Author + information
- Received September 23, 2016
- Revision received January 12, 2017
- Accepted February 10, 2017
- Published online April 17, 2017.
- aDepartment of Medicine, University of Virginia Health System, Charlottesville, Virginia
- bCenter for Advanced Medical Analytics, University of Virginia Health System, Charlottesville, Virginia
- ↵∗Address for correspondence:
Dr. Eric M. Holland, Department of Medicine, University of Virginia Health System, PO Box 800158, Charlottesville, Virginia 22908-0158.
Background Fifty years after the inception of the cardiac intensive care unit (CICU), noncardiovascular illnesses have become more prevalent and may contribute to morbidity and mortality.
Objectives The authors performed multivariate statistical analyses to determine the association of acute noncardiovascular illnesses with outcomes, including length of stay (LOS), mortality, and hospital readmission.
Methods We studied 1,042 admissions between October 12, 2013 and November 28, 2014 to the CICU at the University of Virginia Health System, a tertiary-care academic medical center. Through systematic inspection of individual charts, we identified primary and secondary diagnoses, vital sign measurements, length of stay (LOS), hospital readmissions, and mortality.
Results The most common primary diagnosis was acute coronary syndrome (25%), which consisted of both non−ST-segment elevation acute coronary syndrome (14%) and ST-segment elevation myocardial infarction (11%). Sepsis was the most frequent noncardiovascular primary diagnosis (5%), but it only occurred in 16% of all admissions. Acute kidney injury and acute respiratory failure each occurred in 30% of admissions. One-half of all admissions (n = 524; 50%) were marked by acute respiratory failure, acute kidney injury, or sepsis. Median LOS in the CICU and the hospital were 2 days (interquartile range [IQR]: 1 to 5 days) and 6 days (IQR: 3 to 11 days). Mortality was 7% in the CICU and 12% in the hospital. Of the 920 patients who survived to hospital discharge, 171 (19%) were readmitted within 30 days. Sepsis, acute kidney injury, and acute respiratory failure were associated with mortality. Acute kidney injury, acute respiratory failure, and new-onset subclinical atrial fibrillation, which occurred in 8% of admissions, were all associated with CICU LOS.
Conclusions Many patients in the modern CICU have acute noncardiovascular illnesses that are associated with mortality and increased LOS.
In 1961, Julian (1) presented the first description of the coronary care unit to the British Thoracic Society. Recognizing the association of cardiac arrest with acute myocardial ischemia, these early visionaries established a unit equipped with an electrocardiography (ECG) alarm system and specialized staff trained in prompt treatment of peri-infarction arrhythmias (2). Initial coronary care unit reports received mixed reviews, with challengers believing it was irresponsible to suggest that all myocardial infarction patients should be admitted to a critical care unit (2). However, the coronary care unit concept quickly spread after Killip and Kimbal (3) reported an approximate 20% mortality reduction with this care model.
More than 50 years later, a very different landscape from the one proposed by Julian has evolved. The field of medicine, and in particular, cardiovascular medicine, has progressed exponentially with improvements in diagnostic studies and treatments, which has lead to a population that is living longer, but one that has comorbidities. This once-predominant post–acute coronary syndrome observation unit has evolved into a heterogeneous population rich in both cardiovascular and noncardiovascular illnesses. Katz et al. (4) demonstrated a significant temporal change in the characteristics of cardiac intensive care unit (CICU) patients, with substantial increases in illness severity, comorbid critical illness, and noncardiac procedures.
Yet, as the once-defined lines of the CICU continue to blur with those of the medical and surgical intensive care units (ICUs), we have noted a surprising lack of data on the contemporary CICU related to demographics, length of stay (LOS), mortality, and readmissions. These data are imperative in understanding how to improve patient outcomes, while decreasing overall costs as we transition to a more value-based health care model in which quality measures and readmissions are linked to reimbursements.
We tested the hypothesis that acute noncardiovascular conditions are independently associated with patient outcomes, including LOS, mortality, and readmission.
