Secondary Logo

Journal Logo

Ambulatory Anesthesiology and Perioperative Management: Research Report

Economic and Environmental Considerations During Low Fresh Gas Flow Volatile Agent Administration After Change to a Nonreactive Carbon Dioxide Absorbent

Epstein, Richard H. MD, CPHIMS; Dexter, Franklin MD, PhD; Maguire, David P. MD; Agarwalla, Niraj K. DO; Gratch, David M. DO

Author Information
doi: 10.1213/ANE.0000000000001124

Reducing fresh gas flow (FGF) during general anesthesia is a simple strategy that leads to cost savings from reduced consumption of volatile anesthetics1 and to attenuation of atmospheric pollution from venting these greenhouse gases.2,3 We previously published results from a successful program to reduce the average time-weighted FGF <2 L/min during sevoflurane administration through a post hoc personalized e-mail feedback process.a4,5 Nair et al.1 achieved similar results using near real-time pop-up messages on the anesthesia information management system (AIMS) computer screen.

Some anesthesia providers may be reluctant to further reduce the FGF during sevoflurane administration because of Food and Drug Administration product labeling to not exceed 2 minimum alveolar concentration (MAC) hours at flow rates of 1 to <2 L/min and to maintain FGF at ≥1 L/min.b Health Canada also recommends against flow rates <2 L/min.c The theoretical basis for these recommendations is the inverse relation between FGF and peak concentrations of pentafluoroisopropenyl fluoromethyl ether (compound A) in the breathing circuit.6 Compound A is a sevoflurane degradation product that is generated from an interaction with some carbon dioxide (CO2) absorbents, particularly those containing strong alkali (e.g., potassium hydroxide [KOH] or sodium hydroxide [NaOH]); breakdown is enhanced when the absorbent is highly desiccated.7 Although compound A is nephrotoxic in rats,8,9 it has not been shown to have adverse renal effects in humans under clinical conditions.10–16 Litholyme® (Allied Healthcare Products, St. Louis, MO) and other absorbents lacking strong alkali do not produce compound A when normally hydrated and only subclinical concentrations when completely desiccated.17,18

Carbon monoxide (CO), another degradation product of both sevoflurane and desflurane resulting from the interaction of sevoflurane and strong alkalis,19 also is not produced by absorbents lacking NaOH or KOH.17

Because CO2 absorbents are available that virtually eliminate both compound A and CO production, making flow restrictions moot, we initiated a project to replace the reactive soda lime product we had been using (Sodasorb®; Grace, Columbia, MD) with a premium nonreactive formulation (Litholyme; Allied Healthcare Products). Litholyme contains calcium hydroxide with lithium chloride (LiCl) as a catalyst, does not contain either NaOH or KOH,d and has been reported not to produce either compound A or CO.e The objective of the switch to the nonreactive absorbent was to allow the reduction of FGF during sevoflurane administration to below the current package insert recommendation. The hospital approved the absorbent change based on anticipated (small) net savings (Table 1) and reduced emissions of greenhouse anesthetic gases.

Table 1
Table 1:
Projected Annual Hospital Savings Before the Change to Litholyme®

We evaluated the following 4 hypotheses related to our intervention:

  • Hypothesis 1: Average monthly FGF during sevoflurane administration over the intraoperative portions of cases (i.e., surgery begin to surgery end) and during the interval of agent administration would be reduced.
  • Hypothesis 2: Measured sevoflurane consumption per minute of volatile agent administration would be reduced.
  • Hypothesis 3: Cost savings, from a hospital perspective, due to the reduction in sevoflurane consumption would (modestly) exceed the incremental cost of the nonreactive absorbent.
  • Hypothesis 4: Residual wastage in discarded 250-mL sevoflurane bottles would be <2.5 mL (i.e., <1% of a 250-mL bottle); i.e., reduced consumption during patient care would not be associated with greater wastage.


The Thomas Jefferson University IRB determined that this project did not meet the regulatory definition of human subjects research.

Conversion from Soda Lime to Litholyme and Provider Notifications

Anesthesiologists, nurse anesthetists, and anesthesiology residents (i.e., providers) were notified by e-mail on April 14, 2014 and May 6, 2014 about the impending switch from soda lime to the new absorbent (Supplemental Digital Content 1, Over a several-week period, canisters containing exhausted soda lime were refilled with the nonreactive absorbent. After all canisters had been converted, we distributed e-mails to providers encouraging them to maintain 1 L/min FGF during the maintenance sevoflurane administration instead of increasing the FGF to 2 L/min after administering 2 MAC hours (Supplemental Digital Content 2, An announcement was also made at Grand Rounds. The change in departmental policy was approved and endorsed by the chair and vice chair of the department as well as the vice chair for patient safety and quality improvement in the various communications to the department. As part of our routine feedback process (in place since April 2013 and ongoing), every fourth Monday at 5 AM, an e-mail was sent to every provider listing the average (nontime weighted) intraoperative FGF for his or her most recent 10 cases with each volatile agent (Supplemental Digital Content 3, Percent deviations from target FGF values were provided, along with a reminder of the desired average FGF for each agent. No other efforts at goal setting or providing feedback to reduce FGF were made.

