Hydroxyurea Initiation among Children with Sickle Cell Anemia

Abstract

This study assesses characteristics of children with sickle cell anemia (SCA) associated with hydroxyurea initiation. Medicaid administrative claims from 6 states (2005–2012) were used to identify children with SCA enrolled in Medicaid for ≥2 years. Hydroxyurea use was defined as >30 days’ supply of filled prescriptions. Children were classified as initiators (no use in year 1; use in year 2) or non-users (no use in either year). Logistic regression was used to estimate associations between initiation, healthcare encounters, and demographics. A total of 4,435 children were enrolled for two years during the study period; 885 (20.0%) initiators, and 3,080 (69.4%) non-users. Children had an annual mean of 2.0 sickle cell disease-related inpatient admissions (SD = 2.1), 8.2 sickle cell disease-related outpatient visits (SD = 7.2), and 3.6 emergency department visits (SD = 3.5). The odds of initiating hydroxyurea increased with increasing healthcare utilization, age, and calendar year (all p-values <0.05).

Introduction

Sickle cell disease is a group of inherited red blood cell disorders that predominately affects minority populations in the US; 1 in 365 Black or African American births are diagnosed with this chronic condition.^1–4 There are numerous subtypes of sickle cell disease; the most severe subtype is sickle cell anemia (SCA), which is associated with the highest morbidity of any of the subtypes.^5,6 Children with SCA are at a significantly increased risk of stroke, pulmonary complications, and invasive pneumococcal disease as compared to children without SCA.⁷

The most common challenge facing children with SCA is excruciating, recurrent pain.^5,8 The majority of children with SCA experience at least one pain crisis a year, and these pain crises have a significant negative impacts on children.⁹ For example, children with SCA that experience pain crises are more likely to experience lower quality of life, more frequent school absences, depression, and impaired peer relationships, as compared to those without pain crises.^10–12 In addition, pain crises represent a substantial financial and health care services utilization burden, as they are responsible for the majority of hospitalizations and ED visits children with SCA.¹³

The frequency of pain crises can be reduced through the use of hydroxyurea. Hydroxyurea is a medication that increases the amount of fetal hemoglobin in the blood; increased fetal hemoglobin decreases the likelihood of the red blood cells sickling, which causes many of the complications of sickle cell disease.^7,14 Among children with SCA, use of hydroxyurea therapy is associated with lower rates of initial and recurrent episodes of pain, hand-foot syndrome (dactylitis), acute chest syndrome, and hospitalizations compared with no hydroxyurea therapy.^12,15–18 Given the effectiveness of hydroxyurea to reduce these morbidities, from 2002–2014 the National Heart, Lung, and Blood Institute (NHLBI) guidelines for sickle cell disease management recommended hydroxyurea use among children and adolescents with specific indications, such as frequent pain episodes, history of acute chest syndrome, severe vaso-occlusive events, or severe symptomatic anemia.⁵ The NHLBI released new guidelines in 2014 to broaden the recommendations for hydroxyurea use among children, indicating that all children over the age of 9 months, irrespective of disease severity should be offered hydroxyurea.⁷

Despite improved clinical and economic outcomes of hydroxyurea therapy, initiation of and adherence to hydroxyurea among children with SCA has remained low. Prior studies have shown that only about half of children that should be receiving hydroxyurea based on clinical indication have at least one filled prescription; among those already on hydroxyurea therapy, their hydroxyurea prescriptions only cover about half of the year.^15,19–24 However, very little is known about the characteristics of children with SCA that may influence hydroxyurea therapy initiation.^15,25 Identification of characteristics that potentially predict hydroxyurea initiation may detect specific subgroups of children that could be targeted to increase its use since the introduction of the new NHLBI guidelines. Therefore, the objective of this multistate study was to describe the characteristics of children with SCA that are associated with hydroxyurea initiation.

