Congenital heart defects (CHDs) are a relatively common birth defect affecting approximately 8 to 10 per 1000 infants (Reller et al., 2008; Bjornard et al., 2013). Survival among infants born with CHD has improved over time (Oster et al., 2013). Currently, it is estimated that more adults (over the age of 18 years) than children are living with a CHD in the United States (Warnes et al., 2001); moreover, the number of hospitalizations of adults with CHD is increasing compared with hospitalizations of children with CHD (O’Leary et al., 2013). The medical needs of patients with CHD beyond infancy are largely unknown. A better understanding of the patient-level hospital resource use of children and adults with CHD is needed.

At the hospital level, hospitalizations of individuals with CHD have high costs in both pediatric and adult admissions (Connor et al., 2005; Opotowsky et al., 2009; O’Leary et al., 2013; Simeone et al., 2014). Additionally, hospital costs are influenced by age, CHD complexity, payer status, hospital or surgical center, and complications during treatment (Connor et al., 2005; Benavidez et al., 2007; Pasquali et al., 2011, 2014). A hospital-level analysis of hospitalization costs for CHDs found that across the life-span CHDs accounted for approximately 0.4% of hospitalizations and 2.2% of hospital costs (Simeone et al., 2015).

To address the gap in knowledge, we aim to describe the patient-level hospital use and costs for individuals aged 1 to 64 years with a CHD diagnosis and to identify factors associated with higher hospital costs and multiple hospitalizations among patients with CHDs admitted to Arkansas hospitals between 2006 and 2011. This builds on earlier work to examine hospital-level costs for Arkansas patients of all ages with CHDs (Simeone et al., 2015).

After excluding hospitalizations of infants (n = 269,809), of patients ≥ 65 years of age (n = 947,229), and of out of state residents (n = 63,107), as well as hospitalizations missing data on age at admission (n = 1193), hospital charges (n = 2189), residence (n = 275), length of stay (n = 224), payer status (n = 15,284), and PI (n = 38,118), there were 1,185,868 hospitalizations available for patient-level analysis. These hospitalizations were accumulated by 603,925 individual patients, 2542 (0.42%) of whom had at least one hospitalization with a CHD diagnosis (Supp. Fig. 2).

The total 2006 to 2011 hospital costs of the 2542 individuals aged 1 to 64 years were $126,999,837 and these patients accumulated 7898 hospitalizations with and without a CHD diagnosis between 2006 and 2011. Of those 7898 hospitalizations, 4866 (61.6%) had no CHD diagnosis, 761 (9.6%) included the CHD as the principal diagnosis, and 2271 (28.5%) included the CHD as a secondary diagnosis. Average per-patient cost was $49,961 (standard error [SE]: 1626). Median per-patient hospital costs for all hospitalizations were $28,482 (interquartile range [IQR]: $10,899, $53,530). Among hospitalizations of individuals ever admitted with CHD but with no CHD diagnosis, per-hospitalization mean costs were $10,289 (SE = 322); median costs were $5505 (IQR = $3174, $10,097). Hospitalizations in which the CHD was the principal diagnosis had mean per-hospitalization costs of $36,736 (SE = 1587) and median costs of $29,104 (IQR = $12,722, $44,211). Hospitalizations in which the CHD was a secondary diagnosis had mean per-hospitalization costs of $17,433 (SE = 920) and median costs of $6651 (IQR = $3482, $17,364).

Thirty-eight percent of patients ever admitted with a CHD between 2006 and 2011 were 46 to 64 years old at their first admission (Table 1). Patients with CCHDs comprised 13.4% of all individuals ever hospitalized with CHD over the study period. Of patients ever admitted with a CHD between 2006 and 2011, 45.5% had an expected private payer, 49.1% experienced multiple inpatient hospitalizations within 365 days, 38.6% had ever undergone a prior cardiac procedure, and 28.8% of patients were ever admitted with a CHD as the principal diagnosis. By demographic characteristic, children aged 1 to 3 years at their first admission during this time period had more costly hospitalizations compared with patients at older ages at their first admission; non-Hispanic black patients had more costly hospitalizations than patients of other races; patients with an expected public payer had more costly hospitalizations than patients with private or other expected payers; patients ever admitted with a CHD as the principal diagnosis had more costly hospitalizations than patients who were never admitted with a CHD as the principal diagnosis.

In an adjusted linear regression, several patient factors were associated with increased hospital costs (Table 2). Hospital costs were nearly 1.5 times higher in those with a CCHD compared with those without a CCHD. Multiple hospitalizations within 365 days more than doubled hospital costs compared with patients with no more than one hospitalization within any 365-day period. A history of at least one cardiac procedure during the study period nearly tripled hospital costs. Cardiac procedures were associated with higher hospital costs at all ages (Supp Table 2). Cardiac procedures were more common among patients less than 18 years of age compared with patients 18 to 64 years of age. Similarly, mean per-patient hospital costs were highest in pediatric hospitalizations. Among patients aged 1 to 3 years, hospital costs of patients with a cardiac procedure were over 4 times higher than hospital costs of patients of the same age without a procedure (Supp Table 2). Older age at first admission and female sex were both associated with decreased per-patient costs (Table 2).

In an adjusted logistic regression, patients aged 1 to 3 years and 18 to 45 years at their first admission were 1.84 and 1.63 times more likely to experience multiple hospitalizations within 365 days compared with patients aged 4 to 17 years at their first admission (Table 3). Elevated ORs for multiple hospitalizations within 365 days were also observed among patients with CCHDs, patients with an expected public payer, and patients with a greater number of comorbid conditions. A history of at least one cardiac procedure during the study period was not associated with multiple admissions.

