RESUMO
OBJECTIVE: For years it has been shown that more children die from moderate malnutrition (MM) than severe. Till yet few studies deal specifically with identifying these children. This study attempts to statistically determine the appropriate anthropometric measures and cut-off points for diagnosing moderate malnutrition in preschool children. METHODS: Anthropometric measurements were obtained from 609 preschool children from the cities of Adigrat, Ethiopia; Janampet, India; San Paulo, Brazil. The values were used to determine the sensitivity, specificity, positive predictive value (PPV) and likelihood ratio (LR) of each index studied. The optimum cutoff point for each index was considered to be the cutoff point with the maximum Kappa coefficient for efficiency. The McNemar Test for the significance of changes was used to determine if these findings were in agreement when applied to this data. RESULTS: Weight for height (WFH) at each site had the highest PPV and LR of 4 but was not signficant by the McNemar Test. Mid-upper arm circumference (MUAC) in India had the same PPV (77%) as WFH but a LR of 2. MUAC in India, Brazil and Ethiopia tested significantly for the McNemar Test. The cut-off point for MUAC in India and Brazil was determined to be <15.5 cm in India and Brazil but was <15 cm in Ethiopia. Waist circumference in India tested a significantly PPV of 64%, and a LR of 2. CONCLUSION: These results show that WFH and MUAC could be used with WFA to identify the MM child. The cut-off points for MUAC may vary per location. WC positive data suggests further study is warranted. The McNemar findings yielded significant evidence that statistically determined indicators can be established to identify MM. With further study these methods may prove to be an important component in the efforts to improve child survival.