Background: Acute vaso-occlusive crises (VOC) represent the most common reason for hospitalization among children with sickle cell disease (SCD). Acute chest syndrome (ACS) and severe hemolysis, both of which may require transfusion, are common complications during hospitalizations for VOC. Despite the high morbidity associated with these complications, there are currently no reliable clinical predictors of ACS and transfusion requirement in this setting. Nucleated red blood cells (NRBC), reported as the number of NRBC/100 white blood cells (WBC) on a complete blood count (CBC), may reflect marrow necrosis or increased erythropoetic drive, yet their relationship to VOC and its complications have not been studied. We aimed to evaluate the relationship between admission NRBC and development of ACS or transfusion requirement in children with SCD hospitalized for VOC. We hypothesized that higher admission NRBC is associated with greater risk of ACS/transfusion in this population.
Methods: We performed a single-institution, retrospective review of all hospitalizations from 2011 through 2015 for uncomplicated VOC in children with SCD. Hospitalization encounters were identified by ICD-9 codes for SCD and use of parenteral opioids for at least 24 hours in the electronic medical record. Data extracted from encounters meeting inclusion/exclusion criteria included all CBCs and clinical outcomes. Between-group comparisons were performed by Mann-Whitney U test and Pearson's chi-square test for continuous and categorical variables, respectively, as well as by Spearman's rank order correlation (SPSS V24). The relationship of admission and change in NRBC (∆NRBC) to complications was evaluated by binary logistic regression and adjusted for significant co-variates on bivariate analysis. For ∆NRBC, we used the final CBC obtained prior to a complication or prior to discharge for encounters not resulting in a complication.
Results: We reviewed 271 encounters for uncomplicated VOC in 96 patients (mean age 13.1 years, 47% male, and 76% Hb SS or S/β0 thalassemia). Overall, 48/271 (18%) encounters for VOC resulted in ACS/transfusion, or both. Median NRBC on admission was 2 /100 WBC (range 0 to 137 /100 WBC) for all encounters. Admission NRBC was significantly correlated with age (r = 0.15, p = 0.02), % reticulocytes (r = 0.18, p < 0.01) and hemoglobin (r = 0.39, p < 0.001). Median admission NRBC was higher among patients on hydroxyurea (3 vs. 1.8 /100 WBC, p < 0.01) but did not differ by sex or genotype. In general, median admission NRBC did not differ between hospitalizations that did or did not result in ACS/transfusion (2 vs 2.5 /100 WBC, p = 0.91). In our logistic regression model that included NRBC, % reticulocytes, WBC, hemoglobin, platelets and genotype as co-variates, only lower hemoglobin (p = 0.02) and higher WBC (p < 0.001) on admission were independently associated with ACS/transfusion. A second CBC was obtained in 110/271 (41%) encounters reviewed. Median ∆NRBC was 0 /100 WBC (range -37 to 116 /100 WBC) in these encounters. ∆NRBC was significantly correlated with age (r = 0.2, p = 0.04) but not with change in other lab values. Median ∆NRBC was higher among patients with a history of hypertension (0 vs. 13 /100 WBC, p < 0.01) but did not differ by hydroxyurea status, sex or genotype. Importantly, we found that median ∆NRBC was significantly higher in hospitalizations that resulted in ACS/transfusion compared to those that did not (0 vs 2 /100 WBC, p < 0.01). By logistic regression, greater ∆NRBC (p < 0.01) and lesser ∆platelets (p = 0.04) were independently associated with higher risk of ACS/transfusion after adjustment for ∆hemoglobin and genotype. A receiver operating characteristic curve constructed from a logistic regression model that included only ∆NRBC, admission WBC, admission hemoglobin and ∆platelets resulted in an area under the curve of 0.82.
Conclusions: We found no relationship between admission NRBC and complications during hospitalization for VOC in children with SCD. Greater ∆NRBC, however, was independently associated with development of ACS/transfusion requirement during hospitalization for VOC, suggesting ∆NRBC may represent a useful biomarker for predicting complications in children with SCD hospitalized for VOC. Clinical decision rules that incorporate the careful monitoring of ∆NRBC in this setting should be prospectively studied.
Disclosures
No relevant conflicts of interest to declare.
Implementation of the Logistic Regression Model and its Applications
