Secondary Logo

Journal Logo

Postoperative pain

Association between intra-operative fentanyl dosing and postoperative nausea/vomiting and pain

A prospective cohort study

Mauermann, Eckhard*; Clamer, Damian*; Ruppen, Wilhelm; Bandschapp, Oliver

Author Information
European Journal of Anaesthesiology: November 2019 - Volume 36 - Issue 11 - p 871-880
doi: 10.1097/EJA.0000000000001081

Abstract

Introduction

Postoperative nausea and/or vomiting (PONV) is one of the main side-effects of general anaesthesia and ranks among patients’ top concerns, higher than the fear of surgical pain or postoperative recall without pain.1 Furthermore, PONV reduces patient satisfaction,2 increases time-to-discharge from the postoperative anaesthesia care unit (PACU) or hospital,3 and may even be present after discharge from hospital, leading to rehospitalisation.4

To identify risk factors for PONV and guide patient management, a number of scores, generally based on procedure-related risks or anaesthesia-independent risk factors, have been established.5–7 The most common of these, the simplified Apfel score, sums four risk factors, namely female sex, history of motion sickness or PONV, nonsmoking status and planned/expected postoperative opioid treatment6,8–11: scores of 2, 3 and 4 yield a 39, 61 and 79% risk of PONV, respectively.6 However, this score was developed 20 years ago in a population receiving volatile anaesthesia and no antiemetic prophylaxis. Therefore, today, the Apfel score may not be representative of PONV since the latter is often pre-empted by the use of total intravenous anaesthesia (TIVA) with propofol and prophylactic antiemetic agents. Furthermore, the role of intra-operative opioids on PONV in a contemporary anaesthesia is largely unexplored. This is important because opioids may lead to late rather than early PONV (e.g. from volatile anaesthetics),9,12 intra-operative opioid dosing is modifiable at an early stage, and contemporary anaesthesia regimens have been developed to pre-empt PONV.

The main objective of this prospective cohort study was to examine intra-operative opioid administration and its association both with the incidence of PONV and with postoperative pain scores in patients at risk of PONV in association with a contemporary anaesthetic regimen. Our primary hypothesis was that increased intra-operative fentanyl dosing would be associated independently with an increased incidence of PONV during the first 24 postoperative hours. Specifically, we examined the prognostic value of intra-operative fentanyl dosing in addition to the simplified Apfel score. We also explored whether or not postoperative pain scores differed when taking the dose of intra-operative fentanyl into account.

Methods

Study design, setting and participants

In this single-centre, prospective cohort study we examined consecutive adult inpatients with a simplified Apfel score6 at least 2 undergoing abdominal, gynaecological or otorhinolaryngological surgery from February 2017 to May 2017. Patients with chronic pain (defined as recurring pain requiring intermittent hospitalisation or regular intake of pain medication) or patients not naive to opioids (defined as a history of abuse or having taken opioids within the last 30 days) were not eligible. The local ethics committee granted an ethics waiver (Ethikkommission Nordwest- und Zentralschweiz, Prof Perruchoud, 18 October 2016). For this study, a designated anaesthesia resident (DC) performed the patient follow-ups after 24 h and explained the study. As such, neither patients nor their anaesthetists were aware of the study during treatment and data collection. This article adheres to STROBE guidelines13 and was preregistered on ClinicalTrials.gov (NCT03201315, Mauermann, 14 February 2017).

Endpoints and measurements

The primary endpoint was the occurrence of PONV (nausea and/or vomiting) during the first 24 postoperative hours. Secondary endpoints were the frequency of vomiting within the first 24 postoperative hours (0, 1 to 2, ≥3 times), any PONV in the recovery room (nausea and/or vomiting), the highest pain score on the numeric rating scale (NRS, 11-point scale: 0 no pain, 10 worst imaginable pain) in the recovery room as well as within the first 24 postoperative hours.

Data were obtained by a semistructured interview (both focused and open questions) during the standard postoperative anaesthesia visit 24 h after surgery, as well as by extraction from hospital charts.

General considerations

As a blinded observational study, clinicians’ decisions were not influenced by the study. In other words, there was no protocol for clinical decision-making or drug dosing. In general, however, patients at our institution receive 2 to 3 μg kg−1 of fentanyl, 2 mg kg−1 of propofol and 0.5 mg kg−1 of either rocuronium or atracurium at induction. Maintenance levels of anaesthetic drugs (either propofol or sevoflurane) are guided by clinical signs and bispectral index monitoring.

