Virtual Pediatric Intensive Care Unit Research Articles

Predicting High Flow Nasal Cannula Failure in an Intensive Care Unit Using a Recurrent Neural Network With Transfer Learning and Input Data Perseveration: Retrospective Analysis.

BackgroundHigh flow nasal cannula (HFNC) provides noninvasive respiratory support for children who are critically ill who may tolerate it more readily than other noninvasive ventilation (NIV) techniques such as bilevel positive airway pressure and continuous positive airway pressure. Moreover, HFNC may preclude the need for mechanical ventilation (intubation). Nevertheless, NIV or intubation may ultimately be necessary for certain patients. Timely prediction of HFNC failure can provide an indication for increasing respiratory support.ObjectiveThe aim of this study is to develop and compare machine learning (ML) models to predict HFNC failure.MethodsA retrospective study was conducted using the Virtual Pediatric Intensive Care Unit database of electronic medical records of patients admitted to a tertiary pediatric intensive care unit between January 2010 and February 2020. Patients aged <19 years, without apnea, and receiving HFNC treatment were included. A long short-term memory (LSTM) model using 517 variables (vital signs, laboratory data, and other clinical parameters) was trained to generate a continuous prediction of HFNC failure, defined as escalation to NIV or intubation within 24 hours of HFNC initiation. For comparison, 7 other models were trained: a logistic regression (LR) using the same 517 variables, another LR using only 14 variables, and 5 additional LSTM-based models using the same 517 variables as the first LSTM model and incorporating additional ML techniques (transfer learning, input perseveration, and ensembling). Performance was assessed using the area under the receiver operating characteristic (AUROC) curve at various times following HFNC initiation. The sensitivity, specificity, and positive and negative predictive values of predictions at 2 hours after HFNC initiation were also evaluated. These metrics were also computed for a cohort with primarily respiratory diagnoses.ResultsA total of 834 HFNC trials (455 [54.6%] training, 173 [20.7%] validation, and 206 [24.7%] test) met the inclusion criteria, of which 175 (21%; training: 103/455, 22.6%; validation: 30/173, 17.3%; test: 42/206, 20.4%) escalated to NIV or intubation. The LSTM models trained with transfer learning generally performed better than the LR models, with the best LSTM model achieving an AUROC of 0.78 versus 0.66 for the 14-variable LR and 0.71 for the 517-variable LR 2 hours after initiation. All models except for the 14-variable LR achieved higher AUROCs in the respiratory cohort than in the general intensive care unit population.ConclusionsML models trained using electronic medical record data were able to identify children at risk of HFNC failure within 24 hours of initiation. LSTM models that incorporated transfer learning, input data perseveration, and ensembling showed improved performance compared with the LR and standard LSTM models.

P0358 / #364: ANALYSIS OF 95TH CENTILE LENGTH OF STAY IN PATIENTS WITH RE-ADMISSIONS TO THE PEDIATRIC ICU.

Aims & Objectives: Length of stay (LOS) is an important driver of healthcare costs in the Pediatric Intensive Care Unit (PICU). Historically, long-stay patients (LSPs) have been defined as patients whose length of stay in PICU is longer than the 95th percentile. Though small in number, these patients contribute significantly to cost. The aim of this study is to assess the impact of readmissions on the 95th centile of length of stay in the PICU. Methods: Retrospective cohort study of all patients admitted to PICUs in the Virtual Pediatric Intensive Care Unit Systems (VPS) database from January 1st, 2009 to January 1st, 2012. Intercept only quantile regressions were used to estimate 95% level quantiles and their 95% confidence intervals. The quantile estimation was performed separately for each PICU admission which preserved independent individual contributions to quantile estimation. Results: 224, 960 unique patient encounters were identified. The 95th percentile of LOS for all encounters was 16 days. However, as the number of previous PICU admissions increases, the 95th percentile of subsequent PICU admissions increases up to 26 days. In comparison, the median length of stay increased by approximately one day. Table -1: Length of stay with subsequent pediatric intensive care admissionsConclusions: Readmissions have a disproportionate impact on 95th percentile of LOS and significantly adds to hospital costs.

