Post-operative nausea and vomiting (PONV) is a common complication following elective and emergent caesarean deliveries (CD). The main reason for this complication is the use of intrathecal morphine (ITM) to treat postoperative pain [1]. Studies have shown PONV rates of up to 60% when this medication is given without antiemetic prophylaxis [2]. The routine use of antiemetic prophylaxis can reduce this rate by 50% [2, 3].

Numerous classes of medication have been studied to minimize PONV following elective CD. Serotonin, dopamine, cholinergic and histamine receptor antagonists and corticosteroids such as dexamethasone are commonly used agents. Previous studies demonstrated that all of these drugs except dexamethasone were effective in preventing PONV post CD [3]. This medication did not show benefits when used as a solo medication and showed only marginal gains when combined with droperidol [2, 3, 4, 5, 6].

The purpose of this study was to determine the impact of different combinations of antiemetics given in the intraoperative period on the rate of PONV in parturients undergoing elective CD in the context of ITM. We hypothesized that there would be an inverse correlation between the number of antiemetics utilized and the incidence of PONV. A secondary hypothesis was that dexamethasone, either alone or in combination, would not be efficacious in the prevention of PONV.

This study adheres to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines [7]. The Fiona Stanley Hospital (FSH) is a tertiary obstetric centre serving the South Metropolitan area of Perth in Western Australia, where approximately 3300 births take place per year. Since opening in October of 2014, the hospital has rapidly increased the capacity of its maternity and neonatal units providing tertiary obstetric care to approximately 1 million patients or 35% of the total population of the city.

We conducted a retrospective analysis of women who underwent elective CD where ITM was used as a part of a post-operative multimodal analgesic regimen between the 1st January 2016 and the 31st October 2017. Patients were excluded if additional additives, e.g., clonidine were added to the intrathecal space; when regional anaesthesia failed; when the ITM dose was less than 100 mcg or greater than 150 mcg or if sufficient information could not be obtained from the paper and electronic perioperative records.

Patients were identified through their unique medical record number (MRN) in the hospitals electronic record system. Demographics, American Society of Anesthesiologists physical status (ASA-PS) [8], smoking status (active smoking or not currently smoking), perioperative and post-operative data was collected. Information surrounding the anaesthetic technique, including the total dose of bupivacaine; dose of intrathecal morphine; uterotonic agents and antiemetic medications were extracted.

The primary outcome of this study was to determine the rates of PONV during the first 24 hours following the surgery when different combinations of antiemetics were given in the intraoperative period. PONV was assessed in the patient’s medication chart. Patients who received rescue antiemetics in the postoperative period were considered to have had one episode of PONV. Secondary outcome focused on the efficacy of dexamethasone when added to one, two or three other classes of antiemetics. We also analysed the impact of modifying factors, such as ITM dose, naloxone use, different uterotonics agents and postoperative opioid use in the rates of PONV.

Uterotonic agents to prevent post-partum haemorrhage (PPH) used at FSH were carbetocin or oxytocin. As per institution protocol, carbetocin 100 micrograms is administered as a bolus after the delivery of the baby and oxytocin is administered as a bolus (3–5 units) followed by an infusion of 10 units per hour over 4 hours.

The choice of drugs given in intrathecal space, namely bupivacaine and intrathecal morphine dosage and the post-operative analgesic regimen was at the discretion of the specialist anaesthetist and the obstetric team supervising the care of the patient. Antiemetics commonly used for prophylaxis of PONV included: serotonin (5HT${}_3$) receptor antagonists (e.g., ondansetron), dopaminergic antagonists (e.g., droperidol. metoclopramide), histamine antagonists (e.g., cyclizine) and corticosteroids (e.g., dexamethasone). The selection of the antiemetic prophylaxis was at the discretion of the individual anaesthetist caring for the patient. The most frequently prescribed opioid based analgesics were oral tramadol and sublingual buprenorphine.

Between the 1st January 2016 and the 31st October 2017, 686 women underwent elective caesarean delivery. After the exclusion of 88 patients, 598 (87.2%) patients remained (Fig. 1). PONV occurred in 174 (29%) parturients. There were no differences in the baseline demographics between the groups who did and did not have PONV (Table 1).

Fig. 1.

Retrospective cohort study flowchart.

There was a significant association between the number of antiemetics used and the probability of experiencing PONV (Odds Ratio = 0.70 (95% CI: 0.58–0.84), p 0.001, R${}^2$ = 0.04), indicating that PONV decreases with moderate effect as the number of antiemetics administered increases. Subgroup analysis showed that patients receiving 0, 1, 2, 3 or 4 antiemetic medications had rates of 37%, 37%, 28%, 16% and 23% of PONV (Table 2). There was no statistically significant difference in the incidence of PONV between patients who received zero and one antiemetic (p = 0.56) and one and four antiemetics (p = 0.194) (Table 2). The number of antiemetics administered accounts for 31.8% of the variability in PONV (R${}^2$ = –0.318). There was a statistically significant difference in the incidence of PONV between patients who received one and two antiemetics (p = 0.062); one and three antiemetics (p = 0.001) and two and three antiemetics (p = 0.008). The median number of antiemetics across the entire cohort was 2 (interquartile range 1–2). Those in PONV group received less antiemetic prophylaxis than those in non-PONV group (2 (1–2) versus 2 (1–3), p 0.001).

