Secondary Logo

Sources of Variation in Anesthetic Drug Costs

Wanderer, Jonathan, P., MD, MPhil*,†; Nelson, Sara, E., MPH*; Hester, Douglas, L., MD*; Shotwell, Matthew, PhD*,‡; Sandberg, Warren, S., MD, PhD*,†; Anderson-Dam, John, MD§; Raines, Douglas, E., MD; Ehrenfeld, Jesse, M., MD, MPH*,‡,¶

doi: 10.1213/ANE.0000000000002732
Healthcare Economics, Policy, and Organization: Original Clinical Research Report
Free
SDC

BACKGROUND: Increasing attention has been focused on health care expenditures, which include anesthetic-related drug costs. Using data from 2 large academic medical centers, we sought to identify significant contributors to anesthetic drug cost variation.

METHODS: Using anesthesia information management systems, we calculated volatile and intravenous drug costs for 8 types of inpatient surgical procedures performed from July 1, 2009, to December 31, 2011. For each case, we determined patient age, American Society of Anesthesiologists (ASA) physical status, gender, institution, case duration, in-room provider, and attending anesthesiologist. These variables were then entered into 2 fixed-effects linear regression models, both with logarithmically transformed case cost as the outcome variable. The first model included duration, attending anesthesiologist, patient age, ASA physical status, and patient gender as independent variables. The second model included case type, institution, patient age, ASA physical status, and patient gender as independent variables. When all variables were entered into 1 model, redundancy analyses showed that case type was highly correlated (R2 = 0.92) with the other variables in the model. More specifically, a model that included case type was no better at predicting cost than a model without the variable, as long as that model contained the combination of attending anesthesiologist and case duration. Therefore, because we were interested in determining the effect both variables had on cost, 2 models were created instead of 1. The average change in cost resulting from each variable compared to the average cost of the reference category was calculated by first exponentiating the β coefficient and subtracting 1 to get the percent difference in cost. We then multiplied that value by the mean cost of the associated reference group.

RESULTS: A total of 5504 records were identified, of which 4856 were analyzed. The median anesthetic drug cost was $38.45 (25th percentile = $23.23, 75th percentile = $63.82). The majority of the variation was not described by our models—35.2% was explained in the model containing case duration, and 32.3% was explained in the model containing case type. However, the largest sources of variation our models identified were attending anesthesiologist, case type, and procedure duration. With all else held constant, the average change in cost between attending anesthesiologists ranged from a cost decrease of $41.25 to a cost increase of $95.67 (10th percentile = −$19.96, 90th percentile = +$20.20) when compared to the provider with the median value for mean cost per case. The average change in cost between institutions was significant but minor ($5.73).

CONCLUSIONS: The majority of the variation was not described by the models, possibly indicating high per-case random variation. The largest sources of variation identified by our models included attending anesthesiologist, procedure type, and case duration. The difference in cost between institutions was statistically significant but was minor. While many prior studies have found significant savings resulting from cost-reducing interventions, our findings suggest that because the overall cost of anesthetic drugs was small, the savings resulting from interventions focused on the clinical practice of attending anesthesiologists may be negligible, especially in institutions where access to more expensive drugs is already limited. Thus, cost-saving efforts may be better focused elsewhere.

From the *Department of Anesthesiology, Vanderbilt University, The Vanderbilt Clinic, Nashville, Tennessee

Departments of Biomedical Informatics

Biostatistics, Vanderbilt University, Nashville, Tennessee

§Department of Anesthesiology, UCLA Medical Center, Ronald Reagan UCLA Medical Center, Los Angeles, California

Department of Anesthesia, Critical Care and Pain Medicine, Massachusetts General Hospital, Boston, Massachusetts

Department of Surgery, Vanderbilt University, Nashville, Tennessee.

Published ahead of print December 15, 2017.

Accepted for publication November 8, 2017.

Funding: Departmental.

The authors declare no conflicts of interest.

Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal’s website.

Reprints will not be available from the authors.

Address correspondence to Jonathan P. Wanderer, MD, MPhil, The Vanderbilt Clinic, 1301 Medical Center Dr, Suite 4648, Nashville, TN 37232. Address e-mail to Jonathan.p.wanderer@vanderbilt.edu.

