Abstract
Midazolam, a short-acting benzodiazepine used for preoperative anxiolysis may also have pharmacologic properties that could further reduce the incidence of postoperative nausea and vomiting in high risk patients when included in a multimodal antiemetic protocol. Despite the encouraging research on midazolam’s antiemetic properties, there remains concerns that the sedating properties of midazolam will delay discharge in short outpatient procedures. A retrospective data analysis (n=4,057) investigated the effect of midazolam on postoperative antiemetic administration and postoperative length of stay following outpatient procedures over a period of 15 months. Following an initial univariate analysis, a multivariable model adjusting for Apfel score, service, age, length of surgery and type of anesthesia was created to test these associations. The multivariable analysis demonstrated that midazolam was associated with a reduced need for postoperative antiemetic medications (3.2% lower in midazolam group, 95% CI 0.03%, 6.0%; p=0.032). Furthermore, the multivariable analysis demonstrated no clinically significant effect on postoperative length of stay (length of stay 7.9 minutes shorter in midazolam group, 95% CI −20, 4.4; p=0.2). We have observed potential secondary benefits in reducing the need for rescue antiemetic medications in addition to treating preoperative anxiety in patients undergoing outpatient cancer surgery. However, in patients for whom midazolam is not otherwise indicated, evidence is insufficient to warrant midazolam administration solely to prevent PONV. Randomized trials are needed to provide an accurate estimation of the effect size of midazolam for PONV in these patients.
Keywords: Midazolam, Benzodiazepine, Postoperative nausea and vomiting, outpatient, cancer, surgery
Introduction
Postoperative nausea and vomiting (PONV) continues to be a stressful consequence of outpatient surgery despite pharmacologic advances. Patients often find experiencing nausea and vomiting more distressing than postsurgical pain. PONV can also delay recovery and discharge in patients undergoing outpatient procedures.1–3 As a result of the aforementioned, significant research has been generated to establish appropriate pharmacological and nonpharmacological interventions to reduce the incidence of PONV.
Patients with a higher risk for PONV often require a combination or multimodal approach of 2 or more interventions for effective risk reduction.1,3 Thus, researchers have explored additional non-traditional anti-emetics, such as midazolam, that would aid in the multimodal prevention of postoperative nausea and vomiting.4,5
Midazolam is often administered in the perioperative period to reduce anxiety in addition to causing sedation and amnesia. The pharmacologic qualities allow for a rapid onset, short duration, and short half-life. The clinical effects of midazolam results from an agonist action on the GABAA receptor throughout the central nervous system. Benzodiazepines do not work directly on the GABA receptor, so there is a physiologic ceiling effect which contributes to their safety and low toxicity.6,7
Although the exact antiemetic mechanisms remain unknown, researchers postulate that midazolam works on the chemoreceptor trigger zone by reducing the synthesis, release, and postsynaptic dopamine.7 It remains debatable whether midazolam reduces dopamine directly or by blocking the re-uptake of adenosine leading to an adenosine-mediated reduction of dopamine release. Additionally, the binding of midazolam to the GABA benzodiazepine complex may cause dopaminergic neuronal activity and the release of 5-hydroxytryptamine. The reduction of PONV may also be a secondary effect of the anxiolysis properties of benzodiazepines.8.
Despite literature demonstrating the PONV benefits of midazolam in the perioperative period4,5, published consensus guidelines have not yet included midazolam as a standard intervention for PONV prevention.1 A small retrospective analysis concluded that midazolam had beneficial antiemetic properties in patients undergoing breast cancer related surgery, but only 48 of the patients studied underwent outpatient procedures.9 The study was limited to partial and complete mastectomies and did not include postoperative length of stay, which is a critical value in outpatient surgeries and a clinical concern for anesthesia providers. Also, the researchers did not investigate plastic reconstructive procedures or other types of outpatient cancer surgeries. Therefore, the current literature exhibits a gap in the effect of midazolam on PONV and length of stay in patients undergoing outpatient cancer surgeries. Our study focused on a large outpatient surgery center where the use of midazolam is high because of the anxiety associated with cancer surgery, enabling us to measure these associations.
The primary aim of this study was to determine whether patients undergoing cancer-related outpatient surgeries who received intraoperative midazolam had a decreased incidence of postoperative rescue antiemetic administration compared to those patients that did not receive midazolam. The secondary aim examined whether patients who received midazolam had longer time to discharge compared to patients who did not receive midazolam.
