Abstract
BACKGROUND: High-cost hospitalizations (HCHs) account for a substantial proportion of pediatric health care expenditures. We aimed to (1) describe the distribution of pediatric HCHs across hospital types caring for children and (2) compare characteristics of pediatric HCHs by hospital type.
METHODS: Cross-sectional analysis of all pediatric hospitalizations in the 2012 Kids’ Inpatient Database. HCHs were defined as costs >$40 000 (94th percentile). Hospitals were categorized as children’s, small general, and large general.
RESULTS: Approximately 166 000 HCHs were responsible for 50.8% of aggregate hospital costs ($18.1 of $35.7 billion) and were mostly at children’s hospitals (65%). Children with an HCH were largely neonates (45%), had public insurance (50%), and had ≥1 chronic condition (74%). A total of 131 children’s hospitals cared for a median of 559 HCHs per hospital (interquartile range [IQR]: 355–1153) compared to 76 HCHs per hospital (IQR: 32–151) at 397 large general hospitals and 5 HCHs per hospital (IQR: 2–22) at 3581 small general hospitals. The median annual aggregate cost for HCHs was $60 million (IQR: $36–$135) per children’s hospital compared to $6.6 million (IQR: $2–$15) per large general hospital and $300 000 (IQR: $116 000–$1.5 million) per small general hospital. HCHs from children’s hospitals encompassed nearly 5 times as many unique clinical conditions as large general hospitals and >30 times as many as small general hospitals.
CONCLUSIONS: Children’s hospitals cared for a disproportionate volume, cost, and diversity of HCHs compared to general hospitals. Future studies should characterize the factors driving cost, resources, and reimbursement practices for HCH to ensure the long-term financial viability of the pediatric health care system.
Health care accounts for a large proportion of the gross domestic product of the United States, and hospitalization costs account for a large amount of health care spending.1,2 Within pediatric health care, hospital inpatient care accounted for >40% of expenditures in 2010.3 In recent years, inpatient health care spending grew at a faster rate for patients <17 years old than for any other age group.4
Children’s hospitals treat ∼30% of all pediatric hospitalizations (excluding normal newborns)5 yet care for nearly half of children hospitalized with the highest illness severity and over half of those with complex chronic conditions (CCCs).6 Patient subgroups with disproportionately high inpatient expenditures include children with medical complexity (representing 10% of hospitalizations and 40% of hospital charges7) and preterm infants (accounting for 8% of hospitalizations and 27% of costs8). Although authors of these previous studies of specific clinical subgroups have included their associated hospitalization costs,7,8 pediatric inpatient spending is rarely examined specifically by high-cost hospitalizations (HCHs). Pediatric HCHs represent a distinct type of hospitalization that generates significant costs in pediatrics. The distribution of pediatric HCH by hospital type and the characteristics of patients driving HCHs have not previously been explored. Understanding the distribution of HCHs among different hospital types in which children receive care is crucial to better understand hospitals’ financial risk, the hospital resources required, and the potential impact of payment reform policies.
Using a nationally representative data set to analyze the cost to each hospital of providing care, we aimed to (1) describe the distribution of HCHs across various hospital types that care for children and (2) compare the characteristics of HCHs by hospital type. As a driver of cost, the HCH distribution across hospital types has implications for the sustainability of the pediatric health care system.
