Editor,

I read with interest the article by Chong et al, in which they studied the variations in blood pressure, heart rate, shin and abdominal wall skin temperature in 44 term infants during sleep, in 4 different positions, prone and supine, horizontal and with a 600 head up tilt respectively. They concluded that the variations observed in the 4 different positions, due to vasomotor tone, may have a role in sudden infant death syndrome.(1) However, I feel that the statistical analysis used is inappropriate and may not necessarily support their findings.

The authors used the Student t test to compare, in each of the 4 body positions, the mean values of each measured parameter in the 44 infants. If the measured values do not follow a Normal distribution (this is not mentioned in the study), the use of the t test would be invalid.

Furthermore, the authors measured the unpaired t test when they compared the mean values in all infants, in each body position respectively. However, the samples were paired (not independent), with 4 body positions for each infant. The parameters of each infant should therefore be compared in the 4 different body positions (each infant being his own control), instead of comparing the mean values of all infants in each body position. The unpaired t test they used cannot distinguish variations for each subject from variations due to differences between groups.

It is however still possible that the reported results are correct, but to prove it, the authors need to use either the paired t test for comparing means, if the measured values follow a Normal distribution, or the paired Wilcoxon test if they do not.

References

1. Chong A, Murphy N, Matthews T. Effect of prone sleeping on circulatory control in infants. Arch Dis Child 2000:82;253-256.

Hassib Narchi
Paediatric Department
Sandwell General Hospital
Lyndon, West Bromwich
West Midlands B 71 4HJ
E-mail: hassibnarchi@hotmail.com