KEY POINTS

  • Question: What are the significant contributors to variation in anesthetic drug cost?
  • Findings: The largest sources of variation identified by our models included attending anesthesiologist, procedure type, and case duration; however, the overall cost of anesthetic drugs was small.
  • Meaning: The savings resulting from interventions focused on the clinical practice of attending anesthesiologists may be negligible, especially in institutions where access to expensive drugs is limited.

Over the past decade, increased attention has been paid to health care expenditures. This has resulted in efforts to control costs within anesthesia care. The adoption of anesthesia information management systems (AIMS) has made it possible to compute anesthetic-related drug costs in a highly granular fashion.1 AIMS-driven cost-containment strategies have been adopted to reduce fresh gas flows,2,3 influence the selection of volatile anesthetics,2,4 and deliver provider-specific cost feedback.2,5,6 Education initiatives, departmental policies, and pharmaceutical practice guidelines have also been associated with decreased costs.7–9 Additionally, technology for automated control of end-tidal volatile anesthetics has the potential to further cost reduction.10 While these strategies have been associated with lower drug costs, this does not imply that utilization of less expensive anesthetic agents results in net cost reduction for all practice environments.11,12 Experimental data have demonstrated conflicting trade-offs among anesthetic drug cost, side effects, and recovery time.13–15

Despite the focus on anesthetic-related drug costs, it is not known how much variation in cost is driven by the needs of a particular surgical procedure, the decisions of individual clinicians, patient factors, institutional practice patterns, and drug acquisition costs. Previous study has demonstrated that an analysis of cost per relative value unit (RVU) may be more accurate than cost per case by accounting for some of these factors. However, no insight is given into the weights of these factors or cost variation by institution, nor the explanatory value of RVUs versus procedure type and case duration.16 Understanding the relative contributions of these factors may clarify which types of interventions will be most successful in reducing anesthetic-related drug costs and supportive of cost containment efforts.

Using data from 2 large academic medical centers, we sought to identify significant contributors to variation in anesthetic-related drug costs. We hypothesized that the attending anesthesiologist and surgical case duration would display the largest observed cost variation, and that there would be a significant difference by institution. Additionally, we hypothesized that surgical procedure and case duration would have more explanatory value when compared to the American Society of Anesthesiologists (ASA) relative value scale system.

Back to Top | Article Outline

METHODS

This study received approval by the Vanderbilt University Medical Center Institutional Review Board and the Partners Human Research Committee with a waiver for written informed consent. This manuscript adheres to the applicable Enhancing the QUAlity and Transparency Of health Research (EQUATOR) guidelines.

Back to Top | Article Outline

Patient Population

We identified all patients who had an electronically documented anesthetic for 10 types of surgical procedures between July 1, 2009, and December 31, 2011, at both Vanderbilt University Medical Center and Massachusetts General Hospital. Procedures were chosen by the study investigators as representative examples of common cases performed by a variety of surgical subspecialties (eg, laparoscopic appendectomy was selected for general surgery).

Back to Top | Article Outline

Cost Calculation and Data Collection

Drug acquisition costs were obtained from the central pharmacy at each institution reflecting costs in January 2012. The study interval was chosen as there were no substantial shifts in drug costs for any anesthetic drugs during this period. Case cost calculators were separately developed at each institution because the institutions used separate electronic medical records with different data structures. These calculators determined the total amount of each drug administered for each included case by querying their respective AIMS. We assumed that drug vials were only used for a single patient, which is the standard practice at both institutions. These totals were divided by the administered drug quantity per vial and rounded up to whole numbers of vials, which were then multiplied by the per-vial drug acquisition cost.

At Vanderbilt University Medical Center, drug administration data elements were derived from the Perioperative Data Warehouse, while they were derived from MetaVision (iMDsoft, Waltham, MA) at Massachusetts General Hospital. Drug infusion dosages were calculated using the patient weight as documented in the electronic medical record. For drug infusions that had no documented stop time, a hierarchical system was used to impute the stop time, using procedure end time, extubation time, out of room time and end of anesthesia care, in that order, as available. Volatile anesthetic costs were determined by measuring total fresh gas flows and percentage of inspired volatile anesthetic, recorded at 1- or 5-minute intervals depending on the institution. These measurements and the cost of volatile anesthetic acquisition per milliliter were used along with a previously published formula17 to determine the volatile anesthetic cost per case. Other aspects of anesthetic costs, such as disposable equipment, intravenous fluid, blood products, and personnel costs, were not included in this analysis.

