Author + information
- Received April 13, 2018
- Revision received June 15, 2018
- Accepted July 12, 2018
- Published online October 1, 2018.
- Jonathan B. Edelson, MDa,∗ (, )@JEdelsonMD@ChildrensPhila,
- Joseph W. Rossano, MD, MSa,
- Heather Griffis, PhDb,
- Dingwei Dai, PhDb,
- Jennifer Faerber, PhDb,
- Chitra Ravishankar, MDa,
- Christopher E. Mascio, MDc,
- Laura M. Mercer-Rosa, MD, MSCEa,
- Andrew C. Glatz, MD, MSCEa and
- Kimberly Y. Lin, MDa
- aDepartment of Pediatrics, Division of Pediatric Cardiology, Children's Hospital of Philadelphia, Philadelphia, Pennsylvania
- bHealthcare Analytics Unit, Children's Hospital of Philadelphia, Philadelphia, Pennsylvania
- cDepartment of Pediatrics, Division of Cardiothoracic Surgery, Children's Hospital of Philadelphia, Philadelphia, Pennsylvania
- ↵∗Address for correspondence:
Dr. Jonathan B. Edelson, Division of Cardiology, The Children's Hospital of Philadelphia, 3400 Civic Center Boulevard, Suite 8NW90, Philadelphia, Pennsylvania 19104.
Background Data related to the epidemiology and resource utilization of congenital heart disease (CHD)–related emergency department (ED) visits in the pediatric population is limited.
Objectives The purpose of this analysis was to describe national estimates of pediatric CHD-related ED visits and evaluate medical complexity, admissions, resource utilization, and mortality.
Methods This was an epidemiological analysis of ED visit-level data from the 2006 to 2014 Nationwide Emergency Department Sample. Patients age <18 years with CHD were identified using International Classification of Diseases-9th Revision-Clinical Modification codes. We evaluated time trends using weighted regression and tested the hypothesis that medical complexity, resource utilization, and mortality are higher in CHD patients.
Results A total of 420,452 CHD-related ED visits (95% confidence interval [CI]: 416,897 to 422,443 visits) were identified, accounting for 0.17% of all pediatric ED visits. Those with CHD were more likely to be <1 year of age (43% vs. 13%), and to have ≥1 complex chronic condition (35% vs. 2%). CHD-related ED visits had higher rates of inpatient admission (46% vs. 4%; adjusted odds ratio: 1.89; 95% CI: 1.85 to 1.93), higher median ED charges ($1,266 [interquartile range (IQR): $701 to $2,093] vs. $741 [IQR: $401 to $1,332]), and a higher mortality rate (1% vs. 0.04%; adjusted odds ratio: 1.25; 95% CI: 1.07 to 1.45). Adjusted median charges for CHD-related ED visits increased from $1,219 (IQR: $673 to $2,138) to $1,630 (IQR: $901 to $2,799), while the mortality rate decreased from 1.13% (95% CI: 0.71% to 1.52%) to 0.75% (95% CI: 0.41% to 1.09%) over the 9 years studied.
Conclusions Children with CHD presenting to the ED represent a medically complex population at increased risk for morbidity, mortality, and resource utilization compared with those without CHD. Over 9 years, charges increased, but the mortality rate improved.
Congenital heart disease (CHD) affects >1% of newborns, making it the most common form of major birth defect (1). While the incidence of CHD has remained stable over the past 50 years, interventional options and resultant outcomes for many forms of CHD have changed dramatically. With major advances in prenatal diagnosis, surgical techniques, and intraoperative and post-operative care, patients who once would have died in infancy are now surviving to childhood, adolescence, and adulthood, thus transforming once fatal lesions into new forms of chronic disease in childhood and beyond (2–5). Given the complexity and size of this population, children with CHD consume a disproportionately large share of available resources (6), with yearly expenditures in the United States approaching $6 billion, and notable recent increases in both institutional costs and charges to the patient (7,8).
Much of the current published data characterizing the costs and outcomes of care for CHD patients focuses on the inpatient setting (9). Among children with CHD, there have been efforts to both measure the value of care provided and identify pre-operative risk factors for increased resource utilization while hospitalized (10–12). However, the data regarding visits to the emergency department (ED) among pediatric patients with CHD are lacking, and there are none that offer a national population-based assessment.
