Early Findings from the Development and Validation of the Patient/Caregiver-Reported Hydroxyurea Evaluation of Adherence for Life (HEAL) Scale

Introduction: Until recently, hydroxyurea (HU) was the only FDA-approved treatment for sickle cell disease (SCD). Despite well-documented benefits, HU is underutilized in the US. Not all patients treated with HU have a clinical response, and nonadherence to treatment is a common reason for non-response. While not fully understood, reasons for nonadherence involve concerns about side effects, benefit, and knowledge. We report on the validation of our patient- and caregiver-reported Hydroxyurea Evaluation of Adherence for Life (HEAL) scale to assess adherence among persons with SCD.

Methods: Persons with SCD who had HU prescribed for ≥ 6 months prior to enrollment were eligible for our IRB-approved single site prospective study. A focus group and pilot study helped refine HEAL scale items. The scale consisted of 24 items with 8 subscales (Dose, Remember, Plan, Cost, Understand, Benefit/Harm [one subscale], Laboratory, and Pharmacy). To validate psychometric properties of HEAL we enrolled 40 pediatric and adult participants. We collected demographic information, relevant biomarkers, and medication possession ratio (MPR) at baseline and throughout the study period. Validity measures included a global adherence rating (GAR) from subjects and a visual analog scale (VAS) rating from subjects and from two healthcare providers per subject. Cronbach's alpha (α) was calculated to assess internal consistency reliability of the scale and subscales. Pearson product-moment correlations (r) between adherence measures and scale and subscale scores were used to assess validity. Principal component analysis was used to evaluate correlated factors (using eigenvalues >1) and a pattern matrix was used to create composite scales.

Results: Participants included 24 adults (self-report) and 16 children (caregiver-report). Self- versus caregiver-report subsamples differed in age but not in gender or race. Median participant age was 30 years (range 1.8 - 65.0 years). 24 subjects (60%) were female; and 31 (77.5%) had HbSS.

All items and subscales were negatively skewed. Three subscales (Dose, Cost, Effectiveness) comprising 9 items demonstrated ceiling effects. Three items (one in Dose and two in Cost) had restricted variance, defined as ≥ 75% of cases scoring at ceiling (e.g., score of 7) or skewness >2.

Principal component analysis revealed two composite scales: 1. Administer (Dose [0.539], Remember [.740], Plan [.688], Pharmacy [.683]); and 2. Value (Cost [-.584], Effectiveness [-.916], Understand [-.831]).

Correlations of adherence measures to Dose and Remember subscales, Total scale, and composite Administer scale were r= 0.340, 0.472, 0.380 and 0.321 respectively indicating validity of these HEAL components.

Test-retest reliability of the overall scale was good (kappa = 0.82); and acceptable (kappa > 0.6) on all subscales but Cost (kappa = 0.06).

Patient report VAS correlated significantly with Dose and Remember subscales, Administer composite scale, and Total scale score.

Discussion: Our data show HEAL subscales and composite scales have strong internal consistency and test-retest reliability. HEAL item and subscale distributions indicate high levels of adherence but sufficient variability for analysis and valid use. Data showed that the Of all subscales the Remember subscale correlated most with adherence measures indicating a critical aspect of patient adherence. The Cost subscale data demonstrated poor internal consistency reliability.

Adherence to HU is important because of its positive impact on patient quality of life when taken as directed. A self-/caregiver-report adherence scale that identifies areas where a patient needs additional support or resources may help increase adherence and improve quality of life. In previous research, implementation of a text message reminder for patients prescribed HU was associated with improved HU treatment adherence. The HEAL scale may help identify patients who would benefit most from targeted text message reminders.

While cost was not a barrier among participants in this study, that may not be the case for patients in other clinical settings. A larger study sample will enhance assessment of reliability and validity of HEAL scale and subscales and describe adherence in a broader patient population. Our initial findings support the use of HEAL, the first adherence scale to monitor HU adherence in patients with SCD.