Erratum to “An Analysis of Risk Factors Affecting Cerebrovascular Disease” [Health, 14 (2022) 866-882] ()
1. Background
The risk factors for cerebrovascular disease were analyzed using data from 2,868,568 medical checkups obtained from the JMDC Claims Database.
2. Results
Age and heart disease history were very important nonmodifiable factors. The OR comparing persons aged 70 to those aged 50 was 2.33 with a 95% CI of 2.18 - 2.48. A heart disease history was also an especially important factor (OR 2.12, 95% CI 1.91 - 2.33). Among the modifiable factors, recent large weight change was a very important factor, changing the risk of cerebrovascular disease by about 25%. Other significant modifiable factors were diastolic blood pressure, urine protein and sleeping well; each of these changed the risk of cerebrovascular disease by about 10%. Taking medications to control hypertension, hyperglycemia and hypercholesterolemia respectively increased the risk of cerebrovascular disease. In particular, taking antihypertensive medications nearly doubled the risk (OR 1.73, 95% CI 1.68 - 1.85).
3. Data and Models
Data
Excluding the observations with missing values of covariates, 2,868,568 observations satisfy these criteria, and 6553 observations or 0.23% experienced cerebrovascular disease at year t + 1 among the total observations.
4. Results of Estimation
The results of the estimation are given in Table 2, and we obtained similar results in both models. Among the nonmodifiable variables, the estimates of Age and Heart_D were positive. The t-values were quite large, and they were significant at any reasonable significance level. The estimates of other nonmodifiable variables were not significant at the 5% level. For modifiable variables, the estimates of other nonmodifiable variables were not significant at the 5% level. For modifiable variables, the estimates of DBP, GGP, U_Protein, and Weight_1 were positive and significant at the 1% level in both models. The estimates of SBP, HbA1c and Weight_20 were positive and significant at the 5% level in Model B. On the other hand, the estimates of HDL, ALT, Speed and Sleep were negative and significant at the 1% level in both models. The estimates of LDL were negative and significant at the 5% and 1% level in Model A and Model B, respectively. The estimate of ALT was and significant at the 5% level in Model A.
For three dummy variables represent taking medications, the estimates of M_Antihypertensive and M_Cholestrol were positive, and their t-values were quite large and significant at any reasonable significance level in Model A. The estimate of M_Glucose was positive and significant at the 5% level. Especially, the estimate and t-value of M_Antihypertensive were 0.551 and 16.69, respectively.
SD: Standard Deviation.
SE: Standard Error.
5. Discussion
For an individual without a history of cerebrovascular disease, the overall probability of developing cerebrovascular disease within one year was very small, with a gross rate of just 0.23%. Therefore, the odds ratio (OR) and confidence interval (CI) are approximately equal to the probability ratio (PR) and its CI as shown in Appendix A. As shown in Figure 1, the OR comparing persons aged 60 to those aged 50 is 1.53 with a 95% CI of 1.48 - 1.57, and the OR comparing persons aged 70 to those aged 50 is the 2.33 with a 95% CI of 2.18 - 2.48. The risk of persons aged 70 is almost twice as large as that of those aged 50. The OR for a heart disease history (comparing those with and without a heart disease history) is 2.12 with a CI of 95% 1.91 - 2.33. This means that individuals with a history of heart disease will have cerebrovascular disease at rates more than double those without a heart disease history. It is necessary for these individuals to pay special attention to prevention of cerebrovascular disease. Since cerebrovascular disease, especially stroke, is a medical emergency, it is also important to ensure that medical personnel know these facts for proper treatments.
Figure 2 shows the ORs and 95% CIs of the modifiable variables whose estimates are significant at the 5% level in Model A, except for dummy variables that represent taking medications. For these variables, not only estimates but also their distributions are also important. Let z be a variable of interest. When z is a numerical variable, the OR is calculated by comparing z and (z + one standard deviation). When z is a dummy variable, the OR is calculated comparing z = 0 and 1. For U_Protein, the majority of the values are 1 or 2, so that the OR is calculated comparing U_Protein = 1 and 2. Among these variables, HDL level (HDL) and recent large weight change (3 kg or more within a year, Weight_1) are important factors. The OR of Triglyceride is 0.93 with a 95% CI of 0.92 - 0.96. The fact that a higher HDL level reduces the risk of cerebrovascular disease seems inconsistent with the expected result. The OR of HDL is 0.93 with a 95% CI of 0.92 - 0.96. The fact that a higher HDL level reduces the risk of cerebrovascular disease seems inconsistent with the expected result. For example, the CDC [58] advises on its website to “limit foods high in saturated fat. Saturated fats come from animal products...” However, Sauvaget et al. [59] reported that higher consumption of animal fat and cholesterol appeared to reduce the risk of deaths from cerebral infarction in Japan. Their finding is consistent with the result of this study. The OR of Weight_1 is 1.25 with a 95% CI of 1.21 - 1.28. This means that recent large (≥3-kg) weight change would increase the risk of the cerebrovascular disease by 25% or more, and individuals in this category should recognize this fact. The ORs (95% CIs) of other important variables are 1.10 (1.05 - 1.14) for DBP, 1.15 (1.10 - 1.19) for U_Protein, 0.92 (0.87 - 0.97) for Speed, and 0.89 (0.84 - 0.93) for SLEEP. The risks of cerebrovascular disease change about 10% by these variables. The ORs (95% CIs) for LDL, ALT and GGP are 0.97 (0.95 - 1.00), 0.95 (0.91 - 1.00), and 1.04 (1.02 - 1.07), respectively. The effects of these variables on the risks of cerebrovascular disease are relatively small.
For dummy variables that represent taking medications, the ORs (95% CIs) are 1.73 (1.68 - 1.79) for M_Antihypertensive, 1.13 (1.06 - 1.20) for M_Glucose and 1.33 (1.28 - 1.38) for M_Cholesterol.
6. Conclusion
In this study, we analyzed the risk factors for cerebrovascular disease using data from 2,868,568 medical checkups obtained from the JMDC Claims Database. Among the nonmodifiable factors, age and a history of heart disease are important risk factors. The risk of persons aged 70 is twice as large as that of those aged 50. A heart disease history is also an especially important factor; the risk more than doubles.
Among the modifiable factors, recent large weight change is a very important factor, and it increases the risk of cerebrovascular disease about 25%. DBP, U_Protein, and Sleep are other important modifiable factors, which change the risk of cerebrovascular disease by about 10%.
![]()
Figure 1. Odds ratios and 95% confidence intervals of nonmodifiable variables.
![]()
Figure 2. Odds ratios and 95% confidence intervals of modifiable variables.
![]()
Figure 3. Odds ratios and 95% confidence intervals of dummy variables taking medications.