ORIGINAL COMMUNICATION Food consumption, energy and nutrient intake and nutritional status in rural Bangladesh: changes from to

Transcription

1 (2003) 57, ß 2003 Nature Publishing Group All rights reserved /03 $ ORIGINAL COMMUNICATION Food consumption, energy and nutrient intake and nutritional status in rural Bangladesh: changes from to O Hels 1 *, N Hassan 2, I Tetens 1 and S Haraksingh Thilsted 1 1 Research Department of Human Nutrition, The Royal Veterinary and Agricultural University, Frederiksberg, Denmark; and 2 Institute of Nutrition and Food Science, University of Dhaka, Dhaka, Bangladesh Objective: To determine and evaluate changes in nutritional status, food consumption, energy and nutrient intake in rural Bangladesh, using appropriate statistical analyses. Design: Repeated cross-sectional surveys. Two seasons in and Setting: Two villages with different production patterns. Subjects: Anthropometric measurements of 1883 individuals, food consumption data of 404 households. Methods: Repeated measurements of one-day food weighing and anthropometry in two seasons in and Mixed model analyses were used to evaluate and quantify temporal changes and their interactions with determinants. Results: Prevalence of underweight children decreased from 82 to 70% (P ¼ 0.015), wasted children from 34% to 18% (P ¼ 0.009) and chronic energy deficient adults decreased from 78 to 64% (P < ). Intake of fish and green leafy vegetables increased from (l.s. mean s.e.) to g=person=day (P < 0.001) and from to g=person=day (P ¼ 0.019), respectively. Rice intake remained unchanged: g raw=person=day in and g raw=person=day in (P ¼ 0.355). Calcium and iron intakes increased by 40% (P < ) and 16% (P ¼ ), respectively, whereas vitamin A intake remained unchanged. Conclusions: Nutritional status improved and intakes of nutrient dense food groups, fat, iron and calcium increased from to Sponsorship: Supported by the Danish Council for Research in Developing Countries, the Bilateral Programme for Enhancement of Research Capacity in Developing Countries (ENRECA), Ministry of Foreign Affairs, Denmark and the Ford Foundation. (2003) 57, doi: =sj.ejcn Keywords: dietary trends; temporal change; nutritional status; mixed model analysis; Bangladesh *Correspondence: O Hels, Research Department of Human Nutrition, The Royal Veterinary and Agricultural University, Rolighedsvej 30, 1958 Frederiksberg C, Denmark. ohe@kvl.dk Guarantor: O Hels. Contributors: OH performed the statistical modelling and analyses and wrote the first draft of the manuscript. SHT co-conceived the study and was involved in writing and revising the manuscript. IT was involved in the revisions of the manuscript and in the data processing. NH coconceived the study and was responsible for field work and data collection. All contributors were involved in preparing the final draft of the manuscript. Received 7 January 2002; revised 4 April 2002; accepted 17 June 2002 Introduction Since the 1996 World Food Summit, efforts towards alleviating food insecurity in developing countries have been strengthened. Accurate and relevant information generated from periodic and systematic measurements of food production and consumption and nutritional status provides a good foundation for estimating trends as a basis for taking appropriate actions to achieve food security. To evaluate the effects of political actions, as well as nutrition interventions and changes of food production systems (eg agriculture and aquaculture), changes in food consumption need to be monitored at an appropriate level. Two main types of dietary data which can be used for this purpose at a population level

2 are obtained from national estimates of the food supply and results of dietary surveys involving a larger or smaller number of households or individuals sampled from the population. Some developed countries have established or conducted regular dietary surveys of representative population samples. These are sometimes carried out 5 or 10 y apart or done continuously (Pietinen & Ovaskainen, 1994; Briefel, 1994; Turrini et al, 2001; Kim et al, 2000). Studies in developed countries have also either used national food supply data alone to monitor national developments (Zilidis, 1993; Zizza, 1997) or compared national food supply data and dietary surveys (Crane et al, 1992; Rodriguez-Artalejo et al, 1996; Dobson et al, 1997; Harnack et al, 2000). It is recognized that surveys and food supply data provide partially different and sometimes inconsistent information. In developing countries, data from national dietary surveys are generally scarce. Results are often published in the form of reports that are not easily accessible. Many developing countries rely to a certain extent on food supply data and household budget surveys which may be combined with minor surveys on food consumption and=or health in the monitoring of the food security situation and