Among all patients ever admitted with a CHD, 1750 (68.8%) patients were admitted with at least one Elixhauser comorbidity measure (Table 1). Hypertension-related comorbidities were most common (40.2% of patients), followed by fluid and electrolyte disorders (23.3% of patients) and chronic pulmonary disease (18.8% of patients) (Table 4). Patients ever diagnosed with congestive heart failure had the highest mean per-patient costs and number of hospitalizations. Having multiple admissions within 365 days was most strongly associated with the presence of congestive heart failure (OR=7.5, 95% confidence interval [CI], 5.2–10.2), depression (OR=6.4, 95% CI, 4.6–8.9), and deficiency anemias (OR=5.8, 95% CI, 4.3–7.7).

When a sensitivity analysis was performed excluding individuals whose hospital costs might have been primarily due to other illnesses and not a CHD (n = 524), total per-patient costs among individuals whose hospitalizations were not due to other illnesses were $98,067,250; mean (SE) costs were $48,596 ($1832) with median (IQR) costs of $28,218 ($10,540–$51,650). The total number of hospitalizations in the remaining 2018 patients decreased to 5265 (mean 2.6 [0.7] hospitalizations per patient). Adjusted linear regression of ln(cost) associations and estimated adjusted ORs were similar to those observed in the full analysis (data not shown).

Arkansas children and adults with CHDs experience variation in hospital use and costs by demographic factors and comorbidities. Adults over the age of 45 at first hospitalization between 2006 and 2011 accounted for the largest patient group, but infants and children aged 1 to 3 years had the highest average per-patient costs. Approximately half of patients with a CHD had multiple hospitalizations within 365 days of each other. After adjusting for patient demographic characteristics, factors associated with increased costs were hospital readmission within 1 year, presence of a CCHD, cardiac procedures, and increased number of comorbidities. Additionally, having multiple hospitalizations within 365 days was associated with having a CCHD, having an expected public payer, and having additional comorbidities. Congestive heart failure was associated with high costs and multiple admissions within a 365-day period. While presence of a cardiac procedure at any age increased costs, cardiac procedures occurring in childhood had the largest impact on costs.

Previous studies have demonstrated higher costs and associated resource usage in those patients with CHDs whose hospitalizations included CCHDs (Mackie et al., 2007; Simeone et al., 2014), hospitalizations during infancy and childhood (Connor et al., 2005; Simeone et al., 2014), complex surgical procedures (Connor et al., 2005; Pasquali et al., 2011), or in which a complication occurred (Benavidez et al., 2007). Our results add to the literature by demonstrating that individuals with multiple hospitalizations account for a meaningful proportion of patients with CHD and are associated with greater hospital costs.

Adults with a CHD diagnosis comprise a large proportion of the Arkansas inpatient population with CHD. This pattern is seen at both the patient and hospital level (Simeone et al., 2015). This observation also follows trends of increasing numbers of adult hospitalizations with CHD, which are likely the result of improvements in diagnosis and treatment in infancy and childhood (Warnes et al., 2001; Marelli et al., 2007, 2014; Oster et al., 2013). Understanding the medical needs of this growing patient population will impact resource allocation and planning within hospital and healthcare systems.

This study was limited by several factors. Inclusion and exclusion criteria were developed to be inclusive; patients were included if at least one hospital discharge during this 6-year period included a diagnosis of any CHD. This may have led to an overestimation of costs and patient hospitalizations, particularly among those individuals who were high resource users, but had very few hospitalizations that included a CHD diagnosis. In a sensitivity analysis, we excluded patients whose primary morbidity seemed unlikely be a CHD; total costs and number of hospitalizations were reduced by 22.8% and 33.3%, respectively, but associations between costs, multiple hospitalizations, and patient characteristics did not change. Additionally, our study was constrained by inherent limitations of administrative datasets. ICD-9-CM codes were used to identify any hospitalization and patient for whom a CHD diagnosis was ever coded and CHDs are prone to misclassification (Frohnert et al., 2005; Strickland et al., 2008). Patients might have been misclassified as having or not having a CHD depending on what was recorded on discharge summaries; this is particularly important for the adult population with CHD, where the ICD-9-CM code may be less likely to be noted and nonspecific cardiovascular disease codes might be used instead. Due to lack of encrypted PI among infants, we were unable to estimate patient-level infant costs. Infants accounted for a large proportion of the hospitalizations and costs for CHDs in Arkansas and nationwide (Connor et al., 2005; Simeone et al., 2014, 2015); excluding hospitalizations of patients less than a year old underestimated patient-level CHD costs, particularly for patients with complex surgeries occurring in the first year of life (Moons et al., 2009; Burstein et al., 2011; Pasquali et al., 2011). Finally, cost data in the SID may be underestimated, as physician fees are not included in the discharge summary (Rogowski, 1998).

Despite these limitations, we were able to analyze patient-level data over multiple years from a large, population-based, all-payer system. The availability of patient-level identifiers allowed an investigation of per-patient characteristics and costs, which are often unavailable with administrative data.

In an analysis of state inpatient hospital costs, patients with CHDs were more likely to have higher costs when they had a more severe CHD, comorbidities, cardiac procedures, or multiple hospitalizations within 365 days. CCHDs, presence of comorbidities, and having a public payer were associated with increased odds of multiple stays within 365 days. Future research should investigate factors associated with multiple hospitalizations, need for cardiac procedures, and comorbidities in the CHD population, as these appear to be strongly associated with hospital costs.