Figure 1
Figure 1:
Example of an intraoperative change of the nonreactive absorbent due to excessive rebreathing of carbon dioxide during a desflurane anesthetic. Simultaneous values of inspired carbon dioxide concentrations (PICO2, blue line), fresh gas flow (FGF, red line), and the concentration of the volatile anesthetic desflurane (FiAGENT, green line) are displayed in a case where exhausted absorbent had not been changed before starting the case. There was a rapid rise in the PICO2 beginning within 4 minutes of starting the volatile agent to a level of 11 mm Hg, followed by a transient drop to 7 mm Hg when the FGF was increased from 1 to 2.5 L/min. Subsequently, the PICO2 continued to rise to 12 mm Hg and the absorbent was changed. This was accompanied by the drop in the FiAGENT concentration from 8.4% to 4.4%, which then returned to the vaporizer setting of 6%. The vaporizer was briefly turned off at approximately 50 minutes for refilling, resulting in a transient drop in the FiAGENT.

After the switch to the new absorbent, our anesthesia providers and technicians reported difficulty in assessing the (irreversible) color change in the Litholyme (from the included ethyl violet indicator) as a cue to replace the absorbent.f This resulted in an undesirably high prevalence of having to change the absorbent intraoperatively. We considered such changes undesirable because (1) this introduces the possibility of introducing a leak into the circuit when reseating the canister (even if a quick-change, self-sealing bracket is usedg); (2) end-tidal concentration of volatile agent can be substantially reduced after the change (Fig. 1); and (3) such activities represent a distraction from care of the patient. Therefore, we evaluated real-time inspired CO2 (PICO2) monitoring criteria to mitigate the need to replace the absorbent during delivery of anesthesia due to excessive rebreathing of CO2 (Fig. 1). We initially instructed providers to use PICO2 = 5 mm Hg as the criterion to ask an anesthesia technician to change the absorbent. However, this still resulted in too many intraoperative changes because the PICO2 rapidly increased during low FGF after reaching this threshold. We decided empirically that if the PICO2 reached 3 mm Hg during cases, the absorbent should be changed before the next case. To implement this, we added a notification to change the absorbent to the text message to turn over the operating room (OR), sent from our real-time decision support system to our anesthesia technicians when cases ended.h5 However, technician staffing was insufficient in the late afternoon and evening to refresh all exhausted canisters, resulting in too many intraoperative absorbent changes during first cases of the day. We added an alert to the anesthesia technicians the next morning, before the start of the OR schedule, to check the absorbent in the specified rooms where the last case in those rooms from the previous day had triggered a rebreathing alert. Finally, we sent an e-mail to providers asking them to place a note on the anesthesia machine requesting the absorbent to be changed if they noticed consistent rebreathing >3 mm Hg during the last case of the day (Supplemental Digital Content 4, We subsequently evaluated formally the time to absorbent exhaustion after reaching the PICO2 = 3 mm Hg threshold (see below).

Data Extraction

We extracted inspired CO2 (PICO2) and expired CO2 (PECO2) in mm Hg, oxygen, air, and nitrous oxide FGF in mL/min, inspired volatile agent concentrations (FiAGENT) as percentages, and liquid volatile agent consumption in mL from our AIMS (Innovian®; Dräger, Telford, PA) database (SQL Server 2008 R2; Microsoft, Redmond, WA). These parameters were transmitted by the monitors incorporated within our anesthesia machines (Apollo®; Dräger) to the local workstation running the AIMS client software, and median values over the prior minute were recorded in the central database at approximately 1-minute intervals. When a valid PICO2 or PECO2 measurement was unobtainable (e.g., due to an irregular capnograph), a null value was transmitted to the database. Only cases during which a volatile agent was administered were included in the analysis. Patient age, total case duration (interval from patient entry into and exit from the OR), and total duration of volatile anesthetic administration (interval from first FiAGENT >0.2 MAC to end of surgery) were also retrieved. The intraoperative period was defined as the interval between the recorded events Surgery, BEGIN and Surgery, END. For calculations of FGF, timestamps were normalized from the start of surgery, rounded to the nearest minute. Total FGF was calculated as the sum of the oxygen, air, and nitrous oxide FGF. For calculation of total sevoflurane consumed, measurements from the entire case were used. For determination of the start of a run of PICO2 ≥3 or ≥5 mm Hg, null values at 1-minute intervals were imputed as the most recent valid measurement. For example, if 3 consecutive values were 3, null, and 4, the missing value was set equal to 3.

We compared Apollo sevoflurane consumption measurements with amounts obtained by weighing vaporizers before and after use20 through a Bland-Altman analysis and found these to be equivalent (Fig. 2). We determined that application of the Dion formula21 to estimate the sevoflurane consumption by integrating the product of FGF, and the inspired agent concentration was not equivalent to the values calculated by the proprietary Dräger algorithmi (Fig. 3). Therefore, our evaluation of average sevoflurane consumption before and after the change to the nonreactive absorbent was limited to cases performed using Apollo anesthesia machines, present in approximately 2 of 3 of our anesthetizing locations.

