Association Between Implementation of a Cardiovascular Step-Down Unit and Process-of-Care Outcomes in Patients With Congenital Heart Disease
Background: The Joint Commission’s 2009 National Patient Safety Goals aimed to improve identification of and response to clinical deterioration in hospital-ward patients. Some hospitals implemented intermediate-care units for patients without intensive care–level support needs. No studies have evaluated what effect changes associated with a move to a pediatric cardiovascular step-down unit (CVSDU) has on process-of-care outcomes.
Methods: A retrospective cohort study comparing process-of-care outcomes in units caring for children with congenital heart disease (n = 1415) 1 year before (July 1, 2010–June 30, 2011) and 1 year after (August 1, 2011–July 30, 2012) implementation of a CVSDU following the move to a new hospital building. Units caring for noncardiac tracheostomy and/or ventilator-dependent patients were used as controls (n = 606). Primary outcomes included length of stay (LOS) and transfers to higher levels of care. Secondary outcomes included rapid response team, cardiopulmonary arrest, and code blue rates. Mann-Whitney U and z tests were used for all analyses.
Results: When compared with a medical-surgical unit, cardiac patients admitted to a CVSDU had a significantly decreased total LOS (median 7.0 vs 5.4 days, P = .03), non-ICU LOS (median 3.5 vs 3.0 days, P = .006), and rapid response team/code blue rate per 1000 non-ICU patient days (11.2 vs 7.0, P = .04). No significant differences in primary or secondary outcomes were seen within the control group.
Conclusions: Changes associated with a new CVSDU were associated with decreased LOS and lower rates of rapid response and code blue events for patients with congenital heart disease.
In December 2004, the Institute for Healthcare Improvement recommended a change in the way we approach patient safety and quality, known as the 100 000 Lives Campaign.1 This campaign led to the implementation of 6 evidence-based interventions, focused on creating achievable goals for all children’s hospitals across the country while generating pressure to perform at the highest level. Goal 16 of the Joint Commission’s 2009 National Patient Safety Goals was to improve the identification of and response to clinical deterioration in hospital-ward patients.2
To further reduce morbidity and mortality, some hospitals have implemented intermediate, or subacute, care units for patients who do not warrant intensive care–level support. These patients are often admitted to specific units that have differing levels of nursing support, monitoring capabilities, staff training, and access to intensive care–level assistance. Few studies have evaluated admission and discharge guidelines for these types of units.3–5 One study showed an improvement in overall costs and length of stay (LOS) for patients admitted to a neuro-interventional step-down unit; however, this study only included adults compared with similar patients in an intensive care setting.4 Another study evaluated the cost-effectiveness of a noninvasive respiratory care unit, finding a positive effect.5 However, no pediatric patients were included and only comparisons with an ICU were made. To the best of our knowledge, no studies have evaluated what effect changes associated with a move to a pediatric cardiovascular step-down unit (CVSDU) has on decreasing LOS as well as decreasing cardiopulmonary arrest (CPA) and transfer to higher level of care rates.
At Children’s Hospital Los Angeles (CHLA), we implemented a new CVSDU after moving to a new hospital building in July 2011. We aim to show that patients with congenital heart disease admitted to this specialized non-ICU CVSDU (versus a standard medical-surgical unit) would have decreased LOS, fewer CPAs, and acute respiratory compromises (ARCs), as well as decreased transfers to higher level of care.
This study was performed at CHLA, a tertiary care, pediatric academic center located in an urban setting. CHLA has ∼11 000 inpatient discharges per year and 60 000 emergency department visits. Approximately 75% of the patients have public insurance. Between January 2008 and June 2011, 529 (42%) of 1260 rapid response team (RRT) and code blue events occurred on patients with congenital heart disease. Patients with tracheostomy and/or ventilator dependence comprised the second largest group using these resources. In July 2011, a new hospital building opened, expanding the number of licensed beds from 286 to 365. Most of the increased bed availability was seen in the ICU and oncology unit, with little change seen on the medical-surgical units (126 before the move, 128 after). The new hospital included a dedicated CVSDU, where patients with congenital heart disease were admitted. Tracheostomy and/or ventilator-dependent patients remained cohorted on a medical-surgical unit. With the implementation of an attending-only hospitalist service during our study period, a small percentage of the tracheostomy/ventilator patients were still followed by residents after the new hospital move. Also, because of changes in the patient coverage model in 2011, there was no dedicated inpatient cardiology exposure for residents after the new hospital move. Therefore, no residents followed the cardiac patients in the new CVSDU. Cardiac patients in the PICU or cardiothoracic ICU (CTICU) were followed exclusively by cardiology and/or critical care medicine fellows and attending physicians. Tables 1 and 2 describe the differences before and after the new hospital move between the study and control groups, respectively.
