Adherence to hydroxyurea, health-related quality of life domains, and patients’ perceptions of sickle cell disease and hydroxyurea: a cross-sectional study in adolescents and young adults

We conducted a cross-sectional study using non-probability convenience sampling methodology. We approached AYA who were 12–22 years old, had SCD patients (all genotypes), were English-speaking and on a steady state dose of hydroxyurea, defined as a stable dose for 2 months or more prior to enrollment in the study. Between January 2015 and December 2015, patients were enrolled before or after their scheduled outpatient comprehensive sickle cell clinic or hydroxyurea clinic appointments. Exclusion criteria included patients with SCD on chronic transfusion support, with any haemoglobin disorder other than SCD, or with any cognitive or physical disability.

We evaluated patients’ perceptions using the brief illness perception questionnaire (B-IPQ), which was adapted to reflect hydroxyurea and SCD [28]. In the adapted B-IPQ, we replaced the word “illness” by “sickle cell disease” and the word “treatment” by “Hydroxyurea”. For example, one of the questions was “How much do you think your hydroxyurea can help your sickle cell disease?” instead of “How much do you think your treatment can help your illness?” In addition, two items were deemed not applicable and/or could be confusing for patients with SCD, and therefore were deleted in the adapted B-IPQ version. The B-IPQ has been extensively utilized in earlier studies and used in patients as young as 8 years of age [27, 28]. The adapted B-IPQ consists of 7 items or dimensions including perceived consequences of SCD, personal control of SCD, control of SCD with hydroxyurea, identity (i.e. SCD-related symptoms), concerns about SCD, understanding of SCD, and emotional response secondary to having SCD. The “Identity” domain will be referred to as “symptoms” throughout the manuscript. Each item uses a scale from 0 to 10 to assess each dimension, and each item is evaluated as a separate subscale. Higher scores indicate stronger perceptions of each dimension of the B-IPQ. In particular, higher B-IPQ scores indicate more negative perceptions of SCD-related consequences, identity (i.e. more perceived symptoms), concerns, and emotional response, but more positive perceptions of personal control, treatment control, and understanding of SCD.

Study participants were asked to complete Patient Reported Outcomes Measurement Information System – Computerized Adaptive Testing (PROMIS®-CAT) measures using an electronic tablet [29]. PROMIS-CAT is a novel application of a comprehensive, item-response theory optimized item bank, which enables precise estimation of a PRO domain while simultaneously minimizing burden to participants [30, 31]. In CAT, items administered are selected based on informant’s previous item responses, using a pre-set computerized algorithm based on individual item information functions. Therefore, the total number of items used in different PROMIS measures vary within and in-between patients. A number of PROMIS® measures have been validated in the paediatric and adult populations [32‐35], and have been recently evaluated in SCD [36, 37]. Adolescents (age 12–17 years) were asked to complete paediatric PROMIS®-CAT measures of fatigue, pain interference, physical functioning mobility and upper-extremity, depression, anxiety, and peer relationships. Young adults (age 18–22 years) were asked to complete adult PROMIS®-CAT measures of fatigue, pain interference, physical functioning mobility and upper-extremity, depression, anxiety, and social isolation. PROMIS peer relationships and social isolation measures have different set of questions, however both evaluate patients’ relationships and social support. All paediatric and most adult PROMIS®-CAT measures elicited participants’ responses based on the previous 7 days using 5-point response options ranging from “with no trouble” to “not able to do” for physical functioning measures and from “never” to “almost always” in all other measures. No timeframe was assigned to adult PROMIS physical functioning measure. Higher PROMIS® domain scores indicated better physical functioning (mobility or upper-extremity), and peer relationships, but worse severity for fatigue, pain interference, depression, anxiety, and social isolation. PROMIS®-CAT paediatric and adult measures were scored on a general population based T-score metric with a mean of 50 and a standard deviation of 10 [32, 38].

