Chronic Medication Burden after Cardiac Surgery for Pediatric Medicaid Beneficiaries

Abstract

Background:

Congenital heart defects are the most common and resource intensive birth defects. As children with congenital heart defects increasingly survive beyond early childhood, it is imperative to understand longitudinal disease burden.

Objective:

To examine chronic outpatient prescription medication use and expenditures for New York State (NYS) pediatric Medicaid enrollees, comparing children who undergo cardiac surgery (cardiac enrollees) and the general pediatric population.

Methods:

Retrospective cohort study of all Medicaid enrollees < 18 years using the NYS Congenital Heart Surgery Collaborative for Longitudinal Outcomes and Utilization of Resources database (2006-2019). Primary outcomes were total chronic medications per person-year; enrollees per 100 person-years using ≥1 and ≥3 medications; and medication expenditures per person-year. We described and compared outcomes between cardiac enrollees and the general pediatric population. Among cardiac enrollees, multivariable regression examined associations between outcomes and clinical characteristics.

Results:

We included 5,459 unique children (32,131 person-years) who underwent cardiac surgery and 4.5 million children (22 million person-years) who did not. More than 4 in 10 children who underwent cardiac surgery used ≥1 chronic medication compared to approximately 1 in 10 children who did not have cardiac surgery. Medication expenditures were 10 times higher per person-year for cardiac compared to non-cardiac enrollees. Among cardiac enrollees, disease severity was associated with chronic medication use; use was highest among infants, however, nearly half of adolescents used ≥1 chronic medication.

Conclusions:

Children who undergo cardiac surgery experience high medication burden that persists throughout childhood. Understanding chronic medication use can inform clinicians (both pediatricians and subspecialists) and policymakers, and ultimately the value of care for this medically complex population.

Central Illustration: Prevalence of chronic medications among pediatric NYS Medicaid enrollees.*

*Five medication groups were of highest use amongst enrollees after cardiac surgery (in decreasing order): cardiovascular and renal (including baby aspirin), immunologic, gastrointestinal, respiratory and allergy, and central nervous system (CNS) agents. Because the immunologic medication group primarily represented use of a single drug, palivizumab, in the first year of life, immunologics are not displayed above for presentation clarity. Error bars denote the upper and lower bounds of the 95% confidence interval of the average enrollee prevalence for a given age, cardiac disease severity, and medication type.

Condensed Abstract:

As more children with congenital cardiac surgery survive into adulthood, longitudinal disease burden must be better understood. Children with a history of cardiac surgery have nearly four times higher use of ≥ 1 chronic medication, and 10 times higher medication expenditures compared to children without surgical history. Regardless of cardiac disease severity, chronic medication use persists after surgery and throughout childhood. Considering patients’ quality of life and value of care, generalists, subspecialists and policy makers must weigh the psychological burden, potential risks and costs associated with prescribing “yet another” chronic medication.

Introduction

Congenital heart defects are the most common birth defects, afflicting approximately 1% of live births.¹ Overall mortality for children with cardiac defects is decreasing; national age-adjusted mortality rates have declined from 1.4 per 100,000 person-years in the late 1990s to 0.8 per 100,000 person-years in the mid-2000s.² As the number of children who survive surgery grows, so does the need to understand the longitudinal disease burden for this population.

Chronic medication use can serve as an indicator of disease burden.^3,4 It correlates with overall healthcare utilization and expenditures^5,6,7 and introduces risks of polypharmacy and other adverse events.^8,9 For providers, knowledge of chronic medication use can inform a more nuanced perspective of the disease experience.^10,11 For payors and policymakers, knowledge of chronic medication use can inform data-driven policies that promote safe, high-value care.

The goals of this study were to (1) describe chronic medication use throughout childhood among Medicaid enrolled children across New York State (NYS) who underwent cardiac surgery compared to the rest of the NYS pediatric Medicaid population, (2) evaluate relationships between chronic medication use, outpatient prescription medication expenditures, and clinical/sociodemographic characteristics among children who underwent cardiac surgery.

