Cardiovascular Disease Risk Assessment
Exploratory Data Analysis and Statistical Insights Using R
Bienuel M. Esmeralda
2026-05-24
This project was conducted as an independent analytical exercise intended to provide statistical insights and practical applications of data analytics in cardiovascular disease risk assessment. While this work is not presented as an official research study, it aims to demonstrate the use of exploratory data analysis, correlation analysis, and regression modeling in examining factors associated with cardiovascular disease risk. The dataset utilized in this project was obtained from Mendeley Data and is based on the “CAIR-CVD-2025: An Extensive Cardiovascular Disease Risk Assessment Dataset from Bangladesh” developed by Md Asraful Sharker Nirob and colleagues. The dataset reflects real-world cardiovascular health information and was used solely for educational, analytical, and portfolio development purposes. This project highlights the application of statistical techniques in extracting meaningful insights from clinical and lifestyle-related variables associated with cardiovascular disease risk.
Dataset Summary This dataset comprises 1,529 patient samples collected from Jamalpur Medical College Hospital, Jamalpur, Bangladesh, from January 20, 2024, to January 1, 2025. The data were gathered following ethical guidelines, ensuring patient confidentiality and informed consent. The dataset provides a comprehensive collection of demographic, anthropometric, clinical, biochemical, and lifestyle parameters essential for assessing cardiovascular disease (CVD) risk and overall patient health. The dataset includes a wide range of variables critical for CVD risk estimation, including basic demographic information, anthropometric measurements, clinical values, biochemical markers, and lifestyle factors. These variables are crucial for identifying risk factors, understanding disease progression, and developing preventive health strategies.
This project aims to:
- Explore the demographic and clinical profile of the respondents.
- Identify patterns in cardiovascular disease risk levels.
- Examine the relationship between selected health indicators and CVD Risk Score.
- Demonstrate the use of RStudio for statistical analysis, visualization, and reporting.
1 Exploratory Data Analysis
1.1 Total number of missing values
[1] 1022Preliminary data inspection revealed the presence of missing values within the dataset. A total of 1,022 missing values were identified across several variables. We will check how many missing values per variable.
Sex Age
0 78
Weight (kg) Height (m)
81 67
BMI Abdominal Circumference (cm)
64 67
Blood Pressure (mmHg) Total Cholesterol (mg/dL)
0 73
HDL (mg/dL) Fasting Blood Sugar (mg/dL)
80 67
Smoking Status Diabetes Status
0 0
Physical Activity Level Family History of CVD
0 0
CVD Risk Level Height (cm)
0 74
Waist-to-Height Ratio Systolic BP
79 71
Diastolic BP Mean Arterial Pressure (MAP)
82 0
Blood Pressure Category Estimated LDL (mg/dL)
0 69
CVD Risk Score
70Variables with the highest number of missing observations include Diastolic Blood Pressure, Weight, HDL, Waist-to-Height Ratio, and Age. In contrast, categorical variables such as Sex, Smoking Status, Diabetes Status, Physical Activity Level, Family History of CVD, Blood Pressure Category, and CVD Risk Level contained no missing values. Since the missing observations were distributed across multiple continuous clinical variables, data cleaning procedures were conducted prior to further statistical analyses to ensure data consistency and reliability.
[1] 762To address the issue of missing data, rows containing incomplete observations were removed using listwise deletion (na.omit). The cleaned dataset was then used for subsequent exploratory and predictive analyses. The total number of respondents remaining for further analysis were 762 respondents.
1.2 Sex
Figure above represents the sex distribution of the respondents. The results revealed that female respondents comprised a slightly larger proportion of the sample, with 394 participants, compared to 368 male respondents.
1.3 Smoking Status by Sex
Figure above represents the sex distribution of the respondents that smokes. This also provides how many respondents were not smoking and smoking. The results revealed that female respondents comprised a slightly larger proportion of the sample that smokes, with 205 participants, compared to 191 male respondents. Overall, the distribution also revealed that respondents who smokes comprised a slightly larger proportion with 396 respodents, compare to 366 respondents that do not smoke.