We studied consecutive admissions to the CICU at the University of Virginia Health System with available continuous ECG data from October 12, 2013 to November 28, 2014. Teams of medicine residents, cardiology fellows, and attending cardiologists staff the 10-bed unit. Physician staffing at nights is reduced, but both medicine residents and a cardiology fellow provide care around the clock under the supervision of attending cardiologists. A CICU or advanced heart failure attending cardiologist and corresponding housestaff assume primary responsibility for all patients admitted or transferred to the CICU. Critical care consultation is available, but not mandated. We followed patients through July 21, 2015 to capture all subsequent hospital admissions to the University of Virginia. Each patient in the CICU had continuous ECG monitoring, and an institutional electronic data warehouse archived the complete medical record.
We queried the electronic medical record to obtain demographics, vital sign measurements, comorbid diagnosis codes, LOS, mortality, and all International Classification of Diseases-Ninth Revision diagnoses codes associated with an admission, which we ascribed as acute and chronic comorbid conditions. Rather than rely on administrative coding to select the primary diagnosis for each admission, we carefully reviewed every history, physical examination, and discharge summary. Patients whose hospital course was complicated by cardiogenic shock and cardiac arrest received a primary diagnosis consistent with the principal etiology of the shock or arrest. Thus, a patient presenting with ST-segment elevation myocardial infarction (STEMI) found to be in cardiogenic shock received the primary diagnosis of STEMI. We categorized admissions as medical if no procedures were performed in the 7 days before or after ICU admission (5). The University of Virginia Institutional Review Board approved this study, with a waiver of informed consent.
Severity of illness
To assess severity of illness, we calculated the Oxford Acute Severity of Illness Score (OASIS) for all CICU admissions. OASIS is an abbreviated acute physiology score that has equivalent discrimination and calibration of the Acute Physiology, Age, and Chronic Health Evaluation IV system from which it was derived (6). OASIS scores the worst measurements from the first 24 h of ICU admission and includes pre-ICU LOS, age, Glasgow coma score, heart rate, mean arterial pressure, respiratory rate, temperature, urine output, mechanical ventilation status, and admission type.
We applied a previously validated rhythm classification methodology that detects atrial fibrillation (AF) at a burden of ≥5% (≥90 s of a 30-min segment) (7) to the continuous ECG record. We categorized admissions as having previous AF if there was evidence of pre-existing AF or atrial flutter, as determined by diagnosis code, 12-lead ECG report, or as the first detected rhythm from the ICU bedside monitor. In those without previous AF, if we subsequently detected AF or atrial flutter during ICU monitoring, we categorized the admission as new AF. We categorized all other admissions as being without AF. We considered new AF to be clinical or recognized if it was associated with a diagnosis code or confirmed by 12-lead ECG during the hospitalization.
For categorical variables, we calculated percentages and counts, and for continuous variables, we calculated median (interquartile range [IQR]). We tested for differences in admission characteristics among the 3 types of admissions using a Kruskal-Wallis test. To assess the association of variables on CICU LOS, hospital mortality, and 30-day readmission, we developed regression models, adjusting for demographics, severity of illness, and common acute comorbid conditions. On the basis of previous work that examined the impact of subacute, potentially catastrophic illness and AF in the medical and surgical ICUs, we included age, illness severity (OASIS), and the diagnosis codes that indicated acute kidney injury, hemorrhage, acute respiratory failure, and sepsis as candidate predictors of hospital mortality (8,9). We also included cardiogenic shock and cardiac arrest on the basis of previously published findings (10,11). In addition, for CICU LOS, we included AF and post-operative status as candidate predictors on the basis of previous findings (9). For hospital readmissions, we included risk factors associated with readmission, including diabetes mellitus, heart failure, chronic kidney disease, and previous stroke (12,13). We also examined the impact of CICU LOS, and whether an admission was for an anticipated or unanticipated procedure. For patients with multiple readmissions during the study period, we analyzed only the first admission.
We allowed continuous variables, such as age, to have nonmonotonic associations through use of restricted cubic splines (14). In all multivariable regression models, we always constrained our events per variable ratio to exceed 15 (15). We quantified predictive accuracy using a concordance index (C-statistic) or R2, and validated these models internally using bootstrap resampling to estimate the performance on a new sample of observations from the same study patients (14). We performed all statistical analyses in R 3.2.3 (16).