Materials and Methods
Following Institutional Review Board Approval, we conducted a retrospective chart review of a cohort of 4,417 patients who underwent a total of 4,954 outpatient surgical procedures at Memorial Sloan Kettering Cancer Center - Josie Robertson Surgery Center from January 2016 to March 2017.
Pre-specified exclusion criteria excluded patients with pre-existing benzodiazepine prescriptions (N=294), patients with two surgical procedures on the same day (N=35), preoperatively placed paravertebral nerve blocks (N=17), an ASA classification IV exhibiting significant co-morbidities (N=7), unexpected admissions to the hospital (N=6), or those undergoing only local anesthesia (N=1). Also excluded from the analysis were any subsequent secondary procedures performed on the same patient at a later date (N=537).
Patients undergoing general anesthesia received standard prophylactic antiemetics of dexamethasone, ondansetron, and all patients with an Apfel score of 4 received aprepitant. Patients undergoing Monitored Anesthesia Care (MAC) received prophylactic antiemetics based on clinician judgement. In the PACU, antiemetics were administered for treatment of nausea or vomiting based upon standard of care treatment protocol. Prophylactic antiemetics were not administered prior to opiate administration in the PACU. Data on intraoperative opioid use in MMEs was also collected.
The primary aim was to determine whether the intra-operative administration of midazolam was associated with a reduction in the use of post-operative nausea and vomiting (PONV) as indicated by use of antiemetic rescue medications in the post-anesthesia care unit. To test this hypothesis, logistic regression models were created for the outcome of rescue medication use in the treatment of PONV. A univariate logistic regression model assessed the association between midazolam and use of rescue medication for PONV. A multivariable logistic regression model analyzed whether midazolam reduced the use of PONV rescue medication after adjusting for factors that may influence PONV such as Apfel score, surgical service, age, length of surgery and type of anesthesia (general or monitored anesthesia care). Cubic splines were created for age and included in the multivariable model if significant non-linearity was identified. It was hypothesized that the effect of midazolam on the use of PONV rescue medication may differ depending on the length of surgery, as the effects of midazolam may have dissipated before the post-operative period for patients who have longer surgeries. To test this hypothesis, an interaction between use of midazolam and operative time was tested.
The secondary study aim was to determine whether the use of intra-operative midazolam increased post-operative length of stay. This was calculated as the time in hours from admission to the post-anesthesia care unit to the time of discharge. Linear regression models were created to test this association. The association was tested using both a univariate model and a multivariable model which was adjusted for Apfel score, service, age, length of surgery, surgery start time and type of anesthesia (general or monitored anesthesia care). Cubic splines were also included in this multivariable model if significant non-linearity was identified.
It was hypothesized that intra-operative midazolam use was lower in patients 65 and older. An analysis was conducted on whether the effect of midazolam on the need for post-operative rescue medications for PONV or length of stay differed based on patient age. An interaction term was added between intra-operative midazolam use and patient age in the multivariable models used for the outcomes of use of rescue medications for PONV and length of stay.
We did not include intraoperative opioid use or prophylactic anti-emetic use in our primary analysis, as opioid and anti-emetic use are standardized at our facility and highly correlated with anesthesia type. However, we performed a sensitivity analysis to confirm that these factors did not influence our results by repeating the analyses including intraoperative opioid use (in MMEs) and dexamethasone and ondansetron use (in mg) in the multivariable models. All analyses were performed using Stata 15 (StataCorp, College Station, TX).
Results
A total of 4,057 patients were included in our analysis after removing those meeting pre-specified exclusion criteria. Patient characteristics are presented in Table 1. Most patients received midazolam (76%), with the most common dose being 2mg (92%) and 6.0% of patients receiving <2 mg and 1.7% receiving >2mg. Patients under 65, patients receiving general anesthesia, and patients with ASA scores less than 3, longer operative times, and higher Apfel scores were more likely to receive midazolam. We first investigated whether midazolam administration affected the use of rescue medications for PONV. Most patients did not require any rescue PONV medications (88%). Among the 491 patients who received rescue medications for PONV, the majority (73%) required only one dose, while 23% required 2 doses and 4.2% required 3 or more doses.
Table 1.
