1
|
Kurita T, Oami T, Tochigi Y, Tomita K, Naito T, Atagi K, Fujitani S, Nakada TA. Machine learning algorithm for predicting 30-day mortality in patients receiving rapid response system activation: A retrospective nationwide cohort study. Heliyon 2024; 10:e32655. [PMID: 38961987 PMCID: PMC11219993 DOI: 10.1016/j.heliyon.2024.e32655] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 10/31/2023] [Revised: 06/06/2024] [Accepted: 06/06/2024] [Indexed: 07/05/2024] Open
Abstract
This study investigated the accuracy of a machine learning algorithm for predicting mortality in patients receiving rapid response system (RRS) activation. This retrospective cohort study used data from the In-Hospital Emergency Registry in Japan, which collects nationwide data on patients receiving RRS activation. The missing values in the dataset were replaced using multiple imputations (mode imputation, BayseRidge sklearn. linear model, and K-nearest neighbor model), and the enrolled patients were randomly assigned to the training and test cohorts. We established prediction models for 30-day mortality using the following four types of machine learning classifiers: Light Gradient Boosting Machine (LightGBM), eXtreme Gradient Boosting, random forest, and neural network. Fifty-two variables (patient characteristics, details of RRS activation, reasons for RRS initiation, and hospital capacity) were used to construct the prediction algorithm. The primary outcome was the accuracy of the prediction model for 30-day mortality. Overall, the data from 4,997 patients across 34 hospitals were analyzed. The machine learning algorithms using LightGBM demonstrated the highest predictive value for 30-day mortality (area under the receiver operating characteristic curve, 0.860 [95 % confidence interval, 0.825-0.895]). The SHapley Additive exPlanations summary plot indicated that hospital capacity, site of incidence, code status, and abnormal vital signs within 24 h were important variables in the prediction model for 30-day mortality.
Collapse
Affiliation(s)
- Takeo Kurita
- Chiba University Graduate School of Medicine, Department of Emergency and Critical Care Medicine, 1-8-1 Inohana, Chuo, Chiba, 260-8677, Japan
| | - Takehiko Oami
- Chiba University Graduate School of Medicine, Department of Emergency and Critical Care Medicine, 1-8-1 Inohana, Chuo, Chiba, 260-8677, Japan
| | - Yoko Tochigi
- Smart119 Inc., 7th Floor, Chiba Chuo Twin Building No. 2, 2-5-1 Chuo, Chiba, 260-0013, Japan
| | - Keisuke Tomita
- Chiba University Graduate School of Medicine, Department of Emergency and Critical Care Medicine, 1-8-1 Inohana, Chuo, Chiba, 260-8677, Japan
| | - Takaki Naito
- Department of Emergency and Critical Care Medicine, St. Marianna University School of Medicine, 2-16-1, Sugao, Miyamae-ku, Kawasaki-shi, Kanagawa, 216-8511, Japan
| | - Kazuaki Atagi
- Intensive Care Unit, Nara General Medical Center, 2-897-5, Shichijonishi, Nara-shi, Nara, 630-8581, Japan
| | - Shigeki Fujitani
- Department of Emergency and Critical Care Medicine, St. Marianna University School of Medicine, 2-16-1, Sugao, Miyamae-ku, Kawasaki-shi, Kanagawa, 216-8511, Japan
| | - Taka-aki Nakada
- Chiba University Graduate School of Medicine, Department of Emergency and Critical Care Medicine, 1-8-1 Inohana, Chuo, Chiba, 260-8677, Japan
- Smart119 Inc., 7th Floor, Chiba Chuo Twin Building No. 2, 2-5-1 Chuo, Chiba, 260-0013, Japan
| | - IHER-J collaborators
- Chiba University Graduate School of Medicine, Department of Emergency and Critical Care Medicine, 1-8-1 Inohana, Chuo, Chiba, 260-8677, Japan
- Smart119 Inc., 7th Floor, Chiba Chuo Twin Building No. 2, 2-5-1 Chuo, Chiba, 260-0013, Japan
- Department of Emergency and Critical Care Medicine, St. Marianna University School of Medicine, 2-16-1, Sugao, Miyamae-ku, Kawasaki-shi, Kanagawa, 216-8511, Japan
- Intensive Care Unit, Nara General Medical Center, 2-897-5, Shichijonishi, Nara-shi, Nara, 630-8581, Japan
| |
Collapse
|
2
|
Nelson K, Brooks M, Mead-Harvey C, Quill J, Kiley B, Peworski C, Ritchie A, Sen A. Nurse-led medical emergency response reduces code blue team activations in non-hospitalized patients. Resusc Plus 2024; 18:100642. [PMID: 38689849 PMCID: PMC11059126 DOI: 10.1016/j.resplu.2024.100642] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 02/04/2024] [Revised: 03/23/2024] [Accepted: 04/06/2024] [Indexed: 05/02/2024] Open
Abstract