Methods

Study Design and Data

Using the NYS Congenital Heart Surgeons Collaborative for Longitudinal Outcomes and Utilization of Resources (CHS-COLOUR) we performed a retrospective cohort study, utilizing 14 years (2006-2019) of clinical and administrative data. CHS-COLOUR links locally-held Society of Thoracic Surgeons-Congenital Heart Surgery Database (STS-CHSD) data and New York State Pediatric Congenital Cardiac Surgery (NYS-PCCS) clinical registry data from 10 of 11 pediatric cardiac surgical centers across NYS, to Medicaid claims via direct patient identifiers.¹¹ The STS-CHSD captures >90% of all U.S. congenital heart surgeries with high-level clinical granularity and accuracy.^12,13 The NYS-PCCS is a mandatory NYS Department of Health pediatric cardiac surgical registry.¹⁴ As previously described, 53% of registry cases listed Medicaid as the payor statewide. Of these, 89% were linked to Medicaid claims,¹¹ allowing for study of longitudinal healthcare utilization. The complete NYS Medicaid dataset was used to identify pediatric enrollees who had not undergone cardiac surgery.

Study Population and Definitions

We included all children <18 years of age who were continuously enrolled in NYS Medicaid for at least one year between January 2006 and December 2019.

The STS-CHSD and NYS-PCCS characterize children who undergo cardiac surgery by fundamental diagnosis, “the most complex cardiac anomaly or condition that is carried with a patient through life, through all operations and hospitalizations.”¹⁵ We stratified these fundamental diagnoses into three disease severity groups: (a) critical single-ventricle physiology requiring initial palliation within the first 60 days of life^16,17 (“critical single ventricle”; e.g. hypoplastic left heart syndrome), (b) critical bi-ventricular physiology requiring operation within the first 60 days of life (“critical bi-ventricular”; e.g. D-transposition of the great arteries), or (c) non-critical cardiac disease with initial operation after 60 days of life (“non-critical”; e.g. atrioventricular canal defect). Hereinafter, “cardiac enrollees” will refer to children who underwent (at least one) cardiac surgery, and “non-cardiac enrollees” will refer to all children who did not undergo cardiac surgery. Of note, children with cardiac defects of minimal clinical significance, for whom palliative care was selected, or who died prior to surgery were included as “non-cardiac enrollees.”

Outcome Variables

The primary outcomes were 1) total number of chronic medications per person-year, 2) total number of enrollees per 100 person-years using ≥1 and ≥3 chronic medications, and 3) total Medicaid outpatient prescription medication expenditures per person-year. To account for differences in enrollment time, we used person-years as the denominator for prevalence estimations. Chronic medication use was defined as ≥90 consecutive days of filled prescriptions, consistent with pharmacoepidemiologic standards.⁵ All medications were classified by the American Hospital Formulary Service Classification Compilation into three levels of decreasing granularity: generic drug name (e.g. furosemide), therapeutic class (e.g. loop diuretics), and medication group (e.g. cardiovascular/renal agents).¹⁸ We calculated chronic medications by unique therapeutic class. For example, a proton pump inhibitor and a histamine H2-receptor antagonist were counted as two separate chronic medications, but prescriptions for two different H2-receptor antagonists were counted as a single chronic medication. We excluded all over-the-counter medications except baby aspirin, which was counted as a “cardiovascular and renal agent,” given its specific use in cardiac surgical patients. Expenditures were adjusted to U.S. 2019 dollars using the Medical Consumer Price Index.¹⁹

Explanatory Variables

Explanatory variables included enrollee fundamental cardiac diagnosis, sex, age, race, ethnicity, and presence of one or more non-cardiac complex chronic condition(s). Fundamental cardiac diagnosis was abstracted from the clinical registries. Demographics were abstracted from Medicaid data. Age at time of chronic medication use was categorized as <1, 1-5, 6-12, and 1317 years. Race and ethnicity were combined into a single variable, categorized as: Hispanic, nonHispanic Asian, non-Hispanic Black, non-Hispanic White, and other (including unknown). The variable was used to account for known inequities in utilization across socially constructed racial and ethnic groups. Non-cardiac comorbidities were identified with International Classification of Disease, Ninth and Tenth Revision, Clinical Modification diagnostic codes in Medicaid claims, using the pediatric Complex Chronic Conditions (CCC) classification system.^20,21

Statistical Analyses

We compared outcomes between cardiac and non-cardiac enrollees, stratified by clinical characteristics and medication type. Poisson models were used to compute prevalence ratios of the binary measures of use of specific types of medications, among children who were continuously enrolled for the entire given year of life. The dataset was organized at the child-year level, so no offset term was needed to account for differential enrollment time. We follow Zou’s method²² of calculating prevalence ratios and their confidence intervals by using robust standard errors in the Poisson estimation.” We estimated and presented the interaction between presence of a non-cardiac Complex Chronic Condition and cardiac disease severity with adjusted enrollee prevalence.

