Letters Section Editor: Robert M. Golub,
MD, Senior Editor.
To the Editor: We are concerned that by performing
a large number of statistical tests on the same data set, Dr Hujoel and colleagues1 underestimated the type I error rate. Three infant
outcomes (preterm, LBW, and term LBW) were evaluated for each of 17 dental
procedure characteristics, resulting in 51 possible effects. When we applied
the Bonferroni correction for the evaluation of 51 repeated measurements we
determined that a P value of less than .00009 would
be required to achieve statistical significance.
Henderson CE, Ahmed H. Antepartum Dental Radiography and Low Birth Weight. JAMA. 2004;292(9):1019. doi:10.1001/jama.292.9.1019-c