At our institution, decisions related to drugs for anaesthesia maintenance and antiemetic prophylaxis are at the discretion of the attending anaesthesiologist. Clinicians tailor pre-emptive medication to perceived patient needs and prescribe postoperative agents on an as-needed rather than a fixed scheme basis.

Analgesia is multimodal, generally being comprised of two nonopioid analgesics (ibuprofen, metamizole, ketorolac, paracetamol) and co-analgesics (lidocaine, magnesium, ketamine, clonidine). The main opioid used for intra-operative analgesia is fentanyl, given throughout surgery as needed. However, remifentanil is often co-administered at low levels to achieve an effect site concentration of 1 to 2 ng ml−1 throughout surgery, with the level being increased near the end of surgery to enable early propofol cessation thereby facilitating a timely tracheal extubation. In the event that the anaesthetist considers postoperative morphine to be necessary, a dose of morphine (usually up to 0.1 mg kg−1) may be given 30 min before the end of anaesthesia. Additional postoperative morphine was administered intravenously in the recovery room and orally on the general ward. We used our hospital's opioid conversion factor of 0.3 (i.e. 10 mg oral morphine = 3 mg intravenous morphine)14,15 to obtain total morphine equivalents.

Bias and confounding

To avoid a Hawthorne effect, neither patients nor care-givers were informed about the study until after its completion. Patients were not informed until the 24-h visit. Patients explicitly stating that they did not want their data used were excluded. Study personnel were at no time involved in individual patient care. In an effort to reduce observation bias, all postoperative anaesthesia visits were performed by one researcher asking the same questions in the same order. Data synthesis and analysis were conducted by another researcher. Confounding was addressed in the design by adjusting for known risk factors, embodied by the simplified Apfel score and the exclusion of nonopioid naive patients.

Statistical analysis

Categorical data are presented as absolute numbers (percentage). Continuous data are presented as mean ± SD or median [IQR]. For patient characteristics, potential differences in patients suffering PONV and not suffering PONV as well as patients vomiting and not vomiting were examined by Pearson's χ2 test, Fisher's exact test, Student's t test or Mann–Whitney U test, as appropriate.

Density plots of intra-operative fentanyl administration were plotted and arbitrarily divided into tertiles. For each tertile, PONV and pain scores (as measured by the NRS) were described and possible differences examined by a Kruskal–Wallis test. In the event of statistical significance, explorative post hoc tests were performed with raw P values reported, as recommended.16

The primary outcome (any PONV within the first 24 postoperative hours) was first examined by univariable logistic regression using predefined variables. Subsequently, three multivariable logistic regression models were employed based on the availability of information during a patient's clinical course: Model 1, which included only the simplified Apfel score and intra-operative fentanyl dose; Model 2, which additionally included antiemetic prophylaxis; Model 3, which additionally included antiemetic prophylaxis and postoperative morphine. All three models were run once with the intra-operative fentanyl term in absolute dose (i.e. μg) and once as dose per hour (i.e. in μg h−1). The Akaike information criterion was used as a parameter of model fit. Models 1 and 2 were also examined for PONV in the PACU. To explore the relationship between expected postoperative morphine, intra-operative fentanyl and actually administered postoperative morphine, we examined an interaction term in Model 3, as well as a crude correlation between pairs of terms. In addition, we examined the incremental effect of fentanyl when added to the Apfel score, excluding planned postoperative morphine. Finally, a subgroup analysis of the primary outcome (Model 1) was performed for general anaesthesia only.

Additional analyses were the area under the receiver operator characteristics (ROC) curve for the simplified Apfel score alone as well as each of the three multivariable models. Differences in the area under the ROC curves were assessed by the DeLong test.

We also examined a possible correlation of intra-operative fentanyl dose and pain in both the PACU and within the first 24 h by linear regression. In patients receiving postoperative morphine, we also correlated morphine equivalents within the first 24 h with intra-operative fentanyl administration.

We considered statistical significance to be a two-tailed P value 0.05 or less. All statistical analyses were conducted with R 3.4.2 (The R Foundation for Statistical Computing, Vienna, Austria). Sample size estimation was based on the rule of thumb that 10 to 12 events are required per variable to be examined.17 Assuming a PONV rate of 30 to 40%8 and 11 estimators, 360 patients would need to be recruited.

