Nutrition Causal Analysis in Niger: Report of Key Findings (March 2017)
Background and rationale
Levels of acute malnutrition remain persistently high throughout Niger. The 2012 DHS revealed that the national prevalence of wasting among children under 5 was 18.0%; the more recent 2014 National Nutrition Survey reported a slightly lower level of 14.8%. Despite the fact that parts of Maradi and Zinder are often referred to as the “bread basket” of the country, both regions exhibit levels of child wasting above the national average. Given the large burden of acute malnutrition and the severity of its consequences, a strong understanding of the causes of undernutrition in Niger is critical. Although the UNICEF conceptual framework serves as a useful paradigm for conceptualizing the main categories and levels of causes, there is a clear need to identify specific risk factors and understand their relative importance in order to design and target the most effective interventions to combat acute malnutrition in this particular setting.
Scope and methodology
In collaboration with local partners in Niger, FEWS NET conducted a Nutrition Causal Analysis (NCA) study to better understand the key factors that are driving the perpetually high burden of acute malnutrition in the agropastoral and agricultural livelihood zones of Maradi and Zinder. The specific objectives were to:
Identify characteristics which were independently associated with acute malnutrition during each round of data collection and rank their relative importance according to the strength of their association;
Identify characteristics which were associated with a change in WHZ from round 1 to round 2 of data collection among children who had anthropometric measurements taken during both rounds;
Identify the characteristics that were significantly associated with becoming wasted from round 1 to round 2 vs. remaining non-wasted during both rounds of data collection.
Data were collected in the same communities at two points in time: 1) between December 25, 2014 and January 17, 2015, a period which is considered the post-harvest season; and 2) between August 26, 2015 and September 22, 2015, a period which is considered the pre-harvest or lean season. Anthropometric measurements were performed on a total of 1702 children at both time points.
The quantitative component of the study was based on the SMART methodology; however, a detailed questionnaire was developed to collect information on household demographics and socioeconomic status; household food security; water, sanitation and hygiene; maternal nutrition and health; infant and young child feeding practices; child morbidity and utilization of health services.
Descriptive statistics were used to summarize all variables. Multivariate logistic analysis was employed to identify the correlates of wasting during each round of data collection. Multivariate linear regression analysis, using data from children who were included in both rounds of data collection, was employed for objective 2. To identify risk factors for “becoming wasted”, multivariate logistic regression models were run using a restricted dataset that included only those children who remained non-wasted in both rounds of data collection or who were non-wasted in round 1, but became wasted in round 2.
The prevalence of wasting and mean WHZ among children included in both rounds of data collection is summarized in Table 1 (See report) according to round, region, and livelihood zone.
During round 1 of data collection (i.e. the post-harvest season), four factors were independently associated with an increased risk of child wasting in the multivariate analysis: occurrence of diarrhea in the two weeks preceding the survey, having a caregiver who did not wash her hands before eating, being 6-23 months of age (vs. 24-59 months of age), and having a mother without any formal education. The multivariate analysis of data from round 2 (i.e. pre-harvest season) revealed that occurrence of diarrhea in the previous two weeks, residence in an agropastoral livelihood zone, being 6-23 months of age, and having a fever in the previous two weeks were factors significantly associated with an increased risk of child wasting.
Table ES1. Prevalence of wasting and mean WHZ among children 6-59 months of age included in both rounds of data collection according to round, region, and livelihood zone
In the analysis of change in WHZ, children who lived in an agricultural livelihood zone experienced an average increase of 0.25 WHZ scores between round 1 and round 2. For every 10,000 CFA increase in household income during round 1, there was a 0.05 increase in WHZ. Children who experienced a respiratory infection in the 2 weeks preceding round 1, experienced an average increase of WHZ scores. The variable most strongly associated with the change in WHZ was round 1 WHZ. Every one WHZ increment at baseline was associated with a subsequent decline of -0.41 WHZ scores from round 1 to round 2.
Only three variables were significantly associated with the odds of becoming wasted between round 1 and round 2. As in the analysis of change in WHZ, round 1 WHZ was the strongest correlate: children who had a weight-for-height Z score between -1 and -2 during round 1 were 3.94 times more likely to become wasted than children who had a WHZ greater than -1. Similarly, children who were 6-23 months old during round 1 were 1.84 times as likely to become wasted than older children. Finally, children living in agropastoral areas were 1.76 times more likely than children living in agricultural areas to become wasted.
There is significant seasonal variation in the prevalence of child wasting in the agropastoral and agricultural livelihood zones of Maradi and Zinder, Niger. The average prevalence of wasting across all four study areas increased from 9.1% during the post-harvest season to 14.3% during the lean season. The variation was particularly pronounced in the agropastoral livelihood zone.
Although there were no marked differences in the prevalence of child stunting by region, livelihood zone or season, the average prevalence at both time points exceeded 60%. Such levels are alarmingly high and call for intensified efforts to prevent chronic undernutrition.
Household food insecurity, as defined by the Household Food Insecurity Access Scale or Food Consumption Score, was not independently associated with the odds of being wasted or becoming wasted. In a typical year, the prevalence of child wasting is a poor indicator of the household food security situation. High levels of wasting do not automatically imply a food security crisis.
Child morbidity was a consistent risk factor for wasting, reiterating the importance of the infection-undernutrition cycle. Services to treat and prevent child morbidity and acute malnutrition are critical. It may be most effective to design and target strategies according to the child’s age and livelihood zone.