In Reply.—Dr Alexander suggests our statistical analysis was inappropriate and our conclusions erroneous. He suggests several statistical alternatives as follows: (1) Wilcoxon signed rank test: we felt this was inappropriate since our data were not paired. (2) Intuitive logic: we felt a 0.6 mg/dL difference in peak serum bilirubin level between suppository- and nonsuppository-treated groups where values ranged from 0.1 to 16 mg/dL would logically not be clinically or statistically significant. (3) Analysis of variance: we reevaluated our data using a two-way analysis of variance with multiple replications and found no significant difference between serum bilirubin values in suppository- v nonsuppository-treated patients (MS= 62.2, F = 2.22, and P =.14). All patients had measurements made at the same postnatal ages, so we felt covariate analysis would not be necessary.
Dr Alexander then refers to the work of De Carvalho et al1 that clearly demonstrates that breast-fed babies who were fed more