Abstract
Purpose :
Glaucoma is a leading cause of vision-threatening disease worldwide and its open-angle variant has previously been explored as a risk factor for stroke. Using a large sample population database, we investigated the possible impact of open-angle glaucoma (OAG) both with and without numerous other clinical and non-clinical factors on the risk of developing stroke.
Methods :
Cases of OAG were obtained from the National Inpatient Sample (NIS) database between 2002 and 2013 using ICD-9 codes. Associated morbidities and procedures were assessed in a random sample of cases with a primary hospital admitting diagnosis of stroke, no prior history of stroke, and a history of OAG. Univariate and multivariate logistic regression analyses were carried out in glaucoma cases to determine risk factors for stroke. A second analysis was conducted after propensity matching (1:5 ratio) for other risk factors to determine if glaucoma was independently associated with increased risk of stroke. The Bonferroni correction method was applied.
Results :
The random sample consisted of 395,724 cases which were grouped into stroke (n=4,614) and non-stroke (n=391,110) cohorts. History of OAG (n=4,187) was found to be strongly associated with stroke in univariate analysis (OR: 2.26, 95% CI: 2.07-2.46) but this association was found to be weaker (OR: 1.11, 95% CI: 1.04-1.18) in multivariate analysis (Figure 1). When propensity matching (Figure 2) was performed for clinical and non-clinical risk factors found to be significant for stroke (including age, sex, race, insurance status, hypertension, diabetes, alcohol use, congestive heart failure, and valve disease), glaucoma was no longer significantly associated with stroke (p=0.06).
Conclusions :
A large patient population was utilized to determine whether OAG is independently associated with stroke. Our findings suggest that prior noted association with stroke is likely based on similar risk factors rather than an independent association. Study limitations include lack of evidence for causal relationships in a retrospective study and limitations in coding of historical diagnoses in patient records.
This is a 2021 ARVO Annual Meeting abstract.