1.4 Diabetes Status by Sex
The figure above represents the sex distribution of respondents with diabetes. It also presents the number of respondents classified as diabetic. The results revealed that female respondents comprised a slightly larger proportion of the diabetic sample, with 202 participants, compared to 193 male respondents. Overall, the distribution further revealed that respondents with diabetes comprised a slightly larger proportion of the dataset, with 395 respondents, compared to 367 respondents without diabetes.
1.4.1 Age grouped
Figure above presents the mean cardiovascular disease (CVD) risk score across different age groups, with the labels indicating the mean age of respondents within each category. The results revealed that the Baby Boomer group (61-79) had the highest mean age of 70 years and also exhibited the highest mean CVD risk score among the groups. In contrast, the Gen-Z group (25-28) had the lowest mean age of 27 years and comparatively lower CVD risk levels. Millenial group (29-44) had a mean of 37 years, while Gen-X group (45-60) had a mean of 52 years.
1.4.2 Age grouped that has diabetes
The figure above presents the distribution of respondents with and without diabetes across different age groups. The results revealed that the Millennial age group comprised the largest number of respondents with diabetes, with 179 individuals, followed by the Gen-X group with 164 respondents. In contrast, the Gen-Z group had the fewest diabetic respondents, with only 16 individuals. For respondents without diabetes, Millennials also represented the largest proportion, with 163 respondents, followed by Gen-X with 143 respondents and Baby Boomers with 56 respondents. Overall, the findings suggest that diabetes prevalence appears more common among middle-aged and older age groups compared to younger respondents.
1.4.3 Age grouped categorized by Blood pressure
The stacked bar chart presents the distribution of Blood Pressure Categories across different age groups. The results revealed that Millennial and Gen-X respondents comprised the largest proportion of individuals classified under Hypertension Stage 1 and Hypertension Stage 2. In contrast, Gen-Z respondents exhibited comparatively fewer cases of hypertension-related categories. The findings suggest that elevated blood pressure conditions appear more prevalent among middle-aged and older respondents compared to younger age groups.
1.5 CVD Risk Level
The figure above presents the distribution of CVD risk levels among the respondents. The results revealed that a large proportion of the respondents were classified as having a high risk of cardiovascular disease, comprising 48.3% of the total population. This was followed by respondents with an intermediary risk of cardiovascular disease, comprising 37.5% of the total population. Lastly, only a small proportion of the respondents were classified as having a low risk of cardiovascular disease.
1.5.1 Fasting Blood Sugar Categories
The classification of Fasting Blood Sugar levels into Normal, Prediabetes, and Diabetes categories was based on the diagnostic criteria provided by the Centers for Disease Control and Prevention [2]. According to the CDC, fasting blood sugar levels below 100 mg/dL are considered normal, levels ranging from 100–125 mg/dL indicate prediabetes, and levels of 126 mg/dL or higher may indicate diabetes.The figure above presents the distribution of the Category of Fasting Blood Sugar of the respodents, having a 126 mg/dL after fasting. The figure also revealed that a signficantly large proportion of the respondents were classified as having Normal Fasting Blood Sugar, comprising 34.1% of the total population. This was followed by respondents with a Pre-diabetes, comprising 29.4 % of the total population.
1.6 Blood Pressure Category
The figure above presents the distribution of the Category of Blood Pressure of the respodents. The results revealed that a large proportion of the respondents were classified as having a Hypertension Stage 2, comprising 43.2% of the total population. This was followed by respondents with a Hypertension Stage 1, comprising 31.8% of the total population. 18.4% of the population were classified having an Elevated Blood Pressure. Lastly, only a small proportion of the respondents were classified as having a Normal Blood Pressure.