Figure 2
Figure 2:
Comparison of sevoflurane consumption calculated by the Apollo anesthesia machine and by weighing vaporizers. Sevoflurane vaporizers were weighed before and after the conclusion of a convenience sample of 28 anesthetics using a digital scale (Model 21-EK9310-ETEK, Etekcity, China) accurate to ±1 g with a maximum capacity of 10 kg. The scale was tested and found to be linear between 8534 and 8643 g (Y = 1.00 × X, R 2 = 0.998) by adding weights measured with a calibrated digital scale accurate to ±0.1 g (Acculab 333; Sartorius AG, Göttingen, Germany) to a partially empty vaporizer. The weight difference was then converted to mL of liquid sevoflurane (20°C, density = 1.51 g/mL). A Bland-Altman plot of the difference in mL between the Apollo estimate of sevoflurane consumption and that determined by weighing the vaporizer versus the average consumption in mL is shown (circles). The Apollo had a positive bias of 1.6 mL (95% confidence interval, 0.5 to 2.6 mL, solid green line) with 95% limits of agreement (by Student t distribution) between −4.1 and 7.2 mL (dotted green lines). This bias and precision were sufficient to use the Apollo measurements to determine the impact on the reduction of sevoflurane consumption before and after the change in absorbent because the bias would have been constant during the 2 intervals.
Figure 3
Figure 3:
Bland-Altman comparison of sevoflurane consumption from the Apollo anesthesia machine and as calculated using the Dion formula.21 Consumption of sevoflurane as measured by the Apollo anesthesia machine and that calculated by integrating the product of fresh gas flow and inspired sevoflurane concentration over the entire anesthetic were compared for N = 3180 consecutive cases where sevoflurane was used. A Bland-Altman plot of the difference in mL between the Apollo estimate of sevoflurane consumption and that determined using the Dion formula in mL is shown (circles). The Apollo had a positive bias of 12.3 mL (95% confidence interval, 12.0 to 12.7 mL, solid green line) and 95% limits of agreement (by Student t distribution) between −5.7 and 30.4 mL (dotted green lines). There was an increasing discrepancy between the Apollo measurements and the Dion formula as the total amount of sevoflurane consumed increased, and the precision was poor. This indicates that during low fresh gas flow conditions, the Dion formula cannot be used accurately to estimate the sevoflurane consumption.

Records of sevoflurane and absorbent invoices were provided by the hospital and used to determine purchases by date of order before and after the change to the nonreactive absorbent. A full bag of either absorbent was used to refill a depleted canister, so the number of bags purchased was proportional to the amount of absorbent utilized.

Measuring Impending Absorbent Exhaustion

For each case, we determined whether a run of 3 consecutive PICO2 values ≥3 mm Hg occurred, and, if so, the interval between the start of the run and the start of the first run of 5 consecutive PICO2 values ≥5 mm Hg (which we defined as absorbent depletion, requiring replacement). If the end of surgery occurred with no run of 5 PICO2 values ≥5 mm Hg having occurred, the case was right censored at the end of surgery timestamp. We terminated measurement at this milestone because FGF is typically increased to approximately 10 L/min at this point, effectively ending any rebreathing, even if the absorbent was nearly completed depleted.

Measurement of Sevoflurane Waste from Incompletely Emptied Bottles

We wanted to rule out discarding partially full bottles of sevoflurane by providers as a potential source of wastage to put the anticipated reduction in sevoflurane consumption in perspective. To estimate wastage, convenience samples were collected in batches (i.e., every 2–4 days) from our surgical suites over a 3-week period. Providers were requested by e-mail and via occasional pop-up notices on their anesthesia workstations to cap and save any sevoflurane bottles that would otherwise be discarded in the trash after filling the vaporizer. The anesthesia technicians retrieved the bottles when turning over the ORs and placed them in a bucket in their workroom; providers also occasionally placed their empties here. The amount of liquid sevoflurane remaining was calculated by weighing each bottle with the residual agent 2 times, suctioning the liquid from the bottle, reweighing 2 times, and calculating the difference between the means for each bottle. The weights were converted to mL by dividing by the density of sevoflurane (1.51 g/mL at 20°C). An Acculab 333 digital scale (Sartorius AG, Göttingen, Germany), accurate to 0.1 g, was utilized. Average wastage ± SE per 250-mL bottle was computed.

Statistical Analysis

We considered a difference of 100 mL/min in the mean sevoflurane FGF to be important. During the first N = 4 4-week batches, the mean sevoflurane FGF was 1892 mL/min and the SD among the batches was 32.5 mL. Using these values, N = 8 4-week batches before and after the absorbent conversion was estimated to provide 95% power to detect a 100 mL/min difference in FGF at α = 0.001 using the 2-sided Student t test.

The method of batch means,22–27 with 4-week batches, was used for all determinations, with 8 batches before the change compared with 8 batches after the change (reported as N = 8 of 8).j For financial calculations related to volatile agent and absorbent purchases (received several months after the FGF analysis was completed), 10 batches were used before and after (reported as N = 10 of 10).