We performed two sensitivity analyses. First, we restricted our sample to children continuously enrolled in Medicaid for at least 10 years to account for the impact of Medicaid disenrollment on outcomes.²² Second, in analyses restricted to cardiac enrollees, we replaced the Complex Chronic Condition categorization with a registry-based measure of preoperative comorbidities including non-cardiac anatomical abnormalities, chromosomal anomalies, or syndromes. This variable is used as a standard risk-modifier in STS-CHSD studies. Data are manually abstracted from medical records, rather than from billing codes. It is narrower than the Complex Chronic Condition variable, which includes chronic conditions affecting 9 organ systems.^20,21 Analyses were conducted in SAS version 9.4 (SAS Institute, Cary, NC). Western Institutional Review Board approved this study with waiver of consent.

Results

Sociodemographic and Clinical Characteristics

Our sample included 4.5 million unique children (22 million person-years), of whom 5,459 children (32,000 person-years, 0.1%) underwent cardiac surgery (i.e., cardiac enrollees). Cardiac enrollees tended to be younger and were more likely to have non-cardiac Complex Chronic Conditions (Table 1) than non-cardiac enrollees.

Table 1.

Sociodemographic and clinical characteristics of pediatric NYS Medicaid enrollees, by presence of cardiac surgery.

	Enrollees with history of cardiac surgery	Enrollees with history of cardiac surgery, by diagnostic severity*			Enrollees without any history of cardiac surgery
		Critical Single Ventricle Disease	Critical Bi-Ventricular Disease	Non-Critical Cardiac Disease
Unique children (N)	5,459	742	1,486	3,231	4,475,398
Total Person-Years	32,131	4,161	7,851	20,119	21,698,842
Female (per 100 person-years)	45.4	40.5	41.7	47.9	48.7
Age (per 100 person-years)
<1 year	12.5	14.0	17.0	10.5	8.3
1-5 years	44.5	48.7	49.5	41.7	31.0
6-12 years	33.6	31.8	29.3	35.6	36.9
13-17 years	9.4	5.5	4.2	12.2	23.8
Race & Ethnicity (per 100 person-years)
Hispanic	19.5	20.5	19.7	19.3	16.9
Non-Hispanic Asian	8.3	4.9	7.1	9.5	8.0
Non-Hispanic Black	18.7	21.3	16.3	19.1	19.6
Non-Hispanic White	30.9	34.6	33.9	29.0	25.8
Other	22.8	18.8	23.8	23.2	29.7
Non-Cardiac Complex Chronic Condition (per 100 person-years)	75.5	90.1	78.2	71.4	26.4

^*Critical single ventricle disease indicates single ventricle physiology requiring palliation within the first 60 days of life; critical biventricular disease indicates biventricular physiology requiring operation within the first 60 days of life, and non-critical congenital cardiac disease indicates disease requiring initial operation after the first 60 days of life

Chronic Medication Use and Expenditures

Table 2 describes chronic medication use and total medication expenditures for cardiac and non-cardiac enrollees. On average, cardiac enrollees used 1.0 (SD 1.8) chronic medications per person-year, versus 0.3 (SD 0.8) chronic medications per person-year for non-cardiac enrollees. Among cardiac enrollees, 41 per 100 person-years used ≥1 chronic medication and 14 used ≥3 chronic medications. In contrast, among non-cardiac enrollees, 15 per 100 person-years used ≥1, and 3 used ≥3 chronic medications.

Table 2.

Chronic medication use for pediatric NYS Medicaid enrollees, by presence of cardiac surgery.

