AJA Asian Journal of Anesthesiology

Advancing, Capability, Improving lives

Research Paper
Volume 54, Issue 4, Pages 114-120
HidekiMatsuura , Satoki Inoue , MasahikoKawaguchi
7937 Views


Abstract

Objectives

The current consensus guidelines for managing postoperative nausea and vomiting (PONV) suggest that one of anesthetic risk factors is the use of volatile anesthetics. However, in clinical settings, it is rare to perceive propofol to be superior to volatile anesthetics for the prevention of PONV. To assess whether PONV is related to the type of anesthetic delivered, we compared the incidence and duration of PONV between propofol anesthesia and sevoflurane anesthesia.

Methods

We performed a retrospective review of an institutional registry containing 21606 general anesthesia cases conducted following ethics board approval. Anesthesia for all patients was managed with propofol or sevoflurane. To avoid channeling bias, a propensity score analysis was used to generate a set of matched cases (propofol anesthesia) and controls (sevoflurane anesthesia), yielding 2554 matched patient pairs. The incidence and sustained rate of symptoms were compared as the primary outcomes.

Results

In the unmatched population, a higher incidence of PONV occurred following propofol anesthesia compared to sevoflurane anesthesia (propofol vs. sevoflurane anesthesia: 18.9% vs. 15.3%, respectively, p < 0.0001). The sustained rate of PONV over the course after propofol anesthesia was also higher than that following sevoflurane anesthesia (p < 0.001). Conversely, less PONV occurred after propofol compared to sevoflurane after propensity matching (propofol vs. sevoflurane anesthesia: 20.4% vs. 23.3%, respectively, p = 0.01). However, the sustained rate of PONV over the course after propofol anesthesia did not differ from that following sevoflurane anesthesia (p = 0.09).

Conclusions

Propofol could decrease the incidence of PONV compared with sevoflurane, although the duration of PONV was not affected as f

Keywords

PONV; Propofol; Sevoflurane;


1. Introduction

Despite impressive advances in the field of anesthesia, 25%–30% of patients continue to experience postoperative nausea and vomiting (PONV).1 Multiple factors, including the anesthetic agent delivered, are associated with an increased incidence of PONV. Moreover, the optimal strategy for preventing PONV continues to be debated. For example, the current consensus guidelines for managing PONV suggests that one of the anesthetic risk factors is the use of volatile anesthetics and recommends the use of propofol for the induction and maintenance of anesthesia, in addition to the avoidance of volatile anesthetics.2 However, these statements have not been revised since the former guidelines issued in 20033 and all new recommendations in both guidelines are based on the reports issued before the former guideline.4–6 This may not be inevitable because there have been few new reports that have directly compared the effects of propofol and volatile anesthetics on PONV since the establishment of the current guidelines. Thus, it appears to be difficult to reevaluate these effects using randomized control trials because of the statements in the guidelines, which are liable to believe to be established without doubt. Especially in Japan, an environment for aggressively conducting randomized control trials in this field does not exist and is only achieved by overcoming numerous difficulties. This implies that it is very difficult to gain an understanding of the citizens and dispel their doubts in the wake of scandals found in clinical trials.7 However, in clinical settings, it is rare to perceive propofol as superior to volatile anesthetics to prevent PONV. This is because propofol is frequently used for patients with risk factors for PONV while volatile anesthetics are used for the others.

Instead of randomized control trials, there is a growing interest in the use of propensity score-based methods in observational studies to estimate treatment effects. The propensity score is defined as the conditional probability of assigning a subject to a particular treatment protocol given a vector of measured covariates.8,9 In our institute, surgical patients managed by the anesthesia department undergo a structured postoperative interview with registered anesthesiologists at the postoperative anesthesia consultation clinic. At this time, the occurrence of any perioperative adverse events is assessed, and the patients can critique perioperative management based on the completion of an interview form. Using these interview data and several perioperative variables, we generated a propensity score for the probability of a patient being assigned to a particular anesthesia method (propofol vs. volatile anesthetics [sevoflurane]). With a propensity score matching method, we retrospectively investigated whether the incidence and duration of PONV were associated with the anesthetics delivered.