Agent consumption and FGF data are presented as the mean ± SE. Statistical comparisons between groups were made using the 2-sided Student t test. For comparison with our target FGF of 1.25 L/min, the 1-group, 1-sided Student t test was used. For the analysis of the interval from reaching a PICO2 = 3 mm Hg to reaching a PICO2 = 5 mm Hg, the Kaplan-Meier survival statistic was calculated with absorbent exhaustion (i.e., 5 PICO2 values ≥5 mm Hg) considered equivalent to death. Statistical significance for all comparisons required P < 0.01. Systat® 12 (Systat Software, Inc., San Jose, CA) was used for all statistical calculations.


Characteristics of Cases Before and After the Absorbent Change

A total of N = 20,235 cases were analyzed (80.2% sevoflurane, 15.1% desflurane, and 4.7% isoflurane). There were no statistically significant differences before and after the absorbent change among any of the volatile agents for patient age, total case duration, duration of volatile agent administration, or inspired or expired agent concentration (Table 2). The average monthly number of cases in which desflurane was used for at least a portion of the case decreased significantly after the absorbent switch.

Table 2
Table 2:
Case Characteristics (Mean ± SEM) Before and After the Change from Soda Lime to Litholyme

There was a nonsignificant trend for a reduction in the number of bags of CO2 absorbent purchased during the 10 consecutive 4-week intervals after, compared with before, the switch (N = 10 of 10; difference = −19.1; 95% confidence interval [CI], −40.8 to 2.63; P = 0.08).

Hypothesis 1: FGF During Sevoflurane Administration Was Reduced After the Absorbent Change

There was an immediate and sustained reduction in intraoperative FGF after the announcement of the initiative (Fig. 4). Intraoperative FGF was reduced for cases in which sevoflurane was administered by 435 mL/min (N = 8 of 8 periods; 95% CI, 391 to 479; P < 0.0001) (Table 3). Considering the portion of cases during which sevoflurane was administered, FGF was reduced by 372 mL/min (N = 8 of 8; 95% CI, 334 to 409 mL/min; P < 0.0001) (Table 3). Hypothesis 1 was accepted.

Table 3
Table 3:
Fresh Gas Flow and Volatile Agent Consumption Before and After the Change to Litholyme
Figure 4
Figure 4:
Sevoflurane average total fresh gas flow (FGF) during the intraoperative period (surgery begin to surgery end) during the 8 4-week intervals before and after replacement of soda lime with the nonreactive absorbent. A LOESS curve was constructed for each segment with span = 0.75 and degree = 2 using the R function loess. Within 4 weeks of the announcement to staff that the carbon dioxide absorbent had been switched and that FGF now could be maintained at 1.0 L/min throughout the entire case when sevoflurane was administered (as opposed to our previous policy of increasing to 2.0 L/min after 2 MAC hours had been administered), there was a decrease in the FGF from 1706 ± 11 to 1391 ± 9 mL/min (P < 10−5).

For cases where desflurane or isoflurane were administered, there was a nonsignificant trend for a decrease in the intraoperative FGF (Table 3).

Hypothesis 2: Sevoflurane Consumption Was Reduced After the Absorbent Change

Sevoflurane consumption per minute of administration decreased by 0.039 mL/min (N = 8 of 8; 95% CI, 0.029 to 0.049 mL/min; P < 0.0001) after the change to the nonreactive absorbent (Table 3). Based on the hospital’s wholesale cost for sevoflurane, this would be equivalent to a reduction of $0.89 per hour (95% CI, −$0.67 to −$1.11). Hypothesis 2 was accepted.

There was a negative trend but not a significant decrease in the average number of bottles of sevoflurane purchased during each of the 10 4-week intervals before, compared with after, the change to the nonreactive absorbent decreased from 180.8 ± 6.3 to 160.7 ± 8.5 (N = 10 of 10, difference = −17.1; 95% CI, −17.4 to 16.8; P = 0.07). This was in the context of no significant change in inspired or expired agent concentrations, duration of cases, or number of cases (Table 3).

Hypothesis 3: Net Hospital Savings Would Be Positive After the Absorbent Change

To compare the hospital expenses before and after the absorbent switch, sevoflurane costs were determined by multiplying the number of bottles purchased by the average wholesale cost ($95 per bottle). Similarly, absorbent costs reflected the number of bags purchased times the unit cost. The difference in mean cost for the sum of the sevoflurane and absorbent purchases for each of the 10 4-week intervals before and after the absorbent switch was −$293 (N = 10 of 10, 95% CI, −$2853 to $2266; P = 0.81). Hypothesis 3 was rejected.

Figure 5
Figure 5:
Contour plot for the sensitivity analysis of net cost savings in US dollars (USD). The estimated annual sevoflurane savings from reducing fresh gas flow are plotted on the y-axis and the estimated annual increase in absorbent cost are plotted on the x-axis. The colored bands represent the net difference (grouped from −$50,000 to $50,000 in increments of $25,000) between the sevoflurane savings and the additional absorbent cost. The 3 scenarios in Appendix A are overlaid on this graph. In scenario A, there would be potential for a small net savings, but a small risk of a negative financial outcome. In scenario B, there would be net savings over the entire range of sevoflurane savings and incremental absorbent costs. In scenario C, there is a substantial risk of a net negative savings. See Appendix A for instructions on how to use this graph to conduct a sensitivity analysis representing local conditions.