Objective We describe the creation of a two-tier emergency response system with a nurse-led first responder program titled "MET-RN" (Medical Emergency Team-Registered Nurse) created for ambulatory settings supported by a critical care code blue team for escalation of care. This observational study evaluated the clinical characteristics and effects of a MET-RN program on the code blue response. Methods A retrospective review of the MET-RN response data was assessed from January 2016 to June 2021. Data collected included time of call, call location, patient comorbidities, triage category (minor, urgent, or emergent), activation trigger, interventions performed, duration of the event, and patient disposition. In instances where the patient was admitted to the hospital, the discharge diagnosis and emergency department (ED) triage score were collected. Differences were tested using analysis of variance (ANOVA) F-tests, with Tukey post-hoc testing where applicable. Results MET-RN responded to 6,564 encounters from January 2016 to June 2021. The most frequent trigger call was dizziness/lightheadedness, with a prevalence of 12.0%. 33.9% of the patients seen by MET-RN were transported to the ED for further evaluation. Establishing a MET-RN system led to an estimated median of 58.3% reduction in utilization of the code blue team per quarter. Conclusion The creation of MET-RN first responder system enabled the ambulatory areas to receive minor, urgent, and emergent patient care support, leading to a decrease in utilization of the code blue team for the hospital. A two-tiered response system resulted in an improved allocation of hospital resources and kept critical care teams in high-acuity areas while maintaining patient safety.
Collapse
Affiliation(s)
- Kiley Nelson
- Department of Critical Care Medicine, United States
| | | | | | - Janae Quill
- Department of Critical Care Medicine, United States
| | | | | | | | - Ayan Sen
- Department of Critical Care Medicine, United States
| |
Collapse
|
3
|
Majeed J, Chawla S, Bondar E, Chimonas S, Martin SC, O'Sullivan M, Jones D. Rapid Response Team Activations in Oncologic Ambulatory Sites: Characteristics, Interventions, and Outcomes. JCO Oncol Pract 2022; 18:e1961-e1970. [PMID: 36306480 PMCID: PMC9750547 DOI: 10.1200/op.22.00436] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [MESH Headings] [Grants] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 06/23/2022] [Revised: 08/14/2022] [Accepted: 09/13/2022] [Indexed: 01/05/2023] Open
Abstract
PURPOSE Patients with cancer are vulnerable to clinical deterioration. Rapid response teams (RRTs) identify and manage patients with acute changes in clinical status. Although RRTs have been well studied in the hospital setting, there are limited data on patients who require support in the ambulatory or outpatient oncologic settings. Describe baseline characteristics, reasons for activations, interventions, and outcomes of ambulatory oncologic patients receiving RRT activation in a tertiary cancer center. METHODS We conducted a retrospective review of adult (age ≥ 18 years) patients requiring RRT activation at multiple ambulatory sites between July 2020 and June 2021. Demographic and clinical data captured include age, sex, race, ethnicity, do not resuscitate status, vital signs, receipt of active cancer treatment within 30 days, and cancer type. Using Kaplan-Meier survival analysis and multivariable Cox proportion hazard ratio regression models, outcomes of 90-day mortality and hospitalization were assessed. RESULTS There were 322 RRT activations among 427,734 visits to 10 ambulatory sites (0.75 RRTs/1,000 visits). The most frequent reasons were syncope (25.2%), fall (24.5%), and adverse reaction to cancer therapy or intravenous contrast (16.5%). One hundred thirty-seven (42.5%) required transfer to an emergency department, of which 81 (59.1%) required hospital admission. At 90 days, 51 (15.8%) had died, with 44 (86.3%) receiving comfort measures. Kaplan-Meier survival analysis and multivariable Cox proportional hazard ratio regression showed that heart rate > 100 at RRT presentation and hospitalization after a RRT event were significantly associated with 90-day mortality. CONCLUSION Although uncommon, patients with cancer undergoing care at ambulatory sites can suffer acute clinical deterioration needing RRT review. The rates of hospitalization and mortality among such patients are high, suggesting the need for improved end-of-life care.