	Enrollees with a history of cardiac surgery	Enrollees with a history of cardiac surgery, by diagnostic severity*			Enrollees without any history of cardiac surgery
		Critical Single Ventricle Disease	Critical Bi-Ventricular Disease	Non-Critical Cardiac Disease
Mean total number of chronic medications† per person-year (SD); Median [IQR]	1.0 (1.8) 0 [0, 1]	1.9 (2.2) 1 [0, 3]	0.8 (1.4) 0 [0, 1]	0.9 (1.7) 0 [0, 1]	0.3 (0.8) 0 [0, 0]
Rate of enrollees per 100 person-years ‡
with 1+ chronic medications	40.8	65.4	37.7	36.9	14.6
with 3+ chronic medications	14.3	30.9	11.4	12.0	2.6
Mean total Medicaid expenditures§ per person-year (SD); Median [IQR]
including palivizumab	$2,654 ($11,640) $18 [$0, $579]	$4,100 ($10,279) $162 [$2, $2,202]	$2,408 ($7,909) $11 [$0, $592]	$2,451 ($13,024) $10 [$0, $424]	$328 ($5,679) $0 [$0, $13]
excluding palivizumab	$1,544 ($10,812) $12 [$0, $308]	$1,855 ($8,455) $92 [$0, $640]	$976 ($6,172) $4 [$0, $246]	$1,701 ($12,525) $6 [$0, $267]	$307 ($5643) $0 [$0, $12]

^*Critical single ventricle disease indicates single ventricle physiology requiring palliation within the first 60 days of life; critical biventricular disease indicates biventricular physiology requiring operation within the first 60 days of life, and non-critical congenital cardiac disease indicates disease requiring initial operation after the first 60 days of life

^†Denotes the total number of unique chronic medication classes. For example, proton pump inhibitor and H2 blocker use count as two medication classes, while use of two H2 blockers (or two H2 blocker doses) count as a single medication class.

^‡Denotes the number of unique enrollees per 100 person-years.

^§Adjusted to U.S. 2020 U.S. dollars using the Medical Consumer Price Index

Mean total medication expenditures for cardiac enrollees were $2,654 per person-year (SD $11,640), with a median of $18 per person-year (10^th, 90^th percentile: $0, $7,353). For noncardiac enrollees, mean expenditures were $328 per person-year (SD $5,679), with a median of $0 per person-year (10^th, 90^th percentile: $0, $398). A single medication, palivizumab (Synagis), accounted for 42% of total medication expenditures for cardiac enrollees and 6% for non-cardiac enrollees.

Chronic Medication Use by Medication Type and Age Group

Central Figure and Supplemental Table A describe chronic medication use for cardiac and non-cardiac enrollees by medication type and age. On average, use of cardiac/renal agents among cardiac enrollees was 15.8 times (CI: 15.4-16.2, p<0.001) that of non-cardiac enrollees. With baby aspirin excluded, cardiac/renal agents among cardiac enrollees was 13.1 times higher than non-cardiac enrollees. Across all age groups, use of cardiovascular/renal agents was higher among enrollees with critical single ventricle disease (average 55 enrollees per 100 person-years) compared to those with critical bi-ventricular and non-critical disease (17 and 16 enrollees per 100 person-years).

On average, use of gastrointestinal medications among cardiac enrollees was 7.4 times that of non-cardiac enrollees (CI: 7.1-7.7; p<0.001). Use of respiratory medications among cardiac enrollees was 1.9 times that of non-cardiac enrollees (CI: 1.9-2.0; p<0.001). Use of central nervous system medications was slightly higher for cardiac enrollees, but less pronounced than for other medication groups—only 20% higher (prevalence ratio 1.2, CI: 1.21-3; p<0.001) compared to non-cardiac enrollees. Use of central nervous system medications peaked in adolescence (13-1703 years) at 13 users per 100 person-years for cardiac enrollees and 8 per 100 person-years for non-cardiac enrollees. Central nervous system medication use did not differ by cardiac disease severity. During the first year of life, central nervous system medication use was driven by anticonvulsants (99% of cardiac enrollees; 94% of non-cardiac enrollees), by stimulants (54% of cardiac enrollees; 75% of non-cardiac enrollees) in the school-age years, and by non-stimulant psychotropics, such as antidepressants and antipsychotics, (58% of cardiac enrollees; 60% of non-cardiac enrollees) in adolescence.

