Abstract
Background:
Factors leading to cardiac intensive care unit (CICU) readmission and the impact on mortality have yet to be well delineated. We sought to define the prevalence and outcome for unscheduled CICU readmission. Secondary objectives were to identify indications and risk factors for unscheduled CICU readmission.
Methods:
Retrospective analysis of prospectively collected registry data at a tertiary care children’s hospital. Pediatric and adult patients with congenital and acquired heart disease who survived to initial CICU discharge were included. Patients with unexpected return to the CICU for acute change in clinical status were defined as unscheduled readmissions.
Results:
Of the 645 discharges that met inclusion criteria, 37 resulted in unplanned readmission to the CICU. Patients requiring unscheduled readmission had higher mortality rates (16.2% vs 0.5%, P < .0001). Cardiac symptoms were the most common reason for readmission. On multivariate analysis, genetic anomaly (P = .001) and longer length of stay (LOS) during the index CICU admission (P = .01) were independently associated with readmission. For surgical patients, genetic anomaly (P = .001), single-ventricle anatomy (P = .05), and longer surgical support time (P < .001) were independently associated with readmission.
Conclusion:
Unscheduled readmission to the CICU within the same hospitalization was uncommon but associated with a higher mortality rate. Genetic anomaly and longer initial LOS were important risk factors for the entire cohort. Single-ventricle anatomy and longer intraoperative course were risk factors for surgical readmissions.
Introduction
In recent years, many tertiary pediatric cardiac programs have noted improved survival for critically ill patients with acquired and congenital heart disease. 1–2 However, our quality improvement efforts have shown that after recovery and transfer from the cardiac intensive care unit (CICU), approximately 5% of patients are unexpectedly readmitted to the CICU within the same hospitalization (internal data). Impact on mortality and the factors leading to CICU readmission have yet to be well delineated. There are many variables that may affect the need for readmission to the CICU including intrinsic patient characteristics, such as age, weight, and type of heart disease; complications in their initial CICU course; operative procedures performed; timing of CICU discharge; or human factors such as nurse staffing and physician experience level.
The primary objectives of this study were to define the prevalence and outcome for unplanned CICU readmission. The secondary objectives were to identify indications and risk factors for readmission with the broader goal of determining whether there are modifiable factors which could be addressed to improve outcomes for readmitted CICU patients.
Materials and Methods
This study is a retrospective analysis of prospectively collected registry data. The study was approved by the institutional review board at the Children’s Hospital of Philadelphia. All patients admitted to the CICU at the Children’s Hospital of Philadelphia from July 1, 2011, through June 30, 2012, were eligible and identified by review of the Virtual PICU System (VPS) database. Those patients with unexpected return to CICU for acute change in clinical status were defined as unscheduled readmissions.
Within our institution, we have a 26-bed CICU and 25-bed cardiac care unit (CCU), which serves as our step-down unit to which most CICU patients are transferred when stable. The CICU and CCU are separate teams staffed by different attending physicians, fellows, and frontline providers. Prior to transfer of a patient out of the CICU, verbal handoff is completed from attending to attending, from the CICU fellow to the CCU fellow and frontline providers, and from CICU nurse to CCU nurse. No patient can be transferred out of the CICU without review from the CICU attending. To be eligible for transfer to the CCU, a patient must be stable from a cardiorespiratory standpoint. The CCU staff can initiate high-flow nasal cannula, continuous positive airway pressure, and vasoactive agents, such as milrinone and dopamine, but the patient is then transitioned back to the CICU for change in clinical status. Those weaning off high-flow nasal cannula or on stable inotropic infusions may be transferred to the CCU. Arterial blood pressure monitoring, central venous pressure monitoring, near infrared spectroscopy, and temporary pacing are not performed in the CCU. Procedures requiring sedation are not performed in the CCU. The CCU does care for ventricular assist device (VAD) patients once stable postoperatively and in anticipation of transition to home or while awaiting heart transplantation as an inpatient. The CCU is staffed 24 hours a day with in-house residents, nurse practitioners, hospitalists, and a pediatric cardiology fellow. These frontline providers are supervised by pediatric cardiologists. The CICU is staffed 24 hours a day with an in-house attending cardiac intensivist, pediatric cardiology and critical care fellows, hospitalists, and nurse practitioners. Based on their medical needs, the CICU patients are occasionally transferred from the CICU to other units within the institution.
