A Tale of Two Measures: Self‐Report and Lab‐Assessed Values in Amount of Oil Added to CSB Porridge Prepared by Caregivers of Children with Moderate Acute Malnutrition in Southern Malawi

Abstract

ObjectiveTo compare self‐reported with laboratory‐assessed grams oil per 100 grams Corn‐Soy Blend (CSB) in porridge prepared by caregivers of beneficiary children receiving the food as treatment for moderate acute malnutritionMethodsThis was a secondary analysis of an effectiveness study conducted in Southern Malawi in 2014 assessing 2 interventions designed to increase the amount of oil added to CSB porridge prepared by caregivers, with a target of 30 g added oil per 100 g CSB. The Control Group received standard monthly ration: 1 L oil, 8 kg CSB in bulk. Intervention groups received 2.6 L oil, 8 kg CSB provided either in bulk (Group 1) or in 4 2‐kg packages with printed messages (Group 2), and social behavior change communication to meet added oil target (Groups 1 and 2). Data were collected through structured interviews with caregivers and lab analysis of porridge samples. Paired sample t‐tests compared oil added to CSB porridge (oil g per 100 g CSB) from self‐report and lab analysis within each study group (Wilcoxon when appropriate); ANOVA test assessed the mean difference (self‐report – lab value) between the study groups. Bland‐Altman plots were used to display the discrepancy between the 2 measures.ResultsA total of 584 caregivers participated: n=192 in Group 1; n=196 in Group 2; n=196 in the Control Group. The mean ± SDs of added oil (in g per 100 g CSB) from self‐report and lab analysis, respectively, were: 30±9 and 28±16 (Group 1), 30±9 and 25±15 (Group 2), 15±9 and 12±10 (Control). Estimated added oil from self‐report was significantly higher than lab analysis within each study group (p<0.05 for all). Among the study groups, the mean differences between the measures of added oil (reported – lab) were not significant (p=0.56). A cluster of observations in the intervention groups had abnormally high fat content in lab assessed values; sensitivity analysis excluding this cluster showed mean lab values of added oil were reduced to 22±12 in both intervention groups, causing differences in self‐report versus lab to become significantly larger in the intervention groups compared to control (p=0.002). Bland‐Altman plots revealed a clear bias between the 2 measures in the intervention groups: participants tended to over‐report the amount of oil used, but as the lab‐assessed amount increased (i.e. as their behavior changed towards using more oil) there was a shift towards under‐reporting. This was less evident in the control group. Both self–report and lab‐assessed values showed the same relationships among the study groups, with Groups 1 and 2 having more added oil than the Control.ConclusionCaregivers in the intervention groups reported what they were instructed to do, regardless of whether lab analysis reflected the targeted behavior change. While self‐report was not as reliable as the lab measure, both conveyed that intervention group caregivers added more oil to porridge than the control group. Laboratory analysis was critical to determine the precise magnitude of added oil to CSB porridge.Support or Funding InformationThis research was performed under the extension of the Food Aid Quality Review (FAQR II) project implemented for the USAID Office of Food for Peace, under the terms of Cooperative Agreement AFP‐C‐00‐09‐00016‐00, by the Tufts University Friedman School of Nutrition Science and Policy.