It is with great interest that we read the recent article by Chou and associates comparing intraocular pressure (IOP) measurements after penetrating keratoplasty (PK) obtained by 4 different instruments. There are some points which remain open and need to be addressed.
The clinical usefulness of any device depends strongly on the precision and accuracy of its measurements. If one method has poor repeatability and reproducibility, the agreement between the two methods is bound to be poor too. Assessing method precision is of particularly importance. Previous studies have demonstrated that these 4 devices provide reliable measurements for IOP in normal eyes. However, this conclusion cannot be applied to the irregular cornea, such as that found after PK with significant astigmatism and edema. Therefore, first and foremost, the authors should evaluate and show the within-rater repeatability and between-rater reproducibility of IOP measurement.
Furthermore, the authors suggested that the Tono-Pen (Reichert Inc., Depew, New York, USA) or Pascal dynamic contour tonometer (Swiss Microtechnology AG, Port, Switzerland) were the most suitable alternatives for measuring IOP in PK eyes in which Goldmann applanation tonometer (GAT; Haag-Streit USA, Mason, Ohio, USA) readings were difficult to obtain. However, the 95% limits of agreement were larger, ranging from −8.41 to 7.27 mm Hg and −3.88 to 8.12 mm Hg for Tono-Pen versus GAT and Pascal dynamic contour tonometer versus GAT, respectively, based on Bland-Altman plots, which suggested that they were in poor agreement. Clearly, this range is not acceptable for these 2 instruments to be used interchangeably with GAT in eyes that have undergone PK. We suggest using the same device to measure, screen, and manage follow-up IOP in PK eyes to avoid the significant interdevice variation.
Finally, although the study size is comparable with that of other similar studies in the literature, the sample size calculation should be performed to provide the level of significance (α) and a power (1− β). A large α value and insufficient power will limit the general application of this research.