Table 3.

Predictors of the extent of coronary artery calcium, as measured by log-transformed non-zero coronary artery calcium scores, in three linear regression models.

Model approach	Model results
Predictors	Coefficients (95% CI)	Corresponding percent increase in natural CAC scores*	p-values	Adjusted R2
Age and sex only†
- Age, per 10 years	0.68 (0.63 – 0.73)	97% (88 – 107%)	<0.001	0.14
- Male sex	0.72 (0.61 – 0.82)	105% (85 – 127%)	<0.001
All CHD risk factors†
- Age, per 10 years	0.69 (0.64 – 0.73)	99% (89 – 109%)	<0.001	0.17
- Male sex	0.73 (0.63 – 0.83)	108% (88 – 130%)	<0.001
- Hypertension	0.23 (0.14 – 0.32)	26% (14 – 38%)	<0.001
- Diabetes mellitus	0.48 (0.33 – 0.62)	61% (40 – 88%)	<0.001
- High cholesterol	0.15 (0.05 – 0.24)	16% (4.8 – 28%)	0.004
- Smoking	0.45 (0.35 – 0.54)	56% (42 – 71%)	<0.001
Estimated 10-year risk of CHD‡, only†
- 10-year risk, per 5% increase	0.34 (0.31 – 0.36)	40% (36 – 44%)	<0.001	0.11

* – The percent increase in the natural (non-transformed) CAC score associated with each predictor is calculated by exponentiating the regression coefficient from the linear regression model (when the dependent variable is log-transformed), and subtracting 1.

† – Intercepts were -0.181, -0.705, and 3.17 for each model respectively. The standard deviations of the residuals were 1.682, 1.653, and 1.707.

‡ – 10-year risk of CHD estimated according to Framingham equations [16]; for assumptions used, see Methods.

CAC – Coronary artery calcium; CI – Confidence interval; CHD – Coronary heart disease.