Methods
Study Population, Design, and Data Source
This cross-sectional, observational analysis of weighted data for pediatric inpatient hospitalizations in the United States used the 2012 Kids’ Inpatient Database (KID), which is maintained by the Agency for Healthcare Research and Quality as part of the Healthcare Resource and Utilization Project (HCUP). We included all discharges for patients <21 years of age and excluded hospitalizations for normal newborn deliveries (identified by using All Patient Refined Diagnostic Related Groups [APR-DRGs], ie, groups 626 [neonate birth weight of 2000–2499 g, normal newborn or neonate with other problems] and 640 [neonate birth weight >2499 g, normal newborn or neonate with other problems]).4 Thus, we use “neonate” to refer to all nonroutine birth hospitalizations. The KID contains information on patient demographics, hospital characteristics, diagnoses, procedures, and resource use, including length of stay and total charges.4 Hospital charge information was reported in the KID as the total charge for the hospitalization. The HCUP Cost-to-Charge Ratio Files provided hospital-specific cost-to-charge ratios allowing for the conversion of charges to estimated inpatient cost. “Cost” estimates the expense incurred by the hospital to deliver health care services, which is a more standardized comparator than charges. The Institutional Review Board at Baylor College of Medicine determined this study did not constitute human subjects research.
HCHs
The primary outcome was HCH as defined by costs >94th percentile. In modeling the total cost distribution across all hospitalizations, the HCH cutoff was set at $40 000.
Hospital-Type Classifications
The primary independent variable was hospital type as determined by the number of pediatric (<21 years old) discharges per year and children’s hospital status. Children’s hospitals were defined on the basis of the number of unique diagnostic related groups in their patient population being >334, as developed in previous work (J.C. Gay, MD, MMHC, J. Rodean, MPP, T. Richardson, MS, MPH, PhD, et al, unpublished observations). The children’s hospital category includes both freestanding and nonfreestanding children’s hospitals. Hospitals were then categorized as small general (<900 pediatric discharges per year), large general (≥900 pediatric discharges per year), or children’s hospital of any volume. The cutoff point for large general hospitals was set at the 90th percentile, and 900 pediatric discharges per year represents the 90th percentile of volume for general hospitals in the data set.
Patient and Hospital Characteristics
Patient characteristics included age, payer type, severity index, and number of CCCs. Age was divided into neonate (birth hospitalization, nonroutine), infant (nonbirth hospitalization, 0–1 year), 1 to 4 years (preschool), 5 to 9 years (school age), 10 to 14 years (preadolescent), and 15 to 20 years (adolescent). Age was determined at the time of admission. The neonate category was determined on the basis of the APR-DRG (Supplemental Table 4). The “nonneonate” category is used to describe nonbirth hospitalizations for all age groups collectively. Payer types were classified as private, public (including Medicare, Medicaid, Title V, or other government insurance), self-pay (including uninsured patients and patients who paid directly for their services), and other (including no-charge patients). The severity index is a measure of risk of mortality, hospital severity of illness, and hospital resource intensity and was determined by using pediatric-specific weights that were based on the APR-DRG and severity of illness assignment (hospitalization resource intensity scores for kids).9 CCCs were defined as medical conditions expected to last ≥12 months and involve either several organ systems (including respiratory, renal, and others) or 1 organ system severely enough to require specialty pediatric care and potentially hospitalization in a tertiary care center and were identified by using validated International Classification of Diseases, Ninth Revision, Clinical Modification codes.7,10–13
Hospital characteristics included the median number of unique APR-DRGs for pediatric HCH discharges and transfer status. The median number of unique APR-DRGs was examined to understand the diversity of conditions and services at each hospital type. Similarly, we examined the number of pediatric transfers into and out of the hospital as a marker of hospital service capacity. “Transfer in” indicates whether the point of origin for the admission was from another facility.14 “Transfer out” indicates whether the final hospital disposition was a transfer to another facility.14
Analysis
We described the patient characteristics of HCHs across hospital types. We calculated the proportion of total discharges and costs attributable to HCHs across all hospitals nationally and by hospital type. Further analysis was done to compare neonatal and nonneonatal HCH discharges and costs because of the heterogeneity in patient and hospitalization factors between these 2 groups. We characterized the 10 most common conditions (using APR-DRGs) resulting in HCH among neonates and nonneonates, procedure versus nonprocedure, and subspecialty services.
