Implementing a Clinical Decision Support Tool to Reduce Operating Room Anesthetic Fresh Gas Flow: A Resident-Led, Sustainability-Focused Quality Improvement Initiative Open Access

In recent years, the escalating impact of climate change on global health has become increasingly evident, posing significant threats to human well-being.¹  The health care sector is responsible for an estimated 8 to 10% of all greenhouse gas (GHG) emissions in the United States, much of which are produced through resource-intensive surgical and operative room services.²  Volatile anesthetic agents account for over 50% of operating room carbon emissions, with other waste from electricity and single-use equipment, among others.³  Sevoflurane, isoflurane, and desflurane are the most used vaporized anesthetic agents, with global warming potentials between 130 (sevoflurane) and 2540 (desflurane), relative to CO2 on a 100-year timeline.⁴  These gases are administered to patients using fresh gas flow (FGF) via an anesthesia machine which recycles the volatile agents when used with low-flow techniques (FGF <2 L/min) as only a fraction of the gas is metabolized by a patient with each pass through the circuit. High flows through the anesthesia machine circuit contribute to more anesthetic agent waste than do low flows.

Historically, clinical practice to avoid FGF <2 L/min when using sevoflurane has been driven by theoretical concerns of compound A nephrotoxicity in animal models.^5,6  The US Food and Drug Administration package insert for sevoflurane recommends FGF rates >1 L/min to minimize the exposure to compound A, though this has been explicitly challenged by the American Society of Anesthesiologists due to the poor level of evidence, lack of high-quality nonanimal studies, and a paradigm shift toward more sustainable anesthetic practices.^7,8  Research and decades of clinical use have demonstrated that low-flow gas delivery is safe and not associated with clinically adverse outcomes in humans.^9-11 

The escalating impact of climate change disproportionately affects the health of society’s most vulnerable, exacerbating existing health disparities on already marginalized communities, including those within Zuckerberg San Francisco General Hospital’s (ZSFG) community safety net. The goal of this quality improvement (QI) initiative was to implement a clinical decision support (CDS) tool to encourage clinicians at ZSFG to reduce intraoperative sevoflurane use by lowering FGF, thereby decreasing GHG. The Multicenter Perioperative Outcomes Group (MPOG), which collects data on perioperative outcomes, has created an FGF performance benchmark of a mean intraoperative FGF ≤2 L/min.¹²  Hospital systems have demonstrated success in lowering intraoperative FGF rates with CDS tool use.^13-16  Our baseline percentage of cases with a mean intraoperative FGF ≤2 L/min was very low, at 27%. Thus, our primary aim was to use system-wide CDS tools to increase the percentage of cases with FGF ≤2 L/min to at least 40%. To demonstrate reduced GHG emissions as due to lower FGF, our secondary aim was to reduce overall sevoflurane use per hour of general anesthesia.

The project was conducted at ZSFG, a US large academic level 1 trauma center with 284 beds and 89 anesthesiology residents. The participants included anesthesiology residents (all years), certified registered nurse anesthetists (CRNAs), and attending physician anesthesiologists.

The initiative was introduced in July 2023 through 2 email announcements and 4 QI meetings. In July 2023, a CDS tool within the Anesthesia Information Management System (AIMS) was implemented in the hospital’s electronic health record (EHR), appearing as a non-interruptive, orange-colored box whenever the 10-minute average FGF exceeded 1 L/min during the procedure (Figure 1). The alert was an existing configurable feature of the EHR, activated at our request and configured and tested by ZSFG IT Informatics. FGFs were captured automatically in the EHR every minute. Our sevoflurane vaporizers (Dräger Vapor 2000) are mechanical and lack electronic output data, so per-minute sevoflurane use was estimated by multiplying FGF by the inspired sevoflurane concentration. Data were reviewed weekly by our QI program manager, progress was announced at weekly QI huddles and in emails, and a leaderboard of clinicians with the lowest FGFs was distributed at the end of each month.

Our primary outcome, based on the MPOG SUS-04 metric, was the percentage of cases with a mean FGF ≤2 L/min between intubation and extubation. If these time stamps were missing, we used “anesthesia ready” for intubation and “procedure end,” “out of room,” or “anesthesia stop” for extubation. We included cases with intubation or supraglottic airway placement that received nitrous oxide and/or sevoflurane. Desflurane and isoflurane are not used at ZSFG. Cases ≤30 minutes, MRIs, and total intravenous anesthesia (TIVA) cases were excluded.

For our secondary outcome, we tracked sevoflurane usage per hour of general anesthesia. To fully assess our environmental impact and changes in anesthetic practices, we included all TIVA cases, accounting for instances where clinicians may have completely discontinued sevoflurane after our intervention.

Our analysis compared data from our baseline (July 2022 to June 2023) and intervention (July 2023 to July 2024) periods. We tracked the percentage of cases with FGF ≤2 L/min and the average amount of sevoflurane used per hour of general anesthesia, as well as total sevoflurane usage and mean FGF. The highest and lowest performing clinicians and frequency of CDS activation were also observed.

This project was approved by the University of California, San Francisco Institutional Review Board before we implemented our initiative.

Eighty-nine residents, 22 CRNAs, and 38 attendings participated in our project. We analyzed FGF data for 4573 general anesthesia cases at ZSFG, finding that 2677 (58.5%) had a mean FGF ≤2 L/min—a 116.7% increase from our baseline of 27.0% (1200 of 4446; Figure 2A).

Including TIVA cases, the total number of cases was 5366. Sevoflurane usage dropped from an average of 1.20 L/h during baseline to 0.76 L/h during the intervention period, a 36.7% decrease (Figure 2B).

This study found that an integrated EHR signal alerting anesthesia clinicians to FGF rates above 2 L/min substantially increased the percentage of cases with low flow rates and decreased overall sevoflurane use. These findings are consistent with the success seen by other hospital systems that have similarly implemented CDS tools to reduce GHG emissions by way of reducing intraoperative FGF.^13-16 

This project’s success was facilitated by periodic presentations to staff and dissemination of leaderboard reports. In response to feedback, our EHR CDS was updated to trigger after a shorter period of average FGF >2 L/min. This was done to alert anesthesia clinicians to excess FGF during periods of volatile anesthetic delivery where this is most often seen, such as induction. Overall, general feedback from ZSFG clinicians was positive and the alerts have been continued to date.

This study is limited by the lack of included balancing measures (ie, whether the EHR alert reduced other quality measures while improving FGF rates is not known). The short duration of this project also did not allow discovery of whether alert fatigue might occur over time.

Future directions include surveying anesthesia clinicians to optimize alert features within the intraoperative EHR, to deliver individualized feedback about low FGF performance. The amount and frequency of educational reinforcement, through data and other presentations at department meetings, needs to be studied.

Implementation of an institution-wide, non-interruptive alert within the AIMS, together with feedback to clinicians on their performance, substantially changed FGF anesthesia practices, with a more than 100% improvement in low-flow practice. Sevoflurane use also decreased, with overall high participant acceptability.

The authors would like to thank Romain Pirracchio, MD, PhD, and Omar Salman, MD, for their help in conceiving this quality improvement project.

* Denotes co-first authors.

Funding: The authors report no external funding source for this study.

Conflict of interest: The authors declare they have no competing interests.