Patient characteristics by use of midazolam (N=4057). Data are presented as number (%) or median (quartiles).
| No Midazolam (N=993) | Midazolam (N=3064) | p-value* | |
|---|---|---|---|
| Male | 53 (5.3%) | 88 (2.9%) | 0.0002 |
| Age (years) | 65 (51, 74) | 52 (44, 60) | <0.0001 |
| ASA score | <0.0001 | ||
| 1 | 32 (3.2%) | 137 (4.5%) | |
| 2 | 436 (44%) | 1817 (59%) | |
| 3 | 525 (53%) | 1110 (36%) | |
| Service | <0.0001 | ||
| Breast | 692 (70%) | 1545 (50%) | |
| Gastric | 1 (0.1%) | 8 (0.3%) | |
| Gynecology | 107 (11%) | 508 (17%) | |
| Head/Neck | 57 (5.7%) | 84 (2.7%) | |
| Plastics | 125 (13%) | 906 (30%) | |
| Urology | 11 (1.1%) | 13 (0.4%) | |
| Type of anesthesia | <0.0001 | ||
| General | 346 (35%) | 1641 (54%) | |
| MAC | 647 (65%) | 1423 (46%) | |
| Operative time, in minutes | 79 (62, 105) | 92 (67, 126) | <0.0001 |
| Apfel score | 0.0002 | ||
| 0 | 1 (0.1%) | 1 (<0.1%) | |
| 1 | 11 (1.1%) | 23 (0.8%) | |
| 2 | 107 (11%) | 277 (9.0%) | |
| 3 | 732 (74%) | 2128 (69%) | |
| 4 | 142 (14%) | 635 (21%) | |
| Intraoperative opioid use, in MMEs (N=4054) | 20 (10, 25) | 20 (15, 40) | <0.0001 |
p-values determined by chi-square test for categorical variables and Wilcoxon Rank Sum for continuous variables
On univariate analysis, the use of midazolam was associated with an increased use of PONV rescue medications (rate of PONV medication use 13% in midazolam group and 8.3% in no midazolam group, 5.1% higher in midazolam group, 95% CI 3.0%, 7.2%, p<0.0001). However, when controlling for Apfel score, service, age, length of surgery and type of anesthesia, midazolam was associated with a reduced need for rescue PONV medications (adjusted rate in midazolam group 11.6% (95% CI 10.7%, 12.6%) vs no midazolam group 14.8% (95% CI 12.1%, 17.4%), rate of PONV medication use was 3.2% lower in midazolam group, 95% CI 0.03%, 6.0%; p=0.032. In the sensitivity analysis including opioid use and preventative anti-emetic use, results were similar, with a 3.2% lower risk in the midazolam group (95% CI 0.04%, 6.1%, p=0.020).
Longer length of surgery was associated with increased use of rescue PONV medications on multivariable analysis (OR 1.13, 95% CI 1.11, 1.16; p<0.0001). There was some evidence that the use of rescue PONV medications also differed by service (p=0.066), with higher rates of rescue medication use in plastics than in breast surgeries (OR 1.43, 95% CI 0.98, 2.08). Thus, the increased use of PONV medications seen on univariate analysis was likely due to differences between services, increased midazolam use and increased PONV medication use among patients with longer anesthesia times. We then investigated whether midazolam affected use of rescue PONV medications differently depending on the length of surgery and found no evidence that this association differed based on operative time (p=0.6).
We investigated whether the use of midazolam was associated with an increased length of stay. For our cohort, the median length of stay was 2.6 hours (IQR 2.0, 3.6). On univariate analysis, there was a significant association between midazolam and an increase in postoperative length of stay (3.5 hours in midazolam and 3.1 hours in no midazolam group, with length of stay 22 minutes longer in midazolam group, 95% CI 8.6, 34; p=0.001). However, when controlling for Apfel score, surgical service, age, surgery start time and type of anesthesia, the association was not significant (3.4 hours in midazolam group (95% CI 3.3, 3.5) and 3.5 hours in no midazolam group (95% CI 3.3, 3.7), with length of stay 7.9 minutes shorter in midazolam group, 95% CI −20, 4.4; p=0.2). The confidence interval was narrow, with an upper bound of a 4.4 minute increase in length of stay, suggesting that we can exclude any clinically relevant increase in discharge time. Anesthesia time was longer for patients receiving midazolam (p<0.0001), and use of midazolam differed between services (p<0.0001). In the multivariable model, anesthesia time was also associated with an increase in length of stay (15 minute increase in length of stay per 10 minutes anesthesia time, 95% CI 14, 17; p<0.0001) and there was a significant difference in length of stay between services (p<0.0001). The differences between services, and the increased use of midazolam and longer length of stay among patients with longer surgeries is likely the reason for seeing an association between midazolam use and increased length of stay on univariate analysis. When including intraoperative opioid use and prophylactic anti-emetic use in our model as a sensitivity analysis, results were similar, with an 8.1 minute shorter length of stay in the midazolam group (95% CI −2, 4.2, p=0.2).
