Identifying an Oxygenation Index Threshold for Increased Mortality in Acute Respiratory Failure

• Main Author: Hammond, Brandon
• Other Authors: The University of Arizona College of Medicine - Phoenix
• Language: en_US
• Published: The University of Arizona. 2016
• Online Access: http://hdl.handle.net/10150/603630; http://arizona.openrepository.com/arizona/handle/10150/603630

A Thesis submitted to The University of Arizona College of Medicine - Phoenix in partial fulfillment of the requirements for the Degree of Doctor of Medicine. === Objectives: To examine current oxygenation index (OI) data and outcomes using EMR data to identify a specific OI values associated with outcome. Methods: Retrospective review of electronic medical record (EMR) data for patients age 1 month ‐ 20 years mechanically ventilated for >24 hours in the PICU. Serial, average and maximum OI values were calculated. Length of mechanical ventilation, hospital stay and outcome were assessed. Results: OI was calculated on 65 patients from EMR data, of which 6 died (9.2%). The median maximum OI was 10 for all patients, 17 for non‐survivors (NS), and 8 for survivors (S), (p=0.14 via Wilcoxon rank‐sum test). Odds ratios (OR) indicated 2.1 times increase odds of death (p=.08), 95% confidence interval (0.89–5.03) for each one‐percent increase in maximum OI. Average OI OR also revealed 2.1 times increase in odds of death (p=.14), 95% confidence interval (0.77–5.48). ROC analysis indicated a higher discriminate ability for max OI (AUC = 0.68) than average OI (AUC = .58). OI cut points for mortality were established. Mortality was unchanged until max OI >17, for which mortality nearly tripled at a value of 18% versus 6‐7% for range 0‐17. Conclusions: Serial assessment of OI values may allow creation of alert values for increased mortality risk and aid in development of clinical decision rules. Consideration for escalation of therapies for respiratory failure such as high frequency ventilation or ECMO at lower levels of OI than historically reported may be warranted. This study also helps to validate prior reports that OI is useful as a severity score for clinical research and outcome prediction.