Speaker
Matthew W. Semler, M.D., M.Sc.
Assistant Professor of Medicine and Biomedical Informatics
Medical Director, VICTR Center for Learning Healthcare
Vanderbilt University Medical Center
Keywords
Pragmatic Clinical Trials, Cluster-crossover Trial
Key Points
- More than 3 million adults receive invasive mechanical ventilation each year in the U.S. and providing mechanical ventilation to critically ill adults involved titrating the fraction of inspired oxygen (FiO2) to maintain arterial oxygenation. Yet, even though millions of patients have received oxygen during mechanical ventilation in clinical care for decades, the oxygen target that optimizes outcomes has been unknown.
- There are 3 basic approaches to oxygenation targets in clinical care. There is an approach that targets values in the low end of the range commonly seen in critical care (about 88-92% range of SpO2 values), an approach that targets a middle range (92-96% SpO2), or an approach that targets values at the higher end of the range (96-100% SpO2).
- Each approach has theoretical pros and cons, with oxygen levels on the higher end providing a larger buffer against hypoxia but might increase exposure to hyperoxymia while oxygen levels in the lower range might minimize those risks but might not provide as big a buffer against hypoxia, and oxygen levels in the middle of the range might avoid the risk of hyperoxia and hypoxia or might intermittently expose patients to both sets. Because of the lack of data about these 3 approaches, the recommended targets vary among 3 international guidelines.
- In 2017 the study team developed a pilot trial with the aim of determining the effect of lower intermediate and higher SpO2 targets on clinical outcomes for mechanically validated critically ill adults.
- The pilot trial was a pragmatic, unblinded, cluster-randomized, cluster-crossover trial that began in April 2018, paused from April 1-May 31, 2020 due to the COVID-19 pandemic, and concluded on Aug. 31, 2021. Study sites were in the emergency department and medical intensive care unit at Vanderbilt University. All adults in the medical ICU or in the ED with planned medical ICU admission were eligible and enrolled at the time of the first receipt of invasive mechanical ventilation.
- All patients in the ED and ICU were assigned together to an SpO2 target. Every 2 months, the ED and ICU switched together between SpO2 targets in a sequence generated by computerized randomization using permuted blocks of 3 to minimize impact on seasonal variation of changes over time. There was a washout period the last 7 days of each two-month period. The trial was conducted under waiver of informed consent.
- For the intervention, respiratory therapists titrated FiO2 to achieve SpO2 beginning within 15 minutes of initiation of mechanical ventilation and ending at discontinuation from mechanical ventilation or transfer. If a clinician, patient, or family member determined that an oxygenation target other than the assigned target would be best for the patient, that target was used and the reason was recorded.
- The primary outcome was ventilator-free days, the number of days alive and free of mechanical ventilation through study day 28. The secondary outcome was in-hospital mortality. The sample size was 2,250 patients over 36 months.
- The pilot trial found that for mechanically ventilated critically ill adults, clinical outcomes do not differ between lower, intermediate, and higher SpO2 targets. This was also true for all prespecified subgroups, and the 3 target groups did not differ for the secondary, exploratory, and safety outcomes.
Discussion Themes
-What was the experience doing these repeated crossovers? I think this is philosophically something that our ICU is used to when we don’t know what the right treatment is for a patient. That culture is very deep in our ICU and we benefited from it. The amount of education and interaction with clinical personnel at the start of the trial and at the first crossover is huge to help understand why we are doing this. By the second half of the trial, our team is available but everyone knows what and why we are doing it. For this type of trial all of the work happens in the first 3 months and then it is more streamlined?
–How many times were patients not included because the clinician had a good idea of the target for the patient. About 5% of patients developed a condition where the clinician knew what the target should be. For many patients, this did not happen. It was important to us that clinicians exercise their autonomy when they had the evidence needed to provide treatment
Tags
#pctGR, @Collaboratory1