We summarized categorical variables using frequencies and percentages and continuous variables using medians and interquartile ranges (IQRs). Data weights provided by the HCUP were used to generate national estimates, and weighted data were used in all statistical analyses.14 Comparisons for all patient and hospital characteristics by hospital type and for HCH by hospital type were conducted by using the χ2 test for categorical data and Kruskal-Wallis tests for continuous data. P values <.01 were deemed statistically significant. Analyses were performed by using SAS 9.4 (SAS Institute, Inc, Cary, NC).
Results
Resource Use
Of the 2 649 759 pediatric hospitalizations during the study period (excluding normal newborn births), 165 779 (6.3%) were HCHs (Table 1). HCHs accounted for 50.8% of pediatric inpatient costs nationwide. Approximately two-thirds (65%) of the nation’s HCHs were cared for within children’s hospitals (Supplemental Fig 2), resulting in annual aggregate costs of $12.6 billion across all children’s hospitals (70% of all HCH costs nationally). In 2012, children’s hospitals cared for a median of 559 HCH discharges per hospital compared to 76 per large general hospital and 5 per small general hospital (Table 1). The median HCH cost per hospital was $59.5 million for children’s hospitals, which was ninefold higher than the HCH costs per large general hospitals ($6.5 million) and 191-fold higher than small general hospitals ($311 454).
Volume and Cost Distribution of HCHs (>$40 000) Across Hospital Types
Children’s hospitals cared for 47% of neonatal HCHs and 79% of nonneonatal HCHs (Table 2). Among nonneonatal HCHs, children’s hospitals cared for a median of 350 HCH discharges per hospital compared to a median of 18 at large general hospitals and 3 at small general hospitals. The median cost of nonneonatal HCHs per hospital was $29.8 million for children’s hospitals, compared to $1.4 million at large general hospitals and $171 429 at small general hospitals.
Volume and Cost Distribution of HCHs (>$40 000) Across Hospital Types by Neonatal Status
Hospitalization Characteristics
Of all HCHs, 45% were attributable to neonates, 50% were insured by public payers, and 74% had ≥1 CCC (Table 3). Children’s and large general hospitals had a similar distribution of payer types, whereas small general hospitals had a higher proportion of private payers than any other hospital type. The median patient severity index of HCHs was higher at large general than other hospital types, whereas the proportion of HCH patients with ≥1 CCC was highest at children’s hospitals (82% at children’s versus 59.7% at large general hospitals).
Distribution of HCH (>$40 000) Across Hospital Types by Age, Payer, Illness Severity, and Disease Complexity
Compared with those at general hospitals, HCH patients at children’s hospitals encompassed a broader array of APR-DRGs. Among neonates, HCHs at children’s hospitals had a median of 21 unique APR-DRGs per hospital, compared to 14 and 6 for large and small general hospitals, respectively. Among nonneonates, HCHs at children’s hospitals had a median of 77 unique APR-DRGs per hospital, compared to 9 and 2 for large and small general hospitals, respectively. Over 25% of HCHs in children’s hospitals were transferred in from other facilities, compared to ∼10% of HCHs transferred into large and small general hospitals. Of the total 35 115 HCH patients transferred into hospitals, 82% were to children’s hospitals.
Distribution of Common High-Cost Cases
We examined the most common HCH by hospital type and neonate status (Fig 1). Neonatal HCHs were primarily distributed among large general and children’s hospitals, although the top neonatal HCHs requiring procedures were predominately cared for in children’s hospitals. Nonneonatal HCHs were mostly (70.2%–97.7%) treated in children’s hospitals.
The most common HCH case distribution by hospital type and neonate status. Numbers indicate the percentage of HCHs at each hospital by condition. Colors indicate the percentage of discharges; the highest percent is red, and the lowest percent is green relative to hospitals within a category. Gray shading indicates APR-DRG with procedure. ECMO, extracorporeal membrane oxygenation; MV, mechanical ventilation; RDS, respiratory distress syndrome.
