Sir.—I recently read the article by van de Bor et al1 published in the November 1986 issue of AJDC. They presented the possible relationship between a multitude of perinatal variables and periventricular-intraventricular hemorrhage (PIVH) in 49 infants.
There are four issues that I would like to address. First, they do not indicate whether all of the newborns included in the study were born at their institution or were transported from other centers. Some investigators have shown an increased incidence of PIVH in the transported group.2
Second, it is true that the data were collected prospectively. The problem is that the hypothesis (not clearly stated in the "Patients and Methods" section) was probably chosen after the analysis of the data. Therefore, the results accepted as significant after "fishing through the data" are not valid. The hypothesis needs to be set prior to the actual collection and analysis. The