No Sustained Benefit of High-Flow Nasal Oxygen After Cardiac Surgery: NOTACS

Giving high-flow nasal oxygen therapy (HFNOT) to patients after cardiac surgery does not translate to a gain in clinical benefits in the first 90 days compared with standard oxygen therapy, according to results from the multinational NOTACS trial.

Patients experienced a similar number of days alive and at home, without an increase in the level of respiratory support needed, regardless of the type of oxygen therapy they received after elective or urgent cardiac surgery.

The authors, led by Edward Litton, PhD (Fiona Stanley Hospital, Perth, Australia), say the results are consistent with a systematic review and meta-analysis of several small, mostly single-center trials that showed no consistent clinical benefits of HFNOT.

“That systematic review found a decrease in treatment escalation with HFNOT that we did not observe and no difference in the rates of tracheal reintubation, mortality, or hospital or ICU stay that we also observed,” Litton and colleagues write. “NOTACS recruited a larger sample than all the previous randomized clinical trials of HFNOT in cardiac surgery combined, with the additional strengths of an international multicenter design, measurement of postdischarge patient-centered outcomes, and a comprehensive health economic analysis.”

Despite a lack of difference in the primary endpoint, many patients did in fact have increased respiratory support needs compared with what they required prior to surgery. The study was not powered to determine if this need on returning home was sustained or confined to a short period after hospital discharge. Overall, Litton and colleagues say, these results “do not support routine implementation of prophylactic HFNOT following cardiac surgery.”

A much smaller pilot trial had suggested that using HFNOT routinely might reduce both hospital length of stay and ICU readmission in high-risk groups.

Cardiac surgeon Subodh Verma, MD, PhD (University of Toronto, Canada), said the real question is less about whether prophylactic HFNOT works and more about whether a response-guided window exists. 

“NOTACS randomized to 16 hours of blanket therapy,” he told TCTMD. “It cannot tell us whether HFNOT applied selectively to patients with a falling [respiratory rate and oxygenation; ROX] index at 2 to 6 hours postextubation intercepts deterioration before full escalation.”

The ROX data, he added, hint at some heterogeneity among these patients. 

No Key Differences Seen

The NOTACS investigators randomized 1,280 patients (mean age 63 years; 30% women) from 17 cardiac surgery centers in three countries who were undergoing CABG, valve surgery, CABG plus valve surgery, aortic surgery, or other cardiac surgery. Patients received HFNOT or standard oxygen therapy immediately after postoperative tracheal extubation in the ICU.

For more than 95% of patients in both groups, the cardiac surgery was their first. Approximately 30% of patients were current smokers, 30% had diabetes, 27% had asthma, and more than 17% had chronic obstructive pulmonary disease (COPD). As the study focused on patients at high risk of pulmonary complications, it also allowed for enrollment of those who had a lower respiratory tract infection in the prior 4 weeks.

Under the protocol, warm, humidified HFNOT was started at 40% inspired oxygen and flow of 30 L/min before being increased to 50 L/min for 5 to 10 minutes. After 15 minutes, patients were monitored for oxygen saturation, respiratory rate, and arterial blood gas levels. If oxygen saturation was < 93%, inspired oxygen was increased. Standard oxygen therapy consisted of 30% to 40% inspired oxygen and flow 2 to 6 L/min via nasal prongs or nonrebreathing mask. If oxygen saturation was < 93%, inspired oxygen dose was increased. Both types of oxygen therapy were used for a minimum of 16 hours whether patients remained in the ICU or were discharged to a hospital ward. Beyond 16 hours, ongoing adherence to the allocated oxygen therapy was encouraged.

In both groups, the rate of escalation of respiratory support was approximately 55%, with hypoxia being the most common reason.

The primary clinical effectiveness outcome of days alive and at home in the first 90 days without increased respiratory support (DAH90) compared with baseline was calculated by allocating a score of 1 for each day at home and not receiving increased support. At 90 days, both groups had a median score of 0 (P = 0.75). This finding was similar in sensitivity analyses, with no evidence of any treatment effect.

The median score at 90 days for days alive and out of hospital but with increased support also was the same, 82 days, in the HFNOT and standard oxygen groups. All secondary outcomes were not significantly different between the groups. Additionally, quality of life as measured on the EuroQol 5-Dimension instrument and activities of daily living as measured by the Barthel index were similar at 90 days between groups. Approximately 21% of patients in each group were readmitted within 90 days.

Verma highlighted the median DAH90 score of zero in both arms as important. 

“When half your patients are still hospitalized or escalating at 90 days regardless of therapy, the oxygen device at extubation is not the rate-limiting variable,” he said. “These patients arrive to the OR already with multimorbidity COPD, morbid obesity, and decades of smoking. NOTACS reveals that DAH90 in this population is determined long before the ICU, and no postextubation device changes that.”

Another issue the authors acknowledge is that their methodology relied on patient accuracy when reporting changes in their location and their support needs after discharge, albeit not on a daily basis, which could have introduced errors. Requiring daily reporting of these data in future trials could lessen the chance of these errors, but the extra burden on the patient also could lead to issues of missing data, Litton and colleagues say.

In addition to understanding the length of support needs after discharge, the authors say future studies should separate out recording location and support needs “as distinct, separate variables to enhance data granularity.”