Dataset Information

Predicting Adverse Outcomes for Febrile Patients in the Emergency Department Using Sparse Laboratory Data: Development of a Time Adaptive Model.

ABSTRACT:

Background

A timely decision in the initial stages for patients with an acute illness is important. However, only a few studies have determined the prognosis of patients based on insufficient laboratory data during the initial stages of treatment.

Objective

This study aimed to develop and validate time adaptive prediction models to predict the severity of illness in the emergency department (ED) using highly sparse laboratory test data (test order status and test results) and a machine learning approach.

Methods

This retrospective study used ED data from a tertiary academic hospital in Seoul, Korea. Two different models were developed based on laboratory test data: order status only (OSO) and order status and results (OSR) models. A binary composite adverse outcome was used, including mortality or hospitalization in the intensive care unit. Both models were evaluated using various performance criteria, including the area under the receiver operating characteristic curve (AUC) and balanced accuracy (BA). Clinical usefulness was examined by determining the positive likelihood ratio (PLR) and negative likelihood ratio (NLR).

Results

Of 9491 eligible patients in the ED (mean age, 55.2 years, SD 17.7 years; 4839/9491, 51.0% women), the model development cohort and validation cohort included 6645 and 2846 patients, respectively. The OSR model generally exhibited better performance (AUC=0.88, BA=0.81) than the OSO model (AUC=0.80, BA=0.74). The OSR model was more informative than the OSO model to predict patients at low or high risk of adverse outcomes (P<.001 for differences in both PLR and NLR).

Conclusions

Early-stage adverse outcomes for febrile patients could be predicted using machine learning models of highly sparse data including test order status and laboratory test results. This prediction tool could help medical professionals who are simultaneously treating the same patient share information, lead dynamic communication, and consequently prevent medical errors.

SUBMITTER: Lee S

PROVIDER: S-EPMC7146241 | biostudies-literature | 2020 Mar

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Predicting Adverse Outcomes for Febrile Patients in the Emergency Department Using Sparse Laboratory Data: Development of a Time Adaptive Model.

Lee Sungjoo S Hong Sungjun S Cha Won Chul WC Kim Kyunga K

JMIR medical informatics 20200326 3

<h4>Background</h4>A timely decision in the initial stages for patients with an acute illness is important. However, only a few studies have determined the prognosis of patients based on insufficient laboratory data during the initial stages of treatment.<h4>Objective</h4>This study aimed to develop and validate time adaptive prediction models to predict the severity of illness in the emergency department (ED) using highly sparse laboratory test data (test order status and test results) and a ma ...[more]

PMID: 32213477

Similar Datasets

Project description:INTRODUCTION:The benefit of reducing the time of antibiotic initiation in the emergency department (ED) for neutropenic patients is controversial and the research on the impact of antibiotic adherence to international guidelines in the ED is scarce. We aimed to investigate the effect of antibiotic timing and appropriateness on outcomes in patients with febrile neutropenia (FN) and to assess the performance of the MASCC risk-index to risk-stratify such patients in the ED. METHODS:We prospectively identified patients with FN who presented to our ED and assessed their Multinational Association of Supportive Care in Cancer (MASCC) risk-index. The time to parenteral antibiotic initiation and the appropriateness of the antibiotic regimen according to international guidelines were retrospectively abstracted. The performance of the MASCC risk-index in predicting the absence of complication was assessed with sensitivity, specificity and the area under the receiver operating characteristics curve (AUC). We investigated the effect of the time to antibiotic initiation and the appropriateness of the antibiotic regimen on the outcome (ICU admission or death) by logistic regression analyses. RESULTS:We included 249 patients. Median age was 60 years and 67.9% had hematological malignancies, 26 (10.4%) were admitted to the ICU and 23 (9.8%) died during hospital stay. Among the 173 patients at low risk according to the MASCC risk-index, 56 (32.4%) presented at least one complication including 11 deaths. The MASCC risk-index had a sensitivity and a specificity of 0.78% and 0.43%, respectively, in predicting the absence of complication and the AUC was 0.67. The time to antibiotic initiation in the ED was not associated with the outcome after adjusting for performance status and shock-index. Conversely, an inadequate ED antibiotic regimen was associated with higher ICU admission or death during hospital stay (OR = 3.50; 95% CI = 1.49 to 8.28). CONCLUSION:An inadequate ED antibiotic regimen in patients with FN was significantly associated with higher ICU admission or death during hospital stay.

Project description:Background:Modified systemic inflammatory response syndrome (mSIRS) criteria for the pediatric population together with the provider gestalt have the potential to predict clinical outcomes. However, this has not been studied in low-income countries. We investigated the ability of mSIRS and provider gestalt to predict mortality and morbidity among children presenting to the ED of a tertiary level hospital in Tanzania. Methods:This prospective observational study enrolled a convenience sample of children under 5?years old, presenting to the Emergency Medicine Department of Muhimbili National Hospital from September 2015 to April 2016. Trained researchers used a structured case report form to record patient demographics, clinical presentation, initial provider gestalt of severity of illness, and the mSIRS criteria. Primary outcomes were 24-h mortality and overall in-hospital mortality. Data was analyzed using simple descriptive statistics, Kruskal-Wallis, Mann-Whitney U, and chi-squared tests. Results:We enrolled 1350 patients, median age 17?months (interquartile range 8-32?months), and 58% were male. Provider gestalt estimates of illness severity were recorded for all patients and 1030 (76.3%) had complete data for mSIRS categorization. Provider gestalt classified 97 (7.2%) patients as healthy, 546 (40.4%) as mildly ill, 457 (33.9%) as moderately ill, and 250 (18.5%) as severely ill. Of the patients, classifiable by mSIRS, 411/1030 (39.9%) had ??2 mSIRS criteria. In predicting 24-h mortality, the ??2 mSIRS and gestalt "severely ill" had sensitivities of 82% and 81%, respectively, and specificity of 61% and 84%, respectively. In predicting overall in-hospital mortality, the ??2 mSIRS and gestalt "severely ill" had sensitivities of 66% and 70% with a specificity of 62% and 86% respectively. Conclusion:Both the mSIRS and provider gestalt were highly specific for predicting 24-h and overall in-hospital mortality in our patient population. The clinical utility of these assessment methods is limited by the low positive predictive value.

Dataset Information

Predicting Adverse Outcomes for Febrile Patients in the Emergency Department Using Sparse Laboratory Data: Development of a Time Adaptive Model.

Background

Objective

Methods

Results

Conclusions

Publications

Predicting Adverse Outcomes for Febrile Patients in the Emergency Department Using Sparse Laboratory Data: Development of a Time Adaptive Model.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets