RESUMO
The poor assessment of child malnutrition impacts both national-level trends and prioritisation of regions and vulnerable groups based on malnutrition burden. Namibia has reported a high prevalence of malnutrition among children younger than 5 years of age. The present study's aim was to identify the optimal methods for estimating child stunting and wasting prevalence in Namibia using two datasets with suspected poor data quality: Namibia Demographic and Health Surveys (NDHS) (1992-2013) and Namibia Household Income and Expenditure Survey (NHIES), 2015/16. This comparative secondary data analysis used two prevalence estimation methods: WHO flags and PROBIT. WHO flags is the standard analysis method for most national household surveys, while the PROBIT method is recommended for poor quality anthropometry. In NHIES (n 4960), the prevalence of stunting (n 4780) was 30·3 and 20·9 % for the WHO flags and PROBIT estimates, respectively, and the national wasting prevalence (n 4637) was 11·2 and 4·2 %, respectively. The trends in nutritional status from NDHS and NHIES showed improvement across WHO flags and PROBIT until 2013; however, from 2013 to 2016, PROBIT showed smaller increases in stunting and wasting prevalence (2·5 and 0·6 percentage points) than WHO flags (6·6 and 5·0 percentage points). PROBIT identified the Khoisan ethnic group and Northern geographical regions with the highest stunting and wasting prevalence, while WHO flags identified similar prevalence across most groups and regions. The present study supports the recommendation to use PROBIT when poor data quality is suspected for constructing trends, and for targeting regions and vulnerable groups.