Introduction An increased red cell distribution width (RDW), a routinely available index of the variability of erythrocyte size, has been associated with an adverse prognosis in cardiac and respiratory disease, including COPD.1 We have assessed its value in AECOPD and whether adding RDW improves the predictive power of the DECAF score.2
Methods We studied 2 groups of patients with AECOPD, the “derivation cohort” (n = 920) in whom DECAF was derived2 and the “internal validation cohort” (n = 880) in whom its prognostic value was confirmed.3
In the validation cohort RDW was collected prospectively and relationships to mortality assessed by univariate and multivariate logistic regression. RDW values were dichotomised by visual inspection of the receiver operator characteristic (ROC) curve which showed the optimal prognostic threshold for hospital mortality to be 15.5%, consistent with other studies.1 “RDW score” (15.5% or less=0, greater than 15.5%=1) was added to the DECAF score and the areas under the ROC (AUROC) curves for the DECAF and DECAF-RDW scores were compared by the method of DeLong.
In the derivation cohort RDW was collected from laboratory records and the prognostic utility assessed separately by logistic regression.
Results In the validation cohort RDW >15.5% was a strong predictor of inpatient mortality in both univariate (OR 2.70, 95% CI 1.68–4.32, p < 0.001) and multivariate analysis (OR 2.16, 95% CI 1.28–3.6 4, p = 0.004). However, there was no difference between the AUROC curves for the DECAF and DECAF-RDW scores (Figure 1; p = 0.63).
In the derivation cohort RDW >15.5% showed a non-significant trend towards higher inpatient mortality on univariate analysis (OR 1.55, 95% CI 0.96–2.50, p = 0.07), but there was no association on multivariate analysis (OR 1.05, 95% CI 0.60–1.84, p = 0.86).
Discussion The significant association of RDW with inpatient mortality in AECOPD in one cohort but not the other suggests limited value in this population. When forced into the DECAF model, RDW did not improve its predictive power and is a weaker prognostic index than the component parts of DECAF.