Methods

Data were sourced from the 2001 to 2011 Nationwide Inpatient Sample (NIS). The NIS is the largest all-payor inpatient health care database in the United States, containing data from approximately 8 million hospital discharges per year, yielding a 20% stratified sample of all community hospitals. Previous investigators have used the NIS to study trends in hospital resource utilization, including those trends involving adults with CHD. The Institutional Review Board of the University of Arkansas for Medical Sciences waived the need for investigational approval because the data in the NIS are deidentified.

The data contained in the NIS database include detailed, deidentified information on each inpatient’s demographics, diagnoses, procedures, outcome to hospital discharge, and hospital charges. Data were limited to include only patients aged ≥18 years admitted to the hospital with International Classification of Diseases-Ninth Revision (ICD-9) codes for a diagnosis of either hypoplastic left heart syndrome (HLHS; ICD-9 746.7), tricuspid atresia (TA; 746.1), or common ventricle (CV; 745.3). Patients diagnosed with multiple anatomic subtypes of SV were hierarchically placed into appropriate SV categories. Patients with a diagnosis of HLHS, irrespective of having an additional SV subtype, were categorized as HLHS. The CV category included patients who only had a CV subtype diagnosis. Admissions of patients with previous heart transplants, as determined by the presence of an ICD-9 code for cardiac transplant (ICD-9 V42.1 or 996.83) without the concomitant ICD-9 procedural code for heart transplantation (37.51) was excluded. Collected data included patient demographics (age at admission, gender, race, ethnicity, and payor status), principal and secondary diagnosis codes, mechanical ventilation, duration of hospital length of stay (LOS) in days, and hospital charges. Mortality was defined as failure to survive to hospital discharge. All charge data were adjusted for inflation to 2011 dollars using Consumer Pricing Indices data and were then converted to cost data using appropriate cost-to-charge data.

Co-morbid medical conditions (CMCs) most relevant to patients with SV were derived from principal and secondary ICD-9 diagnosis codes, which were grouped into categories based on organ systems ( Appendices A and B ). CMCs were classified using the method of Elixhauser et al. The original Elixhauser CMC included exclusions based on diagnostic-related groups (DRG). These exclusions operated to exclude many potential co-morbidities, including those in a cardiac DRG. To allow for these co-morbidities to be identified in patients with SV, the DRG exclusions were omitted in identifying a modified set of Elixhauser CMCs for this study. ICD-9-Clinical Modification codes to identify all CMCs were obtained from Elixhauser et al. Because the NIS does not differentiate between PH and GH; hospitals identified as having more than 80% of their inpatients aged ≤17 years at admission were categorized as PH. Previous studies have used a similar method to categorize hospitals.

Primary outcomes of interest were the total number of inpatient admissions for SV subtypes, hospital resource utilization, measured by LOS and costs per day, as well as the distribution of select CMCs in patients with SV. In addition, we compared primary outcomes between inpatients at PH and GH. GHs were subdivided into those with (teaching GH) and those without (nonteaching GH) medical training programs. Furthermore, modeling was used to predict the impact of covariates on hospital resource utilization and in-hospital mortality. Covariates included age, gender, race, ethnicity, primary payor, hospital type, SV anatomical subtype, and CMCs.

All statistical analyses were carried out using SAS, version 9.3, software (SAS Institute Inc., Cary, North Carolina). Because the NIS is complex survey data, we used SAS survey analysis procedures. Continuous and categorical outcomes are presented as means (standard error) and frequency (percentage), respectively. Comparisons of continuous outcomes between PH and TGH and NTGH were done using the analysis of variance test. Likewise, categorical outcomes were compared by use of the Rao–Scott chi–square test. The GLIMMIX procedures produced a multivariate generalized linear mixed model that was used to predict LOS and total hospital costs. The model was adjusted for the covariates mentioned earlier and took into account the nonnormally distributed outcomes and the interactions between hospital type and SV subtype. Sampling weights were used in all analyses to produce nationally representative estimates, while adjusting for clustering by hospital and the complex sampling design of the data. Because of sample weighting, counts of data, such as number of admissions, are reported as the estimated number ± estimated error. Statistical significance was determined at p values <0.05.

Results

There were a total of 8,330 ± 647 admissions of adults with SV (51% women) to US community hospitals during the study period. All SV admissions increased 35% from 602 ± 93 in 2001 to 814 ± 155 in 2011. From 2001 to 2011, admissions of adults with SV to GH increased 57% and decreased 71% in PH. The mean age at admission was 32.7 ± 0.5 years. Those with TA were older (34.8 ± 0.6 years) compared to those with either CV (30.2 ± 0.6) or HLHS (28.0 ± 1.2; p <0.0001 for both). In multivariate analyses, with the exception of a difference in age, there were no differences between SV anatomic subtypes in regard to gender, race, insurance status, CMC, LOS, hospital cost, or in-hospital mortality.

Comparisons of demographic and clinical outcomes between PH and TGH and NTGH are presented in Table 1 . LOS and total hospital costs, controlled for LOS, were highest in PH (p <0.0001; Figure 1 ). As presented in Table 2 , cardiovascular procedures were more common and more costly in PH and accounted for differences in LOS and costs. Hospital SV case volume had no effect on LOS. Both orthotopic heart transplant and procedure-associated mortality did not occur with enough frequency in PH to allow for reporting.

Comparison of hospital LOS and cost in adults with SV CHD by hospital type. There was no difference in LOS between pediatric and TGHs (p = 0.1686) although LOS was higher in both compared with NTGHs (p <0.0001). Total hospital cost was highest in PH compared with either GH type (p <0.0001), and TGHs were more costly than NTGHs (p <0.0001).

