Final elements used in COVID-19 traffic lights
| Pathology marker | No flag | Mild-risk | Medium-risk | High-risk | Suspected | |
| Thromboembolism* | CRP (mg/L) and/or D-dimer (ng/mL) | <1.1 and ≤1000 | 1.1 to 200 and ≤1000 | >200 to 250 and/or >1000 to 3000 | >250 and/or >3000 to 5000 | any with >5000 |
| Cytokine storm | Ferritin (μg/L) | <300 ♂ and <200 ♀ | 300♂/ 200♀ to 1000 | >1000 to 2000 | >2000 to 4000 | >4000 |
| ARDS/secondary infection† | SpO2 (%) and PCT (μg/L) | >96 and ≤0.05 | 94–96 and >0.05 to 0.25 | 92–94 and >0.25 to 0.5 | 90–92 and >0.5 to 1.0 | <90 and >1.01 |
*In the initial analyses, based on pattern of disease in patients with a proven pulmonary embolus (PE) on CT scan, a D-dimer >10 000 ng/mL was used to indicate thromboembolic disease. The single biggest predictor was that of CRP which preceded a subsequent spike in D-dimer: 22% of the patients with a CRP >200mg/L later developed a D-dimer>10 000ng/mL, rising with a CRP >250mg/L. Conversely, of the 30 patients with a pulmonary embolus and CRP exceeding 200mg/mL, 13 (43%) had an initial D-dimer >3 000ng/mL, suggesting that CRP may help as an early warning sign of a future high spike in D-dimer.
†Acute Respiratory Distress Syndrome (ARDS)/ secondary Infection was iterated several times throughout the first COVID-19 infection outbreak, and interpretation depended on the clinical situation. For simplicity and modelling SpO2 <90% on arrival to the emergency department (on room air) was used as a predictive marker for ARDS as identified on neuronal network predictive modelling.
CRP, C-reactive protein; SpO2, oxygen saturation.