended energy intake consumed from August to November. Using the FAO/WHO/UNU (1985) recommendation based on age, children's average intake surpassed 70% of recommended energy consumption only in June and July. Mean energy expenditure gradually increased over the course of the year, as the children aged (Table 3) and was signi®cantly higher for boys than for girls in all rounds. The ratio of energy intake to energy expenditure The second method compared energy intake to total energy expenditure for each individual within round. On average, fathers' energy intake exceeded 97% of energy expenditure (Table 4). Mothers mean energy intake was  93% of the energy expenditure based on observed activity, and this percentage was not signi®cantly different from the fathers Table 2

(Table 4). When the energy needs of pregnancy and lactation were added to the energy expenditure based on physical activity, mothers' energy intake were  89% of expenditure in each round. This percentage was signi®cantly lower than the fathers (P ˆ 0.0001) (Table 4). The number of children included in this analysis was small because only the older children were completely weaned and could provide complete data on energy intake, while only the youngest child in each household was included in the study of energy expenditure. Because this analysis required both energy intake and expenditure data in each round, only 19 children were included in the ®rst round and 60 in the ®nal round. Children's intake ranged from 74±91% of the energy expended in the six rounds of data collection (Table 4). Children's adequacy was consistently lower than that of the men; the relationship with women's adequacy was inconsistent across rounds and differences were not large. The matched analysis The difference in energy adequacy of mother-father and parent-child (mother-child and father-child) pairs was examined among those individuals who were present to share household meals (Table 5). The child's energy adequacy calculated from the weight-based FAO/WHO recommendation was subtracted from the parent's energy adequacy. The analysis was completed for each child, so parents of more than one child could appear in more than one pair. Fathers consumed a larger percent of the FAO/WHO/ UNU (1985) standard for energy intake than their children in every period of data collection. This difference was signi®cant in the June/July and December/January rounds of data collection. Children who were no longer breast fed consistently met a greater percent of their energy require-

Energy intake of children no longer breast fed in rural Bangladesh

Boys energy intake

Period of data collection Months

n

June/July Aug/Sept Oct/Nov Dec/Jan Feb/Mar Apr/May

82 91 107 104 104 115

Girls energy intake

Percent of FAO/WHO Recommendation Based on weighta

Percent of FAO/WHO Recommendation Based on age

Joules mean [s.d.]

Kcal (mean)

Joules mean [s.d.]

Kcal (mean)

Boys mean %

Girls mean %

Boys mean %

Girls mean %

4746 4292 4116 4003 4906 4633

(1130) (1022) (980) (953) (1168) (1103)

4187 3822 3629 4372 4108 4099

(997) (910) (864) (1041) (978) (976)

108 94 90 89 106 97

98 94 86 99 93 95

76 66 63 62 75 69

66 62 57 68 64 64

[1655] [1617] [1445] [1743] [1953] [1583]

[1516] [1441] [1520] [1466] [1562] [1533]

a

World Health Organization. Energy and Protein Requirements. Report of a Joint FAO/WHO/UNU Expert Consultation. Geneva: World Health Organization, 1985. Technical Report Series 724.

Table 3

Total daily energy expenditure of children under 5 years of age in rural Bangladesh All children*

Period of data collection

Girls

Boys

Months

n

Joules mean [s.d.]

Kcal (mean)

n

Joules mean [s.d.]

Kcal (mean)

n

Joules mean [s.d.]

Kcal mean

June/July Aug/Sept Oct/Nov Dec/Jan Feb/Mar April/May

108 126 135 131 134 130

3990 4381 4406 4473 4696 4872

(950) (1043) (1049) (1065) (1118) (1160)

45 56 62 61 62 58

3671 4049 4166 4246 4372 4628

(874) (964) (992) (1011) (1041) (1102)

63 70 73 70 72 72

4221 4645 4607 4666 4973 5065

(1005) (1106) (1097) (1111) (1184) (1206)

[1016] [958] [890] [878] [899] [861]

[1084] [916] [911] [886] [878] [861]

*Pˆ0.001 (Beta coef®cient for boys signi®cantly different compared with girls in linear regression model by round).

[903] [916] [827] [827] [827] [819]

753

Intrahousehold allocation in Bangladesh EM Kramer et al

754

Table 4

Energy intake as a percent of the individual's total energy expenditure among children under 5 years and their parents in rural Bangladesh Women

Period of data collection

Based on observed physical activity only mean % (s.d.)