We analyzed 1,042 CICU admissions with available ECG data (84% of 1,238 admissions screened). The median age was 65.9 years (IQR: 55.6 to 75.9 years), and 62% of admissions were men. Although many admissions originated from the emergency department (44%), more than one-third were transferred from outside hospitals (35%). Severity of illness, as quantified by OASIS, was variable, with scores ranging from 6 to 69, with a median value of 25 (IQR: 20 to 30) (Table 1).
Types of admission
We categorized 70% of admissions as medical, 26% as scheduled procedures, and 4% as unscheduled procedures. Minimally invasive valvular procedures (e.g., transcatheter aortic valve replacement, MitraClip, Abbott Vascular, Menlo Park, California) accounted for 26% of procedure admissions; 11% were for electrophysiology procedures, and the remaining 63% were for surgical procedures in the operating room. Compared with admissions for acute medical illness and unscheduled procedures, those for scheduled procedures had decreased acuity and mortality, and fewer 30-day hospital readmissions (Table 2).
Primary diagnoses and comorbid disease
The most common primary diagnosis was acute coronary syndrome (ACS), which accounted for 25% of all admissions, and comprised non−ST-segment elevation acute coronary syndrome (NSTE-ACS) (14%) and STEMI (11%). Acute heart failure exacerbation (15%) and valvular disease (10%) were also frequent causes of admission. Noncardiovascular primary diagnoses accounted for 14% of admissions, with sepsis most notably the fifth most common primary diagnosis (5%). Figure 1 depicts the most common primary diagnoses for all CICU admissions. Frequently, admissions were complicated by cardiogenic shock (n = 104; 10%) and cardiac arrest (n = 52; 5%), including 1 of every 4 acute coronary syndrome admissions to the CICU. One-half of all admissions (n = 524; 50%) were marked by either acute respiratory failure, acute kidney injury, or sepsis. Most admissions for acute noncardiovascular illness had severe chronic cardiovascular disease (85%). Pre-existing AF was present in 43% of patients, whereas new AF occurred in 9%. Of the 97 patients with new AF, 79 had subclinical AF. Nearly all patients (91%) had a CHA2DS2-VASc score ≥2, with a median score of 4 (Table 1).
Median LOS in the CICU and hospital were 2 days (IQR: 1 to 5 days) and 6 days (IQR: 3 to 11 days), respectively. Overall hospital mortality was 12%, with 7% occurring within the CICU. Of the 920 patients who survived to hospital discharge, 171 (19%) were readmitted to the University of Virginia Medical Center within 30 days, of whom 49 (29%) were readmitted to the CICU. Most readmissions were for cardiovascular diagnoses (n = 99; 58%), with heart failure being the most common (n = 30; 17%). Other common principal diagnoses for readmissions included sepsis (n = 14; 8%), gastrointestinal bleeding (n = 14; 8%), ACS (n = 11; 6%), and atrial tachyarrhythmia (n = 8; 5%).
Figure 2 shows the association between multiple independent predictors in relation to hospital mortality. The model for hospital mortality with only 8 predictors had excellent discrimination (validated C-statistic: 0.90; optimism <0.01). Sepsis, the most frequent noncardiovascular primary diagnosis (5%), but which was present in 16% of all admissions, was associated with an increased risk of mortality, with an odds ratio of 2.1 (95% confidence interval: 1.22 to 3.58; p = 0.0076). Cardiogenic shock, cardiac arrest, acute respiratory failure, and hemorrhage also had strong associations. Regression analysis for CICU LOS (Table 3) demonstrated that acute kidney injury, acute respiratory failure, new subclinical AF, and post-operative states were all significantly associated with longer LOS in the CICU (with marginal effect sizes of 5.5 ± 1.5 days, 6.7 ± 1.7 days, 6.3 ± 2.6 days, and 3.6 ± 1.7 days, respectively). Thirty-day hospital readmission analysis (Table 3) showed that comorbid conditions, such as chronic kidney disease, were associated with increased risk of readmission (C-statistic: 0.61; optimism: 0.04).
