Abstract
Cardiac critical care has evolved into one of the cornerstones of successful congenital heart centers. A deeper understanding of critical care’s impact on surgical outcomes represents crucial knowledge for efforts to continue improving the health of this patient population. Further, as the cohort of critically ill patients slowly shifts away from postoperative care to include more patients with end-stage heart failure, multiorgan disease, and chronic device utilization, measurement of critical care quality must account for this diversity. Multiple databases now exist that include practice and outcome information that could be used to measure the performance of critical care teams. Improving outcomes for children with critical cardiovascular disease depends upon converting these existing data into actionable information that drives practice change. Collaborative learning paradigms are a means to such improvement, and burgeoning efforts within the field of pediatric cardiac critical care hold promise to achieve further reductions in morbidity and mortality for these vulnerable children.
Key Words
Cardiac, Critical care, Outcomes, Quality improvement, Databases, Clinical registry, Collaborative learning
Cardiac critical care has evolved into one of the cornerstones of successful congenital heart centers. Improvements in morbidity and mortality after pediatric and congenital heart surgery observed in the modern era most likely resulted from several advances in perioperative care such as preoperative imaging, surgical techniques, anesthesia management, and mechanical circulatory support but are also due to increasingly specialized intensive care units (ICUs) and expert postoperative care. The degree to which any one of these domains impacts short- and long-term patient outcomes in children and adults with critical cardiovascular disease remains unclear, and this is particularly true when considering the role and contribution of critical care. A deeper understanding of critical care’s impact on surgical outcomes represents crucial knowledge for efforts to continue improving the health of this patient population. Further, as the cohort of critically ill patients slowly shifts away from postoperative care to include more patients with end-stage heart failure (with or without underlying structural heart disease), multiorgan disease, and chronic device utilization (e.g., ventricular assist devices), measurement of critical care quality must account for this diversity.
Robust data infrastructures are essential to measuring cardiac critical care outcomes, assessing quality of care, and testing new treatment approaches or interventions. Multiple databases now exist that include practice and outcome information that could be used to measure the performance of critical care teams. Finally, improving outcomes for children with critical cardiovascular disease depends upon converting these existing data into actionable information that drives practice change. Collaborative learning paradigms are a means to such improvement, and burgeoning efforts within the field of pediatric cardiac critical care hold promise to achieve further reductions in morbidity and mortality for these vulnerable children.
This chapter begins with a framing of concepts and challenges in measuring critical care outcomes and quality, using risk-adjusted mortality as an illustrative example. We then describe the essential features of a clinical data repository for critical care quality and outcomes assessment. The breadth of existing data sources and registries for cardiac critical care data is described, with a subsequent focus on the two primary clinical databases used in North America: the Pediatric Cardiac Critical Care Consortium (PC 4 ) clinical registry and the Virtual PICU System (VPS, LLC, Los Angeles, California) database. The chapter concludes with a discussion of collaborative quality improvement in the field of congenital cardiac care that involves interventions in the critical care environment.
Measuring Outcome and Quality of Critical Care
Defining Patient Outcomes and Quality Metrics
Outcome measures used for pediatric cardiac critical care quality assessment should ideally reflect the competence and performance of the cardiac intensive care unit (CICU) team and be independent of care provided and outcomes realized before and subsequent to the CICU admission. Hospital episode metrics (e.g., discharge mortality) are certainly most important to understanding patient-level outcomes, but these metrics do not inform improvement strategies because they do not necessarily provide granular information on the performance and quality of individual teams that separately care for a patient throughout the hospitalization. Congenital heart centers desiring to improve outcomes for hospitalized patients need quality metrics that disentangle teams’ performance from one another. This is particularly true when considering the quality of care provided in the CICU.