Men

Months

n

June/July Aug/Sept Oct/Nov Dec/Jan Feb/Mar April/May

68 89 95 88 97 94

Mean % (s.d.) 123 121 97 104 102 107

(41) (38) (40) (33) (38) (29)

n 99 122 123 113 115 119

110 93 96 107 106 99

Based on observed activity plus physiological status* mean % (s.d.)

(37) (29) (26) (27) (31) (28)

89 76 80 83 88 83

(32) (23) (21) (21) (25) (25)

Children n

mean % (s.d.)

19 36 45 48 54 60

91 79 79 74 84 78

(40) (39) (35) (30) (36) (29)

*P=0.0001 (Beta coef®cient signi®cantly different for men compared with women in linear regression model. N.S. for model based on observed physical activity only). Table 5 Parent±child and father±mother differences in daily energy intake as a percent of the WHO/FAO recommended intakea,b Period of data collection Months

n

June/July Aug/Sept Oct/Nov Dec/Jan Feb/March April/May

34 51 65 53 62 60

Father-child mean % (s.d.) 24** 3 3 16** 6 9

(48) (41) (35) (38) (47) (37)

n

Mother-Child mean % (s.d.)

n

49 79 84 76 73 78

74 7 19*** 7 11*** 7 13** 7 11* 7 9*

64 84 88 81 88 85

(38) (37) (28) (36) (38) (33)

Father-mother mean % (s.d.) 24*** 18*** 7* 17*** 14*** 18***

(33) (32) (32) (33) (36) (26)

*Signi®cant 0.01 < P < 0.05 by paired t-test. **Signi®cant 0.05 P < 0.001 by paired t-test. ***Signi®cant 0.001 < P < 0.0001 by paired t-test. a World Health Organization. Energy and Protein Requirements. Report of a Joint FAO/UNU/WHO Expert Consultation. Geneva: World Health Organization, 1985. Technical Report Series 724. Children's intake based on recommendations for weight. b This analysis compared energy intake in parent±child and father±mother pairs among those who shared family meals.

ments than their mothers did, and this difference was signi®cant in all rounds except June/July. Fathers consumed a signi®cantly greater percent of recommended energy intake than mothers in every round of data collection (Table 5). Discussion The purpose of this study was to determine whether there was evidence of bias by age or gender in the adequacy of energy intake when measured energy expenditure was included in estimating daily energy requirements. We found that men met their recommended energy needs adjusted for energy expenditure over most of the study year, but women and children did not. The same conclusion was supported by all three analytic methods, and is consistent with the ®ndings of most previous studies in Bangladesh (Haddad et al, 1994b; Carloni, 1981). Observation of dietary intake and physical activity patterns can overcome many of the problems of recall: lapse of memory, uncertainty about the types or amounts of foods eaten, poor motivation, and unawareness of food intake (Witschi, 1990). The training of the observer, the type of food, and portion sizes can limit the accuracy of dietary observation (Gittleson et al, 1994) and the day-long presence of an observer could alter usual patterns. Because one dietary worker and one assistant collected energy intake and energy expenditure data for mother, father and child (or children) in the household, they could not follow each individual who left the household area. Information on foods eaten away from home were collected through recall

twice daily. As children became old enough to wander on their own they may have eaten local foods but failed to report them leading to underestimates of energy intake. The use of the factorial method to estimate energy expenditure is limited by the variability of energy cost between individuals in performing the same task, the dif®culty in estimating the values for tasks that combine a variety of movements and the extent to which body weight affects energy expenditure for a given type of physical activity (FAO/WHO/UNU, 1985). Only the predominant activity was recorded for each 15 min observation period and no attempt was made to capture the intensity of physical activity, due to lack of information across groups in different societies (James & Scho®eld, 1990). Despite the advantages of observational data, it was not possible to overcome all logistical problems and subjects could not be observed when they left the household area. In this Muslim culture, women were unlikely to leave the household area and our observations of physical activity record predomoninantly household-based activities for women. Our data were not coded to permit us to differentiate activities which were directly observed from activity obtained from twice daily recall. In such cases, activity level may have been over or underestimated but the direction of bias is not known and hence the effect on energy adequacy is not known. Because of the large day to day variability in energy intake, a single day of dietary recall is a poor estimate of long term dietary intake and the effect on outcome measures is to decrease the strength of association. A previous study using this same data set concluded that seven days of

