Measuring hydroxyurea adherence by pharmacy and laboratory data compared with video observation in children with sickle cell disease

Background Hydroxyurea nonadherence is common among children with sickle cell disease (SCD), but it is unclear if current adherence measures are valid compared with video directly observed therapy (VDOT), a reference method. The objectives were to evaluate if hydroxyurea adherence by pharmacy records, urine assay, mean corpuscular volume (MCV), and/or fetal hemoglobin (HbF) correlated with and was sensitive and specific compared with VDOT. Methods This was a cross-sectional analysis of adherence data from 34 children with SCD on a single-arm, six-month hydroxyurea adherence study. Spearman correlation coefficient compared participants' adherence by pharmacy records, MCV, and HbF to adherence by VDOT. The sensitivity and specificity of >= 80% adherence by pharmacy records, two urine samples with hydroxyurea, MCV >= 100 fl/L, and HbF >= 20% compared with >= 80% VDOT adherence were also calculated. Results Median pharmacy and VDOT adherence rates were similar (87.8% vs 88.1%, P = 0.75) and mildly correlated (r(s) = 0.45; P = 0.008) but the sensitivity of >= 80% adherence by pharmacy records was 72.7% and specificity was 45.5%. MCV (r(s )= -0.02, P = 0.92) and HbF (r(s )= -0.2, P = 0.33) did not significantly correlate with VDOT adherence. Sensitivity and specificity were 83.3% and 33.3% for having two urine samples with hydroxyurea, 35% and 71.4% for MCV >= 100 fl/L, and 75% and 0% for HbF >= 20%, respectively. Conclusions Commonly used tools to measure hydroxyurea adherence may not correlate with or be valid compared with video adherence. Future studies to refine these measures are needed to effectively target adherence interventions to children with SCD who have the potential to benefit. (ClinicalTrials.gov NCT02578017)