Relating growth rate to environmental factors--methodological problems in the study of growth-infection interaction.

The effect of infection on the growth of children under 2 was studied in a rural (Keneba, n = 176) and an urban (Bakau, n = 122) community in The Gambia, and in a Papua New Guinean village (PNG, n = 24). Weights taken every 4 to 6 weeks (n = 3929) were converted to weight velocities, and these were related to the intensity of infection (coded in ten categories) during each growth period using within-subject multiple regression analysis. Diarrhoea (in Keneba), lower respiratory tract illness (Bakau) and measles (PNG) all had highly significant effects on growth, reducing weight gain by between 17 and 69 g/day. Two statistical problems, i.e. the varying lengths of time between weight measurements, and the negative correlation between successive weight velocities, were discussed. However, redoing the analysis taking them into account made no practical difference to the conclusions.