HarvestPlus Bangladesh Bio-fortified Rice Project 2007-2008

Baseline Dietary Survey

Within each Upazila, all villages or mauzas identified by the 2001 census were included in the sampling universe. During the first sampling stage, 25 clusters (mauzas or villages) were selected from each study site using systematic sampling with probability of selection proportional to estimated population size. Twenty-four clusters in each district were included in the study; an extra cluster was selected for each district in case logistical considerations made recruitment impossible in a cluster. The extra cluster was used in both sites.

At the second stage of sampling, ten households within each cluster were selected using a global positioning system (GPS) sampling method. Digital maps were obtained from the local government or created by walking the boundaries of the cluster with a local guide and a Garmin Etrex Venture hand-held GPS device (Olathe, KS). Each digital map was overlaid with a grid that had lines at 1-sec intervals, so each square on the grid corresponded to a specific set of GPS coordinates. Random pairs of row numbers and column letters were generated in a spreadsheet, each of which corresponded to one square on the grid. The random pairs were tested sequentially until two squares that fell within the boundaries of the cluster digital map were identified as starting points. Two separate starting points were selected to allow estimation of within-cluster variance. A random transect direction for each starting point was generated. In the field, the medical officers in charge of study recruitment used the handheld GPS device to locate the two randomly-selected starting points within each cluster and to follow the designated transect line. Moving along the transect line, the first five households in unique self-identified basis (extended family compounds composed of one or more houses in the same area) with a child 24-48 months of age were identified and invited to participate in the study. This was then repeated at the second starting point for each cluster.

In 11 of 48 clusters (23%) logistics made it impossible to recruit from two starting points, so all households in these clusters were recruited from just one starting point. During the initial contact with a potential study household, the field staff collected basic household intake information. If an eligible child was identified, and the household agreed to participate in the study, consent was obtained using the Bangla version of the consent form, and the intake interview was continued. The names, ages, and dates of birth of the index children, other children (5 years of age), mother, and father were obtained. The ages of the index children were verified by checking the immunization cards. The exact day of the birth month was unknown for nine of the index children and '15' was imputed for the day; for one of the index children the month and day of birth were unknown so the age was not calculated. The first diet study was scheduled during the next week. Approximately 92% of the originally selected eligible households agreed to participate in the study. The sample size in each of the two districts was 240 index children. Data were available from 239 households in Trishal because the forms for one household were lost during transit from the field site to ICDDR,B.