The purpose of this study was to describe nationwide estimates of pediatric CHD-related ED visits over a 9-year period, detail their medical complexity and admissions, and test the hypothesis that, over the time period studied, ED visits in children with CHD resulted in increased resource utilization and decreased mortality.
We used the discharge data from the Nationwide Emergency Department Sample (NEDS), the largest publicly available all-payer ED database that contains a 20% stratified sample of ED visits from across the United States, provided by the Healthcare Cost and Utilization Project (HCUP), Agency for Healthcare Research and Quality (13,14). NEDS data are provided by the HCUP State Inpatient Databases and the State Emergency Department Databases that capture the discharge information on ED visits that do versus do not result in an admission to the same hospital, or transfer to another hospital, respectively. Of note, NEDS does not capture urgent care visits. NEDS provides appropriate weights to obtain weighted national estimates (15).
For this study, we used NEDS data from 2006 to 2014 and limited analyses to visits of patients younger than 18 years of age. The unit of analysis is the ED visit, not a patient. A patient may therefore be represented by multiple ED visits in any given year or multiple years. For each visit in NEDS, up to 15 International Classification of Diseases-9th Revision-Clinical Modification (ICD-9-CM) diagnostic codes before 2014, up to 30 ICD-9-CM diagnostic codes in 2014, and 9 ICD-9-CM procedure codes across all years are provided. We identified CHD-related ED visits using ICD-9 CM-and acute CHD-associated comorbidities, which included heart failure, cyanosis, arrhythmia, pulmonary hypertension, acute respiratory disease, acute gastrointestinal disease, acute neurological disease, sepsis, and acute kidney injury. We classified CHD into 3 groups hierarchically: single-ventricle complex CHD, non–single-ventricle complex CHD, and simple CHD, using ICD-9-CM (Online Appendix).
Outcomes of interest
We examined the following outcomes of interest: 1) admitted or transferred for inpatient hospitalization; 2) mortality (died in ED or during admission associated with ED visit); and 3) total inflation-adjusted ED visit charges.
We examined patient and hospital characteristics as covariates. These included age in years (<1, 1 to 4, 5 to 9, 10 to 14, and 15 to 17 years), sex, insurance status (government, private, and other), residence (urban vs. rural), and time of visit (weekday vs. weekend). Hospital characteristics included geographic region (Northeast, Midwest, South, West), and location/teaching status (metropolitan/teaching, metropolitan/nonteaching, or nonmetropolitan area).
Complex chronic conditions (CCC) were defined as medical conditions that can be reasonably expected to last at least 12 months and that involve either several different organ systems or 1 organ system severely affected enough to require specialty pediatric care and some period of hospitalization in a tertiary care center, and include categories such as gastrointestinal, immunological, and genetic. Given that the majority of CHD patients should have a CCC for cardiovascular disease, the CCC for cardiovascular disease was not included in this analysis. To identify whether an individual was diagnosed with a CCC, we used the previously published classification scheme based on ICD-9-CM (16).
All analyses accounted for the NEDS’s complex survey designs utilizing appropriate survey weights. National estimates of ED visits overall, with CHD and without CHD, by year across 2006 to 2014 were calculated. Demographic and clinical characteristics of ED visits with CHD and without CHD were compared using chi-square tests for categorical variables and analysis of variance or Kruskal-Wallis tests as appropriate for continuous variables. Additionally, national estimates of ED visit outcomes (hospital admission, mortality, total charges) were analyzed by year using weighted generalized linear regression.
The total ED charges were inflated to 2014 U.S. dollars using the medical care services component of the Consumer Price Index (17). To examine the relationship between patient and hospital characteristics and outcomes, we used multivariable logistic regression for hospital admission and mortality, and generalized linear models with a log link and gamma distribution for total charges. Patient demographic and clinical, as well as hospital, characteristics were included in the multivariable models. All national estimates, 95% confidence intervals (CIs), or interquartile range (IQR) and trend tests were conducted by using SAS survey procedures and proc glimmix with sampling weights, as recommended by HCUP to compute accurate variances (15). All data management and statistical analyses were performed using SAS version 9.4 (SAS Institute, Cary, North Carolina). A 2-sided p value <0.05 was used as the threshold for statistical signiﬁcance.