Cost data were separately reviewed at each institution through a manual validation process. To evaluate potential outliers, the top 100 most expensive and 100 least expensive cases were reviewed (total of 200 cases), and cases with suspected data entry errors were excluded from further analysis. These data entry errors were predominately related to selection of the appropriate drug dosage unit, such as documentation of the administration of 1000 g of vancomycin rather than 1000 mg. Supplemental Digital Content 1, Appendix 1, http://links.lww.com/AA/C162, lists a detailed description of the suspected data entry errors. ASA relative value scale units were quantified by determining the units associated with the procedure based on the anesthesia Current Procedural Terminology code and then imputing the time units based on the duration of the procedure. The base and time units were added together to determine the RVUs associated with each case.

In addition to the elements mentioned above, patient age, gender, performed surgical procedure by Common Procedure Terminology code, attending anesthesiologist, in-room provider, and ASA physical status classification were included in the data. These data sets were deidentified and an institution flag was added at each respective hospital. Finally, these data sets were combined and analyzed.

Back to Top | Article Outline

Statistical Analysis

Data analysis was performed using SPSS version 21.0 (IBM, Armonk, NY) and R version 3.3.2 (R Foundation for Statistical Computing, Vienna, Austria). We performed normality assessments using the Shapiro-Wilk analysis. Demographics were then analyzed. Age is expressed using means and standard deviation and compared between hospitals using a 2-tailed independent t test. The variables of ASA physical status and gender are expressed using counts and percentages and were compared between hospitals using χ2 analysis. The significance level was set at α = .05.

The variables were entered into 2 fixed-effects linear regression models, both with anesthetic drug cost as the dependent variable. The first model included duration, attending anesthesiologist, patient age, ASA physical status, and patient gender as the independent variables. We will refer to this as the “duration model.” The second model included case type, institution, patient age, ASA physical status, and patient gender as independent variables. We will refer to this as the “case type model.”

Redundancy analyses showed that when the variables of interest were included in 1 model, case type was highly correlated (R2 = 0.92) with the other variables. More specifically, a model that included case type was no better at predicting cost than a model without the variable when that model contained the combination of attending anesthesiologist and case duration. Therefore, because we were interested in evaluating the association between cost and both case type and duration, the variables were modeled separately—case type in 1 model and duration and attending anesthesiologist in another. Additionally, no attending anesthesiologists practiced at both institutions during the study period; therefore, the institution variable was not included in the duration model.

Moreover, attending anesthesiologist and in-room provider (typically a resident or nurse anesthetist) were initially both included in the analysis. However, redundancy analyses found that these variables were perfectly collinear, again indicating that no additional information was provided by including both variables independently in the regression model. We then considered including the variables in the analysis as attending/in-room provider “teams,” as cost may vary with certain combinations of attending anesthesiologists and in-room providers. However, there was not enough power to do so because of the large number of combinations of attendings and in-room providers. Therefore, to reduce redundancy and avoid overfitting the model (ie, including too many parameters), in-room anesthesiology provider was removed from the analysis. We chose to model attending anesthesiologist rather than in-room anesthesia provider because at both institutions the attending anesthesiologist prescribes the anesthetic plan, including the decision to use more expensive anesthetic agents.

Anesthetic drug cost was not normally distributed; therefore, regression models were developed both with and without logarithmic transformation of cost. Analysis of the residual plots revealed a better model fit after logarithmic transformation; thus, anesthetic drug cost was logarithmically transformed in the final models. For ease of interpretation, all predictor variables in the model were included as categorical. Case duration was divided into deciles, and patient age was grouped according to the United Nation’s designated age categories for health services usage.18

Because cost was log transformed, the average change in cost for each variable compared to the average cost of the reference category was calculated by first exponentiating the β coefficient and subtracting 1 to get the percent difference in cost. We then multiplied that value by the mean cost of the associated reference group. The reference groups for all variables except attending anesthesiologist were set at the lowest category for that particular variable. The reference group for attending anesthesiologists was the individual with the median value for mean cost per case. In this case, the actual mean value fell between 2 providers; therefore, the provider with the larger mean cost per case was chosen as the reference group. Mean values for each category of included variables are provided in Supplemental Digital Content 2, Appendix 2, http://links.lww.com/AA/C163. Reference values are indicated with an asterisk.

