Many of the dropouts from our study were due to the inherent high migration rate among the families. As stated in the paper, an attempt was made to trace these families by writing to all 50 state health departments who answered but did not report deaths among study infants. Admittedly the statistical appraisal of our figures is compromised by the dropouts. However, the fact that there were no additional deaths reported to us would confirm the trend in favor of the GG group.
The disproportionate loss of babies in the two groups, as explained in the paper, was accounted for by painful injections. This reaction was undoubtedly greater in the infants receiving the placebo as determined once the code was broken.