Abstract
Purpose:
We explored whether the risk factor profile for POAG differed when the disease was divided into cases with either incident paracentral or incident peripheral visual field (VF) loss.
Methods:
We used competing risks analysis to identify unique risk factors for POAG with paracentral VF loss or POAG with peripheral VF loss. We used updated prospective data from 79,191 women in the Nurses Health Study (NHS) and 43,311 men in the Health Professionals Follow-up Study (HPFS). We identified 974 incident POAG cases in the NHS (1980-2010) and 425 incident POAG cases in the HPFS (1986-2010). We confirmed that cases had reproducible VF loss on reliable tests and defined VF subtype based on the earliest abnormal VF. We considered the following risk factors in competing risk analyses: gender, African-heritage, family history of glaucoma, body mass index (BMI), hypertension, diabetes mellitus, physical activity, smoking, caffeine intake, and alcohol intake. We used the forward selection approach, where we sequentially identified the factor that showed the most significant differences in associations with POAG subtypes (iteration threshold for testing was p<0.1) Backward selection approach was also used for sensitivity analysis.
Results:
Risk factors with similar significant associations with the POAG subtypes were family history of glaucoma, African-heritage and greater caffeine consumption. An adverse risk factor for POAG with peripheral VF loss but not paracentral VF loss was diabetes (HR=1.35 [95% CI: 1.05-1.72] for peripheral vs. HR=0.86 [95% CI: 0.58-1.28] for paracentral). These HRs for diabetes were significantly different from one another (p=0.02). Risk factors that were more strongly inversely associated with POAG with paracentral VF loss were 10-kg/m2 increments of BMI (HR=0.79 [95% CI: 0.64-0.98] for paracentral vs. HR=0.99 [95% CI: 0.85-1.15] for peripheral) and 10 pack-year increases in smoking (HR=0.89 [95% CI: 0.85-0.95] for paracentral vs. HR=0.97[95% CI: 0.94-1.01] for peripheral). The relation between smoking (p=0.01), but not BMI (p=0.09), and the POAG subtypes were significantly different from one another. Results were similar with a backward selection approach.
Conclusions:
These data show unique risk factor profiles for POAG subtypes indicating that POAG is a heterogeneous condition.
Keywords: 459 clinical (human) or epidemiologic studies: biostatistics/epidemiology methodology •
464 clinical (human) or epidemiologic studies: risk factor assessment •
413 aging