BACKGROUND AND OBJECTIVES: Pulmonary hypertension (PH) has been associated with substantial morbidity and mortality in children, but existing analyses of inpatient care are limited to small single-institution series or focused registries representative of selected patient subgroups. We examined US national data on pediatric PH hospitalizations to determine trends in volume, demographics, procedures performed during admission, and resource utilization.
METHODS: Retrospective cohort study using a national administrative database of pediatric hospital discharges: the Kids’ Inpatient Database.
RESULTS: Children with PH accounted for 0.13% of the 43 million pediatric hospitalizations in the United States between 1997 and 2012, and discharges demonstrated an increasing trend over the study period (P < .0001). Cumulative, inflation-adjusted national hospital charges for PH hospitalizations rose (P = .0003) from $926 million in 1997 to $3.12 billion in 2012. Patients with PH without associated congenital heart disease (CHD) comprised an increasing and majority (56.4%) proportion over the study period (P < .0001), children without associated CHD admitted at urban teaching hospitals comprised the fastest-growing subgroup. In-hospital, all-cause mortality was high (5.9%) in children with PH, but demonstrated a decreasing trend (P < .0001).
CONCLUSIONS: Morbidity and mortality of pediatric PH continue to represent substantial and growing health care burdens. Shifts in case mix toward PH not associated with CHD, toward noncardiac procedures, and toward care in urban teaching hospitals will increase pressure to manage resource utilization in this small but growing patient group and to improve expertise and define excellence in PH care across a wide range of clinical settings.
- CHD —
- congenital heart disease
- CI —
- confidence interval
- ECMO —
- extracorporeal membrane oxygenation
- ICD-9-CM —
- International Classification of Diseases, Ninth Revision, Clinical Modification
- KID —
- Kids’ Inpatient Database
- HCUP —
- Healthcare Cost and Utilization Project
- OR —
- odds ratio
- PH —
- pulmonary hypertension
What’s Known on This Subject:
Although existing analyses of inpatient pediatric pulmonary hypertension (PH) care have established an association with substantial morbidity and mortality, these investigations have been limited to small single-institution series or focused registries representative of selected patient subgroups.
What This Study Adds:
This study provides the first contemporary, national trend analysis of inpatient care for children with PH. Pediatric PH is associated with a rapidly increasing number of hospital discharges and magnitude of resource utilization, and the makeup of this population is changing.
Pediatric pulmonary hypertension (PH) is a severe and often chronic condition1,2 associated with significantly reduced quality of life for patients, parents, and caregivers.3 It typically requires lifelong care and is associated with a significant burden of health care costs.4 Nearly all PH therapies are used off-label in children because evidence for their use is available only from adult data that have been extrapolated to the pediatric population. Assessing treatment efficacy is further complicated by heterogeneity of patient subsets and by the challenges of establishing clinical endpoints that are appropriate to children. For instance, surrogate clinical endpoints used in adults with PH, such as a 6-minute-walk test, are more challenging to standardize and validate in the pediatric population. The gold standard in adult and pediatric PH diagnosis, invasive testing via cardiac catheterization, is resource intensive, requires an experienced center, is complicated by the lack of a uniformly accepted pediatric protocol,5 and involves the risk of repeated exposures to general anesthesia.6
Despite improved treatment, it appears from the limited outcome data that are available that morbidity and mortality from pediatric PH remain high: 5-year survival has been estimated at 74% and 64% in United States (REVEAL7) and United Kingdom (UK Pulmonary Hypertension Service for Children8) registries, respectively. Neither registry shows a significant survival difference between idiopathic PH and PH that is associated with congenital heart disease (CHD), although the estimate of 5-year survival in the UK group of children with PH associated with CHD is as low as 57%.