Multicenter mortality and morbidity associated with pulmonary hypertension in the pediatric intensive care unit.

Despite advances in the diagnosis and management of pediatric pulmonary hypertension (PH), children with PH represent a growing inpatient population with significant morbidity and mortality. To date, no studies have described the clinical characteristics of children with PH in the pediatric intensive care unit (PICU). A retrospective multicenter cohort study of 153 centers in the Virtual PICU Systems database who submitted data between 1 January 2009 and 31 December 2015 was performed. A total of 14,880/670,098 admissions (2.2%) with a diagnosis of PH were identified. Of these, 2190 (14.7%) had primary PH and 12,690 (85.3%) had secondary PH. Mortality for PH admissions was 6.8% compared to 2.3% in those admitted without PH (odds ratio = 3.1; 95% confidence interval = 2.9–3.4). Compared to patients admitted to the PICU without PH, those with PH were younger, had longer length of stay, higher illness severity scores, were more likely to receive invasive mechanical ventilation, cardiopulmonary resuscitation, extracorporeal membrane oxygenation, and more likely to have co-diagnoses of sepsis, heart failure, and respiratory failure. In a multivariate model, factors significantly associated with mortality for children with PH included age < 6 months or > 16 years, invasive mechanical ventilation, and co-diagnoses of heart failure, sepsis, hemoptysis, disseminated intravascular coagulation, stroke, and multi-organ dysfunction syndrome. Despite therapeutic advances, the disease burden and mortality of children with PH remains significant. Further investigation of the risk factors associated with clinical deterioration and mortality in this population could improve the ability to prognosticate and inform clinical decision-making.

Identification of a “VTE-rich” population in pediatrics – Critically ill children with central venous catheters

BackgroundThe incidence of venous thromboembolism (VTE) is rising among inpatients in US hospitals, especially among kids with central venous catheters (CVCs) in the pediatric intensive care unit (PICU). ObjectivesTo identify a sub-group of “VTE-rich” population among PICU children, and to assess the effect of VTE on morbidity and mortality. MethodsData was extracted from a multicenter Virtual PICU Database, or VPS, for children with a CVC and presence of a VTE. The primary outcome variable was all cause mortality and secondary outcome measure was prolonged mechanical ventilation. Primary diagnoses and Pediatric risk of mortality 2 (PIM 2) score were also recorded. ResultsThe database identified 158,299 PICU patients who had a CVC. A total of 1602 patients had VTE (103 per 10,000 PICU patients). Multivariate analysis showed increased risk of VTE in patients who were <1year old (OR 1.48; 1.30–1.68), mechanically ventilated (OR 2.48; 2.07–2.98), had either a diagnostic (OR 2.32; 1.94–2.78) or therapeutic cardiac catheterization (OR 2.06; 1.66–2.55), PICC (OR 3.91; 3.50–4.37), and percutaneous CVC (OR 3.99; 3.48–4.61). Primary diagnoses associated with VTE were endocrinologic, immunologic, and gastrointestinal disorders. PICU patients with CVC and VTE had increased odds of mortality (OR 1.71; 1.47–2.00) after adjusting for factors associated with mortality, and fewer Ventilator Free Days (VFD) than patients without VTE. ConclusionsCritically ill children with CVC have a significant risk of developing VTE. Identification of the above “VTE-rich” population may aid in design of clinical trials aimed at prevention of VTE among hospitalized PICU children.

Predictors of need for mechanical ventilation at discharge after tracheostomy in the PICU.