estimation of changes and trends with time. The costs of regular dietary surveys of representative population samples may be prohibitive for many developing countries when recognized determinants of food consumption and nutritional status such as seasonality and regional differences are also considered. Lower-cost methodology procedures in developing countries such as data from minor periodic food consumption surveys analysed with advanced statistical methods can play a role in quantifying variation and estimates in trends in food security. In Bangladesh, the first national food consumption survey was carried out in , followed by two more in and (Hassan & Ahmad, 1984). The latest large-scale food consumption survey in Bangladesh was undertaken in (Jahan & Hossain, 1998). Results from these surveys suggest that nutritional status improved from to but remained unchanged during the last two decades and intakes of energy and macronutrients in rural Bangladesh declined from to and Intakes of iron, calcium and vitamin A declined between and but increased from to to a level similar to that in Anaemia and vitamin A deficiency have remained unchanged during the last three decades and are persisting to be public health problems in Bangladesh (Ahmed, 1999, 2000). Although the overall food and nutrition situation in Bangladesh and other South-Asian countries has been described (Gopalan, 1996; FAO, 2000), changes in time have not been tested with the use of appropriate statistical methods. The objective of this study is to determine and evaluate changes in nutritional status, food consumption, energy and nutrient intake in two rural villages in Bangladesh by use of appropriate statistical analyses, which adjust for seasonal, locational and socioeconomic variations. Methods Study area The field work was conducted in two selected villages in Bangladesh: Falshatia and Jorbaria. Falshatia, in the Manikganj thana (subdistrict) is a flooded, double-cropped area, west of Dhaka and Jorbaria. Mymensingh is a non-flooded, triple-cropped area, north-east of Dhaka. In the Manikganj village, the agricultural production system can be characterized as traditional, whereas in the Mymensingh village the farming system practised can be characterized as modern, with the use of high-yielding rice varieties, irrigation, fertilizers and pesticides. In Bangladesh, the three rice harvest seasons comprise the Aman (November December), the Aus (March April), and the Boro (May June). The Aman harvest was traditionally the most important, but in areas where high-yielding rice varieties and modern technology have been introduced, the Boro harvest has become more important during the last decade. The present study was undertaken to cover two seasons with respect to rice production: the lean season which includes and the peak season which includes December March. The study was conducted in and Study population and sampling design In , the study population comprised 50 households in each village in both seasons ( 1981 and December 1981 January 1982) and in , 152 households in each village in both seasons (October November 1995 and January March 96). The data collected in was a part of the national nutrition survey of rural Bangladesh in (Ahmad & Hassan, 1983). In , households were sampled and data collected using the same technique as in The households were selected from four socioeconomic subgroups following the probability proportion to size method. The four subgroups were based on the Bangladesh rural population tax system: group A is the poorest, most certainly landless and not required to pay any local tax; group B is the less poor and pays a token tax. Group C may be regarded as a rural upper-middle class and pay an appreciable amount of tax, while group D is the richest group and pays the highest proportion of tax. In Manikganj, the proportions of sampled households in the two wealthiest groups were 14% in and 36% in In Mymensingh, the proportions of the wealthiest groups were 24% in and 10% in , perhaps due to the concentration of landholding on fewer households. Of the 100 households studied in , data were missing from one household in December January. Of the 304 households studied in 1995, data were missing from 13 households in January March. The 587