The heterogeneity of pediatric PH has inhibited inclusive, broad assessments of hospitalization outcomes, trends, and associated resource allocation. Existing insights and information from registries7–9 have defined PH as a pressing health care issue,10 but although important, these studies have limitations. The landmark work by Haworth and Hislop8 from the UK registry provides data from a hub-and-spoke model of regional centers of expertise within a national referral program. Groups in the Netherlands11,12 and Spain13 have focused on broad capture of the many individual diagnostic groups that contribute to the growing problem of pediatric PH. However, these analyses capture selected patient subsets and/or draw from regional centers of expertise that may have specific diagnostic criteria for inclusion, and therefore may not reflect patterns of PH care seen in broader clinical practice at the population level.14
Despite the suggestion that this population is growing, few recognized PH centers for children exist in the United States, and to date, no current published national US data exist on volume or trends in pediatric PH hospitalizations. A more robust understanding of the development of pediatric PH as a growing health care priority in the United States is needed to work toward the critical goal of registry development15 that will underpin PH subtype differentiation and clinical trials, as well as to guide public policy and resource allocation.
We sought to use an established, nationally representative administrative database to examine pediatric hospitalizations in the United States involving children diagnosed with PH, in an effort to inform providers and policymakers with insights regarding shifting demographics and unmet needs that may guide system and policy improvements. Specifically, we sought to assess the hypothesis that PH hospitalizations have become more common, with shifts in patient demographics, in the balance of PH associated with CHD compared with PH without associated CHD, in procedures performed during admission, and in overall resource utilization.
This study was exempt from institutional review board review because it uses data that are deidentified and publicly available. Administrative records were extracted from discharge datasets for 1997 to 2012 from the Kids’ Inpatient Database (KID), part of the Healthcare Cost and Utilization Project (HCUP) of the Agency for Healthcare Research and Quality. KID is the largest publicly available database of inpatient pediatric care in the United States. Datasets are released every third year and include records for a sample of discharges from all nonfederal, nonrehabilitation hospitals in states that participate in HCUP.
KID is designed with sample weights that facilitate nationally representative estimates; these weights adjust for growth in participation from 1997, when 2521 hospitals in 22 participating states submitted 1.9 million records reflective of 6.66 million national discharges, to 2012, when 4179 hospitals in 44 states submitted 3.2 million records reflective of 6.68 million national discharges. KID contains limited data on each inpatient encounter, but its size and sampling frame facilitate the analysis of rare clinical events at a national level.
Diagnostic, comorbidity, and procedural information was based on HCUP-supplied International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes. Pulmonary hypertension was identified (416.×) and divided based on the presence or absence of CHD (745.×, 746.×, and 747.1–4) into 2 subgroups for analysis: PH associated with CHD (hereafter “CHD group”) and PH without associated CHD (hereafter “non-CHD group”). The performance of surgical procedures was identified based on previously described ICD-9 volume 3 procedure codes for cardiac16 and noncardiac17 operations. Post hoc reclassification as part of the CHD group was performed for a small number (0.8%) of admission records that did not contain a CHD diagnosis code but contained a cardiac surgical procedure code that would have occurred only in a patient with CHD (eg, “repair of total anomalous pulmonary venous connection”).
A composite comorbidity point score was calculated based on the van Walraven modification18 of the Elixhauser comorbidity measure,19 which is the basis of the dedicated comorbidity variables included in HCUP databases. Procedure codes also were used to identify the performance of procedures that were viewed by the multidisciplinary author group as clinically relevant to either the diagnosis or sequelae of pediatric PH and that reliably could be identified by procedure codes: cardiac catheterization (37.2×, 38.2×), MRI (88.91–88.97), tracheostomy (31.1, 31.2), hemodialysis or peritoneal dialysis (39.95, 54.93, 54.98, or V45.11), and extracorporeal membrane oxygenation (ECMO; 39.65, 39.66).
Beginning in 2012, freestanding children’s hospitals were identified directly in KID (variable KID_STRATUM=9999). For previous years, we reconstructed this classification based on the National Association of Children’s Hospitals and Related Institutions classifications, provided in KID (variable NACHTYPE=1 or 2).