We performed a retrospective cohort study comparing process-of-care outcomes in hospital units caring for patients with congenital heart disease (n = 1415) 1 year before (July 1, 2010–June 30, 2011) and 1 year after (August 1, 2011–July 30, 2012) the implementation of a CVSDU, which coincided with a move to a new hospital building. Units providing care for noncardiac tracheostomy and/or ventilator-dependent patients (n = 606) were used as our control, as they represented another complex patient population with a similar pre-CVSDU coverage model but did not receive a specialized unit after the new hospital move. Similar to the study population, our control groups were being compared with each other before and after hospital move. Therefore, rather than 1 type of analysis comparing changes in the study group with changes in the control group, 2 separate analyses (1 for our study group and 1 for our control group) were performed during our study period.
Although RRTs may not necessarily be considered an adverse event, they are a surrogate for patient deterioration and were combined with code blue events for the purpose of this study. RRT and code blue events were classified by using the Children’s Hospital Association definition as ARC (requiring noninvasive ventilation during the emergency response, and continued during transfer to higher level of care) or CPA (pulselessness or a pulse with inadequate perfusion requiring chest compressions and/or defibrillation) events. The study protocol was approved by the Committee on Clinical Investigations at CHLA. This institutional review board determined that informed consent was not required.
Before the opening of the new CVSDU, an expert panel at CHLA was tasked with identifying specific International Classification of Diseases, Ninth Revision (ICD-9) diagnosis codes that would require admission to this unit, rather than possible admission to a standard medical-surgical unit. This panel consisted of cardiologists, cardiothoracic surgeons, intensive care physicians, pediatric hospitalists, and nursing leadership. These diagnoses included surgical (pre- and postoperative) and nonsurgical cardiac conditions that were felt to warrant closer monitoring on a specialized unit (Table 3). Patients on inotropes awaiting cardiac transplantation and those with ventricular-assist devices were admitted to the CTICU, whereas patients who were postoperative from a Norwood procedure were still eligible for CVSDU admission. Patients with other conditions not listed in Table 3 did not require CVSDU admission and could be admitted to a standard medical-surgical unit and, therefore, were excluded from our study. Using this completed list of ICD-9 codes that required admission to the CVSDU, we were able to identify all similar cardiac patients who were admitted to the old hospital building up to 1 year before the new hospital move. To verify that all included cardiac patients truly met inclusion criteria, we performed chart reviews on a random sample of 30% of all admissions during January 2011 and January 2012 (n = 50) by using the completed ICD-9 inclusion list. We confirmed a correct primary diagnosis in 49 (98%) of 50 charts reviewed. Specific ICD-9 codes were used to identify noncardiac patients with tracheostomy and/or ventilator dependence (V46.11, V44.0, 519.00–519.02, 519.09, 518.81–518.84, 770.84, 997.31) admitted both before and after hospital move.
All cardiac patients directly admitted to the designated medical-surgical unit (old hospital) or CVSDU (new hospital) during the study period, regardless of whether they were transferred to an ICU later in their hospitalization, were included. All critically ill surgical cardiac patients in the old and new hospital building were directly admitted to the CTICU from the emergency department. Most critically ill nonsurgical cardiac patients were admitted to the PICU in the old hospital and CTICU in the new hospital. Given the difference in location and possible variability in care provided, the nonsurgical cardiac patients transferred from the ICU to the designated medical-surgical unit or CVSDU were excluded from our final analysis. However, surgical cardiac patients transferred out of the CTICU remained in our study population. No data were collected during the month of the new hospital move (July 2011).
