There are no guidelines to determine when bronchoscopy is appropriate in patients with inhalation injury complicated by pneumonia. We reviewed the National Burn Repository from 1998 to 2007 to determine if there is any difference in outcome in burn patients with inhalation injury and pneumonia who did and did not undergo bronchoscopy. Three hundred fifty-five patients with pneumonia did not undergo bronchoscopy, 173 patients underwent one bronchoscopy, and 96 patients underwent more than one bronchoscopy. Patients with a 30 to 59% surface area burn and pneumonia who underwent bronchoscopy had a decreased duration of mechanical ventilation compared with those who did not (21 days, 95% CI: 19-23 days vs 28 days, 95% CI: 25-31 days, P=.0001). When compared with patients who did not undergo bronchoscopy, patients having a single bronchoscopy had a significantly shorter length of intensive care unit stay and hospital stay (35+/-3 vs 39+/-2, P=.04, and 45+/-3 vs 49+/-2, P=.009). The hospital charges were on average much higher in those patients who did not undergo bronchoscopy, compared with those who did ($473,654+/-44,944 vs $370,572+/-36,602, P=.12). When compared with patients who did not undergo bronchoscopy, patients who did have one or more bronchoscopies showed a reduced risk of death by 18% (OR=0.82, 95% CI: 0.53-1.27, P=.37). Patients with inhalation injury complicated by pneumonia seem to benefit from bronchoscopy. This benefit can be seen in a decreased duration of mechanical ventilation, decreased length of intensive care unit stay, and decreased overall hospital cost. In addition, there was a trend toward an improvement in mortality. The aggressive use of bronchoscopy after inhalation injury may be justified.
Length of stay is a frequently reported outcome after a burn injury. Length of stay benchmarking will benefit individual burn centers as a way to measure their performance & set expectations for patients. We sought to create a nationwide, risk-adjusted model to allow for length of stay benchmarking based on data from a national burn registry. Using data from the American Burn Association’s Burn Care Quality Platform, we queried admissions from 7/2015-6/2020 & identified 130,729 records reported by 103 centers. Using 22 predictor variables, comparisons of unpenalized linear regression & Gradient boosted (CatBoost) regressor models were performed by measuring the R 2 & concordance correlation coefficient on the application of the model to the test dataset. The CatBoost model applied to bootstrapped versions of the entire dataset was used to calculate O/E ratios for individual burn centers. Analyses were run on 3 cohorts: all patients, 10-20% TBSA, >20% TBSA. The CatBoost model outperformed the linear regression model with a test R 2 of 0.67 & CCC of 0.81 compared to the linear model with R 2=0.50, CCC=0.68. The CatBoost was also less biased for higher & lower length of stay durations. Gradient boosted regression models provided greater model performance than traditional regression analysis. Using national burn data, we can predict length of stay across contributing burn centers while accounting for patient & center characteristics, producing more meaningful O/E ratios. These models provide a risk-adjusted LOS benchmarking using a robust data source, the first of its kind, for burn centers.
Introduction Length of stay (LOS) is a frequently reported outcome after a burn injury. Previous literature estimates LOS at 1 day per % total burn surface area (TBSA) but this varies considerably across patients & centers. LOS benchmarking will benefit individual burn centers as a way to measure their performance & set expectations for patients. We sought to create a nationwide, risk-adjusted model to allow for LOS benchmarking based on data from a national burn registry. Methods Using data from a national burn registry, we queried admissions from 7/2015-6/2020 & identified 126,129 records with LOS data reported by 103 centers. We selected 23 predictor variables on the basis of completeness (min. 75% required) & clinical significance. Missing data were multiply imputed with a Bayesian Ridge Regression estimator. All statistics were calculated in Python using Numpy & Scikit-Learn libraries. Comparisons of unpenalized linear regression & Gradient boosted (CatBoost) regressor models were performed by measuring the R2 & concordance correlation coefficient (CCC) on the application of the model to the test dataset. The CatBoost model applied to bootstrapped versions of the entire dataset was then used to calculate O/E ratios for individual burn centers. Confidence intervals (CI) for O/E ratios were calculated using a normal distribution parametric model. Analyses were run on 3 cohorts: all patients, 10-20% TBSA, >20% TBSA. Results The CatBoost model outperformed the linear regression model with a test R2 of 0.68 & CCC of 0.81 compared to the regression model with R2=0.52, CCC=0.70. The CatBoost was also less biased for higher & lower LOS durations. Due to the CatBoost model’s superiority in predicting the outcome, this model alone was used for O/E ratio calculations. The O/E ratio data from the model for all 3 cohorts are shown in Figure 1. Conclusions Gradient boosted regression models provided greater model performance than traditional regression analysis. Using national burn data, we can predict LOS across contributing burn centers while accounting for patient & center characteristics, producing more meaningful O/E ratios. These models provide a risk-adjusted LOS benchmarking using a robust data source, the first of its kind, for burn centers.