Inclusion criteria for this study included all medical and surgical CICU patients with congenital and acquired heart disease who survived to initial CICU discharge. Patients undergoing a surgical procedure during the index hospitalization were classified as surgical patients and the remainder was classified as medical patients. Pediatric and adult patients were included. Patients who were discharged to home directly from the CICU were excluded. For patients who had multiple unscheduled readmissions during the same hospitalization, data from their first unscheduled readmission were included. For included patients, the VPS and Society of Thoracic Surgeons Congenital Heart Database (STS-CHD) databases were used to extract baseline data, including demographic data and data from the CICU episodes of care. Details of cardiac procedures were obtained from the STS-CHD database. Further details of the patient’s hospital course, including test results and details of cardiac anatomy, were obtained from the electronic medical record. For time of transfer, night transfer was defined as transfer between 7
Statistical Analysis
Cohort characteristics and outcomes were summarized using standard descriptive statistics and expressed as mean ± standard deviation for normally distributed continuous variables, median (with range) for skewed continuous variables, or count (with percentage of total) for categorical variables. Associations between potential risk factors and primary outcome (unscheduled CICU readmission) were tested initially by univariate logistic regression with the point estimate of the effect size expressed as an odds ratio (with 95% confidence interval). Univariate logistic regression testing of associations between covariates and outcomes was also performed for the mutually exclusive subgroups of medical and surgical patients. For regression testing, natural log-transformed values were used for skewed continuous variables (age, weight, and length of stay [LOS]). Multivariate logistic regression was then performed to identify covariates independently associated with outcome for the overall cohort as well as for the medical and surgical subgroups. Covariates with P < .2 in univariate testing were considered for inclusion in the final multivariate model in a forward stepwise fashion and retained if P < .05. Correlation testing was performed to screen for highly collinear covariates, and only a single covariate from a collinear cluster was considered for inclusion in the final multivariate model. In the event that more than one collinear covariate could potentially be included, the covariate with the best goodness of fit measured in post hoc testing was used. For the medical subgroup, there were two covariates with P < .2 in the univariate model. When entered into a multivariate model, both the P values were >.05. Hence, there was no meaningful multivariate model, so multivariate results will not be presented for that group. Statistical significance was established using a two-tailed P value < .05. All statistical analyses were performed using STATA v10 (Stata Corp, College Station, Texas).
Results
There were 994 CICU discharges for the study period of July 2011 to June 2012 (Figure 1). Three hundred one discharges were excluded due to death during initial CICU stay, discharge directly to home, or need for scheduled CICU readmission after a procedure within the same hospitalization. Patients discharged directly to home were either high-risk patients (such as those with pulmonary hypertension) admitted for a short period of observation after minor procedures requiring sedation such as diagnostic cardiac catheterization and minor noncardiac surgery or otherwise healthy patients status post relatively straightforward cardiac surgical procedures such as atrial septal defect closure. Six hundred eight discharges did not require CICU readmission and served as the control group; 31 discharges required single, unscheduled CICU readmission (accounting for 62 total discharges); and 6 discharges required a total of 23 multiple, unscheduled CICU readmissions. For the study and control groups, 396 patients were surgical and 249 were medical.
Cardiac intensive care unit (CICU) discharges for 2011 to 2012.
Thirty-seven patients had one or more unscheduled CICU readmissions within the same hospitalization. Median time to readmission was 2.9 days (0.1-82.8 days). Twenty-four percent (9 of 37) of those were readmitted within 24 hours. Thirty-eight percent (14 of 37) were readmitted after 7
Readmission Indications.
Abbreviations: AVVR, atrioventricular valve regurgitation; PA, pulmonary artery; MCA, middle cerebral artery; VAD, ventricular assist device.
Table 2 compares patients based on unplanned readmission status. The presence of a genetic anomaly (P = 0.001) and longer LOS during the index admission (P = .01) were independently associated with unplanned readmission in multivariate testing. Overall mortality rate for those with unscheduled readmission was 16.2%. Patients not requiring CICU readmission had a mortality rate of 0.5% (P < .0001).
Differences Based on Readmission Status for the Entire Cohort (N = 645) by Logistic Regression.a
Abbreviations: CI, confidence interval; CICU, cardiac intensive care unit; OR, odds ratio.
aData expressed as median (range) or count (% of total).
bNatural log-transformed values used for regression testing.