Intrahousehold allocation in Bangladesh EM Kramer et al

dietary intake would be needed to estimate average dietary intake for the year within 20% of the true value 95% of the time (Torres et al, 1990). A maximum of six days of dietary data was available for subjects in this study, but this was enough information to discern signi®cant differences in energy intake and expenditure by age and gender. Some data suggests that adult values for BMR are systematically lower in India (FAO/WHO/UNU, 1985). Because this may re¯ect the malnutrition endemic in the population, we estimated energy expenditure based on the FAO/WHO/UNU equations to create BMR, rather than institutionalizing the malnutrition seen in the study population. Choosing the lower BMR regressions would have resulted in lower estimates of energy expenditure for adults. It has been suggested that BMR has a CV of 12.5%, which could also effect the estimates of energy adequacy (FAO/WHO/UNU, 1985). We found that 20± 42% of men and 50±69% of women had less than 85% of recommended energy intake, consistent with the problems of food insecurity in the region (Garcia & Mason, 1993). The paucity of data directly measuring children's energy expenditure limits the validity of the MET values for children in this study (Torun, 1990a). Our results, however, were comparable to ®ndings from studies in other settings, with energy expenditure signi®cantly higher in boys than girls (Baranowski et al, 1993). Furthermore, energy expenditure calculations, based in part on weight, increased with age. Finally, the estimates of total daily energy expenditure gradually increased over the study year as expected, because children spent less time sleeping as they aged, leaving more time for physical activities with a high MET value (Kramer, 1996). The matched analysis was undertaken to determine whether missing data for fathers was affecting study conclusions. The analysis controlled for missing data by round but this method did not change study conclusions. It con®rmed that children's energy adequacy based on their current weight was greater than their mothers but less than their fathers. Nutrition education programs that promote increased dietary intake for children should ensure that increase in children's energy intake is not taken from the mothers' meager allotment of calories (Brown et al, 1994). The severity of the children's calorie de®cit was greatest when the FAO/WHO/UNU standard for age was used. Even when the standard for their current weight was used, children still needed more calories than they were consuming. When children's mean dietary intakes were calculated as a percent of their observed energy expenditure, they still met only 74±91% of their energy needs. Low energy intake can reduce the amount and intensity of physical activity in infants and young children (Torun, 1990b). In this study, the coding allowed for the categories of light and vigorous play, but no child was ever seen in vigorous play, indicating low levels of physical activity. While decreased physical activity can compensate for poor energy intake, it is not a desirable adaptation because of the potential for long term consequences on social, intellectual and economic productivity. Prolonged inactivity has many rami®cations including reduced physical ®tness (with concomitant long term effects on work productivity) and limited social interaction and exploration (impairing learning) (Torun, 1990b). The BMI of fathers averaged 18.3 and the BMI of mothers (including pregnant women) averaged 18.7. Among women over 49 y of age in these same households,

average BMI was 16.9 and 43% had a BMI below 16, indicating severe chronic energy de®ciency (Kramer, 1996). While both men and women exhibit low BMI, the data on older women suggest that the energy demands of frequent pregnancy and extended periods of lactation may be important in explaining why mothers, on average, are not meeting their energy needs. The FAO/WHO/UNU expert committee estimates the energy demands of pregnancy at 285 calories per day and lactation at 500 calories per day averaged over the three trimesters (FAO/WHO/ UNU, 1985), although some subsequent studies suggest the estimates of the pregnancy requirement are too high (Durnin, 1987). To test this, we repeated our analyses allowing only 200, rather than the recommended 285, calories per day but this adjustment changed the mean energy adequacy of women by less than 1% and did not change our conclusions in any way. Moreover, few mothers reported pregnancy for more than 4 or 6 months in a row, implying 1±2 rounds of data collection where women are misclassi®ed as to pregnancy status. Because pregnancy was under-reported in this study, energy intake as a percent of recommendation may be even lower than reported, although under reporting of pregnancy might be expected to have a small effect on energy adequacy. Pregnancy may also have in¯uenced energy intake through nausea and vomiting related to pregnancy and cultural beliefs that women should restrict food intake during pregnancy (Vermury, 1980). Father's energy intake and expenditure was lowest from August to November, when food stocks are low and rains restrict activity. This season in Bangladesh is hot and muggy, discouraging activity, and demand is low for agricultural labor (Kramer, 1996). From December to May father's energy intake and energy expenditure peaked at over 10920 J (2600 calories), supporting the heavy agricultural labor during this period. One explanation for the fact that fathers came closest to meeting their energy requirements is that the food supply of all members of the household is dependent on the agricultural productivity of the men. Studies in Guatemala found that higher energy intake increased performance of heavy physical work, and permitted continued physical activity after work, supporting socially desirable and economically productive activities (Torun, 1990a). Thus, feeding men during periods of peak labor demand, even at the expense of lower intake for their wives and children, may be perceived as essential to the well being of the household. This study was not based on a simple random sample, but was strati®ed to re¯ect households from all socioeconomic strata in rural Bangladesh. Although the data were collected in the late 1970s, they were collected to address the study questions in this paper. Recent data suggest that the nutrition situation in Bangladesh has improved only marginally. Over the last 15 y, the prevalence of under weight children decreased only by about half a percentage point per year (Garcia & Mason, 1993) and women continue to suffer social discrimination, re¯ected in their continuing very low literacy rates as well as suboptimal nutrition status re¯ected in a low birth weight prevalence of 34% (Garcia & Mason, 1993). Conclusions Energy intake of women was insuf®cient to support their energy needs for both observed physical activity and the

