The Risk Adjustment for Congenital Heart Surgery (RACHS-1) classification is an established method for predicting mortality for congenital heart disease surgery. It is unknown if this extends to the cost of hospitalization or if differences in economic and medical outcomes exist in certain subpopulations. Using data obtained from the University HealthSystem Consortium, we examined inpatient resource use by patients with International Classification of Diseases, Ninth Revision, procedure codes representative of RACHS-1 classifications 1 through 5 and 6 from 2006 to 2012. A total of 15,453 pediatric congenital heart disease surgical admissions were analyzed, with overall mortality of 4.5% (n = 689). As RACHS-1 classification increased, the total cost of hospitalization, hospital charges, total length of stay, length of intensive care unit stay, and mortality increased. Even when controlled for RACHS-1 classification, black patients (n = 2034) had higher total costs ($96,884 ± $3,392, p = 0.003), higher charges ($318,313 ± $12,018, p <0.001), and longer length of stay (20.4 ± 0.7 days, p <0.001) compared with white patients ($85,396 ± $1,382, $285,622 ± $5,090, and 18.0 ± 0.3 days, respectively). Hispanic patients had similarly disparate outcomes ($104,292 ± $2,759, $351,371 ± $10,627, and 23.0 ± 0.6 days, respectively) and also spent longer in the intensive care unit (14.9 ± 0.5 days, p <0.001). In conclusion, medical and economic measures increased predictably with increased procedure risk, and admissions for black and Hispanic patients were longer and more expensive than those of their white counterparts but without increased mortality.
Highlights
- •
Administrative data from 15,453 pediatric cardiac surgical admissions were examined.
- •
RACHS-1 predicts increases in medical and economic measures in addition to mortality.
- •
Black and Hispanic patients have longer courses and increased costs without increased mortality.
- •
No disparities in outcomes found on the basis of gender.
Administrative databases are composed of demographic and diagnostic coding information typically used for hospital billing purposes. The demographic data embedded in such databases afford the opportunity to examine not only mortality but racial, ethnic, and gender disparities as well. Databases of this type have been used specifically to evaluate medical and economic outcomes in patients with congenital heart disease (CHD). Although the Risk Adjustment for Congenital Heart Surgery (RACHS-1) classification for determining mortality risk for CHD patients is an established method for risk stratification, it is unknown if this also extends to the cost of hospitalization or if there are differences in economic and medical outcomes in certain subpopulations of children with CHD when accounting for the underlying surgical risk for a particular procedure.
Methods
Data were obtained from the University HealthSystem Consortium (UHC), an alliance of 120 academic medical centers and 307 affiliated hospitals that share diagnostic, procedural, and financial data on all admissions and discharges. Generally, the UHC does not include private, free-standing children’s hospitals. The UHC maintains a patient-level clinical database that contains administrative data on all inpatient hospital discharges for member organizations. This clinical database contains patient demographics, utilization patterns, diagnostic and procedure codes, outcomes (mortality, length of hospital stay, length of intensive care unit [ICU] stay), and financial data including charges and costs.
We examined inpatient resource use by patients with International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) procedure codes representative of RACHS-1 classifications 1 through 6 from 2007 to 2012. Because of the small numbers of category 5 cases, it was combined with category 6. We used a representative sampling of diagnostic and procedure codes to define the RACHS-1 categories as listed in Table 1 . The following data were collected on each patient: date of admission, age, gender, race or ethnicity, discharge status, length of stay (LOS), length of ICU stay, total costs, total charges, and up to 30 separate ICD-9-CM diagnostic and procedure codes for each admission. A total of 62 UHC institutions perform all procedures queried in this analysis.