Results

Descriptive analysis

A total of 686 procedures were performed in abdominal, gynaecological or otorhinolaryngological surgery from February 2017 to May 2017, with 484 patients eligible for inclusion (Fig. 1). Of these, 60 (12%) were not available for postoperative questioning at 24 h, 36 (7%) were unable to answer questions or were otherwise excluded and 20 (4%) requested that their data not be used. Thus, data from 363 patients were analysed.

F1-11
Fig. 1:
Flow chart illustrating potentially eligible patients and patient inclusion. PONV, postoperative nausea and/or vomiting.

Within the first 24 postoperative hours, 163 (45%) patients suffered PONV. Half of these suffered nausea at some time without vomiting (n = 82), 34% vomited one to two times (n = 55) and 16% vomited three times or more (n = 26). During their stay in the PACU, only 10% of patients suffered PONV (31 patients had nausea, five also vomited).

Patient characteristics stratified by both PONV and vomiting within the first 24 h are shown in Table 1. In brief, 166 (46%), 151 (42%) and 46 (13%) patients had Apfel scores of 2, 3 and 4, respectively. TIVA was used for maintenance in 297 (82%) patients. In terms of pre-emptive antiemetic prophylaxis, 126 patients (35%) received no antiemetic prophylaxis, 101 (28%) received one pre-emptive antiemetic agent, 106 (29%) received two agents and 30 (8%) received three agents. The 126 patients receiving no antiemetic therapy (90, TIVA and 36, volatile anaesthesia) were clearly undertreated according to current PONV guidelines.7 Of the 403 doses of antiemetics administered pre-emptively, 209 (52%) received dexamethasone, 121 (30%) received droperidol, 69 (17%) received ondansetron, with four (1%) receiving other agents.

T1-11
Table 1:
Patient characteristics and surgical and postoperative data

In terms of intra-operative opioid administration, the 82% of patients receiving TIVA also received continuous remifentanil, as is common practice in our institution, and 15% received intra-operative morphine. Patients suffering PONV had a higher hourly intra-operative fentanyl dose and, if they received morphine intra-operatively or on the general ward, they received a higher total morphine dose. Remifentanil doses were similar between patients suffering PONV and those not suffering PONV. In addition, pain scores tended to be higher and were more likely to be severe in both the PACU and on the general ward in the first 24 h in patients suffering PONV than in those not suffering PONV. In contrast to patients with PONV, those actually vomiting appear to be a distinct subset. Pain scores at 24 h were higher and more likely to be severe in the group vomiting postoperatively.

Figure 2 relates the relative frequency distribution of intra-operative fentanyl in μg h−1 (top panel), the incidence of PONV (and its relative standard error) both in PACU and during the first 24 h (middle panel) and the highest pain score (median and IQR) both in PACU and during the first 24 h (bottom panel). Differences existed in global PONV rates for the low, middle and high fentanyl doses (P = 0.022) with post hoc tests suggesting differences between the low dose and both the middle and high doses (P = 0.027 and 0.019, respectively). Supplemental Fig. S1, https://links.lww.com/EJA/A221 shows an analogous analysis by total intra-operative fentanyl in μg, which additionally showed increased pain scores during the first 24 h in the highest two tertiles compared with the lowest tertile.

F2-11
Fig. 2:
Density plot of intra-operative mean hourly fentanyl doses, postoperative nausea and vomiting and postoperative pain scores. Incidence of postoperative nausea and/or vomiting is given as a crude rate and the relative standard error; highest pain scores are given as median [IQR]. Dashed lines represent tertiles for the fentanyl dose.

Intra-operative fentanyl and postoperative nausea and/or vomiting

Table 2 shows the univariable and multivariable regression results for the occurrence of PONV during the first 24 h. The multivariable models showed nominally increasing model fit with increasing variables, albeit at later times of a patient's hospital stay. For all three multivariable models the intra-operative dose of fentanyl per hour was associated with an increased risk of PONV, while the absolute dose was not. Antiemetic prophylaxis also significantly reduced PONV, whereas postoperative morphine was NS (P = 0.124). Only 31 patients experienced PONV in the recovery room, severely limiting the value of this predefined multivariable analysis. For completeness, these data are shown in Supplemental Table S1, https://links.lww.com/EJA/A221.