1.7 Weight in Kg and Height in Meters
| Parameter | Weight | Height |
|---|---|---|
| Min. | 50.10000 | 1.503000 |
| 1st Qu. | 67.32500 | 1.661250 |
| Median | 87.45000 | 1.760000 |
| Mean | 86.43526 | 1.754665 |
| 3rd Qu. | 105.95000 | 1.840750 |
| Max. | 119.80000 | 2.000000 |
The summary statistics revealed that respondents had a mean weight of 86.44 kg and a mean height of 1.75 meters. Weight values ranged from 50.10 kg to 119.80 kg, while height measurements ranged from 1.50 meters to 2.00 meters. The findings indicate variability in anthropometric characteristics among the respondents.
1.8 Body Mass Index
| Parameter | Value |
|---|---|
| Min. | 15.00000 |
| 1st Qu. | 22.60000 |
| Median | 28.33750 |
| Mean | 28.48156 |
| 3rd Qu. | 33.96300 |
| Max. | 46.10000 |
The summary statistics revealed that respondents had a mean BMI of 28.48, with values ranging from 15.00 to 46.10. This suggests that, on average, respondents fall within the overweight category, which may increase the risk of cardiovascular disease and other related health conditions.
The figure above presents the distribution of Body Mass Index (BMI) among the respondents. The results revealed that most respondents had BMI values concentrated between 20 and 35, with the mean BMI located at approximately 28.48. This indicates that the respondents, on average, fall within the overweight category.
1.9 Blood Pressure (mmHg) through Mean Arterial Pressure (MAP)
While Blood Pressure (mmHg) was available in the dataset, the study utilized Mean Arterial Pressure (MAP) for further analyses. MAP was selected because it reflects the overall average arterial pressure throughout the cardiac cycle by incorporating both systolic and diastolic blood pressure values. This provides a more stable and clinically meaningful indicator of cardiovascular status for regression modeling.
\[ \text{MAP} = \frac{\text{Systolic Pressure} + 2(\text{Diastolic Pressure})}{3}\]| Parameter | Value |
|---|---|
| Min. | 70.66667 |
| 1st Qu. | 88.33333 |
| Median | 96.33333 |
| Mean | 97.40376 |
| 3rd Qu. | 105.33333 |
| Max. | 133.33333 |
The summary statistics revealed that respondents had a mean arterial pressure (MAP) of 97.40 mmHg, with values ranging from 70.67 mmHg to 133.33 mmHg. The findings suggest variability in blood pressure levels among the respondents, indicating differences in cardiovascular health conditions within the population.
The figure above presents the distribution of Mean Arterial Pressure (MAP) among the respondents. The results revealed that most respondents had MAP values concentrated between 90 mmHg and 110 mmHg, with the mean MAP located at approximately 97.40 mmHg. The distribution suggests variability in arterial pressure levels among the respondents, indicating differences in cardiovascular health status within the population.
1.10 Total Cholesterol
| Parameter | Total.Cholesterol |
|---|---|
| Min. | 100.000 |
| 1st Qu. | 150.250 |
| Median | 199.000 |
| Mean | 199.727 |
| 3rd Qu. | 252.000 |
| Max. | 300.000 |
Table above presents the summary statistics for Estimated LDL, HDL, and Total Cholesterol levels among the respondents. The results revealed that respondents had a mean LDL level of 113.48 mg/dL, a mean HDL level of 56.25 mg/dL, and a mean Total Cholesterol level of 199.73 mg/dL. The findings indicate variability in lipid profile measurements within the population, suggesting differences in cholesterol-related cardiovascular risk factors among the respondents.
The results revealed that most respondents had LDL values concentrated around 113.48 mg/dL, while HDL levels were generally centered around 56.25 mg/dL. In addition, Total Cholesterol levels were concentrated near the mean value of 199.73 mg/dL. The distributions indicate variability in lipid profile measurements among the respondents, suggesting differences in cardiovascular health and cholesterol-related risk factors within the population.