The objective of this study was to determine factors predictive of need for mechanical ventilation (MV) upon discharge from the pediatric intensive care unit (PICU) among patients who receive a tracheostomy during their stay. This was a retrospective cohort study using the Virtual PICU Systems (VPS) database. Patients <18 years old admitted between 2009-2011 who required MV for at least 3 days and received a tracheostomy during their PICU stay were included. A total of 680 pediatric patients from 74 PICUs were included, of whom 347 (51%) remained on MV at the time of PICU discharge. Neonates (30/38, 79%) and infants (129/203, 64%) required MV at PICU discharge after tracheostomy more often than adolescents (66/141, 47%) and children (122/298, 41%). Time on MV pre-tracheostomy was longer among those who required MV at discharge (median 18.3 vs. 13.8 days, P < 0.0001); however, number of failed extubations was similar (median 1 for both groups, P = 0.97). On mixed-effects multivariable regression analysis, the age categories of neonate (OR 2.9, 95%CI 1.1-7.6, P = 0.03), and infant (OR 1.7, 95%CI 1.1-2.8, P = 0.03), and ventilator days prior to tracheostomy (OR 1.01, 95%CI 1.0-1.02, P = 0.01) were significantly associated with increased odds of MV upon PICU discharge, while being a trauma admission was associated with decreased odds (OR 0.45, 95%CI 0.28-0.73, P = 0.001). Younger patients and those with prolonged courses of MV prior to tracheostomy are more likely to continue to need MV upon PICU discharge.

Adults with childhood-onset chronic conditions admitted to US pediatric and adult intensive care units

PurposeThe purpose of the study is to compare demographics, intensive care unit (ICU) admission characteristics, and ICU outcomes among adults with childhood-onset chronic conditions (COCCs) admitted to US pediatric and adult ICUs. Materials and methodsRetrospective cross-sectional analyses of 6088 adults aged 19 to 40 years admitted in 2008 to 70 pediatric ICUs that participated in the Virtual Pediatric Intensive Care Unit Performance Systems and 50 adult ICUs that participated in Project IMPACT. ResultsChildhood-onset chronic conditions were present in 53% of young adults admitted to pediatric units, compared with 9% of those in adult units. The most common COCC in both groups were congenital cardiac abnormalities, cerebral palsy, and chromosomal abnormalities. Adults with COCC admitted to pediatric units were significantly more likely to be younger, have lower functional status, and be nontrauma patients than those in adult units. The median ICU length of stay was 2 days, and the intensive care unit mortality rate was 5% for all COCC patients with no statistical difference between pediatric or adult units. ConclusionsThere are marked differences in characteristics between young adults with COCC admitted to pediatric ICUs and adult ICUs. Barriers to accommodating these young adults may be reasons why many such adults have not transitioned from pediatric to adult critical care.

Frequency, Risk Factors, and Outcomes of Early Unplanned Readmissions to PICUs*

To determine the rate of unplanned PICU readmissions, examine the characteristics of index admissions associated with readmission, and compare outcomes of readmissions versus index admissions. Retrospective cohort analysis. Ninety North American PICUs that participated in the Virtual Pediatric Intensive Care Unit Systems. One hundred five thousand four hundred thirty-seven admissions between July 2009 and March 2011. None. Unplanned PICU readmission within 48 hours of index discharge was the primary outcome. Summary statistics, bivariate analyses, and mixed-effects logistic regression model with random effects for each hospital were performed.There were 1,161 readmissions (1.2%). The readmission rate varied among PICUs (0-3.3%), and acute respiratory (56%), infectious (35%), neurological (28%), and cardiovascular (20%) diagnoses were often present on readmission. Readmission risk increased in patients with two or more complex chronic conditions (adjusted odds ratio, 1.72; p < 0.001), unscheduled index admission (adjusted odds ratio, 1.37; p < 0.001), and transfer to an intermediate unit (adjusted odds ratio, 1.29; p = 0.004, compared with ward). Trauma patients had a decreased risk of readmission (adjusted odds ratio, 0.67; p = 0.003). Gender, race, insurance, age more than 6 months, perioperative status, and nighttime transfer were not associated with readmission. Compared with index admissions, readmissions had longer median PICU length of stay (3.1 vs 1.7 d, p < 0.001) and higher mortality (4% vs 2.5%, p = 0.002). Unplanned PICU readmissions were relatively uncommon, but were associated with worse outcomes. Several patient and admission characteristics were associated with readmission. These data help identify high-risk patient groups and inform risk-adjustment for standardized readmission rates.