Across neonatal and nonneonatal conditions, almost all (>95%) of HCH patients were cared for at children’s hospitals for the following conditions: major cardiothoracic repair of heart anomaly, cardiac valve procedures without cardiac catheterizations, and neonates >2499 g with major cardiovascular procedures.
Discussion
In this study of HCHs across a breadth of pediatric hospitalizations nationally (encompassing 2 649 759 weighted discharges from 4109 hospitals), we found that HCHs account for 51% of pediatric inpatient costs and an annual national aggregate cost of $18.1 billion. Although over half of all discharges (56%) were cared for at general hospitals (as is consistent with previous literature),5 approximately two-thirds of the nation’s pediatric HCH discharges and costs were at children’s hospitals.
Further examination into the 131 children’s hospitals included in the study revealed a higher median number of discharges and costs per children’s hospital compared to other hospital types. This distribution of care and the associated costs raise important questions about the financial risk associated with HCHs posed to specific types of hospitals. Although children’s and large general hospitals had a similar distribution of payer types (with ∼50% publicly insured patients), children’s hospitals were exposed to higher financial risk because of their higher total number of HCHs (108 000 at children’s versus 40 000 at large general hospitals). Recent estimates reveal a potential 30% net loss in caring for hospitalized patients with Medicaid.15 On the basis of this literature and calculations displayed in Supplemental Table 5, we estimate that caring for HCHs with public payers amounts to ∼$9 million annual loss per children’s hospital, compared to a loss of ∼$1 million in large general hospitals and $43 354 in small general hospitals. This is similar to previous literature revealing estimated losses across all pediatric Medicaid hospitalizations of >$10 million per hospital in freestanding children’s hospitals compared to <$50 000 per hospital in nonchildren’s, nonteaching hospitals.15 These losses generate questions regarding the margins at which hospitals operate and the ability of large general and children’s hospitals to absorb these losses. Children’s hospitals serve as a critical safety net for children in poverty, and large financial losses in children’s hospitals could jeopardize care to vulnerable populations. Conversely, if large general hospitals are unable to absorb the losses in caring for pediatric HCHs, this could result in a further shift in care toward children’s hospitals. Consequently, it is important to gain a better understanding of the factors contributing to costs in HCHs and ensure sustainable sources of care for vulnerable populations, including children with Medicaid.
Understanding hospital costs associated with HCHs is especially important in an era of increasing adoption of value-based, pay-for-performance (P4P) models. It is unclear if P4P may create an even larger financial liability for hospitals caring for a disproportionate number of HCHs.16 Although the need to reduce overall health care spending is critical, the impact of new payment models may have a differential impact across different hospital types and patient populations. Payment models that seek to primarily lower costs may be misaligned with the goal of delivering high-quality care to medically complex, critically ill, or extremely subspecialized patients. The multiple factors contributing to the high cost of some hospitalizations require additional study. In particular, to understand the relative value, it is important to examine the differences in outcomes of children with HCHs who are hospitalized at large general hospitals compared to children’s hospitals. Because children’s hospitals already appear to be at risk for financial loss because of their high percentage of HCH discharges with public insurance, further work is needed to study what is driving costs in HCHs and determine if biomedical and psychosocial factors that increase length of stay, costs, or readmissions need to be adjusted in P4P programs. For example, previous work by Sills et al17 and Fieldston et al18 support risk adjustment for social determinants in pediatric P4P measures at the hospital level. Pediatric HCHs are understudied, and upstream social determinants of health factors may contribute to costs. Research and health care funding policy discussions around HCHs are needed to ensure that caring for this patient population is financially sustainable, while meeting national goals for cost containment among all patient populations.