1.11 CVD Risk Score
| Parameter | Values |
|---|---|
| Min. | 10.53000 |
| 1st Qu. | 15.29000 |
| Median | 16.91500 |
| Mean | 17.03475 |
| 3rd Qu. | 18.82950 |
| Max. | 23.88000 |
Table above presents the summary statistics for the CVD Risk Score among the respondents. The results revealed that respondents had a mean CVD Risk Score of 17.03, with values ranging from 10.53 to 23.88. The findings suggest variability in cardiovascular disease risk levels within the population, indicating differences in the respondents’ overall cardiovascular health profiles.
The table above revealed distribution of CVD Risk Scores among the respondents. The results revealed that most respondents had CVD Risk Scores concentrated between 14 and 20, with the mean score located at approximately 17.03. The distribution suggests variability in cardiovascular disease risk among the respondents, indicating differences in overall cardiovascular health profiles within the population.
2 Correlation Analysis
To determine the association between two variables, the Pearson correlation coefficient will be utilized. Before applying the Pearson correlation coefficient, its assumptions must first be assessed. Pearson’s correlation is a parametric test used to identify the relationship between two continuous variables. If at least one of the assumptions is violated, a non-parametric alternative test should be employed instead.
Key Assumptions:
Linear relationship: The relationship between the two variables should be approximately linear, which can be checked with a scatterplot.
No extreme outliers: The data should not contain any extreme outliers that could disproportionately influence the results.
Normality: Both variables should be approximately normally distributed.
2.1 Determine how strongly BMI, Waist-to-Height Ratio, and Blood Pressure correlate with the CVD final risk score.
2.1.1 Checking Linear relationship between BMI, Waist-to-Height Ratio, Blood Pressure, and CVD Risk Score
Visual inspection of the scatterplots revealed generally positive linear relationships between Body Mass Index (BMI), Waist-to-Height Ratio, Mean Arterial Pressure (MAP), and CVD Risk Score. As the values of BMI, Waist-to-Height Ratio, and MAP increased, the CVD Risk Score also tended to increase. The scatterplots further indicated no major deviations from linearity, suggesting that the linearity assumption for Pearson correlation analysis was satisfied.
2.1.2 Checking No extreme outliers between BMI, Waist-to-Height Ratio, Mean Arterial Pressure, and CVD Risk Score
The data points appeared reasonably clustered without observations that could substantially distort the correlation analysis. Therefore, the assumption regarding the absence of extreme outliers was satisfied for Pearson correlation analysis.
2.1.3 Checking Normality of variables between BMI, Waist-to-Height Ratio, Mean Arterial Pressure, and CVD Risk Score
| Variable | Statistic | p.value | Normality |
|---|---|---|---|
| BMI | 0.9741169 | 0.0000000 | Not Normal |
| Waist-Height Ratio | 0.9799669 | 0.0000000 | Not Normal |
| Mean Arterial Pressure | 0.9837004 | 0.0000002 | Not Normal |
| CVD Risk Score | 0.9966865 | 0.1152097 | Normal |
The normality assumption was assessed through Shapiro Wilk Normality Test. The results revealed that the variables were not normally distributed. Therefore, the normality assumption required for Pearson r correlation coefficient analysis was violated. Hence, non-parametric Spearman rank correlation coefficient was utilized instead.
2.1.4 Analysis
| Variable | Statistic | p.value | Signficance | Strength |
|---|---|---|---|---|
| BMI | 0.6120038 | 0.0e+00 | Signficant | Moderate correlation |
| Waist-Height Ratio | 0.1432937 | 7.2e-05 | Signficant | Weak correlation |
| Mean Arterial Pressure | 0.3285645 | 0.0e+00 | Signficant | Weak correlation |
Correlation analysis revealed that Body Mass Index (BMI), Waist-to-Height Ratio, and Mean Arterial Pressure (MAP) were all significantly associated with CVD Risk Score. BMI showed a moderate positive correlation with CVD Risk Score (r = 0.612, p < 0.05), indicating that higher BMI values were associated with higher cardiovascular disease risk scores. Waist-to-Height Ratio showed a weak positive correlation (r = 0.143, p < 0.05), also Mean Arterial Pressure (MAP) exhibited a weak positive correlation with CVD Risk Score (r = 0.329, p < 0.001). These findings suggest that increases in BMI, Waist-to-Height Ratio, and MAP were associated with increased cardiovascular disease risk among the respondents.
