Investigating the Association Between a Risk-Directed Prophylaxis Protocol and Postoperative Nausea and Vomiting: Validation in a Low-Income Setting : Anesthesia & Analgesia

Secondary Logo

Journal Logo

Global Health

Investigating the Association Between a Risk-Directed Prophylaxis Protocol and Postoperative Nausea and Vomiting: Validation in a Low-Income Setting

Tuyishime, Jean de Dieu H. MD, MMed*; Niyitegeka, Joseph MD, MMed*; Olufolabi, Adeyemi J. MBBS, FRCA; Powers, Samuel MA; Naik, Bhiken I. MBBCh, MSCR§; Tsang, Siny PhD§; Durieux, Marcel E. MD, PhD§; Twagirumugabe, Theogene MD, MMed, FCCM, PhD*

Author Information
Anesthesia & Analgesia ():10.1213/ANE.0000000000006251, October 12, 2022. | DOI: 10.1213/ANE.0000000000006251

Abstract

KEY POINTS

  • Question: Is implementation of risk-directed postoperative nausea and vomiting (PONV) prophylaxis based on the Apfel score associated with a decreased incidence of PONV in low-income settings?
  • Findings: Implementation of a protocol was associated with reduced vomiting and time to first oral intake, although longitudinal studies are needed to assess whether these effects are sustained over time.
  • Meaning: Despite major differences in low-income settings, implementing multimodal prevention of PONV based on the Apfel score, a simple preoperative risk-stratification tool, is associated with short-term benefits.

Postoperative nausea and vomiting (PONV) is an unpleasant and potentially harmful complication of surgical procedures that remains poorly addressed.1–5 It typically occurs within 24 to 48 hours after surgery, with 70% to 80% of events occurring within the first 24 hours.6–8 Therefore, PONV is defined as nausea or vomiting occurring during the first 24 to 48 hours after surgery.7,9

According to the 2014 American Society for Ambulatory Anesthesiology guidelines, the incidence of PONV is about 30% for vomiting and 50% for nausea among patients who undergo surgery under general anesthesia.10–12 PONV can reach an incidence of 80% in high-risk patients (Apfel scores 3 and 4); therefore, it is one of the most common causes of patient dissatisfaction after anesthesia.7,10

Unresolved PONV may result in a prolonged postanesthesia care unit (PACU) stay and unanticipated prolonged hospital admission, which results in a significant increase in overall health care cost.9–16

PONV scoring systems and guideline-driven prophylaxis are well established in high-income countries. Well-established risk factors include: female gender, postpubescence, nonsmoking status, history of PONV or motion sickness, childhood and young adulthood, prolonged surgery duration, type of surgery, use of volatile anesthetics, nitrous oxide, large-dose neostigmine, and intraoperative or postoperative opioid use.2,3,9,10 The existing preventive measures are typically guided by a preoperative simplified risk assessment score developed by Apfel et al.10,15 However, little is known about the effectiveness of PONV prevention strategies in variable resource settings.7,13

In such settings, routine PONV prophylaxis often is not well structured. Antiemetic prophylactic medications are not consistently available, drug choices are made without protocol guidance, and risk stratification is infrequently used. Differences between settings, coexisting diseases, surgical procedures, as well as availability and use of anesthetic and analgesic drugs imply that existing models from high-income countries may not be automatically applicable and need to be validated in variable resource settings.

We, therefore, studied the impact of a risk score-directed PONV prophylaxis in Rwanda. We used the Apfel score, which assesses 4 predictors for PONV: gender, history of motion sickness or PONV, nonsmoking status, and the use of opioids in the postoperative period.10,15 The identification of patients at risk for PONV through preoperative risk assessment, combined with prophylaxis based on predicted risk as stratified by Apfel score, has been shown to be a cost-effective means to reduce the incidence of PONV.5,16–19

The objective of this study was to assess the association between introduction of protocolized PONV management and postoperative outcomes. We hypothesized that administering PONV prophylaxis based on the Apfel score would be associated with a decreased incidence of PONV (primary outcome). In addition, we expected earlier oral feeding and improved wound healing, and, hence, a shorter length of stay (secondary outcomes) compared to the preexisting routine practice.