3 588 study covered a total of 510 individuals in 1981, 512 individuals in December January , 1373 individuals in 1995 and 1281 individuals in January March The characteristics of the surveyed households are shown in Table 1. Although the average household size was larger in than in , the difference was not significant. The average number of consumption units for each season and year ranged from 4.4 to 5.3 and decreased from 5.2 (95% CI ) in to 4.5 (95% CI ) in and (Darnton-Hill et al, 1988). The foods consumed were divided into selected food groups on the basis of the above food composition table with the following modifications: pulses and legumes, green leafy vegetables and vegetables were pooled in the food group of vegetables; fats, oils and oil seeds were pooled in the food group of fats and oils and fish, meat, eggs, milk and milk products were pooled in the group of animal products. Fish were divided into small and large fish, of which all the small fish are indigenous and most of the large fish are cultured. Dietary intake Dietary intakes of all members of the selected households were obtained by 24 h food weighing. Training of investigators comprised practising the techniques of food weighing and recording at the Institute of Nutrition and Food Science, University of Dhaka, Bangladesh for 2 weeks, followed by a pilot survey in a village close to Dhaka city. Exactly the same method was used in and All food that went into the household cooking pots was weighed as raw weight. Intake per person was calculated at household level by dividing the total intake of each single food in every household by the total number of consumption units in the household. An individual present at every meal within the one day period was counted as a full-time consumer, with the value of one consumption unit. All family members, including children and breast-fed infants, were valued similarly in terms of consumption unit. Arithmetical adjustments of the household s number of consumption units were made for household members who were partially absent. If a household member was absent at one out of the usual three meals consumed in the household, twothirds of a consumption unit, instead of one, was allocated to this member (FAO, 1967). To avoid changes in energy and nutrient intakes caused by differences in food composition values, the values of all foods were computed using the same food composition table pertinent for the Bangladeshi diet for both Anthropometry Height, weight and mid upper arm circumference (MUAC) of all members of the selected households were obtained by house visits during both seasons in and A portable beam balance was used to measure weight to the nearest 0.5 kg after removal of shoes and heavy clothing. The balances were frequently checked using standard weights. Two types of wooden scales were used for measuring height: one for recording recumbent length to the nearest cm of children under 24 months of age and the other for height to the nearest cm of older children and adults. The investigators collecting the anthropometric data underwent extensive training. The nutritional status of children between 6 and 59 months was expressed as height for age, chosen as the indicator for chronic nutritional status. Weight for height and weight for age were chosen as indicators for acute nutritional status and current malnutrition. Malnourished children were defined as those below 72 s.d. of the median of the WHO s height-for-age (stunting), weight-for-height (wasting) and weight-for-age (underweight) values, respectively (WHO, 1983). The nutritional status of adults (here defined as above 18 y of age) was expressed as body mass index (BMI) and calculated as weight=height 2 (kg=m 2 ). The international criteria on classification of chronic energy deficiency (CED) was employed (BMI < 18.5 kg=m 2 ). Table 1 Characteristics of surveyed households in and in Bangladesh December January January March Number of households Household size, mean (s.d.) 5.6 (2.8) 5.5 (3.1) 5.0 (2.0) 5.0 (2.1) Consumption units, mean (s.d.) 5.2 (2.3) 5.3 (2.8) 4.6 (2.3) 4.4 (2.4) Age groups: 0 9 y 27% 29% 29% 29% y 25% 25% 19% 20% y 37% 35% 40% 39% > 50 y 11% 11% 12% 12% Pregnant females (%) 6% 5% 5% 5% Lactating females (%) 25% 26% 21% 19%

4 Nineteen children in and two in were excluded from the analyses of nutritional status due to recorded negative height difference between seasons. Because of unrealistic seasonal height differences, two adults were excluded in Data analysis All analyses were performed with the Statistical Analysis Systems statistical software package version 8.1 (SAS, 2000). Prevalence of stunting, wasting and underweight among children and CED among adults were analysed by logistic regressions using the SAS procedure GENMOD. Year, village, season, gender and socioeconomic group were included as independent fixed variables. Individuals nested within both year, village and household and households nested within year and village were included as random effects in the statistical model. MUAC for children and adults and BMI for adults were analysed by ANOVA as described below. All main effects and two factor interactions between fixed effects were tested. Only P-values for year and significant fixed effects are presented. Univariate analyses of variance (ANOVA) were performed in the procedure MIXED (Littell et al, 1996). In the statistical model, consumption of food by food group, was evaluated as the dependent variable. Year, village, season and socioeconomic group were included as