Relationships between Chronic Medication Use and Expenditures and Enrollee Characteristics among Cardiac Enrollees

Table 3 describes relationships between chronic medication use, expenditures, and sociodemographic and clinical characteristics among cardiac enrollees. In adjusted models, increased chronic medication use and expenditures were associated with worse cardiac disease severity and the presence of non-cardiac Complex Chronic Conditions. Compared to children with non-critical disease, children with critical single ventricle disease had 2.7 times greater odds per person-year of using ≥1 chronic medication (CI 2.1-3.5) and ≥3 chronic medications (CI 2.03.5). Mean total medication expenditures for children with single ventricle disease was 3.0 times higher than enrollees with non-critical cardiac disease (CI 2.0-4.4). There were no differences in chronic medication use or expenditures between critical bi-ventricular versus non-critical enrollees.

Table 3:

Predictors of chronic medication use and expenditures after cardiac surgery among NYS Medicaid pediatric enrollees^*

	Use of ≥1 Chronic Medications†			Use of ≥3 Chronic Medications†			Chronic Medication Expenditures
	OR‡	95% CI	p-value	OR‡	95% CI	p-value	Ratio§	95% CI	p-value
Cardiac Disease Severity (non-critical)
Critical Single Ventricle	2.71	(2.10, 3.50)	<0.001	2.68	(2.04, 3.53)	<0.001	2.98	(2.03, 4.36)	<0.001
Critical Bi-Ventricular	0.90	(0.79, 1.02)	0.105	0.86	(0.69, 1.08)	0.183	0.81	(0.65, 1.01)	0.059
Female	1.12	(1.01, 1.23)	0.024	1.11	(0.98, 1.25)	0.113	1.19	(1.04, 1.37)	0.014
Age (6-12 years)
<1 year	2.91	(2.61, 3.25)	<0.001	1.96	(1.70, 2.27)	<0.001	9.88	(7.87, 12.41)	<0.001
1-5 years	1.00	(0.91, 1.08)	0.9173	0.92	(0.82, 1.04)	0.1808	1.30	(1.13, 1.51)	<0.001
13-17 years	1.60	(1.43, 1.80)	<0.001	2.08	(1.82, 2.37)	<0.001	1.74	(1.42, 2.12)	<0.001
Race and Ethnicity (Hispanic)
Non-Hispanic Asian	0.73	(0.59, 0.90)	0.004	0.68	(0.49, 0.95)	0.024	0.67	(0.46, 0.98)	0.037
Non-Hispanic Black	0.74	(0.63, 0.85)	<0.001	0.76	(0.59, 0.96)	0.022	0.69	(0.52, 0.90)	0.006
Non-Hispanic White	0.91	(0.81, 1.04)	0.161	0.78	(0.63, 0.95)	0.016	0.88	(0.70, 1.11)	0.280
Other	0.74	(0.64, 0.85)	<0.001	0.68	(0.55, 0.85)	<0.001	0.59	(0.46, 0.75)	<0.001
Non-Cardiac Complex Chronic Condition	2.63	(2.33, 2.96)	<0.001	5.04	(4.11, 6.16)	<0.001	5.37	(4.52, 6.38)	<0.001

^*All models include surgical center and year of surgery fixed effects.

^†Denotes the total number of unique chronic medication classes. For example, proton pump inhibitor and H2 blocker use count as two medication classes, while use of two H2 blockers (or two H2 blocker doses) count as a single medication class.

^‡Logistic regression coefficients are expressed odds ratios for use of ≥1 (or ≥3) chronic medication class. For example, critical single ventricle patients have 2.71 times higher odds of using ≥1 chronic medication use compared to patients with non-critical disease, after adjusting for sex, age, race/ethnicity, complex chronic condition, surgical center and year of surgery.

^§Log-linear regression coefficients are expressed as ratios of average expenditures. For example, critical single ventricle patients have 2.98 times higher average chronic medication expenditures compared to patients with non-critical disease, after adjusting for sex, age, race/ethnicity, complex chronic condition, surgical center and year of surgery.

In an interaction analysis, the presence of a Complex Chronic Condition (CCC) had a smaller influence on chronic medication use among critical single ventricle patients than it did among those with critical bi-ventricular or non-critical disease (interaction term p=0.002). The adjusted prevalence of using ≥1 chronic medication among enrollees with critical single ventricle and CCC was 66 enrollees/100 person-years, versus 58 enrollees/100 person-years among enrollees with critical single ventricle disease and no CCC. The spread in adjusted prevalence between those with versus without a CCC was higher among enrollees with critical bi-ventricular disease (with versus without a CCC: 41 versus 19 enrollees/100 person-years) or non-critical disease (with versus without a CCC: 43 enrollees versus 22 enrollees/100 person-years).