Methods

Data Source

Our target population was drawn from the Medicaid programs for the following 6 states with an average to high prevalence of SCA: Florida, Illinois, Louisiana, Michigan, South Carolina, and Texas. Medicaid Analytic eXtract (MAX) administrative data were acquired from the Centers for Medicare and Medicaid Services (CMS) for 2005 to 2012; at the time of the study, this was the most recent complete data released by CMS. Administrative data included enrollment history and all paid claims for inpatient, emergency department (ED), outpatient, laboratory, and outpatient pharmacy services. Approximately 90% of children with SCA are enrolled in Medicaid at some point in time and as a consequence, we believe that these data capture a large proportion of the children with SCA in these states.²⁶

Study Population

We identified children ages 1 through 17 years with SCA; our case definition was children having at least 3 claims within any calendar year that were SCA-related (ICD-9-CM codes 282.61, 282.62). This case definition was previously demonstrated to have a high sensitivity (91%) and specificity (80%) using state newborn screening records as the gold standard.²¹ Continuous enrollment in the Medicaid program for at least 2 consecutive calendar years within this time period was required. This allowed investigation of the predictors in the first year as related to hydroxyurea in the second year, which preserves temporality of the relationship. We restricted our analysis to children with no other forms of private health insurance during the study period to maximize the completeness of claims available.

Identifying Hydroxyurea Use

To identify all pharmacy claims for hydroxyurea, a list of National Drug Codes (NDCs) for hydroxyurea was generated and verified using RxNorm, a normalized naming system for all drugs produced and maintained by the National Library of Medicine.²⁷ Prescriptions for hydroxyurea included all forms for oral administration, such as tablet or powder, which could be used to create an oral suspension. For each person-year in the study population, the total days’ supply of hydroxyurea was calculated as the sum of the days’ supply of each filled prescription within a calendar year.^28–31 For this study, the authors chose to define hydroxyurea use as more than 30 days’ supply of hydroxyurea within a year, as this is likely reflective of obtaining at least one refill from the pharmacy and may be indicative of actual hydroxyurea use as opposed to a one-time fill.

Predictors of Hydroxyurea Initiation

Children within the study population were characterized into one of the following four mutually exclusive classifications, determined in the following order: 1) no hydroxyurea use in year 1, followed by hydroxyurea use in year 2 [initiators]; 2) no hydroxyurea use in either year [non-users]; 3) hydroxyurea use in year 1, followed by no hydroxyurea use in year 2; or 4) hydroxyurea use in both years. As each category was mutually exclusive and the categorization was sequential, children were only eligible to contribute one two-year time interval for the purposes of this study. We investigated the following potential predictors associated with hydroxyurea initiation: sickle cell disease-related inpatient admissions, ED visits, sickle cell disease-related outpatient visits, and demographic characteristics of sex and age. Inpatient admissions, ED visits, and outpatient visits were identified using HEDIS definitions.³² With the exception of age and calendar year, all potential predictors were characterized in the first year of continuous enrollment.

Statistical Analysis

Frequencies and percentages or means and standard deviations (SDs) were determined for all demographics. To identify potential predictors of initiation, the proportion of children attributed to each 2-year classification of hydroxyurea as described above was calculated. The assessment of predictors of initiation included only those cases classified as initiators (no hydroxyurea use in year 1, followed by hydroxyurea use in year 2) and non-users (no hydroxyurea use in either year); other patterns of hydroxyurea use were excluded from further analysis. Among included children, means and SDs of health services visits in year 1 were assessed. The mean of each type of utilization and age was compared across groups (initiators versus non-users) using t-tests; proportion female was compared using a chi-square test.

Logistic regression was used to estimate the association between each potential predictor and the initiation of hydroxyurea, with non-users serving as the comparison group. The functional form of each continuous variable was assessed. Continuous variables were first broken into quintiles. Quintiles with similar effect estimates were then pooled together based on estimated effect sizes. For example, children with zero inpatient hospitalizations had similar effect estimates as children with one inpatient hospitalization; therefore, these categories were pooled together. The Akaike Information Criterion (AIC) was used to compare models with each variable modeled continuously versus categorically.³³ Based on these results, we determined that all covariates should be modeled categorically. These categories were as follows: inpatient admissions (0–1, 2–3, 4+), ED visits (0–3, 5+), outpatient visits (0–4, 5+), and age (0–1, 2–3, 4+). All models were adjusted for sex, state of residence, and year.