Discharge sample weights were used to create national estimates within the KID sampling frame. Primary outcome measures included inpatient length of stay (variable LOS), total inpatient hospital charges (variable TOTCHG), and in-hospital mortality (variable DIED). Secondary outcome measures included respiratory failure (518.5×, 518.8×, 770.84, 799.1, and V46.1–2) and discharge disposition (routine home discharge, home with home health or nursing, or to a long-term care/skilled nursing facility).
Hospital charges were indexed to inflation and to the broadly rising costs of medical care by adjusting all values to 2012 dollars by using the Bureau of Labor Statistics Consumer Price Index subindex specific to inpatient hospital services.20
Because the sampling frame of KID requires the use of advanced techniques (facilitated by PROC SURVEYMEANS in SAS [SAS Institute, Cary, NC]) to estimate variance, continuous variables are presented as means ± SE. Discrete variables are presented as number (percentage). Intergroup comparisons were carried out using the Mann-Whitney-Wilcoxon test for continuous variables and Fisher’s exact test or the Pearson χ2 test for categorical variables, as appropriate. Odds ratios were calculated for intergroup discrete outcome comparisons; for PH versus non-PH comparisons, the non-PH cohort is the reference group. For CHD versus non-CHD subgroups within the PH population, the non-CHD cohort is the reference group.
Trends over time were examined by month of hospital admission by using a seasonal Mann-Kendall test for trend, a nonparametric test to determine the presence and direction of a trend over time21 that accounts for seasonal variability, as inpatient pediatric care might involve seasonal fluctuations due to patterns of winter illness, surgical scheduling in relation to school calendars, and other similar factors. Records with missing data for the month of admission (7.2% of all records in the 1997–2012 KID datasets) were excluded from trend analyses only; these records were included for all nontrend and descriptive analyses. A predetermined α of 0.05 was used as the threshold of statistical significance. Trend analyses were performed by using R (R 3.1.3; The R Foundation for Statistical Computing, Vienna, Austria). All other analyses were performed by using SAS (SAS 9.4; SAS Institute, Cary, NC).
Children with PH accounted for 56 678 discharges nationwide between 1997 and 2012, which represents 0.13% of the estimated 43.0 million total pediatric discharges reflected in KID.
Cumulative, inflation-adjusted national hospital charges for PH hospitalizations rose from $926 million in 1997 (totaled annually) to $3.12 billion in 2012 (P = .0003, Fig 1A). This increase was associated with increasing trends among PH discharges during the study period in average total charges per hospitalization (P < .0001, Fig 1B) and length of stay (P < .0001). The number of discharges for PH increased in absolute terms and as a proportion of all pediatric discharges (P < .0001 for both, Fig 2). The increase in PH discharges was most pronounced during the period 2006 to 2012.
Table 1 shows demographic characteristics of hospitalized children with PH compared with the reference population of all pediatric inpatients. Children with PH were younger, had a greater average comorbidity score, and were more commonly admitted to urban teaching hospitals and children’s hospitals. A minority of PH hospitalizations (33.9%) occurred in children’s hospitals, although this proportion was higher than in non-PH hospitalizations (7.8%, P < .0001, Table 1). No significant trend was observed in average age of PH discharges during the study period (P = .13). Table 2 shows the procedural and outcome characteristics of discharges for children with PH compared with all pediatric discharges.
Among PH discharges, the CHD group accounted for 43.6% overall (n = 24 706). This proportion demonstrated a decreasing trend over the study period (P < .0001, Fig 3) relative to the non-CHD group (n = 31 972; 56.4%). Figure 4 demonstrates that the increase in PH hospitalizations over the study period largely was driven by growth in hospitalizations at urban teaching hospitals, as hospitalizations at urban teaching hospitals increased more than those in other settings. Within urban teaching hospitals, growth in the non-CHD group outpaced that of the CHD group.
Table 3 shows demographic characteristics of the 2 PH subgroups with and without associated CHD. Compared with the non-CHD group, children in the CHD group were younger and had lower baseline comorbidity scores. Table 4 shows procedural and outcome characteristics of PH discharges among the CHD and non-CHD groups. Children in the CHD group were more likely to undergo cardiac catheterization, cardiac and noncardiac surgery, and ECMO, and had longer lengths of stay and greater total hospital charges.
