Patient population, data source and study period
The patient population consisted of all individuals with chest discomfort and/or dyspnoea and low to intermediate pretest probability of CAD admitted for their first CCTA at the Silkeborg Diagnostic Centre between 1 January 2010 and 31 December 2013 and whose data had been reported to The Western Denmark Heart Registry. The Western Denmark Heart Registry is a validated clinical database within a population-based healthcare system and covers a population of 3.3 million.8 The database was initiated in 1999 and includes invasive procedures and, from 2008, also CCTA.8 ,9 Demographics, vital status, clinical data, data related to the CCTA, and data related to invasive procedures were obtained from the Western Denmark Heart Registry on 11 March 2014, which is also the date that follow-up ended. The Western Denmark Heart Registry holds no information on emigration status. The study was approved by the Danish Data Protection Agency. Danish authorities do not require informed consent when a study is based on registry data. The study therefore did not require approval from an ethics committee.
The study period was chosen because of the nearly complete reporting of patient data to the Western Denmark Heart Registry from 1 January 2010.9
CT technique
Patients were told not to eat for at least 2 h before the examination and to avoid coffee, tea and tobacco. They were asked to take 50–100 mg atenolol in the morning of the day of the examination. For those with contraindications to β-blockers, we used ivabradine 5–7.5 mg. If the heart rate exceeded 65 beats/min during the examination, we added 2.5–20 mg metoprolol intravenously.
We used a 64 multislice spiral CT scanner (Philips Brilliance 64; Philips Medical Systems, The Netherlands) until June 2010 and thereafter a 320 multislice dynamic volume CT scanner (Aquillion One; Toshiba Medical Systems, Japan), which shortened the scan time. The new scanner also uses ECG-synchronised data sampling, which further reduces the radiation dose. In the summer of 2012, the Toshiba scanner was upgraded from a filtered back projection to an iterative reconstruction and dose modulation. This resulted in a 14% reduction in the radiation dose from a mean of 3.1 to 2.7 mSv. In total, the change of CT scanner together with the upgrade resulted in a 50% reduction in the median radiation dose from 4.2 to 2.2 mSv.10
The CCTA examination started with a calcium scan without contrast, which produced a CAC score. Then, 800 μg nitroglycerine was administered sublingually to dilate the coronary arteries 2 min before the next scan. A bolus of 60–90 mL non-ionic contrast (Iomeron 350 mg/mL; Bracco Imaging Scandinavia AB) was then administered at an infusion rate of 5–6 mL/s via a cubital vein. CT recordings with and without contrast were obtained during breath holds of 10 s with simultaneous ECG recording. Subsequently, data were transferred to a Vitrea Advanced Workstation (Vital Images, USA). Coronary calcium was defined as an attenuation of ≥130 Hounsfield units. The CAC score was estimated using the Vitrea Workstation and a calcium analysis programme. A summed calcium score based on data from all coronary arteries, the Agatston Score,2 was calculated for each patient. Contrast infusion was omitted when the Agatston Score exceeded 1000, in which case the patients were admitted for a perfusion scan or invasive coronary angiography.
After completion of the CT scan with contrast infusion, all coronary artery segments were reconstructed on a Vitrea Workstation. The coronary arteries were classified as normal, non-obstructive (stenosis <50%) or obstructive (stenosis ≥50%).
Coronary CT angiograms were analysed by cardiologists with a level 2 certification (American Society of Cardiovascular CT Accreditation).
Statistical analysis
Categorical variables are presented as percentages, and continuous variables as medians with 10th and 90th centiles in parentheses. We used Pearson’s χ2 test for categorical variables and the Mann–Whitney test for continuous variables.
Data were analysed in Cox’s proportional hazards regression models. We adjusted for risk factors and for proxies of risk factors for coronary heart disease such as age, smoking, CAC, diabetes mellitus, medical treatment for hypertension and cholesterol-lowering treatment. An invasive coronary angiography or a coronary revascularisation marked the end of the observation time, and the observation time was censored by death or the date that follow-up ended on 11 March 2014. The proportionality assumptions of the models were evaluated using Schoenfeld residuals and graphically. We considered p<0.05 to be significant. The study power was 90–100% for the detection of a 5% absolute difference between men and women in the range from 5% to 35% of the incidence rates of invasive coronary angiography or coronary revascularisation, given a risk of type I error of 5% and risk of type II error of 20%. All analyses were performed using Stata V.12.0.