Hydroxyurea adherence was evaluated using the ©Modified Morisky Adherence Scale 8-items (©MMAS-8), which was adapted to reflect hydroxyurea and SCD. In the adapted ©MMAS-8, we replaced the words “medicine”, “medication” and “pills” with “Hydroxyurea”. ©MMAS-8 includes 7 yes/no questions and 1 multiple-choice question, and evaluates adherence over the past day and 2 weeks [39, 40]. Total ©MMAS-8 numerical scores were calculated per the assessment manual. Higher ©MMAS-8 scores indicated better adherence to hydroxyurea. Based on the total ©MMAS-8 score, three levels of adherence were also considered: low (0 to <6), medium (6 to <8) and high (8).

Data were directly captured from electronic tablets using two online platforms at Northwestern University Feinberg School of Medicine: a) REDCap for hydroxyurea adherence and perceptions data, supported by the Northwestern University Clinical and Translational Sciences (NUCATS) Institute; and b) the Assessment Center^SM (https://​www.​assessmentcenter​.​net/​) for PROs data, supported by the Department of Medical Social Sciences, Northwestern University. Pilot testing was conducted to ensure tablet and Internet functionality in the outpatient setting. We collected information on SCD-related clinical events (e.g. number of emergency department (ED) visits and/or hospitalization at our institution in the year prior to enrollment), and hydroxyurea indication, dose, and duration of treatment through electronic medical chart review of enrolled participants. One author (SB) conducted chart review using a customized data extraction sheet. We obtained selected laboratory markers at the time of study enrollment, including fetal haemoglobin (HbF) and mean corpuscular volume (MCV).

We used descriptive statistics to report categorical data in frequencies and percentages. All B-IPQ scores were presented as medians, interquartile range, or both. An exploratory analysis was conducted to evaluate the relationship of patients’ perceptions of SCD and hydroxyurea to their demographic and clinical characteristics, domains of HRQOL and adherence to hydroxyurea. We used Wilcoxon rank-sum test to determine if patients’ perceptions (B-IPQ domain scores) were related to different categories of age groups, sex, hydroxyurea adherence level, HbF%, and/or MCV values. We also used Kruskal-Wallis test to assess if patients’ perceptions (B-IPQ domain scores) were related to number of SCD-related hospitalizations. Spearman’s rho (r_s) correlations were used to examine the relationship of patients’ perceptions (B-IPQ domain scores) to number of ED visits in 1 year prior to study enrollment, hydroxyurea adherence level using ©MMAS-8 and HRQOL domain scores using PROMIS measures. Participants were categorized according to their age as adolescents [12–17 years] (n = 25) and young adults [18–22 years] (n = 9); number of hospitalizations in the year prior to study enrollment as none (n = 14), 1–3 (n = 11), and 4 or more hospitalizations (n = 9); and hydroxyurea adherence level as low (n = 25) and moderate/high (n = 9). Given that most of the evidence in the literature on laboratory response to hydroxyurea is based on studies that included mainly patients with HbSS disease, we conducted a sub-analysis in this population to evaluate the relationship between laboratory markers of hydroxyurea adherence and patients’ perceptions of SCD and hydroxyurea. In HbSS patients only (N = 29), patients were categorized according to their median values for HbF% (16%) and MCV (102 fl) to compare HRQOL domain scores across groups. HbF% was low if it was less than 16% (n = 15) and high if HbF was equal to or greater than 16% (n = 14); and MCV was low if it was less than 102 fl (n = 15) and high if MCV was equal to or greater than 102 fl (n = 14). PROMIS® T-scores from the sample were analyzed as a group rather than stratified by age. All tests were 2-sided and a p-value <0.05 was considered statistically significant. We conducted a complete case analysis where missing variables are dropped from the analysis, leaving only complete cases. Statistical analysis was performed using SAS 9.3 (SAS Institute Inc., Cary, NC).

A total of 34 patients were approached, and all agreed to participate and were enrolled in the study (100% participation rate). The majority of participants were male (59%), Black (91%), and had a median age of 13.5 years (IQR 12–18) and a mean (SD) age of 14.8 (3) years. Most study participants had homozygous SCD (85%), and had hydroxyurea prescribed for recurrent pain episodes (53%). Almost 60 % of participants had one or more hospitalizations in the year prior to study enrollment with a median length of SCD-related hospital stay of 22 days (IQR 0–87). Table 1 summarizes participants’ demographic, clinical, and laboratory characteristics.