To test the explanatory value of the Anesthesiologists’ Relative Value Scale as a predictor variable, we constructed an ASA relative value scale unit model using attending anesthesiologist, RVUs as computed above, patient age, patient gender, and ASA physical status as explanatory variables and cost as the outcome variable. Like the previous models, cost was log transformed, and all explanatory variables were categorical. The aim of this part of the analysis was to determine whether economic value, represented by the ASA relative value scale, better predicted anesthetic expenses compared to duration and case type. The decision to perform this analysis was based on prior research performed by Dexter et al,16 which found that using a “cost per unit” scale, such as the one mentioned above, was more accurate than using “cost per case.” We compared the relative value scale model’s coefficient of determination (R2) with that of the duration and case type models’ R2R2 is a measure of how well the regression line approximates the data, where 0 represents no approximation at all and 1 represents a line that perfectly predicts the outcome. Higher R2 values indicate models that explain more variance.

Back to Top | Article Outline

RESULTS

A total of 5504 cases from 10 surgical procedures were identified; a detailed list is presented in Supplemental Digital Content 3, Appendix 3, http://links.lww.com/AA/C164. Of these, 52 cases were excluded due to missing ASA physical status classification, 37 were excluded due to suspected data entry errors, 2 surgical procedure types (aortoiliac bypass graft and prostatectomy, 5 cases and 549 cases, respectively) were excluded due to lack of corresponding cases at 1 institution, 3 cases were excluded due to missing procedure duration, and 2 were excluded due to incorrect procedure coding (Figure). After these exclusions, 4856 cases were included for analysis in this study.

Figure

Figure

Table 1

Table 1

Table 2

Table 2

The demographics were analyzed for each group of patients separately and combined (Table 1). The hospitals had similar demographics, except patients at hospital A were older (54.0 vs 42.1 years; P < .001). In addition, while hospital A represented a slightly larger portion of the study cohort compared to hospital B (53.2% vs 46.8%), the hospitals had a similar number of included attending anesthesiologists (128 vs 118). The median anesthetic drug cost was $38.05 (25th percentile: $24.06, 75th percentile: $61.41) for hospital A, $39.48 (25th percentile: $22.53, 75th percentile: $68.72) for hospital B, and $38.45 (25th percentile: $23.23, 75th percentile: $63.82) for both hospitals combined. The mean, median, and interquartile range for anesthetic case cost were also calculated for each surgical case type in each hospital and are presented in Table 2.

Back to Top | Article Outline

Duration Linear Regression Model

Table 3

Table 3

Parameter estimates from the model containing duration and attending anesthesiologist are presented in Table 3 along with the model’s adjusted R2, the average change in cost attributable to each variable, and the analysis of variance P values. The greatest variation in cost was due to attending anesthesiologists, followed by case duration. With all other variables held constant, the average change in cost due to attending anesthesiologists ranged from a decrease of $41.25 for the attending with the smallest mean cost per case to an increase of $95.67 for the attending with the largest mean cost per case when compared to the provider with the median value for mean cost per case (10th percentile = −$19.96, 90th percentile = +$20.20). Duration ranged from an increase of $6.44 with the second decile to an increase of $44.93 with the tenth decile when compared to the shortest duration period (≤64 minutes). ASA physical status increased in cost as severity increased and ranged from a cost increase of $3.06 with an ASA physical status of II to $14.95 with a status of IV and above. Gender was also significantly associated with cost; being male corresponded to a $4.58 decrease when compared to being female. None of the age categories was significantly associated with cost. The adjusted R2 (adjusted for the number of explanatory variables included in the model) was 0.352, indicating that our model explained 35.4% of the variance in anesthetic drug cost.

