Risk adjustment included the following variables: age, hypercholesterolemia, diabetes, smoking, hypertension, and a family history of premature coronary heart disease. (A) Subsets ranging from 0 to 1,000. The increasing calcium scores were associated with worsening survival. Each increment of calcium score was associated with significant increased risk of all-cause mortality (chi-square = 1,363, p < 0.0001 for variable overall and for each category subset). Model chi square = 2,017, p < 0.0001, and chi-square = 274 for variable (p < 0.0001 overall and for each category subset). Scale on this curve is from 0.80 to 1.00, whereas the remaining curves are plotted within a range of 0.90 to 1.00. (B) Cumulative survival by the coronary calcium extent in the number of vascular territories with scores 100, adjusted for age and risk factors. There was worsening survival with increasing number of calcified vessels (chi-square = 251, p < 0.0001). Model chi-square = 1,290, p < 0.0001, and chi-square = 27 for variable (p < 0.0001 overall and p value for each category subset is stated in figure). The survival curve for 3 vessel (n = 28) was superimposed on the left main subset and, for that reason, the 2 were combined to form 1 category. (C) Cumulative survival in patients with coronary artery calcium (CAC) scores in the range of 11 to 100. Even in patients with small amounts of CAC, increasing number of vessels involved was associated with worsening long-term survival (chi-square = 182, p < 0.0001 for the variable and for each category subset). Model chi-square = 1,148, p < 0.0001, and chi-square = 13 for variable (p = 0.013 overall and p value for each category subset is stated in figure).

Comparison of 5-year mortality rates in 2 different cohorts. The results are quite consistent across different study groups (chi-square = 1,503, p < 0.0001, interaction p < 0.0001). CAC = coronary artery calcium; EBT = electron beam tomography.

These curves note the available data revealing the highest area under the curve for clinical risk factors. In both cases, the addition of the Agatston score resulted in a significant improvement in the area under the curve (p < 0.0001 for the total number of risk factors [A] and for age [B]). Receiver-operating characteristic analysis for other individual risk factors were <0.586 for gender, 0.440 for family history, 0.573 for smoking, 0.577 for diabetes, 0.518 for ethnicity, 0.484 for hyperlipidemia, and 0.562 for hypertension.

This study (Los Angeles) demonstrates similar prevalence to the prior all-cause mortality study from Nashville, Tennessee, and only slightly greater prevalence than the CARDIA (Coronary Artery Risk Development In Young Adults) study and MESA (NIH-NHLBI Multi-Ethnic Study of Atherosclerosis).