In a modern tertiary-care academic medical center CICU, we studied 1,042 patients admitted over the course of a year, and found that although cardiovascular conditions, including acute coronary syndrome, remained common, acute noncardiovascular conditions were strongly associated with outcomes, including mortality and LOS. One-half of all admissions (n = 524; 50%) were marked by either acute respiratory failure, acute kidney injury, or sepsis (Central Illustration). To our knowledge, this was the first study to report hospital readmission data on an unselected cohort of CICU patients.
The CICU and hospital mortality observed in >1 year of consecutive admissions were 7% and 12%, respectively. These findings were consistent with those reported in several previous studies in which CICU mortality ranged from 6% to 9%, and hospital mortality ranged from 11 to 33% (4,17–19). New-onset subclinical AF, which occurred in 8%, was associated with increased CICU LOS, which was consistent with similar findings in medical and surgical ICU cohorts (9,20).
With continued advances in revascularization and pharmacotherapy, hospital mortality in acute myocardial infarctions fell significantly in the 1960s, and has continued to decline from ∼12% in the 1990s to ∼6% in the mid-2000s (3,21). Despite advances in ACS treatment, Katz et al. (4) noted that CICU mortality rates did not change significantly over the 18 years observed due to a significant increase in acute critical illnesses and complex comorbidities. Likewise, the balance of CICU diagnoses shifted, with a significant decline in STEMI and significant increases in both NSTE-ACS and noncardiovascular disease (4,21,22). Another study of 7,869 patients found that there was no difference in clinical outcomes in stable NSTE-ACS among those initially admitted to a ward or to a CICU (23).
Our data echoed these findings, with NSTE-ACS and acute heart failure exacerbation being more common than STEMI. Similarly, noncardiovascular diagnoses and comorbid conditions were both prevalent. Cardiologists staff most CICUs and have relatively limited experience treating the most severe forms of hospital-acquired noncardiovascular conditions compared with intensivists who staff medical ICUs. ICUs (including CICUs) that are either staffed by intensivists or mandate critical care consultations have improved patient outcomes compared with ICUs without such staffing (24,25).
We demonstrated that the strongest predictors of increased CICU LOS were noncardiovascular conditions; patients with acute respiratory failure or acute kidney injury stayed an average of 6 to 7 days longer in the CICU, independent of other risk factors. These findings affirmed recent statements from the American Heart Association calling for staffing by either dedicated cardiac intensivists or co-management with general intensivists, in which the experienced intensivist takes primary responsibility for the care of each critically ill patient in the CICU (e.g., patients with multiorgan dysfunction or those requiring mechanical ventilation), with ongoing collaborative co-management by the cardiologist (26,27). Others have advocated for greater adoption of standard-of-care protocols and practices aimed to prevent or mitigate the risk of these acute noncardiovascular conditions encountered in the CICU (28).
Approximately 1 of every 5 patients in our study was readmitted within 30 days of discharge. In recent years, the U.S. Centers for Medicare & Medicaid Services began publicly reporting readmission rates and reducing reimbursements to hospitals for higher than expected readmission rates. As a result, hospitals have focused on preventing readmissions from acute myocardial infarction and heart failure. Our analysis underscored the difficulty of predicting hospital readmission accurately. A systematic review by Kansagra et al. (29) found that current models performed so poorly that using risk-standardized readmission rates to compare and reimburse hospitals was a potentially unsuitable practice. More recently, Katz et al. (30) described the evolution of the CICU and enumerated several opportunities for further improvements in care delivery, which will require new channels of investigation to examine care models, training pathways, therapeutic strategies, and quality improvement metrics, such as ICU readmission, which has yet to be substantiated among CICU cohorts.