During the study period, there was a decreasing trend in the proportion of PH discharges that involved cardiac surgical procedures (P < .0001, Fig 5). There was also an increasing trend in the proportion of PH discharges that involved noncardiac surgical procedures: from 3.6% in January 1997 to 12.0% in December 2012 (P < .0001, Fig 5).
Mortality occurred in n = 3332 PH discharges (5.9% of these discharges overall). The pattern was one of steady improvement, as the incidence of in-hospital mortality decreased from 11.3% in January 1997 to 6.6% in December 2009 (P < .0001, Fig 6). Mortality was slightly higher in the CHD group than the non-CHD group (6.5% vs 5.4% of discharges, P < .0001).
Mortality among patients undergoing noncardiac surgery was substantially higher in PH patients than in the aggregate of all KID discharges (9.3% vs 1.0%, crude odds ratio [OR] 10.64, 95% confidence interval [CI] 9.71–11.67, P < .0001) and was higher in the CHD group of patients with PH than in the non-CHD group (11.5% vs 7.3%, crude OR 1.66, 95% CI 1.38–2.00], P < .0001).
The principal findings of this study are that hospitalizations of children with PH account for an increasing proportion of all pediatric hospitalizations (1 in 1000 discharges at the beginning of the study period compared with 1 in 500 discharges by its end), with a dramatic increase in resource utilization associated with inpatient pediatric PH care, particularly from 2006 to the present. These trends appear to have resulted from a combination of greater numbers of hospitalizations, longer lengths of stay, and increasing total hospital charges.
The increased number of these children admitted to urban teaching hospitals largely accounts for the increase in discharges. The relative case mix shifted during the study period, with the non-CHD group comprising a majority and growing share of the overall population of pediatric PH discharges. All-cause, in-hospital mortality among children with PH remained high, but it declined significantly during this period and differed only slightly between CHD and non-CHD groups.
Administrative data provide a limited opportunity to elucidate the etiologies of these trends. However, our study period does coincide with an era of significant improvement in neonatology care and outcomes, as well as in the detection and medical management of chronic PH, particularly that in children without CHD.22,23 Improved survival of infants with bronchopulmonary dysplasia and the emergence of pulmonary vasodilator therapy may explain both the decrease in observed mortality per hospitalization as well as the relative increase in the non-CHD group of children admitted with PH. These trends also may relate to increasing success in the early surgical treatment of CHD with a focus on prevention of PH.
Compared with the index population of pediatric discharges, discharges for children with PH were characterized by a younger and considerably sicker population (Table 1) that had a substantially longer average length of stay, greater hospital charges, a greater need for selected diagnostic and therapeutic procedures, a higher rate of respiratory failure and nonroutine discharges, and a substantially higher in-hospital mortality rate (Table 2). These findings are not surprising, but they provide clear, quantitative delineation of the morbidity and mortality associated with pediatric PH.
The non-CHD group had a higher proportion of children who were African American, which may reflect PH from prematurity-related chronic lung disease due to the higher incidence of prematurity in the African American population.24
Resource utilization was greater in the CHD group, with longer lengths of stay and greater average hospital charges compared with the non-CHD group. We suspect, but cannot prove, that the greater resource utilization observed in the CHD group is due to a larger number of surgical procedures.
We observed a decreasing trend over the study period in the proportion of discharges that involved cardiac surgery, likely because of the concomitant growth of the non-CHD group. We also observed an increasing trend in the proportion of discharges that involved noncardiac surgery. It is encouraging that these trends were associated with a decreasing trend in all-cause mortality. Noncardiac surgery has been associated with substantial risk of mortality in children with PH in previous analyses,6,25,26 and this remains evident from the 9.3% incidence of mortality in the current analysis among children who required a noncardiac operation.