Among cardiac enrollees, chronic medication use and expenditures also varied by patient demographics. Both were highest among infants (<1 year) and adolescents (13–17 years), even after excluding palivizumab (Synagis) (Table 3). Compared to school-age children, infants had 2.9 (CI 2.6-3.2) and adolescents had 1.6 times greater odds (CI 1.4-1.8) of using ≥1 chronic medication. Compared to school-age children, infants had 2.0 (CI 1.7-2.3) and adolescents had 2.1 times greater odds (CI 1.8, 2.4) of using ≥3 chronic medications. Hispanic children had higher chronic medication use and expenditures compared to all racial and ethnic groups, except for non-Hispanic White. Compared to Hispanic children, non-Hispanic Black children had 26% (CI: 15-37%) lower odds per person-year of using ≥1 chronic medication, 24% (CI: 4-51%) lower odds per person-year of using ≥3 chronic medications, and 31% (CI: 10-48%) lower total medication expenditures. Compared to Hispanic children, non-Hispanic Asian children had 27% (CI: 10-41%) lower odds per person-year of using ≥1 chronic medication, 32% (CI: 5-51%) lower odds per person-year of using ≥3 chronic medications, and 33% (CI: 2-54%) lower total medication expenditures.

In sensitivity analyses, these patterns of medication use were also observed when restricting to continuously Medicaid-enrolled (Supplemental Table B). Replacing the ICD-based Complex Chronic Condition (CCC) categorization with the registry-based comorbidity variable yielded similar coefficient estimations, though the registry comorbidity measure accounted for slightly less model variation than did CCC (Supplemental Table C).

Discussion:

We evaluated chronic medication use and total medication expenditures across childhood among NYS Medicaid enrollees who underwent cardiac surgery compared to the rest of the NYS pediatric Medicaid population. We found that the prevalence of chronic medication use in the NYS general pediatric Medicaid population was similar to prevalence of chronic medication use in a prior study with a national Medicaid population (15% versus 18% of patients, respectively).⁵ Chronic medication use was significantly higher for cardiac enrollees, which persisted beyond infancy, into adolescence. Critical single ventricle disease, presence of extra-cardiac comorbidity, younger age, and Hispanic ethnicity were associated with greatest use. Although use of most medication types was higher among cardiac enrollees, use of central nervous system medications was more similar between cardiac and non-cardiac enrollees. (Central Illustration). More than 40% of chronic medication expenditures for cardiac enrollees were attributable to palivizumab.

Chronic medication use is a quantifiable metric that can serve as a proxy to understand disease burden in a way that intermittent doctors’ appointments, emergency department visits, and inpatient stays cannot capture.^23,24 Our data reveal that, at any given time, chronic medication use among cardiac enrollees was 4 to 7 times that of non-cardiac enrollees; medication expenditures among cardiac enrollees were 10 times higher than the rest of the pediatric Medicaid population.

These differences in chronic medication use persisted across childhood, regardless of disease severity. Nearly 5 in 10 adolescents on NYS Medicaid who underwent cardiac surgery were still taking at least one chronic medication, versus less than 2 in 100 adolescents in the general pediatric population. In addition, while chronic medication use was highest among children with critical single ventricle disease, 4 in 10 children with either critical biventricular or non-critical cardiac used ≥1 chronic medication at any given time. Thus, even for children with a distant history of cardiac surgery, or with less severe cardiac disease, pediatricians and subspecialists must remain vigilant and weigh the psychological burden and potential risks (e.g. errors, polypharmacy) associated with prescribing “yet another” chronic medication, against the indication.

The probability of chronic medication use among cardiac enrollees increased with a non-cardiac Complex Chronic Condition (CCC), especially for children with less severe (non-critical or critical biventricular) cardiac disease. This interaction between CCC and cardiac disease severity might reflect similarities in medication types—or medicalization rates—among patients with CCCs and critical single ventricle disease. Providers can be proactive about the impacts of CCCs on less severe cardiac disease through early collaboration between subspecialists and pediatricians, early anticipatory guidance (as early as the fetal period), and direction of resources toward long-term care.