Demographics and clinical characteristics
Among an estimated 241.5 million national pediatric visits to the ED over a 9-year period, there were 420,452 CHD visits (0.17% of all ED visits), averaging 46,717 visits of patients with CHD presenting to the ED every year. The vast majority of visits among patients with CHD who presented to the ED were ≤4 years of age (77%), and more likely to be female (54%) than male. Compared with non-CHD patients, CHD patients were more likely to be ≤1 year of age (43% vs. 13%), more likely to pay with Medicaid (61% vs. 53%), and more likely to present to a metropolitan teaching hospital (74% vs. 44%) (p < 0.0001 for all) (Table 1).
The most common acute complaint among patients with CHD presenting to the ED was acute respiratory disease (14%), followed by acute gastrointestinal disease (9.3%). Among cardiac-associated comorbidities, the most frequent presenting diagnoses were arrhythmia (5.8%) and heart failure (4.9%). Those with CHD were more likely to have at least 1 CCC (35% vs. 2%; p < 0.0001), and 13% of patients with CHD had ≥2 CCCs compared with <0.3% in the non-CHD population. The most common CCCs among patients with CHD were congenital/genetic (16.5%), gastrointestinal (12.3%), and respiratory (6.1%) (p < 0.0001 for all) (Table 2).
Patients with CHD had higher rates of admission (46% vs. 4%; adjusted odds ratio [OR]: 1.89; 95% CI: 1.85 to 1.93), longer lengths of stay (3 days [IQR: 1 to 7 days] vs. 2 days [IQR: 1 to 3 days]), higher median ED charges ($1,266 [IQR: $701 to $2,093] vs. $741 [IQR: $401 to $1,332]), and a higher mortality rate (1% vs. 0.04%; adjusted OR: 1.25; 95% CI: 1.07 to 1.45) (p < 0.0001 for all) (Table 3).
Temporal trends in CHD-related visits
Over the observed period of time, total pediatric ED visits as well as visits among patients with CHD remained relatively stable (Table 4). However, the rate of admission or transfer among CHD patients decreased from 58% in 2006 (95% CI: 56.81% to 58.98%) to 46% in 2014 (95% CI: 45.43% to 47.47%) (p for trend <0.0001) (Figure 1). The mortality rate for CHD ED visits decreased over time (1.13% in 2006 [95% CI: 0.90% to 1.36%], 0.75% [95% CI: 0.57% to 0.92%] in 2014; p = 0.0065). There was a concurrent increase in median inflation-adjusted charges for CHD ED visits, from $942 (IQR: $520 to $1,651) in 2006 to $1,630 (IQR: $901 to $2,799) in 2014 (p for trend <0.0001) (Central Illustration).
Factors associated with admission and mortality
Factors independently associated with inpatient admission for CHD in multivariable modeling include age <1 year, patients with single-ventricle disease, having government insurance, presenting to an urban center, and having a CCC (Table 5). The adjusted OR for admission for visits among CHD patients who had one CCC was 1.79 (95% CI: 1.71 to 1.86). Hospital admission was also more likely than emergency room discharge for ED visits among patients with CHD who presented with heart failure, cyanosis, arrhythmia, pulmonary hypertension, acute respiratory disease, or acute neurological disease.
Factors independently associated with mortality included age <1 year, presenting in the Northeast or West regions, presenting to either a metropolitan nonteaching hospital or a nonmetropolitan hospital, having complex CHD, and having a CCC (Table 5). As with admission, the presence of CCCs increased the odds of mortality in CHD ED visits (Figure 2). The adjusted OR for mortality in patients with CHD who also have 1 CCC was 1.73 (95% CI: 1.45 to 2.06), and increased to 2.55 (95% CI: 2.11 to 3.07) for those with ≥2 CCCs.
This study has several main findings. First, to our knowledge, this is the largest known description and analysis of pediatric CHD emergency room visits to date. This study better defines the burden of CHD, with >46,000 CHD ED visits per year. Our analysis demonstrated that infants (age <1 year) represent a substantial portion of CHD ED visits, and that the risk of admission decreased progressively with the age of the patient at the time of presentation. Given the known mortality and resource utilization of the CHD population as a whole, our results highlight the demand that the CHD population places on EDs with respect to volume of visits, acuity of presentation, and resource consumption.