This was a single-center retrospective study in an unselected cohort from the CICU of a rural, tertiary-care academic medical center, and although our findings were consistent with previously reported studies, other hospitals might treat patients with different acuity or provide different specialty services that could affect the generalizability of these results. Continuous ECG data were unavailable for 16% of all CICU admissions due to technical complications with our network processing and archival system. In addition, we might have failed to account for admissions to other hospitals. LOS in clinical studies is challenging to predict and is confounded by nonclinical factors, such as fluctuations in the availability of both floor beds and nurse staffing. We calculated illness severity using OASIS, which failed to account for aggressive resuscitation or interventions that were pursued among the most critically ill patients. Age was nonsignificantly associated with mortality, likely due to our study being underpowered to detect a difference after adjusting for other characteristics and to the heterogeneous age distribution among those admitted with high-risk conditions, such as cardiac arrest and cardiogenic shock. Although we relied on standard diagnosis codes (which have variable validity) to characterize comorbid conditions, we carefully inspected the medical record, including progress notes and discharge summaries, to characterize the primary diagnoses of admissions.
Noncardiovascular illnesses, in particular, acute respiratory failure, acute kidney injury, and sepsis, complicate CICU admissions by contributing to mortality and LOS. Many patients in the modern CICU have acute noncardiovascular illnesses that are associated with mortality and increased LOS. Clinicians caring for these patients increasingly need greater competency in treating complex comorbid noncardiovascular conditions.
COMPETENCY IN PATIENT CARE AND PROCEDURAL SKILLS: Patients in the CICU often have noncardiovascular illnesses that contribute to mortality, lengthy hospital stays, and hospital readmissions. One-half of all cases are marked by acute respiratory failure, acute kidney injury, or sepsis. The extent of comorbidities demands that clinicians caring for patients in the CICU have the competencies necessary to manage patients with both cardiovascular and noncardiovascular conditions.
TRANSLATIONAL OUTLOOK: Additional studies are needed to define optimum models of team-based care for critically ill patients in the CICU.
The authors thank J. Randall Moorman and Douglas E. Lake for mentorship; Gary Huband and Jacalyn M. Huband for design and operation of infrastructure to collect, process, and archive physiological monitoring data; and T. Bruce Hope and John Ainsworth for assistance with data warehouse queries.
The study was sponsored by the University of Virginia Health System. Both authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- atrial fibrillation
- cardiac intensive care unit
- intensive care unit
- interquartile range
- length of stay
- non−ST-segment elevation acute coronary syndrome
- Oxford Acute Severity of Illness Score
- ST-segment elevation myocardial infarction
- Received September 23, 2016.
- Revision received January 12, 2017.
- Accepted February 10, 2017.
- 2017 American College of Cardiology Foundation
- Julian D.G.
- Johnson A.E.,
- Kramer A.A.,
- Clifford G.D.
- Carrara M.,
- Carozzi L.,
- Moss T.J.,
- et al.
- Moss T.J.,
- Lake D.E.,
- Calland J.F.,
- et al.
- Moss T.J.,
- Calland J.F.,
- Enfield K.B.,
- et al.
- Kontos M.C.,
- Scirica B.M.,
- Chen A.Y.,
- et al.
- Donzé J.,
- Lipsitz S.,
- Bates D.W.,
- Schnipper J.L.
- Condon C.,
- Lycan S.,
- Duncan P.,
- Bushnell C.
- Harrell F.E. Jr..
- ↵R Core Team. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. 2015;3.2.0. Available at: https://www.R-project.org. Accessed February 24, 2017.
- Shaver C.M.,
- Chen W.,
- Janz D.R.,
- et al.
- Rogers W.J.,
- Frederick P.D.,
- Stoehr E.,
- et al.
- Sjoding M.W.,
- Prescott H.C.,
- Wunsch H.,
- Iwashyna T.J.,
- Cooke C.R.
- van Diepen S.,
- Lin M.,
- Bakal J.A.,
- et al.
- Na S.J.,
- Chung C.R.,
- Jeon K.,
- et al.
- Morrow D.A.,
- Fang J.C.,
- Fintel D.J.,
- et al.
- O'Malley R.G.,
- Olenchock B.,
- Bohula-May E.,
- et al.
- van Diepen S.,
- Sligl W.I.,
- Washam J.B.,
- Gilchrist I.C.,
- Arora R.C.,
- Katz J.N.
- Katz J.N.,
- Minder M.,
- Olenchock B.,
- et al.