We conducted a sensitivity analysis to explore the financial implications at other hospitals given a range of sevoflurane acquisition costs, the differential costs between the current and a proposed replacement absorbent, and varying percentage reductions in sevoflurane consumption over a realistic range of parameters (Fig. 5, Supplemental Digital Content 4, Hospital cost savings, at best, are likely to be modest, given the relatively low hourly cost of sevoflurane administration.

Hypothesis 4: Sevoflurane Wastage in Discarded 250-mL Bottles Would Be <1%

Fifty-two bottles were collected on 8 different days from the storage bucket in the anesthesia workroom, with samples ranging from 4 to 11 bottles; 1 bottle was missing its cap and was excluded from the analysis. The average amount of residual sevoflurane per bottle was 0.67 ± 0.06 mL (N = 8 collection days; 95% CI, 0.54 to 0.81 mL per bottle; P < 0.0001 vs 2.5 mL).

Monitoring Absorbent Exhaustion

Figure 6
Figure 6:
Kaplan-Meir survival curve for the nonreactive absorbent exhaustion. The interval in minutes from when the inspired carbon dioxide (PICO2) reached at least 3 mm Hg for at least 3 minutes to when the PICO2 reached 5 mm Hg for at least 5 minutes (exhaustion) was determined (solid blue line) with 95% confidence limits (dotted blue lines). Data were right censored if the case ended before reaching the exhaustion threshold. In 50% of instances, the nonreactive absorbent became depleted within 95 minutes of reaching the PICO2 = 3 mm Hg threshold. The curves are nearly identical if 5 minutes of PICO2 ≥3 mm Hg is used as the initial threshold.

Once the PICO2 reached 3 mm Hg for at least 3 consecutive minutes, the absorbent became exhausted within 95 minutes in most (i.e., >50%) canisters (Fig. 6). Because the mean duration of volatile anesthetic administration was 146 minutes during cases where sevoflurane was administered (Table 2), this indicated that the absorbent should be changed between cases if there was persistent rebreathing of CO2 ≥3 mm Hg. Otherwise, there would be a substantive chance that absorbent exhaustion would occur during the subsequent case, requiring an intraoperative absorbent change.


We were able to demonstrate a significant and sustained reduction in both FGF and calculated average sevoflurane consumption per minute. Providers changed their FGF behavior promptly upon the conversion from soda lime to the nonreactive absorbent. However, they did not achieve the desired sevoflurane FGF intraoperative target of 1.25 L/min, only reaching 1.51 L/min (P < 0.0001 compared with 1.25). Given the small additional benefit that would accrue from achieving the FGF goal, we elected not to implement additional measures, such as sending intraoperative pop-up messages to the workstations when the FGF was >1.0 L/min, as described by Nair et al.1 The measured reduction in sevoflurane ordering (9.5%) was less than the reduction in sevoflurane consumption (13.2%). However, when judged from the perspective of the hospital’s variable costs, the project appears to have been cost neutral, as savings from sevoflurane reduction were offset by the additional cost of the premium CO2 absorbent.

The reduction in FGF has a societal benefit from reducing the venting of anesthetic greenhouse gases into the environment. However, in perspective,28 the global contribution of volatile anesthetics released into the atmosphere only represents approximately 0.01% of CO2 released from the fossil fuel consumption.29,30 Furthermore, we were unable to find a life-cycle assessmentk of Litholyme or Sodasorb, so we cannot estimate accurately the differences in the environmental impact between the 2 absorbents. It is not sufficient simply to assess the decreased venting of the anesthetics when choosing an absorbent, as this ignores potential cradle to grave differences due to manufacturing, processing, packaging, transportation, recycling, and disposal.

Managing the process of pre-emptive absorbent replacement to avoid intraoperative changes was unexpectedly challenging. There were behavioral issues related to changing exhausted absorbent at the end of the day. Our providers and anesthesia technicians found that a color change of the indicator was not a reliable cue for when to change the absorbent.l Thus, monitoring PICO2 (a parameter measured by capnometers) during volatile general anesthetics, setting and activating the alarm limit, and responding appropriately when excessive rebreathing is noted are necessary. We initially applied empirical criteria for levels and duration of rebreathing to trigger an alert to our anesthesia technicians to change the absorbent between cases or before the first case of the day and then modified these based on our analysis of impending exhaustion. Although changing the absorbent is neither difficult nor time consuming, intraoperative replenishment creates a distraction from care of the patient. In addition, a circuit leak may occur if the canister does not seal properly, caustic dust may be released from the process of emptying and refilling the canister, and the level of volatile anesthetic may be inadvertently reduced, with the potential for patient movement or recall. Although we have an automated system to notify our technicians to change the absorbent, providers also need to take responsibility to monitor the PICO2 and ensure that nearly exhausted absorbent is replaced before starting the next case.