independent fixed variables. Households nested within both year and village were included as random effect in the statistical model to take into account that repeated measures were done in households within years but the households between years were different. All main effects were evaluated as well as two- and three-factor interactions between all fixed effects. To obtain sufficient numbers of individuals in each socioeconomic group, the two wealthiest groups were pooled (group C and D). Preliminary statistics showed that the two wealthiest groups were not significantly different. Homogeneity of variance and normal distribution among random effects were investigated by plots and histograms of residuals. Shapiro Wilk s test for normal distribution was performed. The investigations showed it necessary to logtransform intakes of protein, fat iron, calcium, vitamin A and the food groups fats and oils and spices. Least square means (l.s. means) and standard errors (s.e.) were estimated and transformed back where necessary before presentation. Significance level was set to 0.05 for main effects, 0.01 for two factor interactions and for three factor interaction to reduce probability of type I error. Only P-values for main fixed effects, significant interactions and the three factor interaction between year, village and season are presented. Results Nutritional status The prevalence of wasting and underweight children aged 6 59 months decreased from to (Table 2). MUAC increased significantly from to and an increase of mm (mean s.e.) was detected between the two seasons in both years. Boys had a mm higher MUAC (mean s.e.) than girls (P ¼ 0.015). The seasonal variation of wasting differed in the years studied: the prevalence was similar in both seasons in (P ¼ 0.654), whereas it decreased from 23 to 13% in (P < 0.010). Taking into consideration both years and seasons, the prevalence of underweight was 69% in the Manikganj village and 78% in the Mymensingh village (P ¼ 0.016), and the prevalence of wasting was 19 and 29%, respectively (P ¼ 0.062). Nutritional status expressed as the prevalence of CED as well as BMI and MUAC for adults improved from to by a decreased prevalence of 14%, increased BMI of kg=m 2 (mean s.e.) and increased MUAC of mm (mean s.e.; Table 3). The Manikganj village had the best nutritional status (63% CED, BMI ¼ 18.2 kg=m 2 and MUAC ¼ 229 mm) compared to the Mymensingh village (72% CED, BMI ¼ 17.4 kg=m 2 and MUAC ¼ 226 mm). Both BMI and MUAC were highest for the wealthiest groups, which also had the lowest prevalence of CED. No significant differences between men and women were seen other than MUAC in women was mm lower MUAC (mean s.e.) than that in men (P < 0.001). 589 Table 2 MUAC, prevalence and OR of stunting, wasting and underweight among children 6 59 months in compared with Estimate , n ¼ 65 Year , n ¼ 149 MUAC a (mm) l.s. mean (s.e.) 133 (1.5) 139 (1.5) Stunting b (height-for-age < 72 s.d.) c Prevalence 74 % 68 % OR (95%CI) 1.4 ( ) reference Wasting d (weight-for-height < 72 s.d.) c Prevalence 34 % 18 % OR (95%CI) 3.2 ( ) reference Underweight e (weight-for-age < 72 s.d.) c Prevalence 82 % 70 % OR (95%CI) 2.1 ( ) reference a Year, P ¼ 0.003; season, P ¼ 0.010, gender, P ¼ b Year, P ¼ c Of WHO reference median. d Year, P ¼ 0.009; season, P ¼ 0.053; village, P ¼ 0.016, yearseason, P < e Year P ¼

5 590 Table 3 Prevalence of CED, BMI (kg=m 2 ) and MUAC (mm) of non pregnant and non lactating adults over 18 y of age in and Estimate , n ¼ 181 Year , n ¼ 708 CED a (BMI > 18.5 kg=m 2 Prevalence 78% 64% OR (95%CI) 2.4 ( ) reference BMI b (kg=m 2 ) l.s. mean (s.e.) 17.3 (0.18) 18.3 (0.11) MUAC c (mm) l.s. mean (s.e.) 223 (1.5) 232 (0.9) a Year, P < 0.001; season, P ¼ 0.001; socioeconomic group, P ¼ 0.006; village, P ¼ 0.017; yearseason, P ¼ 0.008; villagesocioeconomic group, P ¼ b Year, P < 0.001; season, P < 0.001; socioeconomic group, P ¼ 0.018, village, P < c Year, P < 0.001; season, P < 0.001; socioeconomic group, P < 0.001; village, P ¼ 0.013; gender, P < Food, energy and nutrient intakes Mean intakes of food groups (g=day) are shown in Table 4 and mean intakes of energy and nutrients are shown in Table 5. There were no overall significant differences between years, seasons and villages combined in food, energy and nutrient intakes. A statistically significant increased intake of food from to was seen for most food groups except cereals (P ¼ 0.123), rice (P ¼ 0.351), fruit and miscellaneous foods. The intake of fish nearly doubled from to while the proportion of households eating fish increased from 70% (61% in the Manikganj village and 80% in the Mymensingh village) to 82%. Both the frequency of eating large and small fish increased, from 29 to 44% and from 48 to 57%, respectively, from to The seasons covered were not seasons for fruit, and the fruit intake was accordingly low ( g=person= day), with no significant differences. Miscellaneous