Important limitations to this study exist as a result of structural features of the KID database. First, it is important to note that KID is a database of discharges, so this analysis does not provide longitudinal data on patients with PH over time or on outpatient care. No mechanism allows for connecting multiple admissions records for a single child. This also means that the observed decrease in the proportion of PH discharges ending in mortality does not necessarily demonstrate a true decrease in the incidence of PH mortality. This trend also could reflect more frequent hospitalizations earlier in the disease course, with or without related changes in the duration of survival or the proportion eventually experiencing mortality. Similarly, trends in hospital charges and length of stay per admission could reflect a change in the type or acuity of hospital admissions. Thus, cumulative national charges for all PH discharges provide the best reflection of the overall trend in the collective resource utilization associated with PH-related inpatient care.
A second limitation of the current study is that ICD-9-CM codes do not permit accurate identification of subgroups of patients with PH in a way that is consistent with the most current schemata for categorizing PH. Early attempts to classify PH as “primary” or “secondary” have been refined over the past 2 decades through multiple iterations of World Health Organization classifications defined at international consensus symposia in 1998 (Evian, France),27 2003 (Venice, Italy),28 2008 (Dana Point, CA),29 and 2013 (Nice, France)30 to reflect progress in understanding the pathophysiology of PH and identify subgroups of patients with PH who share meaningful aspects of their disease process. Unfortunately, administrative analyses cannot capture these clinical distinctions, because ICD-9 coding has not been modified to reflect these refinements. However, we do believe that it is both clinically meaningful and operationally reliable to identify PH associated with CHD. We therefore designed our analysis to use only this subclassification within the entirety of PH admissions, as previous registries have done.7,8 We should note that our post hoc reclassification of patients otherwise belonging to the non-CHD group on the basis of cardiac surgical procedure codes that should reliably identify them as patients with CHD was likely imperfect, and some classification error may persist. A small proportion of the non-CHD group may represent patients with CHD who simply lack a correct CHD diagnosis code; this would tend to bias the comparison between CHD and non-CHD discharges toward the null.
Lack of granularity about PH subtypes and residual classification error are limitations of the use of an administrative database and must be weighed against the advantages of a large, national analysis. Similarly, a limitation of retrospective studies, such as this one, is that they rely on existing practices of diagnosing PH and do not offer a mechanism for using gold standard (eg, echocardiographic or catheterization-based) diagnostic criteria for inclusion.
We should note that we did not use the persistent PH of the newborn diagnosis code (747.83, “persistent fetal circulation”) as an inclusion criterion for our patient cohort, as we suspect patients with that diagnosis alone differed in important ways from our population of interest. This decision was in keeping with other previous analyses of pediatric PH, including the UK registry.8 Patients with a persistent PH of the newborn diagnosis code were included only in the event of a separate PH diagnosis code (416.×).
A third issue is that total hospital charges reported in KID represent global charges for entire hospital stays. This precludes knowledge of the individual contributors to resource utilization (eg, medication costs, tests, supplies, procedures) or the degree to which the contribution of PH was direct or indirect. Also, the complex relationships between hospital charges, hospital costs, insurance reimbursement, and patient copayments complicate the extrapolation of these results to an estimation of the true societal cost of inpatient pediatric PH care. However, hospital charges are useful for analysis of resource utilization trends, allow for aggregation of data at the national level, and give the best available sense of the magnitude of resource utilization devoted to this patient population. We attempted to minimize the effect of any broader increases in the cost of medical care during the study period by adjusting for inflation to 2012 dollars by using a subindex of the Consumer Price Index that applies specifically to inpatient hospital care; increases in average total hospital charges that we report are therefore above background trends of increasing costs of hospitalization, and not just above the economy-wide inflation.