Our study has several limitations. First, our results cannot be directly applied to other hospitals, because volatile agent and absorbent acquisition costs, patterns of FGF, percentages of volatile agent wasted from discarding partially full bottles, willingness to reduce FGF during sevoflurane anesthesia, and levels of volatile anesthesia provided will vary. In addition, the ability to distribute personalized e-mail FGF reports to providers (as described in our recent review article5) or to implement a near real-time feedback system1 likely is necessary to achieve the reduction in sevoflurane consumption. However, the principles we describe with respect to assessing the cost utility of a program incorporating a change to a nonreactive absorbent and reduction of FGF during volatile agent administration are relevant. Appendix A can be used to conduct a sensitivity analysis on the financial impact of a potential change to a premium CO2 absorbent over a range of input parameters and anticipated changes. Regardless, facilities should not expect large savings, because sevoflurane is relatively inexpensive, nonreactive CO2 absorbents cost substantially more than conventional formulations, and the opportunity to greatly reduce FGF is likely limited. Prefilled disposable or recyclable canisters containing nonreactive absorbents are much more expensive than bulk product, as used in our study, and would have further reduced the net savings.

Another limitation is that the very small amount of wastage of sevoflurane in discarded bottles may have been subject to a Hawthorne effect. However, providers were told that we were collecting bottles to assess sevoflurane usage, not to assess how much they were wasting. In addition, over the 3-week collection period, there was no trend for increased residual volumes in the bottles, as might have been expected due to desensitization had this effect been present. Thus, our application of the method of batch means to measure wastage was appropriate.

Another limitation of our approach is that measurement of agent consumption by the anesthesia machine is vendor and model dependent and may also depend on the ability of one’s AIMS to record transmitted values. For example, Dräger Tiro® anesthesia machines are present in our ambulatory and out-of-OR locations and lack the ability to calculate agent consumption. Weighing vaporizers to determine agent consumption is a cumbersome process. At least during low FGF anesthesia, the Dion method21 of estimating volatile agent consumption is not sufficiently accurate to gauge savings.

A final limitation is that the hospital purchasing records may not have reflected accurately the consumption of either sevoflurane or CO2 absorbent within each 4-week batch. There was limited space for storage of these materials in the hospital’s stock room, so deliveries were made every 1 to 3 days, based on inventory. However, there were considerable supplies stored in the anesthesia drug carts and in the anesthesia work rooms, thereby buffering the impact of any ordering shortfalls.

In summary, we showed that a department can successfully transition to a premium nonreactive CO2 absorbent in a manner that is at least cost neutral by reducing the sevoflurane FGF to below limits recommended by the package insert. This was achieved in part by monitoring electronically for when to change the absorbent and automatically notifying the anesthesia technicians. However, the potential for other departments to achieve the limited success we realized depends on their ability to collect and analyze the necessary data, local policies regarding FGF and absorbent replacement, their ability to alter provider behavior through monitoring and feedback, and local prices.


The net cost of changing to a premium carbon dioxide (CO2) absorbent is the difference between the savings realized through reduced consumption of volatile agent via reduction of fresh gas flow (FGF) and the additional cost of the nonreactive absorbent compared with current absorbent expenses.

There will be uncertainty as to the extent that these potential savings can be realized due to several factors.

First, the new absorbent may have a different CO2-absorbing capacity than the current absorbent, which needs to be accounted for in the projected cost estimate. For example, if the new agent costs 2× as much but lasts 2× as long, then, ideally, the switch would be cost neutral. However, the full benefit of additional absorbent capacity may not be realized, so that needs to be accounted for in the sensitivity analysis.

Second, the total reduction in volatile agent consumption may not be proportional to the reduction in FGF during the intraoperative portion of cases (begin to end surgery). This is because during the induction of general anesthesia, higher gas flows and vaporizer concentrations will be utilized, and when FGF is reduced early into the anesthetic, higher vaporizer concentrations will be needed to maintain the same expired agent concentration compared with previously, when higher FGF was used. These factors are related to the uptake and distribution of volatile agent and will be more pronounced for relatively soluble agents (e.g., sevoflurane and isoflurane) than with more insoluble agents (desflurane). For short cases, the impact of FGF reduction during maintenance on agent consumption will be reduced, compared with longer cases, as the percentage of agent consumed during the early uptake phase will be larger than for long cases.

The following worksheet and the associated graph can be used to estimate the range of potential savings (or loses) that may be associated with the change to a premium CO2 absorbent. Worked examples are provided based for 3 hypothetical hospitals with varying costs for sevoflurane and absorbents, using a range of projected savings.

No title available.

To perform a graphical sensitivity analysis, construct a rectangle on the contour plot in Figure 6 with the vertices at the minimum and maximum values for the annual estimated incremental absorbent costs (x-axis) and the projected annual sevoflurane savings (y-axis). Each colored band represents the net difference between the estimated sevoflurane savings and the additional cost of the premium absorbent. The area within the rectangle represents the range of potential net savings or losses. The 3 scenarios (A, B, and C) presented above are overlaid on this graph.


Name: Richard H. Epstein, MD, CPHIMS.

Contribution: This author helped design the study, conduct the study, analyze the data, and write the manuscript. He is the archival author.

Attestation: Richard H. Epstein attests to the integrity of the original data and the analysis reported in this manuscript and has approved the final manuscript.

Name: Franklin Dexter, MD, PhD.

Contribution: This author helped design the study, analyze the data, and write the manuscript.

Attestation: Franklin Dexter attests to the analysis reported in this manuscript and has approved the final manuscript.

