Search

Purpose Epidemiological studies have reported ocular perfusion pressure (OPP) as a risk factor for open‐angle glaucoma (OAG). OPP is essentially the difference between blood pressure and intraocular pressure (IOP). IOP is thus part of OPP. As IOP itself is a strong OAG risk factor, the crucial question is whether the linear adjustment for IOP as commonly used in statistical models is sufficient to remove all confounding by IOP ‐ a prerequisite to consider OPP an independent risk factor if significant after adjustment for IOP. This question was addressed in our study. Methods We performed Cox regression on longitudinal data from the Rotterdam Study comprising 103 incident OAG cases and 3779 controls. All models were adjusted for age, gender and IOP. First, we entered diastolic OPP (DOPP) in the model as tertiles. Next, we entered systolic OPP (SOPP). Subsequently, we replaced the blood pressure values in DOPP and SOPP by normally distributed random numbers. Based on the diastolic and systolic blood pressure distributions in our data, we used standard deviation (SD) values of 10 and 20 mmHg. For both SD values, the analyses were repeated 10 times. Results The hazard ratio (HR) for the lowest tertile of DOPP was 1.11 (95% confidence interval 0.68‐1.83) and of SOPP 1.21 (0.74‐1.98). The median (range) HR of the lowest tertile of the resampled DOPPs (diastolic blood pressure replaced by noise with a SD of 10 mmHg) was 1.27 (1.03‐1.97) and of the resampled SOPPs (systolic blood pressure replaced by noise with a SD of 20 mmHg) 0.99 (0.72‐1.35). Conclusion Residual confounding by IOP can reveal a significant spurious association between a low DOPP and incident OAG with HRs similar to those reported in the literature. [ABSTRACT FROM AUTHOR]