To the Editor.—In the article by Haque et al1 in the December 1988 issue of AJDC, the authors concluded that mortality from neonatal sepsis (suspected or proved) was significantly reduced in a group of 30 infants treated in a prospective, randomized trial with IgM-enriched intravenous immunoglobulin (Pentaglobin, Biotest Pharma, Frankfurt, West Germany) in addition to antibiotics compared with a control group of 30 infants who were treated with antibiotics alone.
Their conclusion of a statistically significant (P<.001) reduction in mortality between the treated and control groups was based on the use of Student's t test for statistical analysis. Is there some rationale that makes use of the Student t test the appropriate test for analysis of these outcome data?
As outcome is a discrete and not a continuous or numerical parameter, a t test, which is a parametric test of the difference in means between two samples