Name: David P. Maguire, MD.

Contribution: This author helped prepare the manuscript.

Attestation: David P. Maguire attests to the integrity of the original data and approved the final manuscript.

Name: Niraj K. Agarwalla, DO.

Contribution: This author helped conduct the study and prepare the manuscript.

Attestation: Niraj K. Agarwalla attests to the integrity of the original data and approved the final manuscript.

Name: David M. Gratch, DO.

Contribution: This author helped prepare the manuscript.

Attestation: David M. Gratch attests to the integrity of the original data and approved the final manuscript.


Dr. Franklin Dexter is the Statistical Editor for Anesthesia & Analgesia. This manuscript was handled by Dr. Tong J. Gan, Section Editor for Ambulatory Anesthesiology and Perioperative Management, and Dr. Dexter was not involved in any way with the editorial process or decision.


a Patel N, Maguire D, Dexter F, Epstein RH. Reduction of fresh gas flow during administration of volatile anesthetic agents via monthly individualized e-mail feedback. 2014 Annual Meeting of the Society for Technology in Anesthesia. Available at: Accessed January 3, 2015.
Cited Here

b Baxter. Sevoflurane, USP Package Insert. Available at: Accessed January 3, 2015.
Cited Here

c Baxter Corporation. Sevoflurane Product Monograph. Available at: Accessed January 2, 2015.
Cited Here

d Allied Healthcare Products. Material Safety Data Sheet: Litholyme. Available at: Accessed January 3, 2015.
Cited Here

e Allied Healthcare Products. Litholyme: a safer and more cost-effective carbon dioxide absorbent. Available at: Accessed January 3, 2015.
Cited Here

f Most of our absorbent canisters were cloudy from approximately 8 years of exposure to soda lime and were not able to be rendered transparent, despite vigorous cleaning. In addition, fresh nonreactive absorbent has a slightly purplish hue.
Cited Here

g Kuruma Y, Kita Y, Fujii S. Exchanging a CLIC absorber in the middle of surgery. APSF Newsletter, Winter 2013. Available at: Accessed July 6, 2015.
Cited Here

h Gratch D, Maguire D, Epstein RH. Development of a system to pre-emptively identify impending carbon dioxide absorbent depletion during low fresh gas flow anesthesia to mitigate the need for intraoperative replacement during the case to follow. 2015 Annual Meeting of the International Anesthesia Research Society.
Cited Here

i Heesch R. Method for measuring the anesthetic agent consumption in a ventilation system. US Patent 20080029092 A1. Available at: Accessed May 13, 2015.
Cited Here

j The method of batch means (bins) was used, as for analyzing most other operating room managerial data because mean fresh gas flows among successive cases represent a time series of correlated values.22–27 The data are correlated because (1) mean fresh gas flows vary among providers, (2) most providers perform multiple cases on days when delivering anesthesia, and (3) providers are often assigned over several-week intervals to work in the same areas with higher or lower prevalences of volatile anesthetic use. By aggregating cases over 4-week periods, the impact of such unmeasured relationships is mitigated.
Cited Here

k Wikpedia. Life-Cycle Assessment. Available at: Accessed May 8, 2015.
Cited Here

l To put this in perspective, a competing nonreactive absorbent containing pure lithium hydroxide monohydrate in a solid-state matrix (SpiroLith®; Micorpore, Elkton, MD) lacks any color indicator.
Cited Here