2. Materials and methods

Approval for the review of patient clinical charts, access to data of the institutional registry of anesthesia, and reporting of the results was obtained from the Nara Medical University Institutional Review Board, Kashihara, Nara, Japan (Chairperson Prof. N Kurumadani). The requirement for written informed consent was waived by the Institutional Review Board (No. 962 approved on March 19th, 2015).

2.1. Perioperative patient treatment

Patients were fasted for at least 10 h before the surgery but were allowed to drink clear fluids until 3 h before the surgery. No standardization was made for the methods of induction and maintenance of anesthesia. However, general anesthesia was usually induced with intravenous propofol (1–2.5 mg/kg) plus either fentanyl (1–2 μg/kg) or remifentanil (0.2–0.3 μg/kg/min), and neuromuscular blockade was achieved with rocuronium (0.6–0.9 mg/kg). Tracheal intubation was performed using a Macintosh-type laryngoscope. Anesthesia was maintained with sevoflurane (1.5%–2%) in a 40% oxygen and air mixture or with propofol (6–10 mg/kg/h). Nitrous oxide was not used. Fentanyl (1–2 μg/kg/h) or remifentanil (0.1–0.2 μg/kg/min) were used for analgesia. Rocuronium (0.2–0.3 mg/kg/h) was used for the neuromuscular blockade and sugammadex (2–4 mg/kg) (since August 2010) or neostigmine(40 μg/kg) plus atropine (20 μg/kg) until July 2010 was used to reverse the neuromuscular blockade after evaluating the status of the neuromuscular blockade by a nerve stimulator. Occasionally, postoperative analgesia was provided with intravenous fentanyl or epiduralropivacaine combined with fentanyl using a patient-controlled analgesia (PCA) device (Coopdech Syrinjector PCA Device™ for iv, Coopdech Balloonjector PCA Device™ for epidural, Daiken Medical Co. Ltd., Osaka City, Osaka, Japan). When PCA was used, a low dose droperidol (1.25–2.5 mg/day) was combined with a PCA device. PONV prophylaxisa, including a single dose intravenous steroid, 5-HT-3 blocker, low dose droperidol (except for cases with fentanyl-based PCA) or metoclopramide, was not used at the end of surgery. After the completion of anesthesia, the attendant in charge filled out the form for the institutional registry of anesthesia. This form includes the attendant's name, the name of the person who performed intubation, the patient's demographic variables, information regarding the final diagnosis and surgical procedures (later categorized into three classes based on the modified surgical risk stratification),10 medical history (e.g., hypertensiondiabetes mellituscoronary artery disease, history of heart failure, and lung disease), the duration of anesthesia and surgery, ASA physical status, urgency of surgery (emergency or elective), anesthesia technique (inhalational or intravenous with or without regional analgesia), 

2.2. Data handling

Data were collected between January 2009 and December 2013, during which there were 21,606 anesthesia cases. The exclusion criteria for the current study and the subsequent reduction in ineligible patients (initial → final) are as follows: the exclusion criteria for the current study (and reasons for the consequent reductions in ineligible patients) were as follows: 1) cases except for general anesthesia (n = 2,588); 2) cases missing answers on the postoperative questionnaire (n = 2,222); 3) < 15 years old (n = 1,543); 4) cases missing data sets (n = 1,579) (Fig. 1).

Fig. 1. Flow diagram for patient inclusion and exclusion criteria.
Download full-size image
Fig. 1. Flow diagram for patient inclusion and exclusion criteria.

2.3. Statistical analysis

Continuous variables are presented as mean and standard deviation (SD) if normally distributed, or as the median and interquartile range (IQR) if nonparametric. The categorical variables are presented as the number of patients and frequencies (%). Outcomes of the patients anesthetized by propofol or sevoflurane were compared first for PONV using the initial 13,674 patients. For the overall incident rate, a Fisher's exact test was used to estimate the relative risk and 95% confidence interval (CI) of incidence (propofol vs. sevoflurane as reference). PONV that had lasted for longer than a day was defined as a sustained symptom. Plots of the estimated proportion of patients with sustained PONV over time were constructed separately for the two groups (anesthetics delivered) by the Kaplan–Meier method. Patients with sustained PONV when visiting the postoperative anesthesia clinic were treated as censored cases. A log-rank test was used to assess the statistical significance of the treatment effects, and a Cox proportional hazard model was used to estimate the hazard ratio and 95% CI for propofol vs. sevoflurane.