107

Introduction: There is an increasing interest in ICU attending physician’s workload and its effects not only on patient outcomes but also on trainee education, staff and family satisfaction. Some models attempted to optimize attending physician’s day and night availability to improve patient’s care without increasing the cost. The goal of this project is to identify peak periods where ICU interventions are occurring which might benefit from direct supervision by attending physician. Methods: This is a retrospective, observational study utilizing a single tertiary academic center (34-bed, open, multi-specialty) pediatric intensive care unit (PICU) data submitted to the Virtual PICU system (VPS). All admissions between January 1, 2013 and June 30, 2013 were included. A unit of work was defined as an admission, endotracheal intubation, initiation of mechanical ventilation, central venous line placement, or arterial line placement. The mean number of units of work was determined per day of the week and by 4-hr periods during the day. The work of daily rounds, family meetings and other non-clinical academic meetings were not factored into the analysis. Results: For the first 6 months of 2013, there were 1,188 admissions, 169 central venous line placements, 123 arterial line placements, 113 endotracheal intubations, and 416 mechanical ventilation initiations. The average units of worked peaked on Thursday and Friday at ~14 units/day with a low of ~8 units/day on Saturday and Sunday. While not including the work of rounding and non-clinical duties, on average the busiest times of the day were between 1000 to 1400 and 1400 to 1800 with ~2.3 units/4-hr period in each block. The work slowed to only 1 unit/4-hr period from 0200 to 0600. The peak admission time was from 1400 to 1800 with an average of 1.6 admissions/4-hr period. Conclusions: This study identifies our PICU daily and weekly workload cyclical patterns. In future, this analysis may be useful in determining the appropriate physician staffing models at institutional level.

Multi-institutional Profile of Adults Admitted to Pediatric Intensive Care Units

Growing numbers of persons with childhood-onset chronic illnesses are surviving to adulthood. Many use pediatric hospitals for their inpatient needs. To our knowledge, the prevalence and characteristics of adult pediatric intensive care unit patients have not been reported. To estimate the proportion of adults admitted to pediatric intensive care units (PICUs), characterize them, and compare them with older adolescents. One-year cross-sectional analysis. Pediatric intensive care units in the United States that participated in the Virtual Pediatric Intensive Care Unit Systems. Pediatric intensive care unit patients 15 years or older admitted in 2008. We compared adults with adolescents across clinical characteristics and outcomes. Mixed-effects logistic regression was used to estimate the independent association of age with PICU mortality. Seventy PICUs had 67 629 admissions; 1954 admissions (2.7%) were patients 19 years or older; and 9105 admissions (13.5%) were patients aged 15 to 18 years. The proportion of adults (≥19 years) varied considerably by PICU (range, 0%-9.2%). As age increased, the proportion of patients who had a complex chronic condition and planned or perioperative admissions increased; the proportion of trauma-related admissions decreased. Patients aged 21 to 29 years had a 2 times (95% CI, 1.3-3.2; P = .004) greater odds of PICU mortality compared with adolescent patients, after adjusting for Paediatric Index of Mortality score, sex, trauma, and having a complex chronic condition. Being 30 years or older was associated with a 3.5 (95% CI, 1.3-9.7; P = .01) greater odds of mortality. In this multi-institutional study, adults constituted a small but high-risk proportion of patients in some PICUs, suggesting that these PICUs should have plans and protocols specifically focused on this group.