For further analysis of risk factors, the entire cohort was divided into surgical (n = 396) and medical patients (n = 249). Table 3 shows the correlation of surgical factors found to be significantly associated with unscheduled readmission on univariate testing. Due to collinearity, for the purposes of analysis, total support time (cardiopulmonary bypass time + deep hypothermic circulatory arrest time) was used as a measure of operative length and complexity. Society of Thoracic Surgeons-European Association of Cardiothoracic Surgery Mortality Category was used as a measure of surgical complexity. Table 4 compares surgical patients who required unscheduled readmission with those who did not require CICU readmission. The presence of a genetic anomaly (P = .001), single-ventricle physiology (P = .05), and longer total support time (P < .001) were identified as factors independently associated with unplanned readmission.
Correlation Coefficients for the Potentially Collinear Covariates Identified in Univariate Testing as Significantly Associated With Unplanned Readmission Among Surgical Patients.a
Abbreviations: STAT, Society of Thoracic Surgeons-European Association of Cardiothoracic Surgery Mortality Category.
aThe correlation coefficients were generated with Pearson correlation testing. The values in the Table 3 are rho (r) values.
Differences Based on Readmission Status for the Surgical Patients (n = 396) by Logistic Regression.
Abbreviations: CI, confidence interval; CICU, cardiac intensive care unit; OR, odds ratio; STAT, Society of Thoracic Surgeons-European Association of Cardiothoracic Surgery Mortality Category.
aData expressed as median (range) or count (% of total).
b Natural log-transformed values used for regression testing.
Table 5 compares medical patients who required unscheduled CICU readmission with those who did not require readmission. Again, there was a higher prevalence of genetic anomalies among the unscheduled readmissions, although this difference did not attain statistical significance (P = .08). There was an association between weight and risk of readmission (P = .05).
Differences Based on Readmission Status for the Nonsurgical Patients (n = 249) by Logistic Regression.
Abbreviations: CI, confidence interval; CICU, cardiac intensive care unit; OR, odds ratio.
aData expressed as median (range) or count (% of total).
cNatural log-transformed values used for regression testing.
bFrom univariate logistic regression (multivariate modeling not performed as no covariates with P < .05 in univariate testing).
We reviewed physician staffing for all patients in both the readmission and the non-readmission groups. The CICU attending was grouped into categories by experience level. There was no significant difference in experience level of the transferring CICU attending in the readmitted group compared to the non-readmitted group (Tables 2, 4, and 5). Three surgeons performed all cardiac operations during this period, and they were simply labeled by letters. Surgeon was not independently associated with unscheduled readmissions (Table 4).
The time of day for the index transfer was documented. For the unscheduled readmission group, 11% (4 of 37) were transferred out of the CICU after 7
Discussion
In this study of over 600 discharges from the CICU over the course of one year, approximately 4% of patients discharged from the CICU required unscheduled readmission to the CICU within the same hospitalization. These patients had significantly higher mortality rates compared to those not requiring unscheduled readmission.
Patients were most likely to be readmitted to the CICU with cardiac symptoms, such as residual cardiac lesions, arrhythmias, or heart failure. This is not unexpected as cardiac disease was the primary reason for admission to the hospital and residual lesions, such as atrioventricular valve regurgitation or postoperative ventricular dysfunction, are not well tolerated and may not be amenable to conservative medical management.
For the entire cohort, the presence of a genetic syndrome and longer initial LOS were associated with a higher rate of readmission. For patients who required surgery during the index hospitalization, in addition to genetic syndrome, single-ventricle anatomy, and longer intraoperative support time were associated with unscheduled readmissions. Genetic anomalies, single-ventricle anatomy, and longer intraoperative support times have been associated with higher morbidity and mortality during the perioperative phase, so need for CICU readmission likely reflects the need for ongoing management of secondary disease processes. 3 –16 Longer LOS during their index CICU admission likely reflects the complexity of these patients.
Regarding the impact of human factors, the experience level of the transferring CICU attending or surgeon did not affect unscheduled readmissions. We did look at the time of transfer out of the CICU and time of readmission back to the CICU. Most patients were transferred out of the CICU during daytime hours, and most of those requiring readmission were transferred back during daytime hours. Interestingly, 41% were transferred back to the CICU during weekend hours.