Differences in HCH populations across hospital types may also have important policy implications. Several of our findings reveal that children’s hospitals disproportionately cared for specific populations within HCHs. Although HCHs at large general hospitals had a higher severity index, children’s hospitals cared for more of the complex population. The high proportion of neonatal HCH at large general hospitals may contribute to the differences observed in severity index. Children’s hospitals cared for the majority of the most common nonneonatal HCHs as well as the HCH neonatal diagnoses requiring procedures. This resulted in certain APR-DRGs being found almost exclusively at children’s hospitals. Our findings suggest that children’s hospitals may provide certain subspecialty care not currently available at all hospital types. This was particularly notable for cardiac surgeries, in which >95% of the HCHs were cared for in children’s hospitals. In addition to highly specialized services like cardiothoracic surgery teams, the HCHs treated at children’s hospitals required a more diverse array of subspecialties, reflected in nearly 5 times the number of diagnostic groups compared to large general hospitals and >30 times that of small general hospitals. The concentration of HCHs requiring specialized care in children’s hospitals is further elucidated by the high proportion of all “transfers in” directed toward children’s hospitals, representing 82% of all transfers in and one-fourth of the HCHs within children’s hospitals. Recent work by França et al19 demonstrated an increase in pediatric interhospital transfers by 25% between 2006 and 2011 (whereas hospital admissions decreased by 9.3%) and an increasing dependence on pediatric referral centers for pediatric inpatient care. Increased quality for HCHs and specialty care hospitalizations may be gained through further financial support to children’s hospitals for that care, as well as in tangible protocols, such as state-regulated transfer protocols to children’s hospitals for high-risk or highly specialized pediatric admissions. However, the increasing reliance on a small number of children’s hospitals for pediatric HCHs and specialty care hospitalizations may result in these patient populations having limited access to the care that they require. The availability of care and outcomes for HCH populations may benefit more by increasing the capacity and specialization of large general hospitals. The correct policy solution may depend on which approach provides access to sustainable sources of care and delivers equivalent quality at lower costs for HCHs.
Our study has the limitations inherent in using administrative databases, including potential for coding errors, missing variables, and possible misclassifications of hospital types. The KID data occur at the level of the hospitalization rather than the individual patient. We are therefore unable to account for occurrences when a single patient contributes to multiple HCHs. We relied on cost-to-charge variables for the estimation of costs, which may not fully estimate the true cost of care. Additionally, because of the limitations of the HCUP KID, our study only includes inpatient hospitalizations, and observation status hospitalizations are not included in the HCUP KID. However, observation status hospitalizations are rarely drivers of HCHs.20 Most observation hospitalizations would thus be non-HCHs, and they would skew the context around the percent of all hospitalizations and all charges, although we do not know which direction the findings would be skewed. In addition, there may be a proportion of small general hospitals that do not care for pediatric patients and only admit lower-acuity adolescent patients. These hospitals could potentially contribute to a shift toward the decreased costs and lower diagnostic diversity in discharges seen at small general hospitals. Finally, we used the most recent year of available data from the HCUP KID at the time of the study. Although our results reflect HCHs in 2012, we would not expect the distribution of HCHs by hospital type to change dramatically year by year.
Conclusions
Although representing a minority of discharges, HCHs were responsible for 51% of the annual pediatric inpatient costs nationally. Children’s hospitals cared for a disproportionate volume and cost of HCHs as well as a greater diagnostic diversity of HCHs. As a result, children’s hospitals appear to provide care needed in HCHs that does not appear to be provided elsewhere. There is a greater resource capacity and a potential financial risk for hospitals caring for a larger proportion of HCHs, making children’s hospitals especially vulnerable as health care costs continue to rise and payment models evolve. Future studies should characterize the factors driving cost in HCHs and explore reimbursement practices for HCHs to ensure the long-term financial viability of the pediatric health care system.
Footnotes
Dr Macy’s current affiliation is Feinberg School of Medicine, Northwestern University and Ann and Robert H. Lurie Children’s Hospital of Chicago, Chicago, IL.
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: No external funding.
POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.
References
- Copyright © 2020 by the American Academy of Pediatrics
In this issue