Collapse
Affiliation(s)
- Jibran Majeed
- Advanced Practice Provider, Memorial Sloan Kettering Cancer Center, New York, NY
| | - Sanjay Chawla
- Critical Care Medicine Service, Department of Anesthesiology & Critical Care Medicine, Memorial Sloan Kettering Cancer Center, New York, NY
| | - Ellen Bondar
- Department of Medicine, Memorial Sloan Kettering Cancer Center, New York, NY
| | - Susan Chimonas
- Department of Epidemiology and Biostatistics, Memorial Sloan Kettering Cancer Center, New York, NY
| | - Steven C. Martin
- Department of Medicine, Memorial Sloan Kettering Cancer Center, New York, NY
| | | | - Daryl Jones
- University Melbourne, Victoria, Parkville, Australia
- DEPM Monash University, Victoria, Prahran, Australia
- Austin Department of Intensive Care, Victoria, Heidelberg, Australia
- Critical Care Outreach Austin Hospital, Victoria, Heidelberg, Australia
- International Society of Rapid Response Systems, London, United Kingdom
| |
Collapse
|
4
|
Hao J, Huang Y, Su J, Lu Z. Emergency and rapid response systems: a bibliometric analysis. ANNALS OF TRANSLATIONAL MEDICINE 2022; 10:311. [PMID: 35433985 PMCID: PMC9011274 DOI: 10.21037/atm-22-709] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Download PDF] [Figures] [Subscribe] [Scholar Register] [Received: 01/10/2022] [Accepted: 03/04/2022] [Indexed: 11/06/2022]
Abstract
Background The emergency rapid response system (RRS) can reduce the mortality of hospitalized patients, and its core is the activation criteria and the rapid response team (RRT). This study adopted a bibliometric method to analyze the research status of RRSs for hospitalized patients. Methods The Science Citation Index Expanded (SCI-E) database was searched using the keywords "emergency" and "rapid response system", and the search results were analyzed using CiteSpace software. The retrieved data included the annual distribution of studies and literature citations; the source country of the literature; the distribution of institutions and authors of the literature; the cooperation between countries, institutions, and authors; the distribution of journals that published the literature, and the use of keywords in the literature. Results A total of 1,320 research papers were found, with a total of 29,920 citations. The number of papers and their citations increased yearly. The top 5 countries in terms of number of publications were the United States, Australia, China, the United Kingdom, and Canada. The top 5 countries in terms of centrality were the United States, the United Kingdom, Argentina, the Czech Republic, and Switzerland. The research institutions were mainly located in developed countries, such as the United States and Australia. There was relatively little collaboration between researchers. The journals that published the literature mainly specialized in critical care medicine and emergency medicine. The keyword analysis revealed that most studies focused on medical emergency teams (METs) and mortality. Conclusions There were few studies related to the emergency RRS for hospitalized patients. The majority of studies were from developed countries and mainly focused on the impact of team building and the effect of the RRS on mortality.
Collapse
Affiliation(s)
- Jing Hao
- Department of Emergency, The First Affiliated Hospital of Henan University of Science and Technology, Luoyang, China
| | - Yutao Huang
- Intensive Care Unit, Hejin Municipal People's Hospital, Hejin, China
| | - Jianguo Su
- Department of Emergency, Ningxia Chinese Medicine Research Center, Yinchuan, China
| | - Zhaofeng Lu
- Department of Emergency, The First Affiliated Hospital of Henan University of Science and Technology, Luoyang, China
| |
Collapse
|