We then investigated whether the effect of midazolam differed based on patient age. We found no evidence that the effect of midazolam on either use of post-operative PONV rescue medications (p=0.6) or length of stay (p=0.2) differed by age.
Discussion
Our multivariable analysis controlling for Apfel score, service, age, length of surgery and type of anesthesia demonstrated that midazolam was associated with a significantly reduced need for rescue antiemetic medications in the PACU. The administration of midazolam produced no clinically significant effect on the length of post-anesthesia care unit stay in this large cohort of outpatients. Since both PONV and LOS are top clinical concerns in outpatient surgeries, our results are important in reducing anesthesia provider perceptions that midazolam results in longer post-anesthesia care unit stays in outpatients.
This retrospective analysis of ambulatory cancer-related surgical patients suggests the important role that intraoperative administration of midazolam may play towards decreasing PONV rates. The majority of the patients included in this retrospective analysis underwent breast and plastic reconstruction surgeries which are surgeries that are disproportionately represent women who are higher risk for PONV3.
These findings are consistent with the conclusions of a meta-analysis regarding the effects of midazolam on PONV rates4,5. The meta-analysis of 12 randomized controlled trials (N=841) concluded the administration of preoperative or intraoperative intravenous midazolam is associated with a decrease in overall nausea, vomiting, and rescue antiemetic use.4 Another meta-analysis reported a reduction in the incidence of PONV in the early, late, and overall recovery period.5 The results indicated that midazolam treatment can prevent nausea and vomiting in approximately 1 in 3 patients who would otherwise suffer from PONV if given a placebo.5 However, both studies pooled their results from a profoundly heterogeneous surgical population.
Our results are consistent with the small studies included in the meta-analysis that examined the impact of midazolam on PONV in ambulatory surgeries4,5. One study (N=70) reported a significantly lower incidence of antiemetic rescue medication administration in the midazolam group versus placebo in endoscopic outpatient procedures undergoing general anesthesia.10 Our study expanded on these results as our cohort was comprised of 51% monitored anesthesia care (MAC) cases. Furthermore, an additional study (N=88) reported a decreased frequency of postoperative nausea and an increased patient satisfaction compared to a placebo in a prospective study of patients undergoing a variety of intra-abdominal and peripheral ambulatory surgery.11 This study also only included a small number of patients undergoing a variety of procedures which makes drawing conclusions about the broader population difficult. Our study is consistent with the retrospective analysis of women undergoing breast surgery reporting of significantly higher PONV rates in outpatient breast cancer patients who did not receive midazolam.9 In this small study, only 48 of the 196 patients underwent outpatient procedures. A crucial consideration in all outpatient procedures is the length of postoperative stay and none of these studies addressed the impact of midazolam on this important variable.
The major limitation of the study is using rescue antiemetic administration as a surrogate for postoperative nausea and vomiting. Although this measure is consistent with the current literature in capturing the rates of PONV, the actual incidence may be higher. Furthermore, while the study examined the most common type of outpatient cancer surgeries, it did not include all types of outpatient cancer surgeries. The study also included a higher number of females than male patients, making it difficult to generalize the results to male patients. Receipt of chemotherapy was not available for inclusion as a covariate. However, receipt of chemotherapy at the time of surgery was rare, as chemotherapy in this cohort is typically given before or after surgery, and likely did not affect our results.
The retrospective study collected data over the period of 15 months which should address most of the issues related to natural fluctuations related to seasonal or diurnal variation but may not have addressed all possible variables. Due to the selection bias possible in a retrospective design, we cannot exclude the possibility of important differences in confounders influencing the relationship between midazolam and outcomes even after multivariable adjustment. However, these findings can be used to motivate a randomized clinical trial to study the effect of midazolam without the influence of selection bias.