Back to Top | Article Outline

Case Type Linear Regression Model

Table 4

Table 4

Beta coefficients from the case type model are presented in Table 4, along with the model’s adjusted R2, the average change in cost attributable to each variable, and the analysis of variance P values. The greatest variation in cost in this model was attributable to case type. With all other variables held constant, the average change in cost ranged from a decrease of $9.09 for orthopedic surgery cases to an increase of $42.70 for cardiac cases when compared to general surgery cases. Institution was also significantly associated with cost; however, the effect size was small—the average cost per case at institution B was $5.73 higher compared to institution A. Unlike the duration model above, age was significantly associated with cost in this model; the average change in cost ranged from an increase of $6.10 for the 5- to 14-year age category to an increase of $18.21 for the 55- to 64-year age category when compared to the 1- to 4-year age category. ASA physical status of III was significant in the model and corresponded to a $7.06 average cost increase when compared to an ASA physical status of I. Also unlike the aforementioned duration model, gender was not significantly associated with cost in this model. This model explained 33.2% of the variance in anesthetic drug cost (R2 = 0.323).

Back to Top | Article Outline

ASA Relative Value Scale Unit Model

The R2 value for the anesthetic drug cost model with ASA relative value scale units replacing procedure type and case duration was 0.368, indicating that the model explained 36.8% of the variance in anesthetic drug cost. While this was slightly higher than the R2 values for the duration and case type models (R2 = 0.354 and 0.332, respectively), we did not estimate the statistical significance of the observed difference. However, we did not see strong evidence that the ASA relative value scale units did a better job at predicting cost than the duration and case type models.

Back to Top | Article Outline

DISCUSSION

The R2 value for both models indicates that there is still a noteworthy portion of unexplained variation. The factors responsible for variation we identified in our models were predominately attending anesthesiologist, procedure type, and case duration. We suspect the variation observed in the effect of the attending anesthesiologists is attributable to variation in practice patterns, more specifically the anesthetic choice. ASA physical status displayed a small amount of variation in the duration model, with higher acuity patients having an association with increased costs. However, this observation changed somewhat for the case type model, as a significant association with cost was only found for ASA physical status of III. The duration model also found age to be significant, with age exhibiting a bell-shaped relationship with cost. However, age did not have a significant association with cost in the case type model. One explanation for this may be a lower amount of power in the duration model, caused by the high number of attending anesthesiologists included in the model. Additionally, while institution was significantly associated with cost, we suspect the cost variation between institutions is due to differences in pharmacy contracts between the hospitals. Finally, the model utilizing ASA relative value scale units explained a slightly larger amount of variation when compared with the model that included procedure type and case duration. However, we do not believe there was sufficient evidence to conclude that this scale did a better job at predicting cost than the duration and case type models.

Prior described efforts at anesthetic drug cost reduction have had a variety of focal points, ranging from education aimed at an entire department,4 delayed focused feedback for individuals,2 and real-time feedback to the anesthesia provider.3 Oftentimes multiple intervention modalities are implemented. Anesthetic drug cost reduction can also be addressed at the institutional level, by either pharmacy negotiation strategies or utilization of drug cost-reduction programs such as the Medicare Disproportionate Share Hospital 340B Drug Pricing Program.19

A previous study done at Vanderbilt University Medical Center found estimated cost savings of $233,857.57–$359,411.85 USD annually after an intervention to decrease the utilization of cost-prohibitive drugs by substituting with appropriate alternatives.20 An additional study at the University of Washington found annual savings of $104,916 USD by using a real-time decision support tool.2 Finally, a study performed by Atcheson et al21 identified $185,250 USD in potential cost savings from preventable anesthetic drug waste. Thus, implementing an appropriate cost control strategy has the potential for large cost savings.

As stated above, the largest portion of the observed variation was not explained by the cost models. This suggests that there is significant variation introduced by either unmeasured factors or random variation on a per-case basis. The latter supports the clinical use of real-time feedback mechanisms to target cost reduction at the level of specific anesthetics. As procedure type and duration are not characteristics that can usually be adjusted, this reveals the importance of drug waste prevention efforts, as well as reducing access to costly agents that have suitable, more cost-efficient substitutes. The cost differences between attending anesthesiologists are the only directly modifiable sources of variation identified by our model. This variation by attending supports providing feedback to individual clinicians. However, while significant variation did exist, the actual monetary values are small in relation to the cost of the case. Thus, it is likely that the costs savings would be negligible and the efforts not worthwhile, especially at institutions such as ours where expensive drugs are more difficult for clinicians to routinely access. Hospitals should first evaluate the current practice patterns of their providers in relation to anesthetic drug cost before deciding to devote resources to targeting cost reduction in this area.