Larger doses of ITM were associated with higher rates of PONV (p 0.001) (Table 3). There were significant differences in the PONV rate between those who received 100 mcg versus 150 mcg of intrathecal morphine (p 0.001) and those who received 125 mcg versus 150 mcg (p = 0.02). However, no significant difference was found with doses of a 100 mcg and 125 mcg (p = 0.52).

There was no difference in the use of both carboprost (p = 0.15) and ergometrine (p = 0.15) between the two groups. PONV was more frequent in patients who receive opioids for breakthrough pain in the postoperative period (16.1% versus 3.1%, p 0.001). Both smoking (p = 0.03) and the use of naloxone to prevent itch (p 0.001) appeared to be protective against PONV. Although the incidence of PONV was slightly higher in those who received carbetocin (31.5%) when compared with oxytocin (23.5%), no statistical difference was detected between the two drugs (p = 0.06). There were 576 (96.5%) of patients who were ASA-PS2 and the remaining (22) were ASA-PS3 (Table 1). There was no statistical relationship in ASA- PS scores for PONV (p = 0.63).

Sensitivity analyses were completed to account for factors which were known to increase or decrease the rates of PONV in patients undergoing CD [3]. Factors considered to increase the rates of PONV excluded in this analysis were Carboprost (PGF${}_{2\alpha }$), ergometrine and postoperative opioids for breakthrough pain. Factors considered to decrease the rates of PONV excluded in this analysis were Naloxone used in the postoperative period and active smokers. After removing all the potential biases from the main analysis, a significant statistical difference remained between the number of antiemetics given in the intraoperative period and the occurrence of PONV (p 0.001).

To determine the effect of different combinations of antiemetics exploratory analyses were completed. Some of these tests were limited by statistical power and we could not analyse drugs in combination apart from dexamethasone. Overall, 240 (40.1%) patients received dexamethasone. Only 3.1% (5/163) of the patients received this medication as a solo antiemetic. Patients who received two, three or four antiemetics, had dexamethasone as part of the combination in 55.1% (124/225), 77.8% (98/126) and 76.5% (13/17) of the times respectively. There was no difference in the PONV rate between patients who did and did not receive dexamethasone (25.0% versus 31.8%, p = 0.08). In the subgroup analyses no difference was found in the PONV rate between two or three drug combinations when dexamethasone was used or avoided (p = 0.46 and p = 0.56 respectively).

A multivariate model was constructed using backwards elimination to determine the factors associated with PONV in our patient population. Those associated with PONV were number of antiemetics used (Odds Ratio (OR): 0.69 (95% 0.57–0.84), p 0.001); intrathecal morphine dose (OR: 1.01 (1.01–1.02), p 0.001) and perioperative opioid use (OR: 5.5 (2.7–11.2), p 0.001). The R${}^2$ value was 0.22.

This study showed that nausea and vomiting are still common complications in the postoperative period in parturients undergoing caesarean delivery with intrathecal morphine. We found that exposure to a greater number of antiemetics was protective. However, we did not find any benefits of dexamethasone either on its own, or when combined to one, two or three other antiemetics.

In our study, approximately 37% of women who did not receive antiemetic prophylaxis experienced PONV. In comparison, two studies found PONV rates of 67% and 35% when placebo was used in one of the arms [2, 9]. Differences between the doses of ITM given and the management of uterotonic agents in the perioperative period may explain part of these discrepant findings. These same studies demonstrated that the PONV rate decreased by more than 50% when one antiemetic was used. In contrast, patients who received only one antiemetic in our study continued to experience similar rates of PONV as the group who did not have antiemetic prophylaxis. However, these rates significantly decreased to 28%, 16% and 23% in women who received two, three or four antiemetics respectively. To our knowledge, this is the first study to analyse the rates of PONV in patients receiving such a diverse combination of antiemetics in the intraoperative period. The lack of studies comparing various combinations of antiemetics other than dexamethasone is part explained by the difficulties associated with conducting research using multiple arm groups and the element that some of the antiemetics may pose a low theoretical risk to the neonate [10, 11]. As a result, there is a lack of consensus to the best regimen to prevent this complication and combinations selected by anaesthetists are often ad hoc and many avoid using some of the agents.

A few studies assessed the rate of PONV when one or two drugs are used in the context of intrathecal morphine. When compared with placebo, serotonin antagonists, dopaminergic antagonists such as droperidol and histamine H1 antagonists have all shown to be effective as single agents while the evidence for dexamethasone is lacking [2, 3, 4, 5, 6]. Studies comparing dexamethasone as a single agent to placebo could not demonstrate benefits of this medication for PONV prophylaxis [2, 4, 5]. Moreover, when used in combination with another antiemetic, only one study was able to find a marginal benefit of this medication when combined with droperidol [6]. Our study didn’t show any benefit of dexamethasone when combined with either one, two or three other antiemetics. This further reinforces the current literature which suggests there is no evidence supporting the use of dexamethasone for PONV after CD [2, 3, 4, 5].

