Background: Anaemia is a major complication of Plasmodium falciparum malaria among small children in sub-Saharan Africa. We studied the performance of the Integrated Management of Childhood Illness (IMCI) recommended assessment of no/some/severe pallor as predictor of anaemia in health surveys at community level and in clinical practice in an out patient department (OPD) and in a hospital ward in rural Tanzania.
Methods: The study was undertaken among 6-36 months old children. Pallor was evaluated as a combined assessment of conjunctiva, tongue and palms and categorised as no, some or severe pallor. Packed cell volume (PCV) was measured and related to pallor.
Findings: A total of 740 examinations were performed at village, OPD and in the hospital ward. The prevalences of severe pallor were 0%, 1.5% and 7% respectively. The prevalences of any pallor were 14%, 41% and 86%. The prevalences of severe anaemia (PCV<21%) were 1%, 5% and 81% and of any anaemia (PCV<33%) 68%, 73% and 98%. Severe pallor could not detect severe anaemia. The sensitivities were only 0%, 0% and 8%. The sensitivities of any pallor to detect severe anaemia were however 86% and 98% for children at the health care facility level, but still of relatively poor predictive values since the specificities were only 61% and 68%.
Interpretation: Division of pallor into some or severe degrees was of no use at any health care level. The identification of any pallor was of no use at village level, but it may possibly be of some value as a screening test for severe anaemia at health care facilities, if additional assessment is included in view of the low specificity and positive predictive value of the finding.