foods mainly comprised sugar and sweets with no significant differences in intakes ( g=person=day). The intakes of fat, iron and calcium increased from to , whereas the intakes of energy and other nutrients were similar (Table 5). Intake of all food groups, energy and nutrients increased with increasing socioeconomic group. The seasonal variation in the intake of vegetables, energy, protein, carbohydrates and iron was different in compared with In , the intakes remained unchanged between seasons whereas in , they increased significantly from to January March, although there were no significant differences in comparing the intakes in between the years. The intake of vegetables almost doubled from to December March. Consumption of animal foods and fish was lowest in December March. Seasonal variation in the frequency of eating fish and the intake of roots and tubers, vegetables, green leafy vegetables, spices, energy and macronutrients were different for the two villages. In the Mymensingh village, the increase in vegetables, energy and macronutrients from to December March was significantly higher compared with the increase in the Manikganj village, although the intakes in the two villages in were similar. Intakes of roots and tubers nearly tripled and green leafy vegetables doubled from to December March in the Manikganj village. In the Mymensingh village, the opposite tendency was detected: a significant reduction in intakes of both roots and tubers and green leafy vegetables. However, the proportion of households eating green leafy vegetables on the specific day measured in the Manikganj village was 49% in and 33% in December March, compared with 33% and 55%, respectively, in the Mymensingh village. Discussion Nutritional status Although the present study showed that nutritional status improved from to , the prevalence of chronically malnourished children under 5 years and CED adults over 18 y was still high in The crude mean values for MUAC, BMI and estimated prevalence of malnutrition found in other Bangladeshi rural studies (BBS, 1997; Jahan & Hossain, 1998) fall within the 95% confidence intervals of values found in this study. The improvement in nutritional status from to was similar for both socioeconomic groups. Better nutritional status in both children and adults was seen in the sample population in Manikganj village with traditional agricultural technology compared to that in Mymensingh with modern technology. Since a larger proportion of the households belonged to the wealthier group in Manikganj village, this may have resulted in a better nutritional status. In addition, the use of high-yielding varieties of rice resulted in an increased intake of rice and energy in Mymensingh, but at the same time, the workload with respect to irrigation, use of fertilizers and pesticides may have increased the physical activity. The increased energy intake in adults may not have compensated for the increased physical activity. Some studies have reported regional differences in the prevalence of malnourished children (Ahmed, 1993; Jahan & Hossain, 1998), but the differences were not statistically tested. Bangladesh Bureau of Statistics (BBS, 1997) reported

6 591 Table 4 Estimated daily intakes of selected food groups per consumption unit in sampled households in and by season and village (l.s. mean (s.e.)) Food group (g=day) Manikganj Mymensingh December January n ¼ 49 n ¼ 152 January March n ¼ 145 December January n ¼ 152 January March n ¼ 143 P-value a Cereals 484 (20) 459 (21) 419 (12) 474 (12) 475 (20) 533 (20) 448 (12) 514 (12) Rice 451 (21) 444 (21) 399 (12) 454 (12) 446 (21) 513 (21) 446 (13) 502 (13) Vegetables (total) 206 (18) 144 (18) 134 (10) 215 (10) 88 (18) 120 (18) 105 (11) 258 (11) Green leafy vegetables 6 (8.6) 43 (8.7) 28 (4.9) 49 (5.0) 44 (8.6) 17 (8.6) 46 (5.1) 37 (5.2) Roots and tubers 18 (9.6) 41 (9.6) 43 (5.4) 101 (5.6) 51 (9.5) 18 (9.5) 64 (5.6) 37 (5.8) Animal foods (total) 46 (9.0) 22 (9.1) 75 (5.1) 62 (5.3) 40 (9.0) 30 (9.0) 72 (5.4) 69 (5.5) Fish 37 (5.1) 15 (5.2) 52 (2.9) 37 (3.0) 24 (5.1) 16 (5.1) 37 (3.0) 30 (3.1) Fats and Oils 3 (1.8) 2 (1.8) 7 (1.0) 5 (1.0) 4 (1.8) 2 (1.8) 10 (1.1) 9 (1.1) Spices 3 (0.9) 3 (0.9) 5 (0.5) 6 (0.5) 4 (0.9) 2 (0.9) 8 (0.5) 6 (0.5) Fruits 3 (3.8) 1 (3.8) 7 (2.2) 7 (2.2) 3 (3.8) 11 (3.8) 6 (2.2) 9 (2.3) Miscellaneous 1 (5.0) 2 (5.0) 10 (2.8) 13 (2.9) 2 (5.0) 12 (5.0) 5 (2.9) 11 (3.0) Total intake 764 (36) 675 (37) 701 (21) 887 (21) 667 (36) 727 (36) 719 (22) 914 (22) a P yearseasonvillage. Other significant P-values: Pyear < 0.05 for vegetables, green leafy vegetables, roots and tubers, animal foods, fish, fats and oils, spices and total intake. Pseason < 0.05 for all food groups except vegetables. Pvillage < 0.05 for cereals, rice, roots and tubers, animal foods, fish and fats and oils. Psocioecnomic groups < 0.05 for cereals, rice, animal food, fish, fats and oils, spices