Methods
We performed a one-year retrospective review of 172 HbSS and HbSb0
thalassemia patients prescribed HU during 2019 that were cared for in
either the academic center (Birmingham) or satellite clinics
(Montgomery, Tuscaloosa, Opelika). Patients prescribed HU older than age
18 at the beginning of 2019 were excluded. We excluded two patients with
HbSD, twenty-two patients with HbSC, and seven patients with Hb Sb+
thal. We excluded SCA patients on HU and a second sickle cell disease
modifying therapy, such as chronic transfusion therapy or an
investigational or novel SCA drug therapy. We abstracted from the EMR:
age, sex, clinic location, diagnosis, HU dose (mg/kg), white blood cell
count, hemoglobin, mean corpuscle volume, platelet count, absolute
reticulocyte count, hemoglobin F, and absolute neutrophil count from
each clinic visit. We recorded the total number of well-clinic visits
and categorized adherence to clinic visits/year as ≤ 2 or
>2; satellite clinic visits are fixed either monthly to
every three months which limited our ability to reschedule patients with
missed visits. Well-clinic visits were limited to HU follow-up
appointments only. We recorded the number of ED visits and UAB pediatric
pain clinic visits and noted if the chief complaint included sickle cell
related pain. We also recorded the number of admissions to the hospital
for a) pain and b) all other sickle cell complications except for
scheduled procedures. We collected acute visit data from the EMR at the
academic center for Birmingham clinic participants and from patient
recall from satellite clinic participants. We evaluated dosing
adjustment as whether dosing was increased for ANC > 4000
per institutional standard of care. We categorized correct dosing
adjustments if > 65% of visits had a dose adjustment per
institutional protocol; we categorized patients as a deviation from
protocol if the physician did not increase the dose based on ANC. We
defined non-adherence to HU as a HbF% < 5. Socioeconomic
variables were assessed using the home address provided by the
participant and analyzing the information available for that geographic
location. Home address information was linked with the census track data
to determine the poverty level and the mean household income for the
census track of the patient’s residence.
Our primary outcome for this study was percent fetal hemoglobin: our
primary predictors were clinic location, age, and socioeconomic status.
Our secondary outcomes were difference in hydroxyurea monitoring labs,
clinic visits, and acute care utilization. We performed descriptive
statistics including mean and standard deviation (s.d.), t-test for
continuous variables, chi-square for categorical variables, and
regression modelling using JMP Pro 14 (Cary, NC). P-values
<0.05 were considered statistically significant.