3 Multiple Linear regresion
3.1 What health and lifestyle factors significantly predict Arterial Pressure?
3.1.1 Checking Assumptions of Multiple Linear Regression using performance package in Rstudio
Prior to conducting Multiple Linear Regression analysis, the assumptions of normality of residuals, homoscedasticity, absence of multicollinearity, and absence of extreme outliers were assessed using the performance package in RStudio. Diagnostic plots and statistical procedures were utilized to determine whether the assumptions required for Multiple Linear Regression analysis were satisfactorily met.
Warning: Non-normality of residuals detected (p < .001). Warning: Heteroscedasticity (non-constant error variance) detected (p = 0.027). OK: No outliers detected.
- Based on the following method and threshold: cook (0.9).
- For variable: (Whole model) # Check for Multicollinearity
Low Correlation
Term VIF VIF 95% CI adj. VIF Tolerance
Age 1.03 [1.00, 1.41] 1.01 0.97
BMI 1.03 [1.00, 1.40] 1.01 0.97
Waist-to-Height Ratio 1.04 [1.00, 1.29] 1.02 0.97
Fasting Blood Sugar (mg/dL) 1.02 [1.00, 2.38] 1.01 0.98
Total Cholesterol (mg/dL) 1.01 [1.00, 4.79] 1.01 0.99
Smoking Status 1.01 [1.00, 35.31] 1.00 0.99
Diabetes Status 1.02 [1.00, 1.81] 1.01 0.98
Physical Activity Level 1.02 [1.00, 1.59] 1.01 0.98
Sex 1.00 [1.00, 1.70e+07] 1.00 1.00
Tolerance 95% CI
[0.71, 1.00]
[0.71, 1.00]
[0.77, 1.00]
[0.42, 1.00]
[0.21, 1.00]
[0.03, 1.00]
[0.55, 1.00]
[0.63, 1.00]
[0.00, 1.00]The results revealed that the model satisfied the assumptions of absence of extreme outliers, and absence of multicollinearity. However, the residuals showed deviations from normality (p < 0.001), indicating that the normality assumption was violated. Lastly homoscedasticity were violated. To address these issues, a Box-Cox transformation was applied to improve the normality and stabilize the variance of the residuals.
BoxCox Transformation Plot
The Box-Cox analysis indicated that the optimal lambda value was approximately 0, suggesting that a logarithmic transformation was appropriate for the dependent variable. Thus, log-transformed Mean Arterial Pressure (MAP) was utilized in the subsequent regression analysis to address violations of normality and homoscedasticity assumptions.
OK: residuals appear as normally distributed (p = 0.095). OK: Error variance appears to be homoscedastic (p = 0.482). OK: No outliers detected.
- Based on the following method and threshold: cook (0.9).
- For variable: (Whole model) # Check for Multicollinearity
Low Correlation
Term VIF VIF 95% CI adj. VIF Tolerance
Age 1.03 [1.00, 1.41] 1.01 0.97
BMI 1.03 [1.00, 1.40] 1.01 0.97
Waist-to-Height Ratio 1.04 [1.00, 1.29] 1.02 0.97
Fasting Blood Sugar (mg/dL) 1.02 [1.00, 2.38] 1.01 0.98
Total Cholesterol (mg/dL) 1.01 [1.00, 4.79] 1.01 0.99
Smoking Status 1.01 [1.00, 35.31] 1.00 0.99
Diabetes Status 1.02 [1.00, 1.81] 1.01 0.98
Physical Activity Level 1.02 [1.00, 1.59] 1.01 0.98
Sex 1.00 [1.00, 1.70e+07] 1.00 1.00
Tolerance 95% CI
[0.71, 1.00]
[0.71, 1.00]
[0.77, 1.00]
[0.42, 1.00]
[0.21, 1.00]
[0.03, 1.00]
[0.55, 1.00]
[0.63, 1.00]
[0.00, 1.00]The results revealed that the model satisfied all the required assumptions. Therefore, Multiple Linear Regression analysis was deemed appropriate and was subsequently employed for further analysis.