Chronic conditions among children admitted to U.S. pediatric intensive care units

To estimate the prevalence of chronic conditions among children admitted to U.S. pediatric intensive care units and to assess whether patients with complex chronic conditions experience pediatric intensive care unit mortality and prolonged length of stay risk beyond that predicted by commonly used severity-of-illness risk-adjustment models. Retrospective cohort analysis of 52,791 pediatric admissions to 54 U.S. pediatric intensive care units that participated in the Virtual Pediatric Intensive Care Unit Systems database in 2008. Hierarchical logistic regression models, clustered by pediatric intensive care unit site, for pediatric intensive care unit mortality and length of stay >15 days. Standardized mortality ratios adjusted for severity-of-illness score alone and with complex chronic conditions. Fifty-three percent of pediatric intensive care unit admissions had complex chronic conditions, and 18.5% had noncomplex chronic conditions. The prevalence of these conditions and their organ system subcategories varied considerably across sites. The majority of complex chronic condition subcategories were associated with significantly greater odds of pediatric intensive care unit mortality (odds ratios 1.25-2.9, all p values < .02) compared to having a noncomplex chronic condition or no chronic condition, after controlling for age, gender, trauma, and severity-of-illness. Only respiratory, gastrointestinal, and rheumatologic/orthopedic/psychiatric complex chronic conditions were not associated with increased odds of pediatric intensive care unit mortality. All subcategories were significantly associated with prolonged length of stay. All noncomplex chronic condition subcategories were either not associated or were negatively associated with pediatric intensive care unit mortality, and most were not associated with prolonged length of stay, compared to having no chronic conditions. Among this group of pediatric intensive care units, adding complex chronic conditions to risk-adjustment models led to greater model accuracy but did not substantially change unit-level standardized mortality ratios. Children with complex chronic conditions were at greater risk for pediatric intensive care unit mortality and prolonged length of stay than those with no chronic conditions, but the magnitude of risk varied across subcategories. Inclusion of complex chronic conditions into models of pediatric intensive care unit mortality improved model accuracy but had little impact on standardized mortality ratios.

Outcome from multiple organ dysfunction syndrome in chronic illness

Typpo et al from Baylor College of Medicine, in collaboration with 35 US children's hospitals working in the Virtual PICU Performance System database, have studied outcomes of organ dysfunction for children with chronic illness admitted to the PICU. They studied 44 693 children, 52.1% of whom had a chronic diagnosis. Chronic diagnoses increased the risk of multiple organ dysfunction syndrome at admission and also increased mortality, but in general, the children who survived returned to their baseline functional status. These data, while somewhat surprising, are also encouraging to those who care for children with severe acute illness.Article page 108 ▸ Typpo et al from Baylor College of Medicine, in collaboration with 35 US children's hospitals working in the Virtual PICU Performance System database, have studied outcomes of organ dysfunction for children with chronic illness admitted to the PICU. They studied 44 693 children, 52.1% of whom had a chronic diagnosis. Chronic diagnoses increased the risk of multiple organ dysfunction syndrome at admission and also increased mortality, but in general, the children who survived returned to their baseline functional status. These data, while somewhat surprising, are also encouraging to those who care for children with severe acute illness. Article page 108 ▸ Children with Chronic Illness Return to Their Baseline Functional Status after Organ Dysfunction on the First Day of Admission in the Pediatric Intensive Care UnitThe Journal of PediatricsVol. 157Issue 1PreviewTo determine chronic illness outcomes after admission with multiple organ dysfunction syndrome (MODS) for patients in the pediatric intensive care unit (PICU). Full-Text PDF

Review of the 23rd Annual Meeting of the Society for Pediatric Anesthesia

Review of the 23rd Annual Meeting of the Society for Pediatric Anesthesia

Children with Chronic Illness Return to Their Baseline Functional Status after Organ Dysfunction on the First Day of Admission in the Pediatric Intensive Care Unit