| RACHS-1 Class | Diagnostic Code | Procedure Code | Age at Time of Surgery |
|---|---|---|---|
| Group 1 | |||
| Atrial Septal Defect Repair | 7455 | 3551, 3561, 3571 | < 18 years |
| Coarctation of Aorta Repair | 7471 | 3834, 3835, 3845, 3931, 3956, 3958 | > 30 days and < 18 years |
| Group 2 | |||
| Tetralogy of Fallot Repair | 7452 | 3581 | <1 year |
| Ventricular Septal Defect Repair | 7454 | 3553, 3555, 3562, 3572 | <1 year |
| Glenn Procedure | 7467, 7461, 7451 | 3921 | <1 year |
| Group 3 | |||
| Aortic Valve Replacement | 4241, 7463 | 3521, 3522 | < 18 years |
| Fontan Procedure | Any diagnosis | 3594 | < 18 years |
| Arterial Switch Operation | 7451 | 3584 | < 30 days |
| Group 4 | |||
| Truncus Arterious Repair | 7450 | 3583 | Any age |
| Aortic Valvotomy | 7472 | 3511 | < 30 days |
| Total Anomalous Venous Return Repair | 747.41 | 3582 | < 1 year |
| Group 5-6 | |||
| Norwood Procedure | 7467, 7453 | 3929, 3961 | < 30 days |
All statistic analysis was performed using SPSS version 21.0 (IBM Corporation, Armonk, New York). Relations between independent variables (RACHS-1 classification, gender, and race) and primary medical outcomes (discharge status, LOS, length of ICU stay, and mortality) and financial outcomes (total charges and total cost) were examined. Data were analyzed using 1-way analysis of variance as well as multivariate logistic regression analysis to simultaneous evaluate for associations of independent factors mentioned previously with outcomes. Odds ratios (ORs) comparing the levels of these various factors were estimated with corresponding 95% confidence intervals and p values using chi-square tests and risk analysis. Statistical significance was defined as a p value <0.05. Cutoffs were chosen as 1 SD to the nearest day for LOS (30 days) and ICU stay (24 days) and 2σ above the mean for hospital charges and costs. This study was approved with exempt status by the Institutional Review Board at the University of Virginia.
Results
A total of 15,453 hospital admissions were identified with sufficient data to classify by RACHS-1 on the basis of diagnostic and procedural ICD-9-CM codes. As listed in Table 2 , admissions for male patients outnumbered those for female patients (8,799 vs 6,694, respectively). There were significantly more female patients in RACHS-1 categories 1 and 2 than would be expected compared with the male cohort (p <0.001). The distribution of RACHS-1 classifications among different racial and ethnic groups was normal, except for a smaller than expected number of Hispanic patients in RACHS-1 class 1 (p <0.001).
| RACHS-1 Class | 1 | 2 | 3 | 4 | 5,6 | Total | |
|---|---|---|---|---|---|---|---|
| Gender | Male | 765 | 4306 | 2041 | 909 | 758 | 8799 (56.9%) |
| Female | 946 | 3460 | 1200 | 561 | 480 | 6654 (43.1%) | |
| p value | < 0.001 | < 0.001 | 0.300 | 0.754 | – | ||
| Race | White | 1006 | 4295 | 1780 | 794 | 674 | 8549 (55.3%) |
| Black | 259 | 1059 | 457 | 211 | 161 | 2144 (13.9%) | |
| Hispanic | 301* | 1703 | 702 | 330 | 296 | 3332 (21.6%) | |
| Asian/ Pacific Islander | 80 | 350 | 146 | 61 | 46 | 683 (4.4%) | |
| Native American | 56 | 218 | 80 | 43 | 31 | 441 (2.9%) | |
| Other/Unknown | 32 | 148 | 76 | 31 | 17 | 304 (1.9%) | |
| 1731 (11.2%) | 7773 (50.3%) | 3241 (21%) | 1470 (9.5%) | 1238 (8.0%) | 15453 |
The overall mortality among all RACHS-1 classes was 4.5%. The risk for mortality increased with each increase in RACHS-1 category, from 1.6% for class 1% to 21.7% for classes 5 and 6 (see Figure 1 ), although the mortality rate was not significantly different between classes 1 and 2 (p = 0.138) and between classes 2 and 3 (p = 0.063). LOS and ICU stay increased with RACHS-1 class, although there was no significant difference between LOS and ICU stay for classes 2 and 3 (see Figure 2 ). Increases in total hospital charges and cost were correlated with RACHS-1 class ( Figure 3 ). The overall mean cost-to-charge ratio for all RACHS-1 classes was 0.340 ± 0.001. The mean cost-to-charge ratios remained relatively constant between 0.334 ± 0.003 and 0.344 ± 0.003, without correlation to RACHS-1 category.
There was no difference in mortality by gender or race, except for significantly higher mortality in Native American patients (10%, p <0.001). However, compared with their white counterparts, nonwhite patients were more likely to have LOS >30 days (OR 1.21, p <0.001). Hispanic patients had a longer mean LOS (23.0 ± 0.6 days, p <0.001; Table 3 ) and were the most likely to have LOS >30 days (OR 1.37, p <0.001; Table 4 ). Similarly, nonwhite patients were more likely to have longer stays in the ICU compared with the white cohort (OR 1.20, p <0.001). Hispanics had the longest ICU stays (14.9 ± 0.5 days, p <0.001) and were the most likely to have ICU stays >24 days (OR 1.27, p <0.001). Black patients had significantly longer LOS (20.4 ± 0.7 days, p <0.001) compared with white patients, and although there was no difference in mean ICU stay between these 2 cohorts, black patients were more likely have longer ICU stays (OR 1.17, p = 0.017).