T2-11
Table 2:
Univariable and multivariable regression results for postoperative nausea and/or vomiting during the first 24 h

Figure 3 shows the ROC curves for the prediction of PONV within the first 24 postoperative hours by the simplified Apfel score only and the three multivariable models. Although offering only modest differentiation, all three of the models showed significant improvement in terms of the area under the curve compared with the simplified Apfel score alone (P = 0.016, 0.005 and 0.004, respectively); no differences were found among the three models themselves (all pairings, P > 0.05).

F3-11
Fig. 3:
Receiver operating characteristics curve of the simplified Apfel as well as Models 1 to 3 for predicting postoperative nausea and/or vomiting within 24 h. Model 1 was comprised of the simplified Apfel score + intra-operative fentanyl in μg h−1; Model 2 was comprised of Model 1 + antiemetic therapy; Model 3 was comprised of Model 2 + postoperative morphine. AUC, area under the curve; CI, confidence interval.

When comparing propofol TIVA with volatile anaesthesia, there were no differences in the incidence of PONV (P = 0.68) or postoperative pain (P = 0.63) or in morphine administration (P = 0.66) nor total morphine dose (P = 0.68).

Intra-operative fentanyl concentrations, pain scores and morphine administration

Intra-operative fentanyl concentrations in μg h−1 did not correlate with peak pain in the PACU (slope 0.003, P = 0.209, r = 0.07). However, it did correlate with peak pain within the first 24 h (slope 0.008; P = 0.001, r = 0.173). In other words, a 100 μg h−1 increase in intra-operative fentanyl was associated with an increased NRS of nearly 1 point. In the 103 patients receiving postoperative morphine, there was a trend towards an association between intra-operative fentanyl and postoperative morphine (slope 0.025; P = 0.055, r = 0.190). However, in all three linear regressions, the model fit was poor.

Discussion

In this prospective cohort study, intra-operative fentanyl dosing was independently associated with PONV despite a contemporary anaesthesia regimen based mainly on TIVA with propofol and largely guideline conforming in terms of PONV prophylaxis. While PONV rates in the PACU were low, PONV rates within the first 24 h were disappointingly high. In addition, intra-operative fentanyl was associated with higher pain scores and a trend towards higher morphine requirements in the 24 h following surgery.

Postoperative nausea and/or vomiting rates and prophylaxis

Our study showed high PONV rates of 45% in a moderate to high-risk cohort. However, a calculation based on our Apfel score distribution and the published PONV risk rates in the Apfel populations (i.e. volatile anaesthetics and no antiemetics), we would have expected a PONV rate of 53.2%. Although this is significantly higher than we observed (P = 0.025, χ2), our PONV rate remains disappointingly high. At first glance, it appears that the 126 patients without antiemetic prophylaxis may have been undertreated for PONV. However, current guidelines7 suggest treating patients with an Apfel score of 2 to 3 with at least one of the following: propofol for maintenance, dexamethasone, ondansetron, droperidol and so on, and 90 of the 126 patients not receiving prophylactic antiemetic agents had TIVA with propofol. Clearly though, the 36 patients (9.9% of total) receiving neither TIVA with propofol nor antiemetics were undertreated and 20 of these exhibited PONV (56%). However, this does not explain the overall magnitude of PONV observed. A more aggressive pre-emptive antiemetic regimen may have further reduced PONV, and some experts suggest giving PONV prophylaxis to all patients.7 In addition, the evidence-based rationale for using volatile anaesthetics rather than TIVA in patients at risk of cardiac events is decreasing, even in cardiac surgery.18 Although our general institutional policy was to target pre-emptive PONV therapy to patient needs and to not administer three antiemetic agents indiscriminately to each patient, these results have led to a more liberal administration of these drugs in addition to the even more widespread use of TIVA. Worth consideration also is the fact that 60 patients were discharged within the first 24 postoperative hours. This may have led to an overestimation of PONV as patients with (severe) PONV are not generally discharged ahead of plan.

