Date Published: October 15, 2018
Publisher: Public Library of Science
Author(s): Sang Jun Park, Hee Kyung Yang, Seong Jun Byun, Kyu Hyung Park, Jeong-Min Hwang, Simona Sacco.
To determine whether ocular motor cranial nerve (CN) palsy raises the risk of subsequent stroke in the general population.
We investigated the association between ocular motor CN palsy and occurrence of stroke using the National Health Insurance Service-National Sample Cohort database from 2002 to 2013. We included individuals aged ≥ 20 years on January 1st, 2004, and excluded those having any paralytic strabismus, disorders in binocular movement, diplopia and any cerebrovascular diseases before entering the cohort. Incident ocular motor CN palsy was identified by diagnostic codes for third, fourth, and sixth nerve palsies. To determine the effect of incident ocular motor CN palsy on the occurrence of subsequent stroke, we used time-varying covariate Cox regression models. Model 1 included only incident third, fourth, and sixth nerve palsies as a time-varying covariate. Model 2 included Model 1 and defined demographic information. Model 3 included Model 2, comorbidity, co-medication, and the Charlson index score.
Among 727,689 individuals in the cohort, 1,633 patients developed ocular motor CN palsy and 17,657 patients suffered stroke. Cox regression models showed that development of ocular motor CN palsy was associated with an increased risk of subsequent stroke (hazard ratio [HR] = 4.65; 95% confidence intervals [CIs], 3.74–5.80 in Model 1), and the results were consistent even after adjusting for demographic factors and confounders in Model 2 and 3. Men, older age, and individuals not living in Seoul/Incheon area were associated with an increased risk of stroke, while individuals with higher income were associated with decreased risk of stroke in both Model 2 and 3. Sensitivity analyses using propensity score-based matching produced similar results in all three Models (HR = 1.95; 95% CI, 1.55–2.46 in Model 1, HR = 1.91; 95% CI, 1.52–2.41 in Model 2, and HR = 1.63; 95% CI, 1.29–2.06 in Model 3).
The occurrence of ocular motor CN palsy is a significant risk factor of subsequent stroke even after adjusting for demographic factors and confounders in the general population. Physicians may need to educate patients with ocular motor CN palsy regarding the higher risk of future stroke.
Stroke is one of the global leading causes of disability, and its prevalence has increased dramatically. Risk factor modification and identification of predictive factors are essential for prevention of stroke.[1–3]
Finally, 727,689 individuals (370,269 women, 50.9%) in the cohort were examined and 7,008,001 person-years were included in the analysis. Person-years were derived from the sum of years of observation per person during follow-up until death or end of follow-up. Fig 1 shows the flow chart along with the eligible criteria. Among them, 1,633 patients developed ocular motor cranial nerve palsy and 17,657 patients suffered stroke. Table 1 summaries the characteristics of individuals in the cohort. Table 2 shows an increased risk of stroke after development of third, fourth, and sixth nerve palsies (hazard ratio [HR] = 4.65; 95% confidence intervals [CIs], 3.74–5.80 in Model 1) even after adjusting for demographics (HR = 2.19; 95% CI, 1.76–2.73 in Model 2) and confounders (HR = 1.88; 95% CI, 1.51–2.34 in Model 3) by time-varying cox regression models.
This study showed that development of ocular motor CN palsy was associated with an increased risk of subsequent stroke, even after adjusting for demographic factors and confounders. Rim et al reported a higher risk for stroke development in patients with ocular motor CN palsy, and the risk to stroke reduced with time only after third and fourth nerve palsies, but not with sixth nerve palsy. They used the same database (NHIS-NSC database) by comparing the risk of stroke between patients with incident CN palsies and their matched controls. However, in the previous study, patients were included for analysis at the time when ocular motor CN palsy occurred. This approach eliminates the period before ocular motor CN palsy in each patient, which is susceptible to selection bias and length time bias. In a longitudinal study using Cox proportional hazard models, development of intermediate events (e.g. ocular motor CN palsy), which can occur at any time of the study period, could substantially affect outcome assessment (e.g. stroke development). To reduce such bias in the present study, we treated the intermediate event–development of ocular motor CN palsy–as a time-varying covariate. In addition, the baseline characteristics were quite different between ocular motor CN palsy and controls in the study by Rim et al; hypertension and diabetes mellitus were more common in ocular motor CN palsy, which decreased the comparability of outcomes. In order to avoid these problems, we carefully designed the three Cox proportional hazards models to adjust for possible risk factors of stroke. We also performed sensitivity analyses by the use of propensity score matching. Therefore, our method can be a more reasonable way to find any change in the incidence of stroke according to the development of ocular motor CN palsy.