755

Intrahousehold allocation in Bangladesh EM Kramer et al

756

recommended energy for reproduction. While men's energy intake on average met a greater proportion of the recommended intake, information from this study is not by itself suf®cient to support programmatic efforts to encourage reallocation within the household. In this culture of scarcity, there is little margin for error. Well-meaning efforts of nutrition education could deprive men of needed energy, resulting in decreased agricultural production which could undermine food availability for the entire household. Similarly, programs that support increased intake for children could shift calories from mothers' already inadequate intake. Agricultural and income generation programs to increase total energy availability to the household, family planning services to diminish the burden of reproduction, and enhanced opportunities for women to receive education all have potential to help women meet their energy needs. AcknowledgementsÐThe authors gratefully acknowledge Dr. Lincoln Chen and the Bangladesh study team. We also wish to thank Dr. Marito Garcia, Dr. Laurine Brown, and Dr. Guillermo Herrera for their insightful review of this manuscript.

References Abdullah M (1989): The effect of seasonality on intrahousehold food distribution and nutrition in Bangladesh. In Seasonal Variability in Third World Agriculture: the Consequences for Food Security, ed. DE Sahn, pp 57±65. Baltimore: Johns Hopkins University Press. Ahmed AU (1983): Food consumption and nutrition effects of targeted food interventions in Bangladesh. Dhaka: International Food Policy Research Institute, Bangladesh Food Policy Project. Ainsworth BE, Haskell WL, Leon AS, Jacobs DR, Montoye HJ, Sallis JF & Paffenbarger RS (1993): Compendium of physical activities: classi®cation of energy costs of human physical activities. Med. Sci. Sports Exerc. 25, 71±80. Alderman H, Chiappori P-A, Haddad L, Hoddinott J & Kanbur R (1994): Unitary versus collective models of the household: time to shift the burden of proof? April. Branowski T, Thompson WO, Durant RH, Baranowski J & Puhl J (1993): Observations on physical activity in physical locations: age, gender, ethnicity and month effects. Res. Q. Exerc. Sport 64 (2), 127±133. Barkat-E-Khuda (1980): Time Allocation Among People in Rural Bangladesh. Bergen, Norway: Derap Publications. Behrman JR (1992): Intra-household allocation of nutrients and gender effects: a survey of structural and reduced-form estimates. In Nutrition and Poverty, SR Osmani (ed). pp 287±344. Oxford: Clarendon Press. Brown KH, Black RE, Becker S, Nahar S & Sawyer J (1982): Consumption of foods and nutrients by weanlings in rural Bangladesh. Am. J. Clin. Nutr. 36, 878±889. Brown LV, Zietlin MF, Weld LH, Rogers BL, Peterson KE, Huq N & Gershoff SN (1994): Evaluation of the impact of messages to improve the diets of lactating rural Bangladeshi women on their dietary practices and the growth of their breast-fed infants. Food Nutr. Bull. 15(4). Carloni AS (1981): Sex disparities in the distribution of food within rural households. Food Nutr. 7, 3±12. Chen LC, Emdadul H & D'Souza S (1981): Sex bias in the family allocation of food and health care in rural Bangladesh. Popul. Devel. Rev. 7 (1), 55±70. Chowdhury AKMA, Huffman SL & Chen LC (1981): Agriculture and nutrition in Matlab Thana, Bangladesh. In Seasonal Dimensions to Rural Poverty, eds. R Chambers, R Longhurst & A Pacey, pp 52±66. UK: Frances Pinter (Publishers). Durnin JVGA (1987): The energy requirements of pregnancy and lactation. In Chronic Energy De®ciency: Consequences and Related Issues, eds. B Schurch, NS Scrimshaw, pp 135±154. Lausanne: International Dietary Energy Consultancy Group.