Bastero-Minon et al have published the only other study investigating pediatric cardiac patients readmitted to the CICU within the same hospitalization. 17 Over a seven-year period, they found that 112 (2.4%) of 4,625 pediatric patients were readmitted to the CICU from the ward within 72 hours of CICU discharge. They found a lower mortality rate for readmitted patients than that was found in the current study (8% vs 16.2%). They limited their study group to those readmitted within 72 hours, whereas we did not in order to get a more comprehensive look at patients requiring readmission at any point in their hospitalization. They found that most patients were readmitted for respiratory symptoms. Since their study period was 7 years, this is not likely due to seasonal differences. Perhaps the difference is due to the time course of readmission, as residual cardiac lesions may manifest more than 72 hours after CICU transfer. Furthermore, they did not explain how patients with hypoxia were categorized. In our study, if a patient readmitted with hypoxia was found to have a residual anatomic lesion, such as Blalock Taussig shunt obstruction, that patient was categorized as a cardiac readmission rather than respiratory. They did not compare readmitted patients to those not readmitted.
Bastero-Minon et al also looked at the time of transfer and readmission. They similarly found that most of the readmission patients were originally transferred out of the CICU during the daytime. We found a lower number of patients were readmitted to the CICU during the evening hours (51.8% vs 38%), but this is likely due to the differences in the definition, as they defined evening hours as 5
Most studies focusing on CICU readmission look solely at adult surgical patients. Similar to our study, adult patients readmitted to the CICU had higher mortality rates. 18 –20 Predictors of CICU readmission included complex surgery, heart failure, arrhythmias, renal failure, respiratory complications including mechanical ventilation >24 hours during initial CICU admission, reexploration for bleeding, and prolonged initial LOS. These studies reinforce the increased mortality rates associated with readmission. Although some risk factors are similar across the adult and pediatric populations, such as surgical complexity and initial CICU LOS, other factors are difficult to compare as the underlying disease processes, including comorbidities, are different.
The question certainly arises whether or not changes can be made in order to decrease the need for unscheduled CICU readmission and the subsequent mortality. As there is faster recognition of a child in extremis and faster access to rescue treatments within an intensive care unit (ICU), it is possible that observing high-risk patients, such as single-ventricle patients or patients with genetic anomalies, in the CICU for longer periods before transfer to a step-down unit will decrease the mortality rates. It is unclear, however, how long that period of observation should be. Establishing transfer criteria may make timing of transfer less ambiguous. We also noted high rates of patient transfer back to the ICU on weekends. This may be due to changes in staffing, such as fewer physicians, nursing experience level, and/or nurse staffing such as nurse–patient ratios. We did not investigate these factors, but this warrants further study. It is possible that changes in staffing would lead to more expeditious recognition of a child requiring additional care and ultimately improve outcomes.
Strengths of this study include a large number of CICU discharges which were analyzed. This patient population was quite varied, as all ages and all types of cardiac disease were included. We also did not limit the time between initial CICU discharge and readmission in order to get a more complete look at the variables associated with readmission. Limitations include that, although patients were collected prospectively as part of a CICU registry, some data that were not collected for the purposes of the registry were obtained by retrospective chart review. Although a large number of CICU discharges were analyzed, approximately one-third of patients were excluded and most patients did not in fact require readmission. This led to a small number of readmissions eligible for the final analysis. Also, although patients requiring unscheduled readmission had higher mortality rates, this does not prove causality.
Conclusions
Although outcomes of pediatric cardiac patients have improved over the past several decades, interventions are needed to further improve the outcomes of these complex patients. In this study, unscheduled readmission to the CICU within the same hospitalization was uncommon but associated with a higher mortality rate. The presence of a genetic anomaly, single-ventricle anatomy, longer intraoperative course, and longer initial LOS were the most significant risk factors for readmission. This study identifies potential target areas for reducing the risk of CICU readmission. Further studies are also warranted to define the impact of human factors and patient complications on the need for readmission.
Footnotes
Acknowledgment
The authors would like the thank Lisa McCormick Jones, BSN, RN, CCRC, and Elizabeth McBride, BA, CCRC, for their assistance in data collection.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Departments of Anesthesiology/Critical Care Medicine, Cardiology, and Cardiac Surgery at the Children’s Hospital of Philadelphia.