and total intake. P seasonvillage < 0.01 for vegetables, green leafy vegetables and fats and oils. P yearseason < 0.01 for vegetables and total intake. P yearvillage < 0.01 for roots and tubers. Table 5 Estimated daily intakes of energy and nutrients per consumption unit in sampled households in and by season and village (l.s. mean (s.e.)) Food group Manikganj Mymensingh December January n ¼ 49 n ¼ 152 January March n ¼ 145 December January n ¼ 152 January March n ¼ 143 P-value a Energy (MJ=day) 7.9 (0.4) 7.4 (0.4) 7.5 (0.2) 8.5 (0.2) 7.9 (0.4) 8.6 (0.4) 8.0 (0.2) 9.2 (0.2) Fat (g=day) 9.0 (1.3) 7.2 (1.3) 13 (0.8) 12 (0.8) 8.6 (1.3) 9.2 (1.3) 15 (0.8) 18 (0.8) Protein (g=day) 48 (2.3) 43 (2.3) 44 (1.3) 48 (1.4) 47 (2.3) 53 (2.3) 46 (1.4) 54 (1.4) Carbohydrates (g=day) 397 (17) 378 (17) 362 (9.8) 415 (10) 392 (17) 437 (17) 390 (10) 448 (11) Iron (mg=day) 24 (1.6) 22 (1.6) 24 (0.9) 29 (0.9) 26 (1.6) 22 (1.6) 25 (0.9) 29 (1.0) Calcium (mg=day) 335 (34) 189 (34) 339 (19) 348 (20) 272 (34) 259 (34) 387 (20) 398 (20) Vitamin A (IU=day) 2017 (339) 1655 (342) 1365 (193) 1938 (198) 2322 (337) 1043 (341) 1773 (200) 1553 (206) a Pyearseasonvillage. Other significant P-values: P year < 0.05 for fat, iron and calcium. P season < 0.05 for energy, fat, protein, carbohydrates and iron. P village < 0.05 for energy, fat, protein, carbohydates and iron. P socioecnomic groups < 0.05 energy, fat, protein, carbohydates and iron. P seasonvillage < 0.01 for energy, fat, protein and carbohydates. P yearseason < 0.01 for energy, protein, carbohydates and calcium.

7 592 an unchanged prevalence of wasted rural children from to , a larger reduction of stunting and a similar reduction in underweight as compared with this study. From the reported values, it is only possible to calculate difference of crude mean values of MUAC, BMI and prevalence of malnutrition in the estimation of change, which is insufficient for evaluating whether the change with time is statistically significant or whether it interacts with other factors which can affect change such as seasonality and regional variation. Therefore, the results of this study are better than those from previous studies because appropriate adjustments were made for seasonal, locational and socioeconomic variations. Food, energy and nutrient intakes Total intakes of vegetables, intakes of green leafy vegetables, roots and tubers, total animal foods, fish, fats and oils and spices all contributed to an increased total intake of food from to At the same time, a nonsignificant decline in the intakes of cereals, rice, energy and carbohydrates was found, and thus these intakes remained similar in compared with in the two villages surveyed. Two rural villages in Bangladesh are insufficient to represent the total rural part of the country. The villages, however, were purpose chosen for their agricultural production systems and proneness to flooding and the results obtained therefore represent areas of the rural Bangladesh with these characteristics. In comparison, the ranges of intakes of vegetables, animal foods, energy and most nutrients were found to be higher in the present study compared with the large-scale nutrition survey (Jahan & Hossain, 1998), perhaps because the large-scale nutrition survey included more locations and seasons than the present study. Nevertheless, the trends of increased intakes of roots and tubers, animal foods, fats and oils, fat and calcium and decreased intakes of cereals and rice from to described by Jahan and Hossain (1998) are in agreement with the trends found in the present study. From FAO s food balance sheets (FAO, 2001), it is found that the domestic utilization of rice as food per capita increased from 143 kg raw rice=year in to 146 and 152 kg raw rice=y in 1995 and 1996, respectively. Seasonal and regional variation in rice production limits national rice production as a single measure of availability for the sampled population. Other influential measures however, such as falling rice prices and increased income, tend to point towards increased availability of rice (Ahmed, 1993; Delgado et al, 1999). The increased availability of rice and an unchanged intake found in the present study suggest that rice intake in rural Bangladesh has reached and continues to be at a satiation point, although seasonal variation and minor variation by location still exist. The findings of both better nutritional status and increased consumption of nutrient-dense foods such as vegetables, green leafy vegetables and fish point in the direction of an improved diet in the two rural villages. However, the increased intakes of some food groups, total food intake and fat did not result in a significant increased energy intake. That the increase in energy intake was insignificant while nutritional status improved, suggests that other determinants for nutritional status than energy intake have changed. Improvement of