3.1.2 Analysis
Call:
lm(formula = (map_bc) ~ Age + BMI + `Waist-to-Height Ratio` +
`Fasting Blood Sugar (mg/dL)` + `Total Cholesterol (mg/dL)` +
`Smoking Status` + `Diabetes Status` + `Physical Activity Level` +
Sex, data = cvdfin)
Residuals:
Min 1Q Median 3Q Max
-0.081880 -0.021704 0.000414 0.021803 0.084681
Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 2.422e+00 1.089e-02 222.408 < 2e-16 ***
Age 2.301e-04 9.495e-05 2.423 0.01561 *
BMI 1.709e-04 1.640e-04 1.042 0.29769
`Waist-to-Height Ratio` 1.343e-02 1.426e-02 0.942 0.34659
`Fasting Blood Sugar (mg/dL)` 1.490e-04 3.891e-05 3.828 0.00014 ***
`Total Cholesterol (mg/dL)` 4.093e-05 1.995e-05 2.052 0.04054 *
`Smoking Status`Y 3.194e-03 2.323e-03 1.375 0.16953
`Diabetes Status`Y 9.505e-04 2.335e-03 0.407 0.68408
`Physical Activity Level`Low 3.977e-03 2.836e-03 1.402 0.16128
`Physical Activity Level`Moderate 5.279e-04 2.863e-03 0.184 0.85376
SexM -1.752e-03 2.316e-03 -0.756 0.44968
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Residual standard error: 0.0319 on 751 degrees of freedom
Multiple R-squared: 0.04418, Adjusted R-squared: 0.03145
F-statistic: 3.471 on 10 and 751 DF, p-value: 0.0001768Multiple Linear Regression analysis was conducted to determine whether age, BMI, Fasting Blood Sugar, Total Cholesterol, Smoking Status, Diabetes Status, Physical Activity Level, and Sex significantly predict Mean Arterial Pressure (MAP). The overall regression model was statistically significant, F(10, 751)=3.471,p<0.05, indicating that the predictors collectively explained a significant proportion of the variability in MAP. The results revealed that Age, Fasting Blood Sugar, and Total Cholesterol were significant positive predictors of MAP. Specifically, increases in age, fasting blood sugar, and total cholesterol were associated with increases in Mean Arterial Pressure, holding other variables constant. The model explained approximately 4.42 % of the variance in Mean Arterial Pressure, suggesting that while the selected predictors contributed significantly to the model, additional physiological and lifestyle-related factors not included in the analysis may also influence blood pressure levels.
These findings greatly support Singh, J.N. et.al. [5] where Physiologic changes associated with aging lead to an increase in systolic blood pressure (SBP), mean arterial pressure (MAP), and pulse pressure (PP). They also stated that increased blood pressure seen with aging is most likely related to arterial changes, as aging results in the narrowing of the vessel lumen and stiffening of the vessel walls through a process known as atherosclerosis. Furthermore, the findings were consistent with the earlier descriptive analysis in the present study, which revealed that Millennial and Gen-X respondents comprised the largest proportion of individuals classified under Hypertension Stage 1 and Hypertension Stage 2 categories.