Next, to minimize the effect of selection bias on the outcomes, we used propensity score matching to the clinical characteristics to reduce distortion caused by confounding factors. Using a propensity score analysis, we generated a set of matched cases (propofol) and controls (sevoflurane). As the results of the propensity score matching, 8,566 patients were excluded from the PONV analysis. A propensity score was generated for each patient from a multivariable logistic regression model based on the covariates using data from the institutional registry as independent variables, with the treatment type (propofol vs. sevoflurane) as a binary dependent variable. Factors reported to influence the incidence of PONV in the registered variables were included, which consisted of the preoperative physical condition, gender, age, type of surgery (e.g., gynecological, ear–nose–throat, neurosurgical, breast, and ophthalmic surgery), duration of anesthesia, and postoperative analgesia, including opioids.2,3 The prophylactic use of antiemetics was not monitored. However, other factors may have varied and were included as potential confounders. As suggested by a review of statistical research on propensity score development, we used a structured, iterative approach to refine this model with the goal of achieving a covariate balance between the matched pairs.11 The covariate balance was measured using the standardized difference, in which an absolute difference of >0.1 was taken as a meaningful covariate imbalance.12 We matched the patients using a greedy-matching algorithm with a caliper width of 0.001 of the estimated propensity score. A matching ratio of 1:1 was used. This procedure yielded 2,554 patients anesthetized by propofol propensity matched to 2,554 patients anesthetized by sevoflurane. For statistical inference, methods that accounted for the matched nature of the samples were used. For the overall incident rate, the Cochran–Mantel–Haenszel test, stratified on the matched pair, was used to estimate the relative risk and the 95% CI of incidence (propofol vs. sevoflurane as a reference). Plots of the estimated proportion of patients with PONV over time were constructed using the Kaplan–Meier method separately for the two groups. The censored cases were treated similar to an unmatched case analysis. A stratified (by matched pair) log-rank test was used to assess the statistical significance of the treatment effects, and the stratified (by matched pair) Cox proportional hazard model was used to estimate the hazard ratio and 95% CI for propofol vs. sevoflurane.

2.4. Sample size calculation

For the purpose of calculating the sample-size, we assumed a 40% incidence of PONV based on previous reports (30%–50%).2,3 We estimated that 2,474 patients in each group were required to provide 95% power to detect a 5% difference in the incidence of PONV (with an overall incidence of 40%) between propofol and sevoflurane anesthesia with a type I error probability of 0.05. Thus, our sample size was sufficient to detect a difference in the outcome. Analyses were computed using R (version 3.0.3, R Foundation for Statistical Computing, Vienna, Austria). A p < 0.05 was considered statistically significant.

3. Results

The clinical characteristics of the two groups (i.e., patients anesthetized by propofol and patients anesthetized by sevoflurane) based on the 13,674 patients with PONV are presented in Table 1. Only one-third of the variables were similar between the groups (standardized difference < 0.1) before matching. However, variables including gender, age, height, weight, BMI, ASA physical status, concomitant use of regional analgesia, and the rate of coexisting disease, the rate of emergency case, the rate of neurosurgical, gynecological, ear–nose–throat, breast, or ophthalmic procedures, the rate of postoperative ICU admission, and the rate of postoperative analgesia were imbalanced; most of these factors were previously reported factors influencing the incidence of PONV. Patient outcomes are summarized in Table 2 and Fig. 2. PONV occurred more frequently following propofol compared to sevoflurane anesthesia (propofol vs. sevoflurane anesthesia: 18.9% vs. 15.3%, respectively). The relative risk (95% CIs) for propofol anesthesia was 1.36 (1.24–1.48) for PONV (p < 0.0001). Incidences of PONV stratified by age and sex are also presented in Table 2. The rate of sustained PONV following propofol anesthesia was higher than that after sevoflurane anesthesia (The log-rank p < 0.001). The number of censored PONV cases was 135 (0.99%). The rate of sustained PONV on the 12th day post-anesthesia was very low (0.00037% for propofol anesthesia and 0.00009% for sevoflurane). A hazard ratio and 95% CI for propofol anesthesia was 0.93 (0.89–0.97) for PONV.

