Participants and Methods The data used in this analysis were collected from households that participated in the NSP in Bangladesh from February 2000 to December 2005. The NSP was based on UNICEF’s conceptual framework on the causes of malnutrition (10), with the underlying principle of providing information to guide policies and programs to improve nutrition in Bangladesh and other developing countries (11). The NSP was established by Helen Keller International and the Institute of Public Health Nutrition of the Government of Bangladesh in 1989 (12). The NSP was based on stratified multistage cluster sampling of households in subdistricts of administrative divisions of the country in rural areas and slum areas of large cities. For the purpose of this analysis, only the data from the rural areas was considered. Rural data were collected from 4 subdistricts of each of 7 divisions of Bangladesh: Barisal, Chittagong, Chittagong Hill Tracts, Dhaka, Khulni, Rajshahi, and Sylhet. Data were collected every 2 mo by 2-person field teams employed by the Institute of Public Health Nutrition and various nongovernmental organizations. New households were selected every round. A structured coded questionnaire was used to record data about children aged 0–59 mo, including anthropometric measurements, date of birth, and sex. Maternal weight and height were also measured. The mother of the child or other adult member of the household was asked to provide information on the household’s composition, maternal age, parental education, and weekly household expenditures, along with other socioeconomic, environmental sanitation, and health indicators. The NSP included questions on weekly expenditures on food. In each household, data were gathered regarding the expenditures in the previous week on rice, other staple foods, eggs, vegetables and other plant sources of food, fruits, cooking oil, chicken, fish, other meat, sugar, instant noodles, milk, snacks, clothes, housing, education, cigarettes, savings, social activities, medicine, production activities, recreation, transportation, pocket money, water, and other (gasoline, electricity, etc). Expenditures were divided into 3 categories: rice, nonrice, and other household expenditures. Expenditure and price variables were collected on Bangladesh taka.
In this study, families who had at least 1 child who was aged 6–59 mo were included. Families were excluded if there was only 1 child,6 mo of age. The field teams measured and recorded the weight of each child aged 6–59 mo to a precision of 0.1 kg and the length/height to a precision of 0.1 cm. Birth dates of the children were obtained from birth certificates or other records. When not available, the birth dates were estimated using a calendar of local and national events and converted to the Gregorian calendar. Children with Z-scores, 22 for height for-age were considered stunted (13,14). Malnutrition in children was defined using the child growth standards and criteria of the WHO for stunting (15). Maternal underweight was defined as having a BMI,18.5 kg/m2. In analyses where child-level indicators were involved, such as malnutrition, 1 child in the household, the youngest, was used as the index of child malnutrition for that particular household (i.e. households were not counted more than once).
The participation rate of families in the surveillance system was .97% and the main reason for nonresponse was that the family had moved out of the area or was absent at the time the interviews were conducted. Nonresponse because of refusal to participate in the surveillance system was very low (,1%). The study protocol complied with the principles enunciated in the Helsinki Declaration (16). The field teams were instructed to explain the purpose of the NSP and data collection to each child’s mother or caretaker and, if present, the father and/or household head; data collection proceeded only after written informed consent was given. Participation was voluntary and all participants were free to withdraw at any stage of the interview. The protocol was approved by the ethical review committee of the Bangladesh Medical Research Council and the plan for secondary data analysis was approved by the Institutional Review Board of the Johns Hopkins School of Medicine.
Categorical variables were compared using chi-square tests. Maternal age was divided into quartiles. Parental education was divided into none, 1–6 (primary), 7–9 (junior high), and $10 (high school or greater). The quintile cutoffs for weekly per capita household expenditure were 0.59, 1.13, 1.75, and 2.79 U.S. dollars. The survey did not collect data on total income, because nearly all of these poor rural families spend whatever they earn each week. In the multivariate models, we adjusted for total weekly per capita household expenditure, which is a good proxy for income. Separate multivariate models were used to examine the relationship between rice and nonrice food expenditures and child stunting by age categories 6–11, 12–23, and 24–59 mo. Variables that were significantly associated with child stunting or maternal underweight were included in the respective, final multivariate models. The P-values for trend tests of odds ratios (OR) were obtained for logistic regression models by fitting the ordered independent variable as a continuous variable. Population-based weighting was used to account for differences in population size of the various divisions. The level of significance in this study was P, 0.05.