The result showed Fasting Blood Sugar significantly predicts Mean Arterial Pressure (MAP). The findings are also supported by the study conducted by Utama, J.E.P. et al [6], which reported a significant positive relationship between blood glucose levels and Mean Arterial Pressure (MAP) among patients with Type II Diabetes Mellitus (p=0.012, r=0.398). However, unlike the supporting study, which utilized random blood glucose levels, the present study employed Fasting Blood Sugar (FBS). According to Roengrit, T et al. [4], which was also supported by American Diabetes Association criteria, respondents having a Fasting Blood Glucose (FBG) 100-125 mg/dL were included in the Impaired Fasting Blood Glucose, and < 100 mg/dL were included in the normal fasting blood glucose. This aligns with our result earlier that only 34.1 % of the respodents had a normal fasting blood sugar. While the remaining respondents were categorized having a Pre-Diabetes, and Diabetes level of Fasting Blood Sugar. Furthermore, This result signficantly supports the study of Lin, Y. et al [3], it was found that elevated Mean Arterial Pressure (MAP) was significantly associated with an increased risk of impaired fasting glucose (IFG). Also, according to Roengrit, T et al [4] that the study further revealed that fasting blood glucose levels were positively correlated with systolic blood pressure, suggesting that elevated glucose levels may contribute to vascular dysfunction and increased arterial pressure. An increase in both Fasting Blood Sugar (FBS) and Mean Arterial Pressure (MAP) may serve as an important clinical warning sign of underlying metabolic and cardiovascular dysfunction. Elevated fasting blood glucose levels were associated with increased arterial pressure and may contribute to vascular dysfunction and impaired cardiovascular regulation.
Lastly, the findings of this study revealed that an increase of total Cholesterol signficantly increases Mean Arterial Pressure. This finding supports the study of Chen, H et al [1]. They stated that Hypercholesterolemia and hypertension are the two most common risk factors for cardiovascular diseases and often co-occur. Using Mediation Analysis, their result also showed that Total Cholesterol (TC) induced Systolic Blood Pressure (SBP) elevation was mediated by arterial stiffness in more than half of the whole cohort (indirect effect = 0.73; percent mediated = 54.5%). Furthermore, the TC induced SBP elevation was mediated by arterial stiffness in less than half of the males (indirect effect = 0.70; percent mediated = 47.9 %). This suggests that Higher TC increase Blood Pressure because it makes the arteries stiffer.
Regression Coefficient Plot
The regression coefficient plot revealed that most predictors exhibited relatively small effect sizes on log-transformed Mean Arterial Pressure (MAP). Waist-to-Height Ratio showed the largest positive estimated effect, although with the wide confidence interval, it suggests a greater uncertainty in the estimate, while Fasting Blood Sugar and Total Cholesterol demonstrated modest positive associations. Several predictors exhibited confidence intervals close to or crossing zero, indicating weak or non-significant effects within the model.
References
[1] Chen, H., Chen, Y., Wu, W., Cai, Z., Chen, Z., Yan, X., & Wu, S. (2021). Total cholesterol, arterial stiffness, and systolic blood pressure: a mediation analysis. Scientific Reports, 11(1), 1330. https://doi.org/10.1038/s41598-020-79368-x
[2] Diabetes testing. (2024, May 15). Diabetes. https://www.cdc.gov/diabetes/diabetes-testing/index.html
[3] Lin, Y., Zou, J., Hong, M., Huang, X., & Wu, J. (2025). Elevated mean arterial pressure and risk of impaired fasting glucose: a multicenter cohort study revealing age and sex interactions. Frontiers in Endocrinology, 16, 1580036. https://doi.org/10.3389/fendo.2025.1580036
[4] Roengrit, T., Sri-Amad, R., Huipao, N., Phababpha, S., & Prasertsri, P. (2023). Impact of Fasting Blood Glucose Levels on Blood Pressure Parameters among Older Adults with Prediabetes. The Scientific World JOURNAL, 2023, 1–7. https://doi.org/10.1155/2023/1778371
[5] Singh JN, Nguyen T, Kerndt CC, et al. Physiology, Blood Pressure Age Related Changes. [Updated 2023 Aug 28]. In: StatPearls [Internet]. Treasure Island (FL): StatPearls Publishing; 2026 Jan-. Available from: https://www.ncbi.nlm.nih.gov/books/NBK537297/
[6] Utama, Julvainda & Ridha, Monika & Windyarti, Mei. (2023). Bloods Sugar Levels Towards Mean Arterial Pressure (MAP) In Type II Diabetes Mellitus Patients. Jurnal Smart Keperawatan. 10. 7. 10.34310/jskp.v10i1.793.