Heart-Brain 346-7 Score: the development and validation of a simple mortality prediction score for carbon monoxide poisoning utilizing deep learning

Abstract

Introduction

Acute mortality from carbon monoxide poisoning is 1–3%. The long-term mortality risk of survivors of carbon monoxide poisoning is doubled compared to age-matched controls. Cardiac involvement also increases mortality risk. We built a clinical risk score to identify carbon monoxide-poisoned patients at risk for acute and long-term mortality.

Methods

We performed a retrospective analysis. We identified 811 adult carbon monoxide-poisoned patients in the derivation cohort, and 462 adult patients in the validation cohort. We utilized baseline demographics, laboratory values, hospital charge transactions, discharge disposition, and clinical charting information in the electronic medical record in Stepwise Akaike’s Information Criteria with Firth logistic regression to determine optimal parameters to create a prediction model.

Results

In the derivation cohort, 5% had inpatient or 1-year mortality. Three variables following the final Firth logistic regression minimized Stepwise Akaike’s Information Criteria: altered mental status, age, and cardiac complications. The following predict inpatient or 1-year mortality: age > 67, age > 37 with cardiac complications, age > 47 with altered mental status, or any age with cardiac complications and altered mental status. The sensitivity of the score was 82% (95% confidence interval: 65–92%), the specificity was 80% (95% confidence interval: 77–83%), negative predictive value was 99% (95% confidence interval: 98-100%), positive predictive value 17% (95% confidence interval: 12–23%), and the area under the receiver operating characteristic curve was 0.81 (95% confidence interval: 0.74–0.87). A score above the cut-off point of −2.9 was associated with an odds ratio of 18 (95% confidence interval: 8–40). In the validation cohort (462 patients), 4% had inpatient death or 1-year mortality. The score performed similarly in the validation cohort: sensitivity was 72% (95% confidence interval: 47–90%), specificity was 69% (95% confidence interval: 63–73%), negative predictive value was 98% (95% confidence interval: 96–99%), positive predictive value was 9% (95% confidence interval: 5–15%) and the area under the receiver operating characteristic curve was 0.70 (95% confidence interval: 60%–81%).

Conclusions

We developed and validated a simple, clinical-based scoring system, the Heart-Brain 346-7 Score to predict inpatient and long-term mortality based on the following: age > 67, age > 37 with cardiac complications, age > 47 with altered mental status, or any age with cardiac complications and altered mental status. With further validation, this score will hopefully aid decision-making to identify carbon monoxide-poisoned patients with higher mortality risk.

Keywords:

• Complications of poisoning
• carbon monoxide
• heart
• CNS/psychological
• metabolic
• other

Acknowledgements

Dr. Jason J. Rose and Dr. Michael S. Zhang contributed equally to this manuscript. The authors acknowledge Dr. Michael Donahoe and Dr. Mark Gladwin for their contributions.

Disclosure statement

Dr. Rose is a coinventor on patents and applications related to using carbon monoxide scavenging molecules as therapies for carbon monoxide poisoning, licensed to Globin Solutions. Globin Solutions have a license for technologies using sodium nitrite as a therapeutic against cardiovascular disease from the National Institutes of Health and the University of Pittsburgh. Dr. Rose is a coinventor on a patent of using nitrite as a treatment for halogen gas inhalation and smoke inhalation injuries. Dr. Rose’s role in Globin Solutions disclosed above does not have direct conflicts of interests with the contents of this manuscript. Globin Solutions is developing a carbon monoxide poisoning antidote. Globin Solutions did not provide financial support to this study. The remaining authors have disclosed that they do not have any potential conflicts of interest.

Additional information

Funding

National Institutes of Health, Parker B Francis Family Foundation, Robin B. Martin Family Foundation, U.S. Department of Defense. These funding sources were not involved in the collection, management, analysis or interpretation of the data or the preparation, review, or approval of the manuscript. This work was supported by Francis Family Foundation; National Heart, Lung, and Blood Institute; Robin B. Martin Family Foundation.