Intra-operative fentanyl a modest but modifiable postoperative nausea and/or vomiting risk factor

Many risk factors for PONV are known but few are modifiable. If one considers the simplified Apfel score (female sex, smoking status, history of PONV or motion sickness and expected postoperative opioids), it is clear that most factors are not modifiable. The current guidelines for the management of PONV also include younger age, type of surgery, duration of anaesthesia, general vs. regional anaesthesia, use of volatile anaesthetics and nitrous oxide, and postoperative opioids as positive risk factors for PONV in adults.7 Age and type of surgery are nonmodifiable and, at least in our population, regional anaesthesia alone is frequently not an option. Similarly, the duration of anaesthesia is dictated by the length of surgery. Furthermore, many institutions – such as ours – no longer have nitrous oxide, and volatile agents are often limited to patients with cardiac risk (who are generally older and often smokers). This then leaves clinicians with the risk factors of expected postoperative opioids, actually administered postoperative opioids and, as this analysis suggests, the intra-operative fentanyl dose. Ziemann-Gimmel et al.19 observed that compared with general anaesthesia using volatile anaesthetics and opioids an opioid-free TIVA (with propofol, ketamine and dexmedetomidine) reduced both the absolute risk of developing PONV by 17.3% and the severity of PONV. Similarly, an older study of 80 women undergoing gynaecological interventions, randomised to receive either intra-operative fentanyl or ketorolac, showed a higher need for postoperative analgesia (84 vs. 56%, P < 0.05) and higher PONV requiring intervention (29 vs. 10%, P < 0.05) in the group receiving fentanyl.20 No adjustment for pre-operative risk was made and maintenance was by propofol and nitrous oxide.20 More recently, a study of 502 nonsmoking women undergoing gynaecological interventions with maintenance by volatile anaesthetics showed fentanyl-induced cough to be associated with PONV [odds ratio (OR) 2.08, 95% confidence interval 1.41 to 3.07].21

It is uncertain which of these opioid-based variables should be used in a PONV assessment, and it is unclear how they relate to one another. Actually administered postoperative morphine is a rather late risk factor which, although strongly associated with PONV, is of limited value for guiding antiemetic prophylaxis. At an earlier stage, expected postoperative morphine could be used as a risk factor, but this is a variable term, subject to both variation in a caregiver's assessment and also subject to changes over time, for example the current movement towards opioid reduced (or even opioid free) anaesthesia. Our results suggest that administered intra-operative fentanyl is a quantifiable risk factor which, because of its intra-operative occurrence, may be helpful in guiding PONV prophylaxis near the end of surgery. Our models were conceptually designed to combine complete pre-operative information (i.e. the Apfel score) with early intra-operative management choices. Although higher fentanyl doses were associated with higher expectations of postoperative morphine needs, OR 1.008 (1.004 to 1.013), intra-operative fentanyl did improve the Area under the receiver operating characteristics curve (AUROC) significantly when adjusting for the Apfel score. Furthermore, analysing fentanyl dose in addition to a modified Apfel score but excluding the variable ‘expected postoperative morphine’ yielded a similar OR of 1.006 (1.002 to 1.010). Apart from the later time when the morphine information becomes available, the addition of ‘actually administered postoperative morphine’ did not improve the AUROC, nor was it a significant factor in the multivariable analyses. Clinicians should have PONV in mind when administering intra-operative fentanyl.

Intra-operative fentanyl and postoperative pain

We did not find a clear association between intra-operative fentanyl administration (μg h−1) and pain scores in the PACU. However, by 24 h, higher intra-operative fentanyl doses were clearly associated with higher NRS scores and there was a trend towards higher postoperative morphine use. It is important to remember that this is an association, which does not necessarily indicate causality. A common explanation may be that more painful surgical procedures evoke more opioid administration by care-givers, either with or without accounting for the levels of pain expressed by patients. However, an alternative explanation worth considering may be that fentanyl causes opioid tolerance and hyperalgesia,22,23 but this cannot be answered in an observational study. Randomised studies involving gynaecological and obstetric surgery have shown that higher intra-operative opioid doses (either intravenous or intrathecal) are associated with higher pain scores,20,24 increased postoperative opioid requirements,25 or both.26 Nonetheless, some uncertainty regarding the extent24 or even existence27 of this relationship remains. In addition, the precise time point when hyperalgesia develops remains uncertain. In a previous study, we were able to show that high doses of fentanyl induced more hyperalgesia after 4 to 6 h, but not after 1 h.22 This window of about 4 h has also been shown in a study examining fentanyl-induced hyperalgesia and surgical incisions in rats.28 Potentially, this finding may explain why patients in the PACU did not suffer higher pain scores as they may still have been protected by a residual analgesic effect from the fentanyl. At a later time, with decreasing fentanyl plasma concentrations, this analgesic effect may have dissipated, while hyperalgesia may have started to manifest as higher pain scores and a trend towards higher morphine consumption.