Table 1. Clinical characteristics of the two unmatched study groups.
Download full-size image
BMI; Body mass index, ASA; American society of anesthesiologists, Neuro-Gyne-Ear-Opth-Brest; Surgical procedures by department of Neurosurgery, Gynecology, Ear, nose and throat, Ophthalmology, or Brest surgery, Emergency; emergency case, ICU; Intensive care unit.
Table 2. Patient outcome prior to matching.
Download full-size image
PONV; Postoperative nausea and vomiting. It is presented as the relative risk on incidence of PONV for propofol against sevoflurane.
Fig. 2.
Download full-size image
Fig. 2. Curves for the rate of sustained PONV before matching. The solid line indicates propofol anesthesia. The broken line denotes sevoflurane anesthesia. Tick marks highlight the censored cases.

The clinical characteristics of the two matched groups extracted by a propensity analysis are presented in Table 3. The covariates were well balanced after matching. Notably, imbalanced variables in the unmatched populations were balanced after matching. For the items of supine position and surgical risk stratification, the P values became considerably smaller. However, this was due to the large sample size population. Thus, the effect sizes(standardized differences) remained sufficiently small, indicating that they were well balanced. The patient outcomes are summarized in Table 4 and Fig. 3. In contrast, PONV occurred less frequently after propofol compared to sevoflurane anesthesia after the propensity matching (propofol vs. sevoflurane anesthesia: 20.4% vs. 23.3%, respectively). The relative risk and 95% CIs for propofol anesthesia were 0.88 (0.79–0.97) for PONV (p = 0.01). The incidences of PONV stratified by age and sex are also presented in Table 4. However, it was found that propofol anesthesia did not decrease the rate of sustained PONV, and a Kaplan–Meier curve for the sustained rate of PONV was similar (The log-rank p = 0.09). The number of censored cases was 53 (1.04%) for PONV and similarly, the rate of sustained PONVs on the 12th day was very low (0.00039% vs. 0%). A hazard ratio and 95% CIs for propofol anesthesia was 1.03 (0.97–1.09) for PONV.

Table 3.
Download full-size image
BMI; Body mass index, ASA; American society of anesthesiologists, Neuro-Gyne-Ear-Opth-Brest; Surgical procedures by department of Neurosurgery, Gynecology, Ear, nose and throat, Ophthalmology, or Brest surgery, Emergency; emergency case, ICU; Intensive ca
Table 4.
Download full-size image
PONV; Postoperative nausea and vomiting. It is presented as the relative risk on incidence of PONV for propofol against sevoflurane.
Fig. 3.
Download full-size image
Fig. 3. Curves for the rate of sustained PONV after matching. The solid line indicates propofol anesthesia. The broken line denotes sevoflurane anesthesia. Tick marks highlight the censored cases.

4. Discussion

In the present study, we investigated whether propofol anesthesia could decrease the incidence and duration of PONV compared with sevoflurane anesthesia. Based on raw patient data, propofol anesthesia increased both the incidence and duration of PONV compared with sevoflurane anesthesia. Conversely, based on this propensity score analysis of the patient data, propofol anesthesia was found to decrease the incidence of PONV, although it did not change the duration of PONV.

We frequently encounter the patients suffering from PONV despite the administration of propofol anesthesia according to the guidelines.2,3 Moreover, it appears that the more frequently propofol is used, there is a greater incidence in PONV. Therefore, we consider there to be a discrepancy between our perception and the established guidelines. In the clinical setting, propofol has a greater opportunity to be used for patients with risk factors for PONV according to the guideline's recommendations.2,3 This may explain why the raw data analysis in this study found an increasing incidence and duration of PONV compared with sevoflurane. In the comparison between the two anesthetic methods using the raw data, coexisting PONV risk factors (e.g., gender, age, postoperative analgesia including opioid, and type of surgery) were very imbalanced, and there were a greater number of risk factors observed in the patients that received propofol anesthesia. This imbalance likely affected the patient outcome, which may explain the strange phenomenon that we frequently encounter the patients suffering from PONV, despite receiving propofol anesthesia.