FAO/WHO/UNU (1985) Energy and Protein Requirements. Report of a Joint FAO/UNU/WHO Expert Consultation. Technical Report Series 724. Geneva: World Health Organization. Garcia M & Mason J (1993): Second Report on the World Nutrition Situation Volume II Country Trends Methods and Statistics. United Nations ACC/SCN. Gittleson J, Shankar AV, Pokhrel RP & West KP (1994): Accuracy of estimating food intake by observation. J. Am. Diet. Assoc. 94 (11), 1273. Gopalan C (1978): Nutritive Value of Indian Foods. New Delhi: National Institute of nutrition. Haaga JG & Mason JB (1987): Food distribution within the family evidence and implications for research and programmes. Food Policy May 1987, 146±160. Haddad L (1994a): Strengthening food policy through intrahousehold analysis. Food Policy 19 (4), 347±356. Haddad L, Pena C & Slack A (1994b): Poverty and Nutrition within Households: Review and New Evidence. The Nutrition Unit, WHO, Geneva and IFPRI, Washington, DC. Harris B (1990): Food distribution, and disease in South Asia. In: Diet and Disease in Traditional and Developing Societies, eds. GA Harrison, JC Waterlow, pp 290±306. Cambridge: Cambridge University Press. James WPT & Scho®eld EC (1990): Human Energy Requirements. Oxford: Oxford University Press. Jelliffe DB (1966): The Assessment of Nutritional Status of the Community. Geneva: World Health Organization, 1966. WHO Monograph Series No. 53. Kumar S & Bhattarai S (1992): Effects of Infrastructure Development on Intrahousehold Caloric Adequacy in Bangladesh. Washington, DC: International Food Policy Research Institute. Kramer EM (1966): Energy Intake, Energy Expenditure and Body Mass Index Among Children Under Five Years and their Parents in Rural Bangladesh. A thesis submitted to the faculty of the Harvard School of Public Health, Boston, MA. Lipton M (1987): Variable access to food. In Food Policy Integrating Supply, Distribution, and Consumption, eds. JP Gittinger, J Leslie, C Hoisington, pp 385±392. Baltimore: Johns Hopkins University Press. Pitt MM, Rosenzweig MR & Hassan N (1990): Productivity, health and inequality in the intrahousehold distribution of food in low-income countries. Am. Econ. Rev. 80 (5), 1139±1156. Rahman M (1986): Tradition, development and the individual: a study of con¯icts and supports to family planning in rural Bangladesh. Asian Population Change Series No. 1, Department of Demography, Australian National University, Canberra. Rogers BL (1990): The internal dynamics of households: a critical factor in development policy. In Intra-household Resource Allocation: Issues and Methods for Development Policy and Planning, eds. BL Rogers, NP Schlossman, pp 1±19. Tokyo: United Nations University Press. SAS (1988): SAS/STAT User's Guide. Cary, NC. USA: SAS Institute, Inc. Torres A, Willett W, Orav J, Chen LC & Huq E (1990): Variability of total energy and protein intake in rural Bangladesh: Implications for epidemiological studies of diet in developing countries. Food Nutr. Bull. 12 (3), 220±227. Torun B, Flores R, Viteri F, Immink M & Diaz E (1990a): Energy supplementation and work performance: summary of INCAP studies. Korea: Proceedings of the 14th International Congress of Nutrition. Torun B (1909b): Energy cost of various physical activities in healthy children. In Activity, Energy Expenditure and Energy Requirements of Infants and Children, eds. B Schurch, NS Scrimshaw, pp 139±184. Lausanne, Switzerland: International Dietary Energy Consultancy Group. Torun B (1990): Short and long term effects of low or restricted energy intakes on activity of infants and children. In Activity, Energy Expenditure and Energy Requirements of Infants and Children, eds. B Schurch, NS Scrimshaw, pp 335±361. Lausanne: International Dietary Energy Consultancy Group. Witschi JC (1980) Short term dietary recall and recording methods. In: Nutritional Epidemiology, ed. W Willett, pp 52±68. New York: Oxford University Press. Van Esterik P (1984): Intra-family food distribution: its relevance for maternal and child nutrition. Prepared for the Maternal and Child Health Amendment to the Cornell Nutritional Surveillance Program, No. 18. Vermury M (1980): Environmental and social in¯uence on food consumption patterns in six developing countries. New York: CARE.