Introduction The burn care community has demonstrated a long-standing commitment to quality of care, improved outcomes, and research by collecting and sharing clinical data through burn registries. The key to optimizing the value creation from these registries is data quality. In 2019 a new registry platform incorporating the latest, standardized, data definitions and sophisticated audit controls was piloted with 12 burn centers. The following year it was introduced to the broader burn care community along with a robust registrar education and training program. We sought to evaluate the effect of this new system on quality of data collection. Methods We compared data from 27 centers that collected data on the new system against their data collected on prior registry systems. We analyzed 26 data elements, across four different variable categories (Admissions (n=6), Demographics (n=5), Injury (n=10), and Outcomes(n=5)) that have been consistently collected over time, for data completeness. A two-proportion z test was used to assess statistical significance of differences in data completeness rates. Results The 27-center cohort entered data on 4,524 inpatient burn admissions between January 1, 2020 and June 30th 2021 and 7,259 admissions between July 1st, 2019 and June 30th, 2020 in their legacy system. The data elements were harmonized to maximize longitudinal comparability. When comparing data from the same centers but using different registry software (graphic 1) 16 of 26 variables profiled had a higher percentage of completeness in the new registry system (all statistically significant), 5 variables were statistically significantly lower, and 5 variables were not significantly different. Overall, this subset of data elements showed a 14% improvement in data completeness between the new and legacy systems (90% vs 75%). Conclusions The development of a clinical registry requires significant commitment and leadership from the healthcare association and broader clinical community. The resources and effort that the burn community has expended to improve data quality has produced important gains in data quality with the introduction of a new, high-quality registry. The burn care community should continue to emphasize education, innovation, and collaboration to collect the highest quality data at the lowest burden from all burn centers.
Introduction Risk adjusted statistical modeling of deaths in burns has two major purposes. One is to enable comparison of outcomes between centers (Benchmarking). That requires as precise a model as possible and is applied retrospectively. The other objective is to inform patient and family discussions about prognosis and plans of care. These models are applied prospectively based on limited clinical data. The purpose of this study was to derive a model that could be the basis for such a risk calculator. Methods We identified 128,252 records in a national burn registry for initial patient admissions to 103 burn centers between July 2015 through June 2020. Cases from centers with < 100 admissions annually were omitted. We compared a logistic regression model based on the revised Baux score (RBS) (age, burn size, inhalation injury) with a logistic regression model involving age, age2, burn size, presence of 3rd degree burn, inhalation injury, respiratory failure, burn etiology, gender, and admission year. We compared the Adjusted R2, c statistic and average precision for each model. All calculations were done using CatBoostClassifier in Python. Results There were 127,018 patients that served as the basis for these analyses. The RBS model had an Adjusted R2 of 0.41 compared with 0.54 for the more detailed model, a c statistic of 0.95 vs 0.98, and an average precision of 0.69 vs 0.76. Conclusions Both statistical models of mortality following burn injury demonstrated good accuracy. The model with the most predictor variables had better precision. Both models could serve as useful risk calculators for patients following burn injury.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.