Second, our findings highlight the vulnerability of several groups of CHD patients including infants, those with single-ventricle heart disease, and those with government insurance. Age <1 year and the presence of single-ventricle heart disease were both independent risk factors for admission and mortality. These at-risk groups are now a major focus of quality improvement efforts such as the National Pediatric Cardiology Quality Improvement Collaborative Initiative (NPCQIC) (18); the effect that such initiatives will have on CHD ED visits will be an area for further investigation. Our finding that government insurance was also an independent risk factor for admission draws attention to another at-risk group within the CHD population who may benefit from more focused strategies to efficiently use resources. Although Medicaid may improve access to care for children, it has also been associated with high resource utilization among children requiring congenital heart surgery (8) and worse outcomes for those patients (19). Nearly two-thirds of patients with CHD who present to the ED pay with Medicaid, which is higher than the non-CHD cohort. Significant changes to Medicaid funding would affect the majority of this cohort, further underscoring the vulnerability of the CHD population.
Third, this study indicates that the population of children with CHD who present to the ED are perhaps more medically complex than previously recognized, with a marked proportion of children with CHD also having at least 1 comorbid CCC (20). There is evidence that adults with CHD have increased resource utilization during hospitalizations, even for those that were noncardiac in nature, in part because of comorbidities (21). Our data show that the presence of a comorbid chronic medical condition was an independent risk factor for admission and mortality, and that both of those risks increase with multiple CCCs. This further suggests that comorbid chronic medical conditions may be an important contributing factor to the increased resource use of pediatric patients with CHD, and that 1 way to improve the value of care provided is to target these noncardiac comorbidities in the CHD population. Our study supports other studies, which have shown that patients with CHD often coexist with other forms of chronic disease, and warrant increased attention because of their high risk of negative outcomes.
We observed an admission rate among children with CHD 10× that of those children who present to the ED without CHD. Given that visits among children with CHD were more frequently associated with acute presenting complaints such as sepsis, neurological injury, and acute kidney injury, and that they more frequently had complex CCCs, it seems likely that the increased admission rate is due to both the acuity and medical complexity of the CHD group. There may also be a provider influence on resource utilization based on comfort level with assessing CHD in the ED. Interestingly, the admission rate for patients with CHD has gradually but persistently decreased over the 9 years studied, while the overall number of visits remained static, perhaps due to improved care delivery in the ED. Although the mortality rate observed was higher than that of the non-CHD population, it remained low overall, supporting evidence that shows improved post-operative outcomes in patients with CHD (3).
Finally, CHD ED visits represent a substantial financial burden when compared with non-CHD ED visits. Each CHD ED visit, on average, results in nearly twice the charges of a non-CHD visit. These findings correspond with what has been published with regards to the cost of caring for children with CHD in the inpatient setting (7). Also of note, the median inflation-adjusted charges for our patient cohort has gradually and continually increased over the time period studied. This is similar to the increase in costs and charges, which has been reported with respect to inpatient visits, and those with heart failure who present to the ED, which have served to highlight the significant financial burden of pediatric patients with heart disease (22,23). Although the resource utilization of CHD is significant and continues to increase, there is some suggestion that the devotion of extensive resources is not without benefit, as demonstrated by the concurrent decrease in mortality. This trend has been demonstrated in outpatient initiatives as well—for example, telemedicine and interstage monitoring programs provide up-front investment in technology and increased CHD surveillance on the premise that these resources will result in improved outcomes. Programs such as the Cardiac High Acuity Monitoring Program have shown shorter delays in care and lower resource utilization at readmissions with implementation of a tablet PC-based remote monitoring program (24). The effect of such programs on CHD ED visits is unknown at this time. Although the overall CHD ED visit mortality rate demonstrated in this study may be decreasing for any number or reasons, including improved surgical technique and more refined medical therapies, it is at least equally plausible that higher resource devotion and medical attention has appropriately led to decreased mortality in this vulnerable population.