Our data also revealed differences by race and ethnicity among cardiac enrollees. NonHispanic Black and non-Hispanic Asian children used chronic medications significantly less than Hispanic children. Whether these variations represent disparities in health status or in the care delivered is unclear. However, we know there are race- and ethnicity-based disparities in clinical outcomes for this population. We have previously shown that Hispanic and non-Hispanic Black children spend more days per year in hospitals and emergency rooms and fewer days in primary care offices than non-Hispanic White children²² and mortality has repeatedly been demonstrated to be higher for non-Hispanic Black versus Hispanic and non-Hispanic White children.^2,22 Further elucidating associations between care utilization and longitudinal outcomes might lead to understanding of causal mechanisms underlying these disparities.

Although increased prevalence of neuropsychiatric disorders in those with congenital heart defects is known,^27,28 prevalence of psychotropic medication use in this population has yet to be quantified. In adolescents, we found a 20% higher prevalence of chronic central nervous system medication use among cardiac enrollees compared to non-cardiac enrollees. However, with an estimated prevalence of neuropsychiatric disease in the cardiac surgery population ranging 19-46%,^27,28 we would have expected the gap in neuropsychiatric medication use to be larger. On one hand, the similar observed rates of neuropsychiatric medication use among cardiac and non-cardiac enrollees might reflect improved access to psychiatric medications for all publicly insured children, secondary to targeted efforts by Medicaid programs over the past two decades, including pediatrician training programs and live pediatric psychiatrist consultations for physicians and patients.^29,30 On the other hand, this finding might reflect a higher prevalence of psychiatric disorders in the non-cardiac Medicaid population.²⁵ It might also reflect differences in foci on or comfort with pharmacologic treatment of medical, rather than psychiatric, diseases for cardiac enrollees. No matter the explanation, potential provider oversight of mental healthcare needs in the face of physical healthcare needs among children with cardiac disease is a potential target for care improvement.

We note that a single medication, palivizumab, was responsible for 40% of expenditures among cardiac enrollees. Palivizumab is a monthly injection used to prevent severe disease from respiratory syncytial virus. In 2019, $3,783 was spent per Medicaid enrollee per monthly injection, for a total of more than $13 million, on this drug alone. Due to these high costs, eligibility criteria for palivizumab are modified yearly based on cost-effectiveness.²⁶ The case of palivizumab demonstrates the utility of assessing multiple measures of utilization—filled prescriptions and total expenditures—and that patient-level medication burden may be different than the system-level cost burden.

Limitations

There are several limitations to our study. First, we assessed filled prescriptions and refills as proxies for use. While 90-day filled prescriptions imply use, prescribing behaviors and patient adherence could bias (i.e. confound) or lie in the causal pathway (i.e. mediate) associations between prescriptions and use. Second, while we capture total medication expenditures, we do not include out-of-pocket expenses. Third, our results may not generalize to patients privately insured or residing outside NYS. That said, more than 50% of pediatric cardiac surgical patients nationwide are covered by Medicaid.^11,27 Fourth, sicker patients are more likely to be enrolled in Medicaid longer. However, sensitivity analyses on continuously enrolled patients did not meaningfully change our conclusions. Fifth, sample sizes for some racial and ethnic groups were small and collapsed as “Other”; a larger multistate sample might facilitate a disaggregation.

Conclusions:

While survival after cardiac surgery has improved, children who undergo cardiac surgery experience a high medication burden that persists throughout childhood. Compared to the rest of the NYS pediatric Medicaid population, children who underwent cardiac surgery were 4 and 7 times as likely to use ≥1 and ≥3 chronic medications per person year and had average chronic medication expenditures that were 10 times higher per person year. Understanding this burden can inform anticipatory guidance and policies directed at improving the value of care.

Clinical Perspectives:

Competency in Systems-Based Practice:

Pediatric patients who have undergone surgery for congenital heart disease require many more medications than patients without cardiac disease, entailing costs 10 times higher than the rest of the pediatric Medicaid population.

Translational Outlook:

More research is necessary to understand the relative contributions of pharmacologic treatment of medical versus psychiatric disease in pediatric Medicaid enrollees with and without heart disease.

Abbreviations:

CCC

pediatric Complex Chronic Condition

CHS-COLOUR

Congenital Heart Surgery Collaborative for Longitudinal Outcomes and Utilization of Resources

NYS

New York State

NYS-PCCS

New York State Pediatric Congenital Cardiac Surgery Database

STS-CHSD

Society of Thoracic Surgeons-Congenital Heart Surgery Database