Agreement between current status and retrospective data for prevalence and duration of exclusive breast feeding from low- and middle-income countries surveys.

Abstract

Indicators to assess exclusive breast feeding (EBF) status are based on current status data according to World Health Organization (WHO), specially to avoid recall bias or imprecise reporting. To analyse the agreement between current status and retrospective data for prevalence and duration of EBF in low- and middle-income countries. Cross-sectional study using Demographic and Health Surveys (DHS) datasets of infants under six months from 10 low- and middle-income countries. It was applied two kinds of data about EBF: (1) current status data: variables about breast milk and foods offered in the previous day (yes or no) and (2) retrospective data: variables about age at which foods were offered the first time. The prevalence of EBF was estimated the same way for current status and for retrospective data. The median duration for current status data was calculated according to WHO recommendation, and retrospective data were calculated using survival analysis. The Kappa coefficient was applied to assess the agreement of prevalence using both kinds of data. Prevalence of EBF was higher using current status data and differences between both data ranged from 0.5 to 6.4 percentage points. Kappa coefficient ranged from 0.74 (substantial) to 0.94 (almost perfect) in nine countries. Medians were higher using retrospective than current status data for two countries and lower for another two; however, differences were small (from 0.08 to 0.44month). For six countries, breast-feeding practice was so low that it was not possible to calculate the median using either data. The difference between prevalence and median estimated using current status and retrospective data is little and the agreement ranged to substantial and almost perfect. We suggest the use of retrospective data to estimate duration of EBF in cross-sectional surveys.