Abstract
Purpose :
The development of refractive error is driven by genetic and environmental factors. These factors present limited interventional scope, because of their unchangeability (genes) or intervening with known environmental factors would be unacceptable (reducing education). The aim of this work was to investigate associations between refractive error and modifiable risk factors such as medication use.
Methods :
A subsample of 117,279 individuals from the UK Biobank cohort underwent an extensive ophthalmological examination that included measurement of refractive error (RE) represented by mean spherical equivalent (SPHE). The information on prescription and over-the-counter medicines was collected using a touchscreen questionnaire and face-to-face interviews. The statistical analysis was restricted to individuals of European descent with complete data on both RE status and drug intake. The associations between SPHE and medications were tested in mixed linear models (LMM) adjusted for age, sex, educational attainment and Townsend deprivation index. The medications that showed statistically significant associations with RE in LMMs were subjected to sensitivity analysis in order to control for confounding by disease diagnosis.
Results :
The study identified significantly strong associations between RE and most nonsteroidal anti-inflammatory drugs (NSAIDs) as well as other medications commonly prescribed for pain management. In particular significant differences in SPHE between non-medicated participants and subjects receiving paracetamol (β=0.28, SE=0.04, p<0.0001), diclofenac (β=0.047, SE=0.13, p=0.0002), ibuprofen (β=0.23, SE=0.05, p <0.0001). Even stronger effects were observed for other classes of pain controlling medications (tramadol: β=1.073, SE=0.17, p<0.0001; co-codamol: β=0.81, SE=0.11, p<0.0001 ). A sensitivity analysis revealed that pain medications were associated with higher SPHE values even among subjects without significant medical history.
Conclusions :
This is the first report of sustained effects of NSAIDs and other pain control medications over refractive error in the general population. Further work is required to explore these findings, given the associations remained after we attempted to control for education, socioeconomic factors and age and for painful and other medical conditions. As these data are cross-sectional, we cannot exclude additional confounding, so replication is required.
This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.