Perioperative care is an illustrative example of the conundrum regarding team quality assessment; it is challenging to determine how the quality of intraoperative and postoperative care determines individual patient and aggregate population outcomes. For example, duration of mechanical ventilation after a particular operative procedure depends on ventilator management, administration of sedatives and analgesics, and weaning practices of the CICU team. However, despite any efforts made by the CICU to rapidly and safely extubate a postoperative patient, this metric is profoundly impacted by the preoperative physiology, presence of residual anatomic defects, anesthetic practices, and complicating surgical morbidities (e.g., phrenic nerve injury), which includes the quality of care provided by other provider teams. Thoughtful approaches to risk adjustment (see later) and outcome measurement are necessary to understand the unique impact of pediatric cardiac critical care team performance on surgical patients. Presenting existing data on overall program performance (e.g., hospital mortality after cardiovascular surgery) alongside CICU performance (e.g., CICU “attributable” mortality; see later discussion) may provide deeper insights to hospitals on where strengths and weaknesses lie in their overall perioperative care process.
Outcomes of medical (nonsurgical) CICU encounters may better reflect the interventions and quality of care provided by the CICU team. Establishing outcome benchmarks for commonly used quality metrics in general pediatric critical care (e.g., catheter-associated bloodstream infections, unplanned readmissions, frequency of cardiac arrest) is necessary for measuring performance in populations of patients with critical cardiovascular disease. Determining how to appropriately risk adjust outcomes specifically for surgical and medical patients in the CICU presents a challenge, but this approach also holds promise to provide useful, granular information to CICU providers; by measuring performance in these two distinct populations, CICUs may identify areas requiring improvement efforts that would not be elucidated by looking at the patient population as a whole.
Risk Adjustment in Critical Care Outcomes and Quality Assessment
Risk adjustment, broadly defined, is a methodologic approach to measuring outcomes while accounting for unique patient characteristics that impact those outcomes and are unrelated to the quality of care provided by the hospital or provider team. In order for CICUs to understand their performance, adjusted quality metrics must reflect the unique patients they care for and the illness severity of those patients at the time they assumed care of the patient ( Fig. 7.1 ). Multi-institutional clinical registries provide an excellent source of data for generating risk-adjustment models and for applying those models to calculate adjusted outcome measures that can be reported back to hospitals.
Risk adjustment after congenital heart surgery represents the most thorough and successful effort to date within the field of congenital cardiac care. The Society of Thoracic Surgeons Congenital Heart Surgery Database Mortality Risk Model represents the current gold standard in surgical mortality risk adjustment. This empirically derived model accounts for patient characteristics and operative complexity before surgery. However, examination of two hypothetical patients undergoing the same operation highlights why additional tools are needed to assess CICU quality.
Consider two patients with no comorbidities or preoperative complications undergoing arterial switch operation for d-transposition of the great arteries. The first patient undergoes a straightforward operation and arrives at the CICU on low-dose inotropic support and mechanical ventilation. The second patient’s intraoperative course is complicated by several bypass runs to revise the coronary buttons and arrives at the CICU with an open sternum and on extracorporeal membrane oxygenation (ECMO). Clearly the challenges to the CICU team differ significantly in these two patients, and operative mortality is much more likely in the second case than the first independent of the quality of care provided by the CICU team. According to the existing Society of Thoracic Surgeons (STS) risk-adjustment model, these patients would have identical predicted risk of mortality, reflecting none of the complexity faced by the second CICU patient. Measuring performance in the CICU must include markers of physiologic derangement and illness severity at the time of care transfer to the CICU team to understand how CICU care impacts eventual patient outcome. Thus complementary risk-adjustment approaches to disentangle quality of CICU care must be developed.
Existing risk-adjustment models used in general pediatric critical care outcomes assessment have proven insufficient for understanding the quality of pediatric CICU care, particularly in the setting of postoperative care. Databases specifically designed to capture cardiac critical care outcomes have been used to develop new risk-adjustment methods that may solve this difficult problem of isolating CICU team performance. The first such attempt was performed using the VPS database cardiac module. Jeffries et al. developed the Pediatric Index of Cardiac Surgical Intensive Care Mortality from a cohort of 16,574 cardiac surgery patients, and it predicted postoperative mortality in the ICU with an area under the curve of 0.87 and good calibration. However, important questions remained regarding this approach. The model included some postoperative variables that were collected up to 12 hours after admission from the operating room. Some of these predictor variables, such as use of ECMO within 12 hours of surgery, may be related more to CICU performance rather than baseline severity of illness upon arrival to the CICU and thus may lead to erroneous conclusions about quality. Further, this model is applied at the time of CICU admission, not when the patient returns from the operating room. Thus, in cases where patients are admitted preoperatively (e.g., neonates with ductal dependent circulations), illness severity is not assessed in the early postoperative period, and analysis of CICU postoperative care quality may be inaccurate.