Strengths and limitations

The strengths of this cohort study were the inclusion of a substantial number of moderate to high-risk patients who were managed by a contemporary anaesthesia regimen and who were undergoing a range of surgical procedures generally considered to have high PONV rates. Determination of PONV was prospective, systematic and involved both a semistructured interview as well as an extensive examination of hospital records. In addition, both patients and care-givers were unaware of the study during the data collection period. Finally, our main interest was in assessing the effect of intra-operative fentanyl when adjusting for pre-operative risk.

The current study also had some limitations. First, despite the number of patients, our single centre results may be biased by our patient management practice and should not be extended uncritically to other management schemes. As an example, the fact that our data were not able to confirm the protective effect of smoking or male sex on PONV merits attention. As we routinely use the simplified Apfel score pre-operatively, our choice of anaesthetic agents and use of antiemetics was targeted to at-risk patients and may reflect (moderately) successful management. The lower rate of PONV in the PACU may also be indicative of this. Second, in our institution, fentanyl, remifentanil and occasionally morphine, are coadministered intra-operatively, potentially introducing bias. However, in relation to remifentanil, which was coadministered as a target-controlled infusion in the 82% of patients receiving TIVA, whether a patient received or did not receive remifentanil made no difference to the incidence of PONV (P = 0.492). Furthermore, even when excluding the 56 (15%) patients who received intra-operative morphine (which was associated with PONV) intra-operative fentanyl doses per unit time still remained higher in patients suffering PONV (P = 0.031). In addition, our observational, nonprotocolised approach enabled us to observe patients and care-givers without their knowledge (i.e. no Hawthorne effect). Third, 22 patients received combined anaesthesia (either neuraxial or regional) but, when these were excluded from the analysis the OR was similar when adjusting for the Apfel score: OR 1.006 (1.002 to 1.010). Nevertheless, we would be hesitant to apply our results to this combined anaesthesia population without further study. Fourth, our method of ascertaining PONV in the PACU and at 24 h may have influenced PONV rates. PONV episodes in the PACU were only confirmed using patient documents and the structured interview on the first postoperative day, thus possibly underestimating early postoperative nausea. Conversely, specifically asking patients about PONV on the first postoperative day may have led to the inclusion of patients with low levels of PONV, which otherwise may not have been detected or recorded in charts. Finally, 60 patients (12%) were discharged before follow-up on the first postoperative day and another 65 (13%) were excluded, mostly due to an inability to answer questions or to a request that their data not be used. It is uncertain whether or not these groups may be a source of bias.

In conclusion, this prospective cohort study employing a contemporary anaesthetic regimen, showed disappointingly high PONV rates within the first 24 h despite 90% following the guidelines for PONV prophylaxis. In addition, there was a moderate association between intra-operative fentanyl dosing (a modifiable factor) and PONV and pain scores at 24 h. While awaiting further randomised clinical trials for confirmation, this study supports a growing body of literature suggesting we reconsider our daily clinical practice of giving (excessively) high doses of intra-operative fentanyl and examine alternatives.29

Acknowledgements relating to this article

Assistance with the study: none.

Financial support and sponsorship: this work was supported by and attributed to the Department for Anesthesia, Surgical Intensive Care, Prehospital Emergency Medicine and Pain Therapy, University Hospital Basel.

Conflicts of interest: none.

Presentation: none.