To reduce the risk of bias owing to potential confounders in the dataset, we conducted a propensity score analysis to verify the superiority of propofol over sevoflurane regarding the incidence and duration of PONV. The analysis consisted of a large cohort of 5,108 matched patients, a sample size that could detect a 5% difference in the incidence of PONV between propofol anesthesia and sevoflurane. Covariate balances were well-maintained. With well-balanced matched data, it was observed that propofol anesthesia could decrease the incidence of PONV; however, propofol did not affect the duration of PONV. It has been reported that volatile anesthetics may not be the primary cause of delayed postoperative vomiting.13 Apfel et al. divided the first postoperative 24 h into two periods: 1) the early (0–2 h); and 2) the delayed (2–24 h) periods. However, we recorded the duration of PONV daily, but not hourly. A systematic review and meta-analysis also reported that propofol did not reduce post-discharge nausea and vomiting following ambulatory surgery.14 Therefore, it appears that the anesthetic delivered did not affect the duration of PONV on the daily basis. This may be explained by the pharmacokinetic profile propofol, in which therapeutic anti-emetic plasma levels are unlikely to persist several hours or days after the administration of the anesthesia since it has a short half-life.14

There are several limitations of the present study that merit discussion. To minimize the effect of selection bias on the outcomes, we used propensity score matching to the clinical characteristics of the patients to reduce any distortion caused by confounding factors. However, this study was retrospective in nature; thus, unmeasured variables could still confound the results. We used data from the institutional registry of anesthesia, which includes only minimally essential information about each case but does not include precise details. Therefore, we did not obtain several variables which might have affected PONV. For instance, important variables (e.g., the usage of neostigmine, smoking status, and history of PONV or motion sickness) which are included in the Apfel score, are missing.15 First, we had employed neostigmine equally in both groups during a certain early period of the study before sugammadex was available in daily clinical practice. Considering that the majority of patients should have been prohibited from smoking due to the elective surgery and that most had no history of operation (no history of PONV), we can conclude that this variable had a negligible impact on the overall population. Regarding the history of motion sickness, we cannot comment on this issue because the patients had never been asked about motion sickness during a medical interview. However, the selection bias based on motion sickness is thought to have been very small because there is a lack of available information regarding this issue. Collectively, our anesthesia practices were relatively constant during the sampling period, and therefore, the effects of unmeasured variables were likely minimal. In addition, we did not identify a precise follow-up period; thus, patients who continued to complain of PONV were censored when they visited the postoperative anesthesia consultation clinic. The majority of patients were symptom-free when they filled out the interview form, even if they had suffered from symptoms. It has been suggested that plausible assumptions regarding the outcomes of patients lost to follow-up could change the interpretation of results even when the percentage lost is near the median (∼6%).16 Therefore, it is reasonable to think that the censored cases had not had a significant impact on our study results while considering that approximately 1% of the patients were lost to follow-up as censored data. Additionally, this study relied on patient self-reports to determine the duration of symptoms, which were based on memory. Furthermore, this self-reporting system might have caused the relatively small difference in the overall incident rate of PONV between propofol and sevoflurane after propensity score matching. Finally, a considerable number of patients were excluded from the study.

In conclusion, both the actual incidence and duration of PONV were more evident in patients that received propofol compared to sevoflurane anesthesia in an ordinary clinical setting. These findings may be matched with our perception; however, it was observed that propofol could decrease the incidence of PONV compared with sevoflurane, although it did not affect the duration of PONV as noted by previous reports.

Funding/Support statement

This study was supported only by departmental source. The manuscript was edited by an English editing service (ENAGO).

Conflicts of interest

The authors declare no conflicts of interest.

Acknowledgement

None.