vaccination programmes for children and delivery of primary health care facilities between and might have improved public health and thereby contributed to the improved nutritional status despite unchanged energy intake. Other factors contributing to the improved nutritional status despite unchanged energy intake may have been decreased energy expenditure due to decreased physical activity between and However, there are no data on changes in physical activity available to illustrate this point. To improve adequacy of energy and nutrients in accordance with the goal of the World Food Summit, much larger increases are needed. If micronutrients are considered, the estimated intakes of calcium and iron increased significantly by 40% (P < ) and 16% (P ¼ ), respectively, whereas vitamin A intake remained unchanged. These changes remain valid even though they are based on the local food composition table, in which there are missing values for some foods as well as too high values for the iron content of some foods. Micronutrient malnutrition still remains a major problem in rural Bangladesh (Ahmed, 1999, 2000) and the detected improvements could be taken further by focussing on the potential of important food groups rich in micronutrients and increased intakes, in amounts as well as frequencies. Small fish, of which some are very high in micronutrient contents (Roos, 2001), and green leafy vegetables are good examples. At the same time, these foods can increase bioavailability of minerals by supplying increased amounts of enhancers such as fish protein, vitamin C and vitamin A (García-Casal et al, 1998; Hallberg & Hulthen, 2000). Looking at the seasonal differences by location in relation to differences in production patterns, minimizing seasonal differences by extending the growing season of nutrient dense foods may also be a possibility for further improvement. Study methodology strengths and limitations As in the major food consumption surveys in rural Bangladesh, the method used for assessing food and macronutrient intakes at the two points in time was one-day food weighing. Recognized determinants of food consumption and nutritional status such as seasonality, regional variation, socioeconomy and their interactions were considered in this study, improving the estimates and tests by making appropriate adjustments. This has not been done before in a developing country. Even the most recent large-scale food consumption surveys from other countries such as the Baltic

8 Republics (Pomerleau et al, 2001) and Italy (Turrini et al, 2001) have not quantified variation in the evaluation in intakes among recognized determinants, although they are to some extent considered by calculating simple averages and=or medians. However, some studies have compared differences in crude mean values with or without standard errors at different points in time, supplemented with estimates from food supply data in order to evaluate trends (Dobson et al, 1997; Harnack et al, 2000; Kim et al, 2000). One of the strengths of the above studies is the large number of households=individuals included and the representativeness. The comparisons done, however, lack proper statistical modelling which utilizes the design of the surveys and quantifies and evaluates variation. The reported decline in rice, energy and carbohydrates found through surveys (Hassan & Ahmad 1984; BBS, 1997; Jahan & Hossain, 1998) and detected trends in food production (Gopalan, 1996) may perhaps be erroneous and this uncertainty could have been avoided if an appropriate statistical model, such as the one used in this study, was employed. The statistical models and analyses used in the present study could have been strengthened if certain considerations were taken into account. By the definition of consumption unit, all individuals were valued equally (FAO, 1967) and the dependent variables in the overall statistical model were adjusted for household size and absent household members. In the comparison of food consumption, energy and nutrient intakes in and , the error introduced by use of this definition is assumed to be of minor importance as the household size and composition remained unchanged. The decrease in the average number of consumption units per household between years, however, suggested a change in meal pattern towards an increased number of meals being taken outside the household. Data on meals taken outside the home were not included in this study, but adjusted for by the use of consumption units. An improvement of the method used can be achieved by collecting information on foods eaten outside home and including it in the food consumption data as well as adjusting the consumption unit so that age and gender of household members are taken into account. In order to compare results from different surveys, the same method of adjustment must be used. Such a standard has not yet been developed. In this