Results

In the 6 states studied, a total of 4435 children were enrolled in Medicaid for at least one two-year interval from 2005–2012. The mean age overall was 6.5 years (SD = 5.4), and 48% of the population was female (Table 1). The state with the most children with SCA was Florida (31.0%), followed by Texas (16.7%), Louisiana (15.8%), Illinois (14.8%), Michigan (11.8%), and South Carolina (10.0%). Of those with a two-year interval, 20.0% (n=885) were initiators (no hydroxyurea use in year 1, followed by hydroxyurea use in year 2), 69.4% (n=3080) were non-users (no hydroxyurea in either year); 10.1% (n=446) had hydroxyurea use in both years, and 0.5% (n=24) had hydroxyurea use in the index year, followed by no hydroxyurea use in year 2.

Table 1.

Demographic Characteristics of Children with Sickle Cell Anemia with a Two-year Interval of Medicaid Enrollment from 2005–2012 (N=4435)

Age, mean (SD)	6.6 (5.4)
Gender
Female, n (%)	2148 (48.4)
Male, n (%)	2287 (51.6)
State
Florida, n (%)	1373 (31.0)
Illinois, n (%)	654 (14.8)
Louisiana, n (%)	700 (15.8)
Michigan, n (%)	522 (11.8)
South Carolina, n (%)	445 (10.0)
Texas, n (%)	741 (16.7)

Overall, the 3965 children in the sub-population of initiators and non-users (89% of the overall study population) had an annual mean of 2.0 sickle cell disease-related inpatient admissions (SD = 2.1), 8.2 sickle cell disease-related outpatient visits (SD = 7.2), and 3.6 ED visits (SD = 3.5) in the index year. Average age was 8 years old, and approximately half of the children were female. Bivariate analysis indicated that the mean number of each type of visit, as well as age, was significantly greater for initiators versus non-users (all p-values <0.001) (Table 2). The final multivariable model indicated that the odds of initiating hydroxyurea increased with increasing sickle cell disease-related inpatient admissions, ED visits, outpatient visits, age, and calendar year as compared to the reference groups (Table 3).

Table 2.

Demographics and Annual Healthcare Encounters among Initiators and Non-Users of Hydroxyurea (N=3965 children)

	Overall	Initiators n = 885	Non-Users n = 3080
Variable	Mean (SD) or %	Mean (SD) or %	Mean (SD) or %	P-value
Inpatient Admission	2.0 (2.2)	2.6 (2.4)	1.9 (2.1)	<0.001
Emergency Department Visit	3.6 (3.5)	4.4 (4.1)	3.4 (3.3)	<0.001
Outpatient Visit	8.2 (7.2)	8.5 (6.7)	8.1 (7.3)	<0.001
Age	7.9 (5.4)	10.0 (4.8)	7.3 (5.5)	<0.001
Female	48.9%	48.7%	48.9%	0.90

Table 3.

Multivariable Model Predicting Hydroxyurea Initiation among Children with a 2-year Interval of Initiation or Non-Use (N = 3965 children)

Variable	N	OR	Lower 95% CI	Upper 95% CI
Inpatient Admissions
0–1	2014	1.00	Reference	Reference
2–3	1243	1.61	1.33	1.96
4+	708	3.05	2.41	3.86
Emergency Department Visits
0–3	2406	1.00	Reference	Reference
4+	1559	1.43	1.19	1.72
Outpatient Visits
0–4	1332	1.00	Reference	Reference
5+	2633	1.49	1.24	1.80
Age
0–4	1425	1.00	Reference	Reference
5+	2540	5.36	4.35	6.60
Sex
Male	2027	1.00	Reference	Reference
Female	1938	0.94	0.80	1.10
State
Florida	1261	0.87	0.68	1.13
Illinois	575	1.00	Reference	Reference
Louisiana	634	1.14	0.85	1.53
Michigan	473	0.96	0.69	1.32
South Carolina	381	0.79	0.56	1.11
Texas	641	1.10	0.82	1.46
Year
2006	1000	1.00	Reference	Reference
2007	381	3.67	2.67	5.04
2008	459	2.93	2.13	4.03
2009	464	3.95	2.90	5.38
2010	570	4.50	3.35	6.04
2011	522	4.85	3.59	6.55
2012	569	5.39	4.04	7.20

Discussion

In this multistate analysis, hydroxyurea use was low across all geographies among children with SCA. Further, children that experienced more frequent, as well as more intensive, interactions with the healthcare system were more likely to initiate hydroxyurea. Interventions to increase hydroxyurea use among children with SCA should focus on two distinct processes with significant opportunities for improvement: 1) initiation of hydroxyurea; and 2) adherence to hydroxyurea following initiation.

