Professor Myriam Hunink from the Erasmus University Medical Centre in Rotterdam, Netherlands, retrospectively analysed data from 5,677 patients, of whom 3,283 were men and 2,394 were women.
Obstructive coronary artery disease was present in 1,634 of the patients.
The researchers compared the predictive accuracy of an extended risk model that included coronary calcium scores, a basic model and a clinical model.
The basic model included patients' age, sex and details of chest pain.
The clinical model included details of whether patients had diabetes, hypertension, or dyslipidaemia, or were smokers.
The researchers found that the extended model was twice as good as the basic model at predicting coronary artery disease. The clinical model was 35% better than the basic model.
Professor Hunink and her team also compared their models with the Duke clinical score, which the UK National Clinical Guideline Centre for Acute and Chronic Conditions recommends for estimating probabilities of coronary artery disease in patients who have reported chest pain.
The researchers also found that the Duke clinical score significantly overestimated the probability of coronary artery disease.
The researchers said that their model was not only more accurate than the Duke score, but did not require ECG findings, so would be more convenient in primary care.
'A refined estimate of the probability of coronary artery disease allows doctors to make better decisions as to which diagnostic test is best in a particular patient, according to NICE guidelines, and to decide on further management based on the results of such tests,' they said.