study, the use of mixed model analyses showed that there was an improved nutritional status, and increased intakes of nutrient dense food groups, fat, iron and calcium, while intakes of cereals and vitamin A remained unchanged between and in two locations representative of rural Bangladesh. Acknowledgements The contributors wish to thank Dr Jóhanna Haraldsdóttir from The Royal Veterinary and Agricultural University, Denmark (RVAU) for fruitful discussions at the final stages of preparing the manuscript and Dr Annette K. Ersbøll, also from RVAU, for statistical supervision. The members of the surveyed households as well as the field staff who conducted the surveys are thanked. References Ahmad K & Hassan N (1983): Nutrition Survey of Rural Bangladesh, Dhaka: Institute of Nutrition and Food Science, University of Dhaka. Ahmed AU (1993): Patterns of Food Consumption and Nutrition in Rural Bangladesh. Washington DC: International Food Policy Research Institute. Ahmed F (1999): Vitamin A deficiency in Bangladesh: a review and recommendations for improvement. Publ. Health Nutr. 2, Ahmed F (2000): Anaemia in Bangladesh: a review of prevalence and aetiology. Publ. Health Nutr. 3, BBS (1997): Child Nutrition Survey of Bangladesh Dhaka: Bangladesh Bureau of Statistics. Briefel RR (1994): Assessment of the US diet in national nutrition surveys: national collaborative efforts and NHANES. Am. J. Clin. Nutr. 59 (Suppl 1), S164 S167. Crane NT, Lewis CJ & Yetley EA (1992): Do time trends in food supply levels of macronutrients reflect survey estimates of macronutrient intake? Am. J. Public Health 82, Darnton-Hill I, Hassan N, Karim R & Duthie MR (1988): Tables of Nutrient Composition of Bangladeshi Foods. Dhaka: Helen Keller International, Bangladesh. Delgado C, Rosegrant M, Steinfeld H, Ehui S & Courbois C (1999): Livestock to Food, Agriculture, and the Environment Discussion Paper 28. Washington, DC: International Food Policy Research Institute. Dobson A, Porteous J, McElduff P & Alexander H (1997): Dietary trends: Estimates from food supply and survey data. Eur. J. Clin. Nutr. 51, FAO (1967): Manual on Household Food Consumption Surveys. FAO Nutritional Studies No. 18. Rome: FAO. FAO (2000): The State of Food Insecurity in the World. Rome: FAO. FAO (2001): Food Balance Sheet of Bangladesh 1981 and Statistics Division. ¼ nutrition. García-Casal MN, Layrisse M, Solano L, Barón MA, Arguello F, Liovera D, Ramírez J, Leets I & Tropper E (1998): Vitamin A and b-carotene can improve nonheme iron absorption from rice, wheat and corn by humans. J. Nutr. 128, Gopalan C (1996): Current food and nutrition situation in South Asian and South-East Asian Countries. Biomed. Environ. Sci. 9, Hallberg L & Hulthen L (2000): Prediction of dietary iron absorption: an algorithm for calculating absorption and bioavailability of dietary iron. Am. J. Clin. Nutr. 71, Harnack LJ, Jeffery RW & Boutelle KN (2000): Temporal trends in energy intake in the United States: an ecologic perspective. Am. J. Clin. Nutr. 71, Hassan N & Ahmad K (1984): Studies on food and nutrient intake by rural population of Bangladesh: comparison between intakes of , and Ecol. Food Nutr. 15, Jahan K & Hossain M (1998): Nature and Extent of Malnutrition in Bangladesh. Bangladesh National Nutrition Survey, Dhaka: Institute of Nutrition and Food Science, University of Dhaka. Kim S, Soojae M & Popkin BM (2000): The nutrition transition in South Korea. Am. J. Clin. Nutr. 71, Littell RC, Milliken GA, Stroup WW & Wolfinger RD (1996): SAS 1 System for Mixed Models. Cary, NC: SAS Institute Inc. Pietinen P & Ovaskainen MJ (1994): Gaps in dietary-survey methodology in Western Europe. Am. J. Clin. Nutr. 59 (Suppl 1), S161 S163. Pomerleau J, McKee M, Robertson A, Kadziauskiene K, Abaravicius A, Vaask S, Pudule I & Grinberga D (2001): Macronutrient and food intake in the Baltic republics. Eur. J. Clin. Nutr. 55,

9 594 Rodriguez-Artalejo F, Banegas JR, Graciani A, Hernandez-Vecino R & del Rey-Calero J (1996): Food supply versus household survey data: nutrient consumption trends for Spain, Eur. J. Epidemiol. 12, Roos N (2001): Fish consumption and aquaculture in rural Bangladesh. Ph.D. thesis. Frederiksberg, Denmark: Research Department of Human Nutrition, The Royal Veterinary and Agricultural University. SAS (2000): SAS=STAT User s Guide, Version 8.1. Cary, NC: SAS Institute Inc. Turrini A, Saba A, Perrone D, Cialfa E & D Amicis A (2001): Food consumption patterns in Italy: the INN-CA study Eur. J. Clin. Nutr. 55, WHO (1983): Measuring Change in Nutritional Status. Geneva: World Health Organization. WHO (1995): Physical Status: the Use and Interpretation of Anthropometry. WHO Technical Report Series no Geneva: WHO. Zilidis C (1993): Trends in Nutrition in Greece: use of international data to monitor national developments. Publ. Hlth 107, Zizza C (1997): The nutrient content of the Italian food supply Eur. J. Clin. Nutr. 51,