18 Advantages and disadvantages of incorporating C-reactive protein levels, age and D-dimer levels in diagnosing pulmonary embolism
Crop, M.J., Siemes, C., Berendes, P., Straaten, F. van der, Levin, M-D.
Locatie(s): Zaal 0.4
Categorie(ën): Parallelsessie
Introduction: Over the past several years the number of performed radiological scans to diagnose pulmonary embolism (PE) increased dramatically, while the incidence of PE hardly increased. This resultated in an increased number of patients exposed to radiation and higher costs. Recently, the use of an age-adjusted D-dimer cut-off value was suggested to increase specificity of D-dimer testing.
Aim of the study: This study investigated the impact of incorporating age and levels of C-reactive protein (CRP) and D-dimer on the sensitivity and specificity to diagnose PE.
Methods: This observational study (2003-2007) included all consecutive patients suspected for PE presenting on the Emercency Department and with simulatanously measured levels of CRP and D-dimer. Only patients who received pulmonary CT angiography were further analysed. We studied the correlation between age, CRP, D-dimer and the effect of using an age-adjusted cut-off values for D-dimer (age in years/100 mg/L). Moreover, the predictive value of these parameters for PE was calculated.
Results: Of 4609 patients suspected for PE, 1164 patients underwent radiological imaging. PE was demonstrated in 309 patients (26.5%). Increased levels of D-dimer were modestly positively correlated with CRP levels (rs = 0.42; p < 0.001). In addition, D-dimer levels were positively correlated with age (rs = 0.33; p < 0.001). Interestingly, in patients < 50 years D-dimer and CRP were strongly correlated (rs = 0.47; p < 0.01), whereas in elderly the correlation was only 0.24 (p < 0.01). Thus, in young patients 22% of elevated D-dimer levels can be explained by an elevated level of CRP.
While age, D-dimer levels and age*D-dimer (i.e. age-adjusted D-dimer) significantly contributed to the prediction model of PE. Based on area-under-the-curve odds-ratios (OR, 95% confidence interval) were 1.008 (0.995-1.023), 5.536 (3.303-9.529) and 0.987 (0.979-0.996), respectively. CRP did not significantly contribute to the model. Using an age-adapted D-dimer cut-off value for patients ≥ 50 years the negative predictive value decreased from 96% to 93%, while specificity increased from 37% to 50%, Moreover, in patients (> 70 years), 40% more patients had D-dimers levels below the age-adjusted cut-off level compared to the conventional cut-off level, thus reducing the need of performing CT scans.
Conclusion: In the prediction of PE, age and dimer levels are relevant, while CRP levels are not. CRP levels only modestly effected D-dimer levels. Using an age-adjusted D-dimer cut-off in patients ≥ 50 years modestly reduces the negative predictive value, but dramatically increases the specificity of D-dimer testing.