Among our study population, nearly 70% of children had no evidence of hydroxyurea use. Initiators were more likely to have more inpatient admissions or ED visits, consistent with other studies.^34,35 These results are not surprising given the NHLBI guidelines during the study time period suggested hydroxyurea among children with indications that might reflect more severe disease complications.^5,20,36 However, the updated NHLBI guidelines now recommend that beginning at 9 months of age, hydroxyurea therapy should be offered to children and adolescents with sickle cell anemia, irrespective of clinical severity, to reduce disease-related health complications.⁷ Focusing on initiation of hydroxyurea among children that may appear to have less severe disease complications will be key to increasing hydroxyurea use in light of the new NHLBI guidelines. Initiation of a medication regimen among children that may be asymptomatic requires special consideration of specific challenges that families face in this situation. For example, previous research has indicated that in patients with asymptomatic chronic diseases, interventions to increase initiation of medication need to be tailored to patients’ individual needs, including scheduling frequent health care visits according to the stages of behavioral change, monitoring risk factors, targeting perception of risk, and simplifying drug regimens.^37–39 Following an emergency department visit for a SCA complication, clinic intervention involving a hematologist-led discussion of hydroxyurea and a direct offer to start hydroxyurea was found to increase hydroxyurea use.⁴⁰ In addition, interventions focused on initiation among younger children may impact rates overall, as these children were less likely to initiate hydroxyurea as compared to older children. This is particularly important given evidence that earlier initiation of hydroxyurea may reduce the burden of pain experienced by a child over the lifespan.⁴¹

Additional studies indicate that even among children with evidence of hydroxyurea use, the mean number of days’ supply is not sufficient to cover a year.^28,42–44 For example, in the state of New York, the mean number of days covered in the first year of hydroxyurea therapy was 53%.²¹ Further, the annual number of days’ supply increased as the years of hydroxyurea use increased. Increasing medication adherence after initiation is critical. Interventions that have been previously successful in increasing medication adherence among this population include utilization of community health workers, electronic medication container-monitor-reminder devices, electronic or mobile direct observed therapy (DOT) techniques, mobile applications, and text messaging.^45–52

Comprehensive interventions to increase the use of hydroxyurea among children with SCA should consider barriers specific to hydroxyurea use among this population on the sides of the patient, provider, and parent, which collectively affect a child’s use of hydroxyurea. Previous research has demonstrated that patients with negative perceptions of hydroxyurea, including lower perceived benefit of the therapy, and worse emotional response to sickle cell disease are less likely to adhere to the therapy regimen than those with more positive views of the therapy and sickle cell disease.^25,53–55 Additionally, patients that experience difficulty swallowing, depression, distrust of health care providers due to discrimination, more severe pain, and recall and access barriers tend to have worse adherence than their counterparts with sickle cell disease.^25,53,54,56 Negative attitudes of providers and lack of provider knowledge, nursing support, and infrastructure for monitoring hydroxyurea therapy also impedes initiation and adherence among this population.^55,57 For example, parents pointed to a lack of provider recommendation as the most common reason their child was not on hydroxyurea.⁵⁸

This study has several limitations. While the majority of children with SCA are enrolled in Medicaid at some point in time, this study population may not be reflective of the entire population of children with SCA. These data are also subject to the inherent limitations of using administrative claims, such as incomplete or missing codes or claims. It is possible that rates of hydroxyurea prescribing by providers to children with SCA are higher than the rates reported here, since medication claims reflect only those prescriptions that were actually filled. Finally, we did not assess if each child met the necessary indicators for initiation of hydroxyurea as based on NHLBI guidelines from 2002–2014. Although this enabled identification of subgroups that may be targeted to initiate hydroxyurea in light of the 2014 NHLBI guidelines, our low rates of initiation may be reflective of a smaller group of children who were eligible for hydroxyurea at the time of the study.

In conclusion, hydroxyurea use was low among this population. Given recent changes to guidelines, interventions to increase hydroxyurea use and adherence among children with SCA are critical.