Development and Validation of a Pediatric Severity Index for Sickle Cell Patients.

Background: The clinical picture in patients with sickle cell disease (SCD) is heterogeneous. In order to study determinants of a severe clinical course well designed etiological studies are needed, using a valid outcome measure for disease severity. At this moment there is no generally accepted instrument to measure severity in pediatric SCD patients. Objective: The aim of this study was to develop and validate a severity index (SI) for SCD in children. Methods: Item selection for the SI was based on an item pool (n=50) retrieved from a systematic review of all disease severity assessment instruments for SCD. Items from this pool were rejected in case they were not applicable to children, non-specific for SCD or confounded. To address differences in severity among the items we included a transparent weighting process in the summation of the score. Three different weighting systems to summarize the items of the SI were used. First, all items were summed with an equal weight of 1 (score A). Secondly, acute life-threatening events and neurological complications were assigned more weight, receiving a score of 10, with all other items receiving a score of 1 (score B). Finally, items were weighted according to the severity of the different complications and the frequency of occurrence, ranging from 5–50 points (score C). We tested the validity of the SI using data from 92 patients (mean age 9.7 years, range 2–18 years) with SCD who were regularly seen at the study center. We evaluated whether different scores were obtained for patient groups classified according to severity subjectively by two pediatric hematologists (expert classification) and objectively by an existing score. Furthermore, we tested whether the index could differentiate patients classified by genotype (HbSS/HbS-β0-thalassemia versus HbSC/HbS-β+-thalassemia) or the number of alpha-gene deletions. Finally, we evaluated whether the SI correlated with age. Results: The resulting SI consisted of the following 23 items: acute chest syndrome, aplastic crisis, avascular bone necrosis, cardiomyopathy, overt and silent cerebral infarction, cerebral vasculopathy, enuresis nocturna or renal concentration problems, failure to thrive, gall stones, hemolytic crisis, hepatic and splenic sequestration, leg ulcers, painful crisis, pneumococcal septicemia and/or meningitis, priapism, retinopathy and 5 laboratory values (Hb, HbF, bilirubin, leucocytes, reticulocytes). For all weightings there was a significant difference in the scores of patient groups classified as mild, moderate and severely subjectively by experts or by the existing score (p<0.001). The index clearly differentiated patients by genotype (p<0.001) or alpha-gene deletions (p<0.001). Unexpectedly, there was no correlation with age (Spearman’s ρ = 0.19). Score C, with the most extensive differentiation in weighting of the items, discriminated best. Conclusion: In conclusion, we developed and validated a SCD severity index by a transparent and rigorous process. Further validation is of this SI is needed in a larger prospective cohort study of patients diagnosed by neonatal screening. After further refinement and adaptation of this index, it may contribute in achieving consensus on outcome assessment in etiological research in pediatric patients. This is important to enhance the comparability of study results and enable statistical pooling in meta-analyses.