1. Nair BG, Peterson GN, Neradilek MB, Newman SF, Huang EY, Schwid HA. Reducing wastage of inhalation anesthetics using real-time decision support to notify of excessive fresh gas flow. Anesthesiology. 2013;118:874–84
2. Sherman J, Le C, Lamers V, Eckelman M. Life cycle greenhouse gas emissions of anesthetic drugs. Anesth Analg. 2012;114:1086–90
3. McGain F, Story D, Kayak E, Kashima Y, McAlister S. Workplace sustainability: the “cradle to grave” view of what we do. Anesth Analg. 2012;114:1134–9
4. Dexter F, Maguire D, Epstein RH. Observational study of anaesthetists’ fresh gas flow rates during anaesthesia with desflurane, isoflurane and sevoflurane. Anaesth Intensive Care. 2011;39:460–4
5. Epstein RH, Dexter F, Patel N. Influencing anesthesia provider behavior using anesthesia information management system data for near real-time alerts and post hoc reports. Anesth Analg. 2015;121:678–92
6. Bito H, Ikeda K. Effect of total flow rate on the concentration of degradation products generated by reaction between sevoflurane and soda lime. Br J Anaesth. 1995;74:667–9
7. Neumann MA, Laster MJ, Weiskopf RB, Gong DH, Dudziak R, Förster H, Eger EI 2nd. The elimination of sodium and potassium hydroxides from desiccated soda lime diminishes degradation of desflurane to carbon monoxide and sevoflurane to compound A but does not compromise carbon dioxide absorption. Anesth Analg. 1999;89:768–73
8. Morio M, Fujii K, Satoh N, Imai M, Kawakami U, Mizuno T, Kawai Y, Ogasawara Y, Tamura T, Negishi A, Kumagai Y, Kawai T. Reaction of sevoflurane and its degradation products with soda lime: toxicity of the byproducts. Anesthesiology. 1992;77:1155–64
9. Keller KA, Callan C, Prokocimer P, Delgado-Herrera L, Friedman MB, Hoffman GM, Wooding WL, Cusick PK, Krasula RW. Inhalation toxicity study of a haloalkene degradant of sevoflurane, Compound A (PIFE), in Sprague-Dawley rats. Anesthesiology. 1995;83:1220–32
10. Frink EJ Jr, Malan TP, Morgan SE, Brown EA, Malcomson M, Gandolfi AJ, Brown BR Jr. Quantification of the degradation products of sevoflurane in two CO2 absorbents during low-flow anesthesia in surgical patients. Anesthesiology. 1992;77:1064–9
11. Bito H, Ikeda K. Closed-circuit anesthesia with sevoflurane in humans. Effects on renal and hepatic function and concentrations of breakdown products with soda lime in the circuit. Anesthesiology. 1994;80:71–6
12. Bito H, Ikeuchi Y, Ikeda K. Effects of low-flow sevoflurane anesthesia on renal function: comparison with high-flow sevoflurane anesthesia and low-flow isoflurane anesthesia. Anesthesiology. 1997;86:1231–7
13. Kharasch ED, Frink EJ Jr, Zager R, Bowdle TA, Artru A, Nogami WM. Assessment of low-flow sevoflurane and isoflurane effects on renal function using sensitive markers of tubular toxicity. Anesthesiology. 1997;86:1238–53
14. Higuchi H, Sumita S, Wada H, Ura T, Ikemoto T, Nakai T, Kanno M, Satoh T. Effects of sevoflurane and isoflurane on renal function and on possible markers of nephrotoxicity. Anesthesiology. 1998;89:307–22
15. Obata R, Bito H, Ohmura M, Moriwaki G, Ikeuchi Y, Katoh T, Sato S. The effects of prolonged low-flow sevoflurane anesthesia on renal and hepatic function. Anesth Analg. 2000;91:1262–8
16. Ebert TJ, Arain SR. Renal responses to low-flow desflurane, sevoflurane, and propofol in patients. Anesthesiology. 2000;93:1401–6
17. Keijzer C, Perez RS, de Lange JJ. Compound A and carbon monoxide production from sevoflurane and seven different types of carbon dioxide absorbent in a patient model. Acta Anaesthesiol Scand. 2007;51:31–7
18. Struys MM, Bouche MP, Rolly G, Vandevivere YD, Dyzers D, Goeteyn W, Versichelen LF, Van Bocxlaer JF, Mortier EP. Production of compound A and carbon monoxide in circle systems: an in vitro comparison of two carbon dioxide absorbents. Anaesthesia. 2004;59:584–9
19. Baxter PJ, Garton K, Kharasch ED. Mechanistic aspects of carbon monoxide formation from volatile anesthetics. Anesthesiology. 1998;89:929–41
20. Luria I, Lampotang S, Schwab W, Cooper LA, Lizdas D, Gravenstein N. Automated, real-time fresh gas flow recommendations alter isoflurane consumption during the maintenance phase of anesthesia in a simulator-based study. Anesth Analg. 2013;117:1139–47
21. Dion P. The cost of anaesthetic vapours. Can J Anaesth. 1992;39:633
22. Ledolter J, Dexter F, Epstein RH. Analysis of variance of communication latencies in anesthesia: comparing means of multiple log-normal distributions. Anesth Analg. 2011;113:888–96
23. Dexter F, Epstein RH, Marcon E, Ledolter J. Estimating the incidence of prolonged turnover times and delays by time of day. Anesthesiology. 2005;102:1242–8
24. Dexter F, Epstein RH. Increased mean time from end of surgery to operating room exit in a historical cohort of cases with prolonged time to extubation. Anesth Analg. 2013;117:1453–9
25. Dexter F, Marcon E, Epstein RH, Ledolter J. Validation of statistical methods to compare cancellation rates on the day of surgery. Anesth Analg. 2005;101:465–73
26. Dexter F. High-quality operating room management research. J Clin Anesth. 2014;26:341–2
27. Austin TM, Lam HV, Shin NS, Daily BJ, Dunn PF, Sandberg WS. Elective change of surgeon during the OR day has an operationally negligible impact on turnover time. J Clin Anesth. 2014;26:343–9
28. Sulbaek Andersen MP, Nielsen OJ, Wallington TJ, Karpichev B, Sander SP. Medical intelligence article: assessing the impact on global climate from general anesthetic gases. Anesth Analg. 2012;114:1081–5
29. Sulbaek Andersen MP, Sander SP, Nielsen OJ, Wagner DS, Sanford TJ Jr, Wallington TJ. Inhalation anaesthetics and climate change. Br J Anaesth. 2010;105:760–6
30. Friedlingstein P, Houghton RA, Marland G, Hackler J, Boden TA, Conway TJ, Canadell JG, Raupach MR, Ciais P, Le Quéré C. Update on CO2 emissions. Nat Geosci. 2010;3:811–2

Supplemental Digital Content

© 2016 International Anesthesia Research Society