We also found that larger doses of intrathecal morphine and the use of postoperative opioids for breakthrough pain led to greater rates of PONV. It is known that increasing the dose of ITM will increase the risk of side-effects such as PONV, pruritus and urinary retention [12]. Moreover, ergometrine and carboprost used as second line agents to promote uterine contraction can also precipitate nausea and vomiting through their diffuse mechanisms of action including agonism at serotonin and dopamine receptors and augmentation of oxytocin release respectively [1]. These factors were considered potential biases that could have increased the rates of PONV in our study. On the other hand, factors that could have potentially decreased the rates of this complication were antiemetics given in the intraoperative period, the use of the opioid receptor antagonists naloxone in the postoperative period and active smoking patients, which is a known protective factor for PONV [13]. After correcting for these cofounders, a statistically significant difference in the number of antiemetics administered and the rates of PONV remained.

In our population both carbetocin and oxytocin were used as uterotonic agents following the delivery of the foetus to promote uterine contraction and prevent PPH. Both these drugs are a known risk factor for intraoperative emesis [1]. Despite this association, little is known about the impact of these drugs on PONV. Their use is mandatory in those undergoing caesarean deliveries and therefore, comparisons to those who did not receive the drug was not possible. In this study we found no differences in the rate of post-operative nausea and vomiting between carbetocin and oxytocin. These results may however be institution specific due to the variability in oxytocin bolus and infusion regimens which exist between centres [14].

In this study, dexamethasone was the most frequent medication used in combination with multiple other classes of antiemetic. For this reason, this research lacks the statistical power and sample size to draw stronger conclusions including suggestions as to which combinations of antiemetic drugs are the most efficacious in this setting. Although the use of three or even four agents may reduce the rate of PONV further, these results may be confounded by patients who experienced intraoperative nausea and vomiting or patients who had significant risk factors for PONV which may make the data more complex to interpret. It is difficult to ascertain the reason why multiple drugs were used as the medical record in place often lacked comments explaining why various drug combinations were used. Independently, patients who received more antiemetics had lower rates of nausea and vomiting in the postoperative period.

The external validity of this study may be limited by some factors. We did not collect data on patient comorbidities such as gestational diabetes or hypertension to determine their effect on PONV rates. As a surrogate we used the ASA-PS score. Most patients in this cohort were ASA-PS 2 which may indicate a discrepancy on how clinicians judge which patients are ASA-PS 2 or ASA-PS 3 or the patients which were included had a normal pregnancy or had controlled gestational hypertension, diabetes or preeclampsia [8]. Therefore, patients who have a complicated gestation (ASA-PS 3) may require different antiemetic therapy. Moreover, with the worldwide trend towards an increase age at the time of first pregnancy, our results may not be extrapolated outside of this population since most of the cohort were in their fourth decade of life. Finally, this study examined the rate of PONV in a group of patients undergoing elective CD where ITM was used for post-operative analgesia. These results may not be valid in those undergoing emergent CD or where alternative additives are used in the intrathecal injectate.

We have shown that the rate of postoperative nausea and vomiting after elective caesarean deliveries when intrathecal morphine is used was inversely related to the number of antiemetics given in the intraoperative period. Moreover, dexamethasone did not show to provide any benefit to prevent PONV when added to one, two or three antiemetics. Future research is warranted to determine the optimal combination of drugs to reduce PONV without maternal and neonatal complications associated with their use.

MEK First author—Study design, data collection and manuscript writing. NJL—Study design, Statistical analysis, data analysis and manuscript writing. YO—Study design, data collection and manuscript writing. JYC —Data collection and manuscript writing. HSM—Study design and data collection. DWH—Study design and data collection. EJO, principal investigator—Study design, data collection and manuscript writing. All authors contributed to editorial changes in the manuscript. All authors read and approved the final manuscript.

This project received approval as a quality assurance project through the Quality Improvement office at the Fiona Stanley Hospital via the GEKO (Governance, Evidence, Knowledge, Outcomes) system—reference number 23664. Informed consent was not obtained as this was a retrospective cohort study.

The authors would like to recognise all the support offered for this research from the Mackay Institute of Research and Innovation (MIRI). Special thanks for Dr. David Farlow. Also, the authors would like to express gratitude for the Fiona Stanley Hospital Department of Anaesthesia, especially Dr Edmond J O’Loughlin and his vision for research that made this project possible.

Funding to conduct this study was provided by the Mackay Hospital and Health Service. Grant: Mackay Institute of Research and Innovation (MIRI) granted 5000 Australian dollars for research. Reference: MIRI2020-04.

The authors declare no conflict of interest.