To determine chronic illness outcomes after admission with multiple organ dysfunction syndrome (MODS) for patients in the pediatric intensive care unit (PICU). We evaluated consecutive PICU admissions from 35 US children's hospitals from January 2004-December 2005 in the virtual PICU Performance System database. We excluded hospitals with >10% missing values for MODS variables and patients<1 month or>18 years of age. MODS was identified by laboratory and vital sign values from day of admission with International Pediatric Sepsis Consensus Conference criteria. Chronic illness was identified by secondary diagnoses, classified by modified Delphi method. We evaluated functional outcomes with pediatric overall performance category and pediatric cerebral performance category scores from PICU admission and discharge. Of 44 693 admissions, 52.1% had a chronic diagnosis. Chronic diagnoses increased MODS at PICU admission (24.6% vs 12.0%, P<.001) and mortality rates (3.7% vs 1.9%, P<.001). Patients with a chronic diagnosis had similar changes in pediatric overall performance category and pediatric cerebral performance category scores from PICU admission to discharge as previously healthy children. However, outcome in different chronic diagnosis categories was variable. Chronic illness increased MODS incidence at PICU admission and impacted all-cause PICU mortality rates. Although, in aggregate, children who survive return to baseline functional status, this varies by chronic illness category.

Day 1 multiple organ dysfunction syndrome is associated with poor functional outcome and mortality in the pediatric intensive care unit

The epidemiology and outcomes of multiple organ dysfunction syndrome (MODS) are incompletely characterized in the pediatric population due to small sample size and conflicting diagnoses of organ failure. We sought to describe the epidemiology and outcomes of early MODS in a large clinical database of pediatric intensive care unit (PICU) patients based on consensus definitions of organ failure. Retrospective analysis of a contemporaneously collected clinical PICU database. Virtual Pediatric Intensive Care Unit Performance System database patient admissions from January 2004 to December 2005 for 35 U.S. children's hospitals. : We evaluated 63,285 consecutive PICU admissions from January 2004 to December 2005 in the Virtual Pediatric Intensive Care Unit Performance System database. We excluded patients younger than 1 month or older than 18 years of age, and hospitals with >10% missing values for MODS variables. We identified day 1 MODS by International Pediatric Sepsis Consensus Conference criteria with day 1 laboratory and vital sign values. We evaluated functional status using Pediatric Overall Performance Category and Pediatric Cerebral Performance Category scores from PICU admission and discharge. Student's t test, chi-square test, Mann-Whitney rank sum, Kruskal-Wallis, and linear and logistic regression. We analyzed 44,693 admissions from 28 hospitals meeting inclusion criteria. Overall PICU mortality was 2.8%. We identified day 1 MODS in 18.6% of admissions. Patients with day 1 MODS had higher mortality (10.0% vs. 1.2%, p < .001), longer PICU length of stay (3.6 vs. 1.3 days, p < .001), and larger change from baseline Pediatric Overall Performance Category and Pediatric Cerebral Performance Category scores at time of PICU discharge (p < .001). Infants had the highest incidence of day 1 MODS (25.2% vs. 16.5%, p < .001) compared with other age groups. Using the largest clinical dataset to date and consensus definitions for organ failure, we found that children with MODS present on day 1 of intensive care unit admission have worse functional outcomes, higher mortality, and longer PICU length of stay than children who do not have MODS on day 1. Infants are disproportionally affected by MODS.

Databases for assessing the outcomes of the treatment of patients with congenital and paediatric cardiac disease – the perspective of critical care