References

1. Macario A, Weinger M, Carney S, et al. Which clinical anesthesia outcomes are important to avoid? The perspective of patients. Anesth Analg 1999; 89:652–658.
2. Hofer CK, Zollinger A, Buchi S, et al. Patient well being after general anaesthesia: a prospective, randomized, controlled multicentre trial comparing intravenous and inhalation anaesthesia. Br J Anaesth 2003; 91:631–637.
3. Wu CL, Berenholtz SM, Pronovost PJ, et al. Systematic review and analysis of postdischarge symptoms after outpatient surgery. Anesthesiology 2002; 96:994–1003.
4. Gold BS, Kitz DS, Lecky JH, et al. Unanticipated admission to the hospital following ambulatory surgery. JAMA 1989; 262:3008–3010.
5. Koivuranta M, Laara E, Snare L, et al. A survey of postoperative nausea and vomiting. Anaesthesia 1997; 52:443–449.
6. Apfel CC, Laara E, Koivuranta M, et al. A simplified risk score for predicting postoperative nausea and vomiting: conclusions from cross-validations between two centers. Anesthesiology 1999; 91:693–700.
7. Gan TJ, Diemunsch P, Habib AS, et al. Consensus guidelines for the management of postoperative nausea and vomiting. Anesth Analg 2014; 118:85–113.
8. Apfel CC, Greim CA, Haubitz I, et al. A risk score to predict the probability of postoperative vomiting in adults. Acta Anaesthesiol Scand 1998; 42:495–501.
9. Apfel CC, Kranke P, Katz MH, 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 2002; 88:659–668.
10. Eberhart LH, Morin AM. Risk scores for predicting postoperative nausea and vomiting are clinically useful tools and should be used in every patient: con – ‘life is really simple, but we insist on making it complicated’. Eur J Anaesthesiol 2011; 28:155–159.
11. van den Bosch JE, Kalkman CJ, Vergouwe Y, et al. Assessing the applicability of scoring systems for predicting postoperative nausea and vomiting. Anaesthesia 2005; 60:323–331.
12. Watcha MF, White PF. Postoperative nausea and vomiting. Its etiology, treatment, and prevention. Anesthesiology 1992; 77:162–184.
13. von Elm E, Altman DG, Egger M, et al. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. Lancet 2007; 370:1453–1457.
14. Aeschlimann A, Buettner UW, Desmeules J, et al. Recommendations for opioids in chronic pain. Swiss Med Forum 2005; 5:1107–1113.
15. Eychmüller S. Considerations in therapy and selection of potent opioids. Primary Care 2004; 4:528–564.
16. Althouse AD. Adjust for multiple comparisons? It's not that simple. Ann Thorac Surg 2016; 101:1644–1645.
17. Bagley SC, White H, Golomb BA. Logistic regression in the medical literature: standards for use and reporting, with particular attention to one medical domain. J Clin Epidemiol 2001; 54:979–985.
18. Landoni G, Lomivorotov VV, Nigro Neto C, et al. Volatile anesthetics versus total intravenous anesthesia for cardiac surgery. N Engl J Med 2019; 380:1214–1225.
19. Ziemann-Gimmel P, Goldfarb AA, Koppman J, et al. Opioid-free total intravenous anaesthesia reduces postoperative nausea and vomiting in bariatric surgery beyond triple prophylaxis. Br J Anaesth 2014; 112:906–911.
20. Sukhani R, Vazquez J, Pappas AL, et al. Recovery after propofol with and without intraoperative fentanyl in patients undergoing ambulatory gynecologic laparoscopy. Anesth Analg 1996; 83:975–981.
21. Li CC, Chen SS, Huang CH, et al. Fentanyl-induced cough is a risk factor for postoperative nausea and vomiting. Br J Anaesth 2015; 115:444–448.
22. Mauermann E, Filitz J, Dolder P, et al. Does fentanyl lead to opioid-induced hyperalgesia in healthy volunteers?: A double-blind, randomized, crossover trial. Anesthesiology 2016; 124:453–463.
23. Vinik HR, Kissin I. Rapid development of tolerance to analgesia during remifentanil infusion in humans. Anesth Analg 1998; 86:1307–1311.
24. Carvalho B, Drover DR, Ginosar Y, et al. Intrathecal fentanyl added to bupivacaine and morphine for cesarean delivery may induce a subtle acute opioid tolerance. Int J Obstet Anesth 2012; 21:29–34.
25. Cooper DW, Lindsay SL, Ryall DM, et al. Does intrathecal fentanyl produce acute cross-tolerance to i.v. morphine? Br J Anaesth 1997; 78:311–313.
26. Chia YY, Liu K, Wang JJ, et al. Intraoperative high dose fentanyl induces postoperative fentanyl tolerance. Can J Anaesth 1999; 46:872–877.
27. Cooper DW, Garcia E, Mowbray P, et al. Patient-controlled epidural fentanyl following spinal fentanyl at caesarean section. Anaesthesia 2002; 57:266–270.
28. Chang L, Ye F, Luo Q, et al. Increased hyperalgesia and proinflammatory cytokines in the spinal cord and dorsal root ganglion after surgery and/or fentanyl administration in rats. Anesth Analg 2018; 126:289–297.
29. Mauermann E, Ruppen W, Bandschapp O. Different protocols used today to achieve total opioid-free general anesthesia without locoregional blocks. Best Pract Res Clin Anaesthesiol 2017; 31:533–545.

* Eckhard Mauermann and Damian Clamer contributed equally to the article as first authors.

Supplemental Digital Content

© 2019 European Society of Anaesthesiology