Conclusion
Our study focused on a large outpatient cancer surgery center where the use of midazolam is high because of the anxiety associated with these procedures. In addition to treating preoperative anxiety in a highly anxious patient population, we have observed potential secondary benefits in reducing the need for post-operative rescue anti-emetics when controlling for Apfel score, service, age, length of surgery and type of anesthesia. Additionally, we demonstrated no significant increase in time to discharge after adjusting for these factors. In patients for whom midazolam is not otherwise indicated to treat anxiety, evidence is insufficient to warrant midazolam administration solely to prevent PONV; randomized trials are needed to provide a more accurate estimation of the effect size of midazolam. Our findings have important clinical implications as both postoperative nausea and vomiting and length of stay are top concerns in outpatient surgeries.
Acknowledgements:
This research was funded in part through the NIH/NCI Cancer Center Support Grant P30 CA008748 and the Kimmel Center for Prostate and Urologic Cancer. The authors would like to thank Gregory W. Fischer, MD and Andrew J. Vickers, PhD for their guidance in the development of this manuscript.
Contributor Information
Jennifer E. Ross, Memorial Sloan Kettering Cancer Center – Josie Robertson Surgical Center, New York, NY..
Emily Vertosick, Memorial Sloan Kettering Cancer Center in the Department of Epidemiology and Biostatistics, New York, NY..
Michael Long, Marian University in the Leighton School of Nursing, Indianapolis, IN..
Christian Cansino, Memorial Sloan Kettering Cancer Center and an Assistant Professor of Clinical Nursing at Columbia University in the Nurse Anesthesia Program, New York, NY..
Melissa Assel, Memorial Sloan Kettering Cancer Center in the Department of Epidemiology and Biostatistics, New York, NY..
Rebecca Twersky, Josie Robertson Surgical Center at Memorial Sloan Kettering Cancer Center, New York NY..
References
- 1.Gan TJ, Diemunsch P, Habib AS, et al. Consensus guidelines for the management of postoperative nausea and vomiting. Anesthesia & Analgesia. 2014;118(1):85–113. [DOI] [PubMed] [Google Scholar]
- 2.Koivuranta M, Laara E, Snare L, Alahuhta S. A survey of postoperative nausea and vomiting. Anaesthesia. 1997;52:443–449. [DOI] [PubMed] [Google Scholar]
- 3.Apfel CC, Korttila K, Abdalla M, et al. A factorial trial of six interventions for the prevention of postoperative nausea and vomiting. N Engl J Med. 2004;350:2441–51. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Grant MC, Kim J, Page AJ, Hobson D, Wick E, Wu CL. The effect of intravenous midazolam on postoperative nausea and vomiting. Anesthesia & Analgesia. 2016;122(3):656–663 [DOI] [PubMed] [Google Scholar]
- 5.Ahn EJ, Kang H, Choi GJ, Baek CW, Jung YH, Woo YC. The effectiveness of midazolam for preventing postoperative nausea and vomiting. Anesthesia & Analgesia. 2015;122(4):664–676. [DOI] [PubMed] [Google Scholar]
- 6.Nagelhout JJ, Elisha S. Nurse anesthesia. 6e ed. St. Louis, MO: Elsevier; 2017. [Google Scholar]
- 7.Stoelting RK, Hillier SC. Pharmacology & Physiology in Anesthetic Practice. Philadelphia: Lippincott Williams & Wilkins; 2006. [Google Scholar]
- 8.Rodola F. Midazolam as an anti-emetic. Eur Rev Med Pharmacol Sci. 2016;10:121–126. [PubMed] [Google Scholar]
- 9.Wilson S, Meyer H, Fecho K. Postoperative nausea and vomiting after inpatient and outpatient breast surgery: Incidence and effects of midazolam. Ambulatory Surgery. 2009;15(4):68–72. [Google Scholar]
- 10.Qadirullah M, Iqbal Z. Postoperative nausea and vomiting in ambulatory surgery: Comparison of midazolam with normal saline. J Med Sci. 2016; 24(3):149–154. [Google Scholar]
- 11.Bauer KP, Dom PM, Ramirez AM, O’Flaherty JE. Preoperative intravenous midazolam: Benefits beyond anxiolysis. J Clin Anesth. 2004;16(3):177–83. [DOI] [PubMed] [Google Scholar]