Table 3.
Principal Component Analysis Score: Cross-Sectional Associations with Cardiometabolic Risk Factors
| A | ||||||
|---|---|---|---|---|---|---|
| Model 1a | Model 2b | |||||
| Risk Factor | Beta | Standard Error | P value | Beta | Standard Error | P value |
| Age (years) | 9.21 (8.60–9.82) | 0.31 | 8.52E-189 | 8.69 (8.10–9.28) | 0.30 | 1.43E-178 |
| Sex (Male) | −1.31 (−12.25–9.63) | 5.58 | 8.15E-01 | −18.51 (−29.21 - −7.81) | 5.46 | 7.07E-04 |
| Never vs Current Smokers | 41.93 (26.80–57.06) | 7.72 | 5.81E-08 | 43.01 (28.31–57.71) | 7.50 | 1.03E-08 |
| Never vs Former Smokers | 0.24 (−12.50–12.98) | 6.50 | 9.71E-01 | −3.64 (−15.95–8.67) | 6.28 | 5.62E-01 |
| Former vs Current Smokers | 49.94 (31.67–68.21) | 9.32 | 8.90E-08 | 57.4 (39.68–75.12) | 9.04 | 2.38E-10 |
| B | ||||||
| Model 1a | Model 2b | |||||
| Risk Factor | Beta | Standard Error | P value | Beta | Standard Error | P value |
| Triglycerides (mg/dL) | > 0.000 (>0.000 - >0.000) | > 0.000 | 3.96E-49 | > 0.000 (>0.000 - >0.000) | > 0.000 | 8.60E-18 |
| Total Cholesterol (mg/dL) | −0.003 (−0.007–0.001) | 0.002 | 1.25E-01 | −0.009 (−0.013 - −0.005) | 0.002 | 1.47E-05 |
| HDL-Cholesterol (mg/dL) | −0.009 (−0.011 - −0.007) | 0.001 | 6.43E-32 | −0.004 (−0.006 - −0.002) | 0.001 | 1.11E-08 |
| Glucose (mg/dL) | 0.015 (0.013–0.017) | 0.001 | 1.64E-33 | 0.008 (0.006–0.010) | 0.001 | 3.89E-12 |
| SBP (mmHg) | 0.012 (0.010–0.014) | 0.001 | 2.62E-47 | 0.007 (0.005–0.009) | 0.001 | 6.00E-19 |
| DBP (mmHg) | 0.003 (0.001–0.005) | 0.001 | 2.42E-07 | −0.001 (−0.003–0.001) | 0.001 | 1.50E-01 |
| BMI (kg/m2) | 0.006 (0.006–0.006) | > 0.000 | 1.15E-102 | * | * | * |
| C | ||||||
| Model 1a | Model 1 + Risk Factor Adjustmentsc | |||||
| Outcome | OR | 95% CI | P value | OR | 95% CI | P value |
| Hypertension | 1.41 | (1.32–1.50) | 1.85E-25 | 1.35 | (1.18–1.54) | 7.07E-06 |
| Diabetes | 1.62 | (1.48–1.78) | 4.30E-24 | 1.07 | ** | 1.00E+00 |
| Obesity | 1.70 | (1.59–1.81) | 5.59E-58 | 1661.74 | ** | 9.94E-01 |
Notes: Two logistic regression models were used utilizing the PCA score as the independent variable and the dichotomous risk factors as the dependent variable: Model 1 (age, sex, and cohort adjusted) and Model 2 (age, sex, cohort, and BMI adjusted). In both models, the appropriate covariate (age or sex) was removed as a covariate when that variable is investigated as the outcome. Two linear regression models were used utilizing the risk factors as independent variables and the PCA score as the dependent variable using the same list of covariates. Only participants with fasting measurements were included in triglycerides, total cholesterol, HDL, and glucose analyses. Triglyceride levels were log transformed (log10) in order to obtain normality. For age and sex analyses, the relevant model excluding the adjustment for that respective risk factor was utilized. Bonferroni corrected p value = 3.33E-03 (0.05/15). For the disease outcomes, logistic regression was used, in accordance with Models 1 and 2. There were 3526 cases of hypertension at baseline of the 7270 participants at risk, 484 cases of diabetes at baseline out of the 7284 participants at risk, and 1881 cases of obesity at baseline of the 7287 participants at risk. Odds ratios (OR) are presented per 1-SD increase in the respective inflammatory score. aModel 1: PCA Score = Age + Sex + Cohort; Risk Factor = PCA Score + Age + Sex + Cohort. bModel 2: PCA Score = Age + Sex + Cohort + BMI; Risk Factor = PCA Score + Age + Sex + Cohort + BMI. cModel: Outcome = PCA Score + Age + Sex + Cohort + Relevant Risk Factor (SBP and DBP for hypertension, glucose for diabetes, and BMI for obesity). * Model 2 includes an adjustment for BMI. ** Upper and lower confidence limits approach infinity and negative infinity.
Abbreviations: SBP, Systolic Blood Pressure; DBP, Diastolic Blood Pressure.