This study has several limitations. We examined only a subset of surgical procedure types, and thus our conclusions may not be generalizable to other surgical procedure types. Both of the institutions included are large academic medical centers, which may limit how applicable these conclusions are to other practice models such as private practices or institutions with different standard practices. Also, while the cost calculations for cases at the extremes of the distribution were manually reviewed and those with suspected data entry errors were identified and excluded, we did not perform a manual review of all cost calculations performed. Thus, we are unable to be sure that all data entry errors were identified. However, we believe that our methodology should have reliably identified those that significantly impacted the cost calculation.

Additionally, we were not able to include a detailed list of patient comorbidities in our models, as we did not have shared definitions of comorbidities that we could utilize across both institutions. While we captured patient comorbidities at a high level with ASA physical status, it is possible that a more detailed list would have explained more of the variation. Furthermore, we did not prespecify any interaction terms; therefore, none was included in the models. However, given the nature of the anesthetic practice patterns at both institutions, we did not believe any interaction terms were needed. Also, as can be seen in Table 2, there was a wide variability in the median costs between procedure types of the 2 hospitals. Because we were interested in determining if there were significant differences overall between institutions and in teasing apart differences at a procedural level, we no longer had drug-level data from this time period available to us at the time of the analysis. Therefore, we were not able to investigate the reason for this variation. Finally, the analysis did not include any calculation of drug waste. Instead anesthetic cost was calculated using the amount that was given to the patient, instead of the value that was dispensed. This could have potentially resulted in a calculated case cost that was lower than the actual case cost. However, it is not the behavior of clinicians at either institution to pull vials in excess of the anticipated need.

Back to Top | Article Outline

CONCLUSIONS

In summary, we have performed an analysis of real-world anesthetic-related drug costs and identified significant sources of variation. While the majority of the variation was not described by the models, possibly indicating high per-case random variation, significant sources of variation that were identified included attending anesthesiologist, procedure type, and case duration. The cost variation between institutions was significant but minor. Many prior studies have found significant savings resulting from cost-reducing interventions. However, our findings suggest that the savings resulting from interventions focused on the clinical practice of attending anesthesiologists may be negligible, especially in institutions where access to more expensive drugs is already limited. Therefore, we suggest that hospitals first evaluate the practice patterns of their providers in relation to anesthetic drug cost before deciding to devote resources to targeting cost reduction in this area.

Back to Top | Article Outline

ACKNOWLEDGMENTS

The authors would like to acknowledge the technical support and assistance of Michaelene Johnson, Database Analyst, Vanderbilt University Medical Center, in developing and validating the cost calculator at that institution. Additionally, we would like to thank Dr Franklin Dexter, MD, University of Iowa, for his valuable input on our statistical analysis.

Back to Top | Article Outline

DISCLOSURES

Name: Jonathan P. Wanderer, MD, MPhil.

Contribution: This author helped design the study, analyze and interpret the data, and draft and revise the manuscript.

Name: Sara E. Nelson, MPH.

Contribution: This author helped design the study, analyze and interpret the data, and draft and revise the manuscript.

Name: Douglas L. Hester, MD.

Contribution: This author helped design the study, analyze and interpret the data, with critical manuscript review, and approve the final version.

Name: Matthew Shotwell, PhD.

Contribution: This author helped design the study, analyze and interpret the data, with critical manuscript review, and approve the final version.

Name: Warren S. Sandberg, MD, PhD.

Contribution: This author helped with critical manuscript review and approval of the final version.

Name: John Anderson-Dam, MD.

Contribution: This author helped with critical manuscript review and approval of the final version.

Name: Douglas E. Raines, MD.

Contribution: This author helped with critical manuscript review and approval of the final version.

Name: Jesse M. Ehrenfeld, MD, MPH.

Contribution: This author helped design the study, analyze and interpret the data, with critical manuscript review, and approve the final version.

This manuscript was handled by: Nancy Borkowski, DBA, CPA, FACHE, FHFMA.