Summary data from each unit were provided to the study team, and outcomes before and after July 2011 were compared within each group. Primary outcomes included LOS (total and non-ICU) and transfers to higher level of care. LOS data were collected in minutes (ie, admitted at 9:30 on 8/4/12 and discharged at 14:35 on 8/7/12). Secondary outcomes included RRT, code blue, CPA, and ARC rates, which were calculated as number of events per 1000 non-ICU patient days. Because of heavily skewed, non-normal distributions, changes in LOS were analyzed by using Mann-Whitney U tests. Changes in transfer, RRT, code, and CPA rates were analyzed by using z tests.6,7 The minimum detectable effect size for the study group, based on the number of patients available for analysis, corresponded to differences in LOS of ∼0.4 days (9 hours 36 minutes) and event rates of 3 to 4 per 1000 non-ICU patient days. Our smaller control group allowed us to detect effects corresponding to differences in LOS of ∼1.25 days (30 hours) and event rates of 5 to 7 per 1000 non-ICU patient days. We also examined individual patient severity by comparing the mean case mix index (CMI) within each unit before and after the hospital move.
A total of 681 patients (5262 non-ICU patient days) with congenital heart disease were admitted within 1 year before the new hospital move. A total of 735 patients (4694 non-ICU patient days) were admitted within 1 year after. A total of 284 control patients (3651 non-ICU patient days) and 322 (3650 non-ICU patient days) were admitted 1 year before and 1 year after the new hospital move, respectively. Table 4 compares the demographic differences between the study and control groups. The patients in the study group were significantly younger than the controls; however, that was expected given the nature of the disease processes being studied. The patients in the study group were also significantly more likely to be white and privately insured than control subjects, who were more likely to be Hispanic and publicly insured. Most study participants within the “other” group under race in Table 4 likely reflects the large Latino community served by our institution. There were no other differences in gender distribution between the 2 groups. The list of the top 9 surgical and/or nonsurgical cardiac diagnoses admitted before and after implementation of the CVSDU is located in Table 5; there were no statistical differences in the distribution of these diagnoses over time (P = .85). Top diagnoses for tracheostomy and/or ventilator-dependent patients were also similar (data not shown).
Compared with cardiac patients treated in the medical-surgical unit before the hospital move, cardiac patients admitted to a CVSDU had a significantly decreased non-ICU LOS (median 3.5 vs 3.0 days, P = .006), total LOS (7.0 vs 5.4 days, P = .03), and RRT/code blue rate per 1000 non-ICU patient days (11.2 vs 7.0, P = .04) (Table 6). There were no differences in before/after hospital move transfer rates within the study or control population. The control group showed a nearly significant difference in total LOS (P = .051). No other differences in primary or secondary outcomes were seen within the control group (Table 7). There was a half-day decrease in non-ICU LOS within the control group (7.4 days before move vs 6.8 days after move; P = .15), but this was not statistically significant. Similarly, a decrease of 1.8 RRT/code blue events per 1000 non-ICU patient days within the control group did not reach statistical significance (P = .54). The mean CMI for the CVSDU was 5.1, compared with the pre-move cardiac unit mean CMI of 4.8 (P = .20). The mean CMI for our control group was 3.5 before the hospital move and 3.3 after (P = .59).
While evaluating the effect of changes associated with a new pediatric CVSDU, children with congenital heart disease admitted to such a unit had a significantly shorter total and non-ICU LOS and a lower rate of rapid response and code blue events compared with similar cardiac patients admitted to a standard medical-surgical unit. No comparable differences were seen in tracheostomy and/or ventilator-dependent patients, suggesting against general improvements in care after the new hospital move. Few studies describe the effects of changes associated with new step-down units on patient outcomes, and no study has looked at the effects on a complex pediatric population, such as congenital heart disease. Replicating these results in other institutions would require the presence of a fully staffed intermediate-care unit. It may be difficult for hospitals that do not have such a unit, or cannot afford to implement one. However, it does provide a framework for improving process-of-care outcomes.
The significant decrease in non-ICU LOS must be evaluated in the context of total LOS data. For the study group, ICU LOS and total LOS both decreased after implementation of the CVSDU, with only total LOS reaching statistical significance. This suggests that the improvement in non-ICU LOS was not at the cost of worsening ICU stays. Similarly, there was no significant change in the ICU or total LOS for our control population.
Another factor that may have contributed to the decreased LOS in our study population was that patients admitted to the CVSDU were exclusively followed by hospitalists or nurse practitioners. Previous literature has shown that pediatric hospitalist or nurse practitioner models can lead to decreased LOS for common as well as medically complex patient populations.8–13 However, we would have expected to have seen a significant decrease in non-ICU LOS for our control group as well, given that 90% of those patients were followed exclusively by hospitalists after the new hospital move.
