Dual probability approach for risk adjustment in patients with a low clinical likelihood of coronary artery disease

To investigate whether the PROMISE Minimal Risk Score (PMRS) enables adjustment of the risk factor-weighted clinical likelihood of obstructive CAD.

Two cohorts of stable patients with new-onset chest pain were established: a diagnosis cohort (n = 4,298) and a prognosis cohort (n = 14,013). Patients were stratified by the risk factor-weighted clinical likelihood model, and patients with low (>5 to 15%) clinical likelihood were further stratified by the PMRS using a ≥ 34% cut-off. For the diagnosis cohort, obstructive CAD was defined invasively by fractional flow reserve ≤0.80. For the prognosis cohort, the primary endpoint was non-fatal myocardial infarction or all-cause death.

In the diagnosis cohort, 1,669 (39%) patients had low (>5 to ≤15%) clinical likelihood, of whom 301/1,669 (18%) patients had a PMRS ≥34%. In these patients, the prevalence of obstructive CAD was 14/301 (4.7%), similar to patients with very-low (≤5%) clinical likelihood [64/1,667 (3.8%), p = 0.21]. In the prognosis cohort, 6,187 (44%) patients had low (>5 to ≤15%) clinical likelihood, of whom 993/6,187 (16%) patients had a PMRS ≥34%. In these patients, event rates were similar to patients with very-low (≤5%) clinical likelihood [hazard ratio, 0.91 (95% confidence interval, 0.52–1.52), p = 0.77]. Compared to patients with low (>5 to ≤15%) clinical likelihood and a PMRS <34%, the prevalence of obstructive CAD and risk were lower in patients with low (>5 to ≤15%) clinical likelihood and a PMRS ≥34% (p < 0.01 for both comparisons).

In patients with low (>5 to ≤15%) clinical likelihood of obstructive CAD, the PMRS enables safe down-classification of 1 in 6 patients to a very-low (≤5%) clinical likelihood category.

Clinicaltrials.gov identifiers: NCT02264717, NCT03481712, NCT04707859, NCT01174550 and NCT01149590.