References

1
J. Lerman
Surgical and patient factors involved in postoperative nausea and vomiting
Br J Anaesth, 69 (1992), pp. S24-S32
2
T.J. Gan, P. Diemunsch, A.S. Habib, A. Kovac, P. Kranke, T.A. Meyer, et al.
Consensus guidelines for the management of postoperative nausea and vomiting
Anesth Analg, 118 (2014), pp. 85-113
3
T.J. Gan, T. Meyer, C.C. Apfel, F. Chung, P.J. Davis, S. Eubanks, et al.
Consensus guidelines for managing postoperative nausea and vomiting
Anesth Analg, 97 (2003), pp. 62-71
4
R. Sukhani, J. Vazquez, A.L. Pappas, K. Frey, M. Aasen, S. Slogoff
Recovery after propofol with and without intraoperative fentanyl in patients undergoing ambulatory gynecologic laparoscopy
Anesth Analg, 83 (1996), pp. 975-981
5
J.J. Wang, S.T. Ho, J.I. Tzeng, C.S. Tang
The effect of timing of dexamethasone administration on its efficacy as a prophylactic antiemetic for postoperative nausea and vomiting
Anesth Analg, 91 (2000), pp. 136-139
6
K. Visser, E.A. Hassink, G.J. Bonsel, J. Moen, C.J. Kalkman
Randomized controlled trial of total intravenous anesthesia with propofol versus inhalation anesthesia with isoflurane-nitrous oxide: postoperative nausea with vomiting and economic analysis
Anesthesiology, 95 (2001), pp. 616-626
7
K. Hirota
Retraction note: notice of formal retraction of articles by Dr. Yoshitaka Fujii
J Anesth, 27 (2013), p. 322
8
R.B. D’Agostino Jr.
Propensity score methods for bias reduction in the comparison of a treatment to a non-randomized control group
Stat Med, 17 (1998), pp. 2265-2281
9
D.B. Rubin
Estimating causal effects from large data sets using propensity scores
Ann Intern Med, 127 (1997), pp. 757-763
10
K.A. Eagle, P.B. Berger, H. Calkins, B.R. Chaitman, G.A. Ewy, K.E. Fleischmann, et al.
ACC/AHA guideline update for perioperative cardiovascular evaluation for noncardiac surgery—executive summary a report of the American College of Cardiology/American Heart Association Task Force on Pr
Circulation, 105 (2002), pp. 1257-1267
11
P.C. Austin
Propensity-score matching in the cardiovascular surgery literature from 2004 to 2006: a systematic review and suggestions for improvement
J Thorac Cardiovasc Surg, 134 (2007), pp. 1128-1135
12
P.C. Austin, P. Grootendorst, G.M. Anderson
A comparison of the ability of different propensity score models to balance measured variables between treated and untreated subjects: a Monte Carlo study
Stat Med, 26 (2007), pp. 734-753
13
C.C. Apfel, P. Kranke, M.H. Katz, C. Goepfert, T. Papenfuss, S. Rauch, et al.
Volatile anaesthetics may be the main cause of early but not delayed postoperative vomiting: a randomized controlled trial of factorial design
Br J Anaesth, 88 (2002), pp. 659-668
14
G. Kumar, C. Stendall, R. Mistry, C. Goepfert, T. Papenfuss, S. Rauch, et al.
A comparison of total intravenous anaesthesia using propofol with sevoflurane or desflurane in ambulatory surgery: systematic review and meta-analysis
Anaesthesia, 69 (2014), pp. 1138-1150
15
C.C. Apfel, B.K. Philip, O.S. Cakmakkaya, A. Shilling, Y.Y. Shi, J.B. Leslie, et al.
Who is at risk for postdischarge nausea and vomiting after ambulatory surgery?
Anesthesiology, 117 (2012), pp. 475-486
16
E.A. Akl, M. Briel, J.J. You, X. Sun, B.C. Johnston, J.W. Busse, et al.
Potential impact on estimated treatment effects of information lost to follow-up in randomised controlled trials (LOST-IT): systematic review
BMJ, 344 (2012), p. e2809

References

Close