In our cohort, participants’ perceptions of SCD and hydroxyurea varied across B-IPQ domains by age, sex, and number of hospitalizations in the year prior to study enrollment (Table 2). Median scores indicated that older participants had more positive perceptions of personal control of SCD compared to younger ones (7 vs. 5, p = 0.04), while females perceived more SCD symptoms compared to their male counterparts (6 vs. 1, p = 0.02), respectively. None of the B-IPQ domain scores was significantly different by SCD genotype (data not shown).

When compared to those with 1–3 or no hospitalizations, participants with 4 or more hospitalizations perceived more SCD-related consequences (8 vs. 5 vs. 1, p < 0.001) and symptoms (8 vs. 5 vs. 0, p < 0.001) as well as worse emotional response to SCD (10 vs. 7 vs. 0, p < 0.001), and they also perceived hydroxyurea as less beneficial in controlling their SCD (5 vs. 9 vs. 10, p = 0.01) (Table 3). Longer total length of stay (LOS) significantly correlated with more perceived SCD-related symptoms (r _s = 0.72, p < 0.001), consequences (r _s = 0.73, p < 0.001), concerns (r _s = 0.34, p = 0.04), and worse emotional response to SCD (r _s = 0.75, p < 0.001). Longer LOS also correlated significantly with less perceived personal control of SCD (r _s = −0.34, p = 0.05) and benefits of hydroxyurea controlling their SCD (r _s = −0.54, p < 0.01). More frequent emergency department (ED) visits were associated with more perceived SCD-related symptoms (r _s = 0.68, p < 0.001), consequences (r _s = 0.68, p < 0.001), worse emotional response (r _s = 0.65, p < 0.001) and less perceived benefits of hydroxyurea controlling their SCD (r _s = −0.46, p = <0.01).

In our cohort, 25 (74%) participants reported low adherence to hydroxyurea using ©MMAS-8 scale. Participants with higher ©MMAS-8 scores, indicating greater adherence to hydroxyurea, perceived more benefits of hydroxyurea (r _s = 0.44, p < 0.01) and had better emotional response to SCD (r _s = −0.44, p = 0.01) (Table 4). Participants with moderate or high adherence to hydroxyurea perceived more benefits of hydroxyurea (10 vs. 7, p = 0.04), less symptoms (0 vs. 6, p = 0.03) and better emotional response related to SCD (0 vs. 7, p < 0.01), when compared to those with low hydroxyurea adherence.

In a sub-group of HbSS patients, we examined the relationship of MCV and HbF, as additional surrogates for adherence, to B-IPQ scores. Participants with high MCV values reported less negative perceptions of SCD-related symptoms (1 vs. 6, p = 0.01), better emotional response (0 vs. 7, p < 0.01), and more perceived benefits of hydroxyurea (9 vs. 6, p = 0.01), compared to those with low MCV values (Table 5). We found no significant differences in participants’ perceptions of SCD and hydroxyurea using their median HbF level (16%) as a cutoff. However, using a HbF% cutoff of 10%, participants with low HbF% reported worse emotional response related to SCD (10 vs. 0, p < 0.001) and less perceived benefits of hydroxyurea (5 vs. 9, p = 0.04), when compared to those with high HbF%, respectively.

About half of participants’ B-IPQ scores significantly correlated with different HRQOL domain scores as shown in Table 6. Participants with more SCD related consequences, concerns, and emotional response reported worse fatigue (r _s = 0.68, p < 0.001; r _s = 0.44, p = 0.01; r _s = 0.74, p < 0.001), pain (r _s = 0.56, p < 0.01; r _s = 0.54, p < 0.01; r _s = 0.63, p < 0.001), anxiety (r _s = 0.55, p < 0.01; r _s = 0.58, p < 0.001; rs = 0.56, p < 0.01), and depression (rs = 0.64, p < 0.001; rs = 0.49, p < 0.01; rs = 0.7, p < 0.001), respectively. In addition, participants with less perceived benefits of hydroxyurea reported worse fatigue (r _s = −0.6, p < 0.001), pain (r _s = −0.39, p = 0.03), anxiety (r _s = −0.47, p < 0.01), and depression (r _s = −0.62, p < 0.001).