To address remaining knowledge gaps and improve on existing methods, investigators from PC 4 developed a new risk-adjustment model to assess postoperative care quality in the CICU, again using mortality as the quality metric (personal communication of data under peer review). The important new features of this model include (1) that it is always applied at the time postoperative care begins in the CICU, providing a consistent assessment of patient illness severity at that time point, and (2) illness-severity measures are collected only within the first 2 postoperative hours, reducing the likelihood that variables like postoperative vasoactive support or ECMO utilization reflect the quality of CICU care. From a sample that included 8543 postoperative encounters across 23 dedicated CICUs, this model demonstrated excellent discrimination for CICU mortality with a c-statistic of 0.92. The model is being used to provide real-time information to PC 4 hospitals on adjusted CICU (“CICU attributable”) surgical mortality for benchmarking and quality improvement purposes (see later discussion).
The approaches described earlier could and should be applied more widely to investigate CICU quality metrics that go beyond mortality. Several efforts are under way at the time of this writing within PC 4 to develop risk-adjustment models for cardiac arrest, duration of mechanical ventilation, and CICU/hospital length of stay accounting for illness severity at the time of CICU admission in postoperative encounters. It remains unclear whether new CICU-specific risk-adjustment models are necessary and will outperform existing methods in use for general pediatric critical care in the measurement of outcomes for nonsurgical encounters.
Key Components of Effective Cardiac Critical Care Databases
The ideal cardiac critical care database for improving quality of care would manage the challenges described previously around heterogeneity of patients, separating CICU care from other domains, and risk adjustment. In addition, these databases should include two key components: a standard nomenclature and mechanisms that facilitate linkage between registries.
Common Nomenclature
Accurate measurement of clinical outcomes is dependent on a common nomenclature and standardized data collection. The International Society for Nomenclature of Paediatric and Congenital Heart Disease (ISNPCHD; http://www.ipccc.net/ ) and the Multi-Societal Database Committee for Pediatric and Congenital Heart Disease (MSDC), developed a consensus-based, comprehensive nomenclature for the diagnosis, procedures, and complications associated with the treatment of patients with pediatric and congenital cardiac disease. This nomenclature has been adopted by several clinical databases, including most notably:
- •
The Society of Thoracic Surgeons (STS) Congenital Heart Surgery Database
- •
The European Association for Cardio-Thoracic Surgery (EACTS) Congenital Heart Surgery Database
- •
The IMPACT Interventional Cardiology Registry ( IM proving P ediatric and A dult C ongenital T reatment) of the National Cardiovascular Data Registry of the American College of Cardiology Foundation and the Society for Cardiovascular Angiography and Interventions (SCAI)
- •
The Joint Congenital Cardiac Anesthesia Society–Society of Thoracic Surgeons Congenital Cardiac Anesthesia Database
- •
The Virtual PICU System (VPS)
- •
The Pediatric Cardiac Critical Care Consortium (PC 4 )
A common nomenclature allows comparison of reported outcomes from different databases and registries and more importantly facilitates data sharing and integration across these sources.
Methods for Linking Databases
As is true in the clinical care of patients with pediatric and congenital cardiac disease, outcomes assessment benefits from multidisciplinary collaboration. Linking subspecialty databases (e.g., surgery and critical care) can facilitate sharing of longitudinal data across temporal, geographic, and subspecialty boundaries. Clinical and administrative databases have been successfully linked using indirect identifiers, and similar techniques could be used to link clinical databases. Innovative new software platforms can also facilitate direct sharing of data variables between registries and will promote more effective approaches to indirect linkage. Careful thought must be given during the design phase when new registries are developed to ensure the most efficient and seamless harmonization across databases.
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