Through improved proximity to the CTICU, specific patient concerns could be addressed in a more real-time environment with cardiac intensivists. This may have led to quicker assessments and interventions, possibly leading to fewer acutely deteriorating patients. This possible decrease in acute care management, along with the significant decrease in RRT and code blue events in the CVSDU, may have contributed to decreased LOS. Without the ability to have ongoing communication with the intensivists, cardiac patients in the old hospital building may have progressed to clinical states that required interventions resulting in prolonged hospitalizations.
There are some limitations to this study. Data were unit-based rather than individual, so analytic approaches were limited. Therefore, it was not possible to examine effects of individual patient characteristics (eg, diagnosis) on outcomes. Additionally, whenever there is a new hospital as part of the study design, there are some inherent changes strictly due to the new building that cannot be analyzed. However, by including a complex patient population as controls that did not receive their own step-down unit, we feel that we were able to substantiate our findings. With no comparable reductions in event rates seen within the control unit, we have some evidence that the new hospital move itself did not affect the care received or provided.
Because of a significant discrepancy in bed availability between the old and new CTICU (15 vs 24 beds), there was often pressure to transfer patients out of the former CTICU, and this tendency was reduced in the new building with more CTICU beds. However, there was no difference in the ICU LOS between the 2 time periods, suggesting patients were not being kept longer in the new CTICU. The lack of statistical significance for CMI within each group before and after the new hospital move suggests that patients were not any less ill after the expansion. However, given that it can be difficult to gauge severity of illness with 100% accuracy, there remains the possibility that this may have affected the LOS and RRT rate data in the CVSDU. At times, cardiac patients would be transferred to the ICU without a preceding RRT or code blue. Given the low occurrence of these transfers between the 2 groups of cardiac patients both before and after hospital move, it likely did not have a major impact on the overall data; however, we were unable to fully account for this possibility.
Given the smaller number of patients available in the control population, we were unable to detect statistical significance in differences of event rates <5 to 7 per 1000 non-ICU patient days or LOS <30 hours. It is likely that the lack of statistical significance for the LOS differences in the control unit may be attributed to lower statistical power in this group, potentially indicative of an overall LOS reduction accompanying the move to the new hospital. It may further be the case that the difference in event rates seen within our control group is as clinically meaningful as the differences seen in the study group, despite the lack of statistical significance in the control group. We also were unable to differentiate surgical versus medical admissions within our study group before versus after new hospital move. This could have allowed us to better understand the differences seen in event rates within our study group.
Despite these limitations, results from this study suggest that increased hospitalist coverage, more telemetry monitoring capabilities, easier access to intensive care–level assistance, and dedicated respiratory care practitioners within our specialized step-down unit could provide a model for efficiency and improved process-of-care outcomes for patients with congenital heart disease.
It will be important to show that improvement in specific process-of-care outcomes, such as LOS, can have long-term sustainability. Evaluating cardiac-specific pediatric early warning sign algorithms in addition to interventions already analyzed in our study would provide additional information into how best to maintain desired outcomes. Plans to create a specialized unit in our institution for tracheostomy and/or ventilator-dependent patients will provide an opportunity to demonstrate improvements across different subsets of complex patient populations. A separate cost-benefit analysis also will be very informative in better understanding the reasons for the associations identified in our study.
Changes associated with the implementation of a CVSDU for patients with congenital heart disease can be associated with decreased LOS as well as decreased rates of RRT or code blue events. There are no other studies evaluating what the effect of changes associated with an intermediate-care unit such as ours can have on certain patient outcome measures. Our study can provide a framework for improving process-of-care outcomes in a complex patient population.
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.
- acute respiratory compromise
- Children’s Hospital Los Angeles
- case mix index
- cardiopulmonary arrest
- cardiothoracic intensive care unit
- cardiovascular step-down unit
- International Classification of Diseases, Ninth Revision
- length of stay
- rapid response team
- Richards BF,
- Fleming JB,
- Shannon CN,
- Walters BC,
- Harrigan MR
- Weiers RM
- Jekel JF,
- Katz DL,
- Elmore JG,
- Wild DMG
- Bellet PS,
- Whitaker RC
- Dwight P,
- MacArthur C,
- Friedman JN,
- Parkin PC
- Landrigan CP,
- Conway PH,
- Edwards S,
- Srivastava R
- Srivastava R,
- Landrigan CP,
- Ross-Degnan D,
- et al
- Copyright © 2015 by the American Academy of Pediatrics