Back to Top | Article Outline

REFERENCES

1. Lubarsky DA, Sanderson IC, Gilbert WC, et al. Using an anesthesia information management system as a cost containment tool: description and validation. Anesthesiology. 1997;86:1161–1169.
2. Body SC, Fanikos J, DePeiro D, Philip JH, Segal BS. Individualized feedback of volatile agent use reduces fresh gas flow rate, but fails to favorably affect agent choice. Anesthesiology. 1999;90:1171–1175.
3. 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;118:874–884.
4. Wanderer JP, Ehrenfeld JM, Sandberg WS, Rathmell JP. Infographics in anesthesiology: a volatile tale of two institutions. Anesthesiology. 2013;118:1456.
5. Berman MF, Simon AE. The effect of a drug and supply cost feedback system on the use of intraoperative resources by anesthesiologists. Anesth Analg. 1998;86:510–515.
6. Malapero RJ, Gabriel RA, Gimlich R, et al. An anesthesia medication cost scorecard: concepts for individualized feedback. J Med Syst. 2015;39:48.
7. Lubarsky DA, Glass PS, Ginsberg B, et al. The successful implementation of pharmaceutical practice guidelines: analysis of associated outcomes and cost savings. SWiPE Group: systematic withdrawal of perioperative expenses. Anesthesiology. 1997;86:1145–1160.
8. Hanci V, Yurtlu S, Ayoğlu H, et al. Effect of low-flow anesthesia education on knowledge, attitude and behavior of the anesthesia team. Kaohsiung J Med Sci. 2010;26:415–421.
9. Ryu HG, Lee JH, Lee KK, et al. The effect of low fresh gas flow rate on sevoflurane consumption. Korean J Anesthesiol. 2011;60:75–77.
10. Tay S, Weinberg L, Peyton P, Story D, Briedis J. Financial and environmental costs of manual versus automated control of end-tidal gas concentrations. Anaesth Intensive Care. 2013;41:95–101.
11. Chernin EL. Pharmacoeconomics of inhaled anesthetic agents: considerations for the pharmacist. Am J Health Syst Pharm. 2004;61(Suppl 4):S18–S22.
12. Agoliati A, Dexter F, Lok J, et al. Meta-analysis of average and variability of time to extubation comparing isoflurane with desflurane or isoflurane with sevoflurane. Anesth Analg. 2010;110:1433–1439.
13. Fleischmann E, Akça O, Wallner T, et al. Onset time, recovery duration, and drug cost with four different methods of inducing general anesthesia. Anesth Analg. 1999;88:930–935.
14. Dolk A, Cannerfelt R, Anderson RE, Jakobsson J. Inhalation anaesthesia is cost-effective for ambulatory surgery: a clinical comparison with propofol during elective knee arthroscopy. Eur J Anaesthesiol. 2002;19:88–92.
15. Becker KE Jr, Carrithers J. Practical methods of cost containment in anesthesia and surgery. J Clin Anesth. 1994;6:388–399.
16. Dexter F, Lubarsky DA, Gilbert BC, Thompson C. A method to compare costs of drugs and supplies among anesthesia providers: a simple statistical method to reduce variations in cost due to variations in casemix. Anesthesiology. 1998;88:1350–1356.
17. Dion P. The cost of anaesthetic vapours. Can J Anaesth. 1992;39:633.
18. Department of International Economic and Social Affairs, Statistical Office. Provisional Guidelines on Standard International Age Classifications. Series M No 74. 1982.New York, NY: United Nations.
19. Health Resources and Services Administration. Drug Pricing Program Ceiling Price and Manufacturer Civil Monetary Penalties Regulation. Fed Regist. 2017;82:45511–45514.
20. Tabing AK, Ehrenfeld JM, Wanderer JP. Limiting the accessibility of cost-prohibitive drugs reduces overall anesthetic drug costs: a retrospective before and after analysis. Can J Anaesth. 2015;62:1045–1054.
21. Atcheson CL, Spivack J, Williams R, Bryson EO. Preventable drug waste among anesthesia providers: opportunities for efficiency. J Clin Anesth. 2016;30:24–32.

Supplemental Digital Content

Back to Top | Article Outline
© 2018 International Anesthesia Research Society