CMC was common in the study cohort. As presented in Table 3 , a number of CMCs were more common in GH, but none of the CMCs was more common in PH. Multivariate regression revealed the presence and number of CMCs had a direct impact on hospital resource utilization. As shown in Figure 2 , the LOS and hospital costs increased with each additional CMC. Likewise, the odds ratio for in-hospital mortality was increased in the presence of 2 or 3 CMCs ( Figure 3 ). In addition, multivariate analysis demonstrated a number of risk factors for increased hospital resource utilization and death ( Table 4 ).

Impact of number of CMC on hospital LOS and cost by hospital type in adults with SV CHD. (Panel A) the mean LOS increases nearly linearly for all hospital types with each additional co-morbid medical condition (p <0.0001 for comparison between each co-morbid medical condition group). * indicates p <0.006 compared with PH; † indicates p <0.04 compared with general teaching hospitals; ‡ indicates p <0.02 compared with PH; § indicates p <0.01 compared with general teaching hospitals. (Panel B) total hospital cost increases with each additional co-morbid medical condition with significant differences between hospital types with the exception of those admissions with 3 CMCs.

Impact of number of CMC on in-hospital mortality in adults with SV CHD. The odds of in-hospital mortality increase in a linear fashion with each additional co-morbid medical condition and were significantly higher in patients with 2 (p = 0.0001), 3 (p <0.0001), or 4 (p <0.0001) CMCs.

Table 4

Multivariate predictors of hospital resource utilization and death in adults with single ventricle congenital heart disease, 2001 to 2011

Variable	Length of Stay ∗ (Days)	p-value	Cost †	p-value	Odds Ratio of Death	95% CI	p-value
Age
18-30 years	6.4	Ref.	$26,642	Ref.	Ref.	N/A	Ref.
31-44 years	6.6	0.6880	$27,531	0.6176	0.591	0.360, 0.971	0.0550
45-64 years	7.4	0.1185	$30,994	0.1285	5.250	2.825, 9.758	<0.0001
≥65 years	10.7	0.0002	$46,630	0.0002	74.409	13.994, 395.636	<0.0001
Gender
Male	7.4	0.3697	$34,309	0.0158	2.720	1.804, 4.103	<0.0001
Female	7.8	Ref.	$34,009	Ref.	Ref.	N/A	N/A
Payor status
Private	8.0	Ref.	$36,018	Ref.	Ref.	N/A	N/A
Government	7.7	0.4967	$31,771	0.0394	0.943	0.604, 1.473	0.7946
Other	7.1	0.2285	$28,869	0.0484	0.414	0.160, 1.073	0.0997
Race
White	7.6	Ref.	$36,787	Ref.	Ref.	N/A	N/A
Black	9.5	0.0175	$39,464	0.5010	1.103	0.530, 2.296	0.4127
Hispanic	7.0	0.4054	$22,168	< 0.0001	0.302	0.099, 0.919	0.0097
Other	6.6	0.0797	$32,940	0.1778	0.347	0.119, 1.011	0.3325
Type of hospital
Children’s	9.7	Ref.	$56,671	Ref.	Ref. ‡	N/A	N/A
General, non-teaching	5.6	< 0.0001	$18,255	<0.0001	0.882	0.003, 240.755	0.9692
General, teaching	8.1	0.1391	$31,934	<0.0001	1.858	0.010, 329.428	0.8022
Hospital admission volume
1-12 cases	6.8	Ref.	$27,038	Ref.	Ref.	N/A	N/A
13-24 cases	7.5	0.4772	$32,362	0.245	1.966	0.022, 176.982	0.7386
25+ cases	8.7	0.1027	$37,756	0.0676	12.943	0.034, >999	0.4050
Single ventricle anatomy
HLHS	7.8	Ref.	$34,066	Ref.	Ref.	N/A	N/A
Tricuspid atresia	7.2	0.3809	$31,340	0.4101	0.227	0.106, 0.485	<0.0001
Common ventricle	7.7	0.8728	$30,893	0.3572	0.235	0.110, 0.505	< 0.0001
Primary admission diagnosis
Single ventricle anatomy	9.5	<0.0001	$47,839	<0.0001	1.847	1.134, 3.008	0.0077
Other medical diagnosis	6.1	Ref.	$21,522	Ref.	Ref.	N/A	N/A
Elixhauser comorbidity count
0	4.5	Ref.	$23,307	Ref.	Ref.	N/A	N/A
1	6.8	<0.0001	$28,734	0.0085	1.266	0.581, 2.756	0.5622
2	8.9	<0.0001	$33,307	<0.0001	4.549	2.193, 9.436	0.0005
3	8.7	<0.0001	$33,728	< 0.0001	17.20	8.172, 36.200	<0.0001
≥4	10.8	<0.0001	$45,213	< 0.0001	6.146	2.824, 13.376	<0.0001

Only gold members can continue reading. Log In or Register to continue

Share this:

• Click to share on Twitter (Opens in new window)
• Click to share on Facebook (Opens in new window)

Related

Related posts:

1. Inflammatory/Hemostatic Biomarkers and Coronary Artery Calcium Progression in Women at Midlife (from the Study of Women’s Health Across the Nation, Heart Study)
2. Effect of Diabetes Mellitus on Frequency of Adverse Events in Patients With Acute Coronary Syndromes Undergoing Percutaneous Coronary Intervention
3. Usefulness of the Sum Absolute QRST Integral to Predict Outcomes in Patients Receiving Cardiac Resynchronization Therapy
4. Syncope and Dyspnea and Pulmonary Embolism

Stay updated, free articles. Join our Telegram channel

Join