The present study has several limitations associated with the design of retrospective studies. As a retrospective analysis, the findings may differ from a prospectively enrolled cohort. Identification of patients was made using ICD-9-CM codes given at the time of presentation in the emergency room, thus excluding, and including, patients who were either misdiagnosed or miscoded. The deidentification of data limits our ability to assess patients longitudinally or assess for repeat ED visits. Although HCUP performs several analyses to ensure internal consistency and data validity, some of the cases may have misclassified information. Additionally, there are other mechanisms by which children enter the hospital, including direct admission and visits through urgent care centers, neither of which are captured in the NEDS database, or any other comparable database of which we are aware. This deficiency limits our ability to fully capture the transition of all children from the outpatient to the inpatient setting. Finally, we have evaluated charges that serve as a proxy for resource utilization over time, but this does not give us information on actual cost.
Children with CHD who present to the ED are medically complex, with higher resource utilization and mortality than their non-CHD peers, likely in part due to their noncardiac comorbidities and their acuity at presentation. While the cost of a single presentation from a patient with CHD has gradually increased over time, during the same period of time there has been a decrease in rate of admission or transfer, and perhaps most notably, a decrease in the mortality rate. This suggests that increased resources devoted to patients with CHD are having a positive effect on their care. Several subgroups deserve even further attention due to their high risk of admission/transfer and mortality, including infants, patients with single-ventricle disease, and children with other CCCs.
COMPETENCY IN SYSTEMS-BASED PRACTICE: Children with CHD presenting to the ED have higher resource utilization and mortality those without CHD. Infants and those with comorbid conditions are at highest risk.
TRANSLATIONAL OUTLOOK: Better understanding of the causes of mortality may guide better resource utilization and improve outcomes for children with CHD in the ED.
Dr. Rossano has served as a consultant for Novartis and Amgen. Dr. Mascio has served as a consultant to and received honoraria from HeartWare. Dr. Mercer-Rosa has received support from National Institutes of Health grant NIH K01HL125521, and from the Pulmonary Hypertension Society. Dr. Lin has served as a consultant for Pfizer. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- complex chronic condition
- congenital heart disease
- emergency department
- Nationwide Emergency Department Sample
- Received April 13, 2018.
- Revision received June 15, 2018.
- Accepted July 12, 2018.
- 2018 American College of Cardiology Foundation
- Marelli A.J.,
- Mackie A.S.,
- Ionescu-Ittu R.,
- Rahme E.,
- Pilote L.
- Nasr V.G.,
- Faraoni D.,
- Valente A.M.,
- DiNardo J.A.
- Khairy P.,
- Ionescu-Ittu R.,
- MacKie A.S.,
- Abrahamowicz M.,
- Pilote L.,
- Marelli A.J.
- Khairy P.,
- Fernandes S.M.,
- Mayer J.E.,
- et al.
- Opotowsky A.R.,
- Siddiqi O.K.,
- Webb G.D.
- Pasquali S.K.,
- Sun J.-L.,
- d’Almada P.,
- et al.
- Connor J.A.,
- Gauvreau K.,
- Jenkins K.J.
- Simeone R.M.,
- Oster M.E.,
- Hobbs C.A.,
- Robbins J.M.,
- Collins R.T.,
- Honein M.A.
- Shin A.Y.,
- Hu Z.,
- Jin B.,
- et al.
- Mackie A.S.,
- Tran D.T.,
- Marelli A.J.,
- Kaul P.
- ↵HCUP Nationwide Emergency Department Sample (NEDS). Healthcare Cost and Utilization Project (HCUP). Rockville, MD: Agency for Healthcare Research and Quality. Available at: https://www.hcup-us.ahrq.gov/nedsoverview.jsp/. Accessed February 6, 2018.
- Healthcare Cost and Utilization Project
- Healthcare Cost and Utilization Project
- U.S. Bureau of Labor Statistics
- Tuomela K.E.,
- Gordon J.B.,
- Cassidy L.D.,
- Johaningsmeir S.,
- Ghanayem N.S.
- Seckeler M.D.,
- Thomas I.D.,
- Andrews J.,
- et al.
- Dean P.N.,
- Hillman D.G.,
- McHugh K.E.,
- Gutgesell H.P.
- Mejia E.J.,
- O’Connor M.J.,
- Lin K.Y.,
- et al.
- Bingler M.,
- Erickson L.A.,
- Reid K.J.,
- et al.