The development of databases to track the outcomes of children with cardiovascular disease has been ongoing for much of the last two decades, paralleled by the rise of databases in the intensive care unit. While the breadth of data available in national, regional and local databases has grown exponentially, the ability to identify meaningful measurements of outcomes for patients with cardiovascular disease is still in its early stages. In the United States of America, the Virtual Pediatric Intensive Care Unit Performance System (VPS) is a clinically based database system for the paediatric intensive care unit that provides standardized high quality, comparative data to its participants [https://portal.myvps.org/]. All participants collect information on multiple parameters: (1) patients and their stay in the hospital, (2) diagnoses, (3) interventions, (4) discharge, (5) various measures of outcome, (6) organ donation, and (7) paediatric severity of illness scores. Because of the standards of quality within the database, through customizable interfaces, the database can also be used for several applications: (1) administrative purposes, such as assessing the utilization of resources and strategic planning, (2) multi-institutional research studies, and (3) additional internal projects of quality improvement or research.In the United Kingdom, The Paediatric Intensive Care Audit Network is a database established in 2002 to record details of the treatment of all critically ill children in paediatric intensive care units of the National Health Service in England, Wales and Scotland. The Paediatric Intensive Care Audit Network was designed to develop and maintain a secure and confidential high quality clinical database of pediatric intensive care activity in order to meet the following objectives: (1) identify best clinical practice, (2) monitor supply and demand, (3) monitor and review outcomes of treatment episodes, (4) facilitate strategic healthcare planning, (5) quantify resource requirements, and (6) study the epidemiology of critical illness in children.Two distinct physiologic risk adjustment methodologies are the Pediatric Risk of Mortality Scoring System (PRISM), and the Paediatric Index of Mortality Scoring System 2 (PIM 2). Both Pediatric Risk of Mortality (PRISM 2) and Pediatric Risk of Mortality (PRISM 3) are comprised of clinical variables that include physiological and laboratory measurements that are weighted on a logistic scale. The raw Pediatric Risk of Mortality (PRISM) score provides quantitative measures of severity of illness. The Pediatric Risk of Mortality (PRISM) score when used in a logistic regression model provides a probability of the predicted risk of mortality. This predicted risk of mortality can then be used along with the rates of observed mortality to provide a quantitative measurement of the Standardized Mortality Ratio (SMR). Similar to the Pediatric Risk of Mortality (PRISM) scoring system, the Paediatric Index of Mortality (PIM) score is comprised of physiological and laboratory values and provides a quantitative measurement to estimate the probability of death using a logistic regression model.The primary use of national and international databases of patients with congenital cardiac disease should be to improve the quality of care for these patients. The utilization of common nomenclature and datasets by the various regional subspecialty databases will facilitate the eventual linking of these databases and the creation of a comprehensive database that spans conventional geographic and subspecialty boundaries.

Risk Adjustment of Clinical Performance Quality Data to Facilitate Outcomes Comparison of Physicians and Hopitals

Despite the growing national concerns for healthcare quality and patient safety, there is limited success in using clinical performance data to compare the performance of physicians and hospitals. This results in a chilling effect on achieving transparency of quality of care for the healthcare system in the U.S. A significant reason for this limitation is the argument that patient outcomes cannot be meaningfully compared because of clinical differences in patient-mix, including the underlying severity of illness. The limitation to make such comparisons valid and clinically meaningful has also adversely impacted the ability to use such data in many legal settings including litigation in medical malpractice, obtaining informed consent, and public reporting of healthcare quality. Risk adjustment provides a mechanism to overcome this limitation. This paper includes the results of the experience from the National Outcomes Center, using the pediatric critical care setting as an illustration, for successfully applying risk adjustment when making comparisons of Pediatric Intensive Care Units (PICUs) across the U.S. The National Outcomes Center has partnered with the National Association of Children's Hospital and Related Institutions (NACHRI), and the Children's Hospital of Los Angeles to develop the Virtual Pediatric Intensive Care Unit System (VPS). The VPS currently includes clinical performance data from 56 U.S. hospitals representing over 56,000 patients. Outcomes analyses comparing the performance over time for PICUs within specific hospitals, and also comparisons of performance between hospitals, have been completed (N=12). This comparison includes risk adjustment using unconditional logistic regression modeling with severity of illness scoring systems. Results emerging from this national endeavor have been met with significant buy-in and acceptance by physicians and administrators. The findings from this study support that risk adjustment provides a mechanism to overcome the barrier of using clinical performance data to make comparisons of the quality and outcomes of physicians and hospitals.