To The Editor: We would like to thank Dr. Vavilala and colleague for their interest in our manuscript. Their letter raises a number of points primarily concerning the methodology of our study (1).

The first question they raise is how CO2 reactivity was examined. The multiple-regression model used paired values of blood pressure and PaCO2 measured contemporaneously with CBF measurements (1). In this way, we hoped to determine the individual effects of these predictor variables on our outcome variable (CBF).

The question of Type II error would be relevant if statistical comparisons were made between CBF reactivity of the two groups (which we do not do). In the main, we only compared biographical and physiological data between the two groups (Table 2 and 3), and it is in these cases only that one could not exclude Type II errors.

In the final paragraph of the Methods we define CBF-reactivity and in the Discussion we speculate with others as to whether autoregulation remains intact. As stated in our paper (1), this is based on whether the confidence limits encompass zero in the case of MAP-CBF reactivity, implying no change in CBF (2).

Dr. Vavilala is correct in that our study was of a small sample size. This reflects the difficulties of conducting clinical research in a population of the sickest preterm infants; inevitably, this leads to larger confidence limits.

Perhaps the most pertinent question raised by Dr. Vavilala is that of the lower limit of autoregulation in preterm infants. There exists a multiplicity of normograms used to define the normal range of blood pressure of preterm infants. It follows that there is no universally agreed definition of hypotension, and hence no definition of the lower limits of autoregulation. In essence, our paper tested one definition (3) and found impaired CBF reactivity in infants considered to be hypotensive by this definition. Other authors have tested other definitions (4).

The final point made by Dr. Vavilala is well taken as the first sentence of the results is somewhat misleading, and should have stated that the study was conducted over a 23-month period. However, throughout the rest of the paper it is made quite clear that we studied a population of preterm infants during the first week of life (Table 2).