Fourth, KID provides data only triennially, and its sampling frame does change slightly from year to year. Discharge weights adjust to produce national estimates for each year, and we also attempted to address this limitation by ensuring that the ICD-9 codes used to identify PH discharges did not change significantly during the study period.31 It remains possible that shifts in awareness of PH in certain patient subgroups (eg, preterm infants with bronchopulmonary dysplasia) have contributed to the increase in the number of pediatric hospitalizations associated with PH. Also, it should be noted that all figures showing trends have a nonhomogeneous time axis because KID provides triennial data (ie, there are gaps between available dataset years). Therefore, the figures provide information as to the nature and direction of trends over the study period, but should not be interpreted to provide precise quantitative trend information (eg, the slope of a line) or be used to interpolate or extrapolate these trends. As such, we have limited our trend analyses to time-ordered tests and have not used linear regression to assess the slopes of any trends or extrapolate trends into the future.
Last, trend analyses reported here were limited by a small amount of missing data (7.2% of all records did not include the month of admission, only the year). However, we believe the benefit of greater data granularity outweighs the disadvantages of this limitation, as use of admission year alone would significantly limit the power of trend analyses and would prohibit any adjustment for seasonal fluctuations that are particularly important in pediatrics. Sensitivity analyses revealed that the small number of missing records did not change the results of nontrend analyses, including descriptive data or comparisons between the PH group and all KID discharges or between PH subgroups.
Despite these limitations, the current study is useful in demonstrating the burgeoning number and nature of pediatric PH hospitalizations and the implications of these trends for resource utilization and public policy.
This study highlights the substantial and growing health care burden that pediatric PH represents, while also documenting the declining trajectory of morbidity and mortality associated with inpatient PH care. These results have practice and policy implications at the institutional, state, and national levels, particularly in the face of increasing pressure to restrain costs while caring for a population with increasingly complex medical needs.
Our finding that most children were not cared for in dedicated children’s hospitals further underscores the need for increased awareness of PH among providers. Additionally, accelerated efforts are needed to foster multidisciplinary collaboration, to build consensus on standardized treatment protocols, strengthen the evidence base that supports clinical guidelines, and to define requisite expertise for clinical excellence in pediatric PH care across a wide range of settings.
- Accepted May 13, 2015.
- Address correspondence to Bryan G. Maxwell, MD, MPH, Department of Anesthesiology and Critical Care Medicine, Johns Hopkins University School of Medicine, 1800 Orleans St, Zayed 6208P, Baltimore, MD 21287. E-mail: .
Dr Maxwell conceptualized and designed the study, carried out the initial analyses, drafted the initial manuscript, and reviewed and revised the manuscript; Drs Nies, Ajuba-Iwuji, and Coulson provided input about study design and data analysis/interpretation, and critically reviewed the manuscript; Dr Romer helped conceptualize and design the study, reviewed and refined data analysis and interpretation, and reviewed and revised the manuscript; and all authors approved the final manuscript as submitted.
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.
COMPANION PAPER: A companion to this article can be found on page 392, and online at www.pediatrics.org/cgi/doi/10.1542/peds.2015-1697.
- Barst RJ,
- McGoon MD,
- Elliott CG,
- Foreman AJ,
- Miller DP,
- Ivy DD
- Haworth SG,
- Hislop AA
- van Loon RLE,
- Roofthooft MTR,
- Hillege HL,
- et al
- Lévy M,
- Celermajer D,
- Szezepanski I,
- Boudjemline Y,
- Bonnet D
- ↵Maxwell BG, Powers AJ, Sheikh AY, Lee PHU, Lobato RL, Wong JK. Resource use trends in extracorporeal membrane oxygenation in adults: an analysis of the Nationwide Inpatient Sample 1998–2009. J Thorac Cardiovasc Surg. 2014;148(2):416–421.e1
- ↵Bureau of Labor Statistics. Consumer price index detailed reports 1998–2009. Available at: www.bls.gov/cpi/cpi_dr.htm. Accessed November 11, 2014
- ↵National Center for Health Statistics and the Centers for Medicare & Medicaid Services. Conversion table of new ICD-9-CM codes. 2013. Available at: www.cdc.gov/nchs/data/icd/ICD-9-CM_FY14_CNVTBL_Final.pdf. Accessed December 17, 2013
- Copyright © 2015 by the American Academy of Pediatrics