METHODS

Study Setting and Approval

This study was performed in the Centre Hospitalier Universitaire de Kigali (CHUK; University Teaching Hospital of Kigali), one of the major referral hospitals in Rwanda. The study was approved by the University of Rwanda (UR) College of Medicine and Health Sciences (CMHS) institutional review board (IRB; UR/CMHS-IRB; No211/CMHS-IRB/2019) and the CHUK ethics committee (EC/CHUK/100/2019). All patients were informed about the study purpose and provided both their verbal and written consent before enrollment.

Design

This was a single-blinded (patients were blinded, and investigators/providers were unblinded) prospective study.

We assessed PONV and short-term surgical outcomes in periods before and after implementation of an Apfel score-based PONV prevention guideline. During the preintervention period, no systematic assessment of PONV risk and no risk stratification for preventative measures were performed. The intervention consisted of an initiation of PONV preventive measures based on the simplified Apfel score as presented in Table 1. Postintervention outcomes were assessed immediately after implementation of the intervention.

Table 1. - PONV Prevention Protocol as Related to Apfel Score
Apfel Score 18 (Total PONV risk factors) Antiemetics
0–1/4 No prophylaxis 23
2/4 Ondansetron (4 mg IV) 30 min before the end of anesthesia
  Alternative: metoclopramide (10 mg IV) 26
3/4 Ondansetron (4 mg IV) 30 min before the end of anesthesia
  Alternative: metoclopramide (10 mg IV) + dexamethasone (8 mg IV) during induction of anesthesia
4/4 Ondansetron (4 mg IV) 30 min before the end of anesthesia
  Alternative: metoclopramide (10 mg IV) + dexamethasone (8 mg IV) during induction of anesthesia + haloperidol or domperidone (0.5 mg)
 Alternative: propofol induction 4,25
Abbreviations: IV, intravenous; PONV, postoperative nausea and vomiting.

Study Periods and Participants

The preintervention period was between April 1, 2019, and June 30, 2019, and the postintervention period was between July 1, 2019, and September 30, 2019. All patients ≥18 years of age who were scheduled for elective open abdominal surgery under general anesthesia at CHUK were eligible to participate; operation schedules were reviewed each workday by the principal investigator (PI) or associates, and eligible patients were approached.

Intervention

Nausea was defined as any unpleasant sensation with awareness of the urge to vomit, and vomiting was defined as successful or unsuccessful (retching) expulsion of gastric contents.15 PONV risk factors were determined at the preoperative visit (either the day before surgery or on the day of surgery) by the PI, a trained anesthesia resident, or a trained research assistant. Apfel score was calculated for each patient by the PI based on the PONV risk factors. Motion sickness was defined in local context as a history of nausea and vomiting, hypersalivation, or malaise during travel in a bus, boat, or train (or an airplane for those who traveled abroad).17 Past experience of nausea and vomiting after previous surgery was also included in this category.

During preintervention, PONV prophylactic medications were administered at the discretion of the anesthesia provider. During postintervention, PONV prophylactic medications were administered according to the level of PONV risk based on the Apfel score18 (Table 1). A patient scored 1 point for each of the 4 criteria (“female,” “nonsmoking,” “PONV history and/or motion sickness,” and “anticipated need for opioids”) and 0 in the absence of them. Patients with no or 1 risk factor did not receive prophylactic medication. In the presence of 2 risk factors, 4 mg intravenous (IV) ondansetron was administered 30 minutes before the end of the procedure. If 3 risk factors were present, 8 mg IV dexamethasone was added to the ondansetron regimen. In the presence of 4 risk factors, 0.5 mg IV haloperidol was added to this regimen of 2 antiemetics. If any of the above medications were not available, substitutions were made as follows: ondansetron by 10 mg IV metoclopramide and haloperidol by induction of anesthesia with propofol instead of the usual induction agents thiopental or ketamine. Therapeutic ondansetron (4 mg IV) was planned to be given up to 4 times daily when PONV occurred during the 48 hours of follow-up.

Outcome Variables

The primary outcome was to assess the association of protocol implementation, guided by the Apfel score, on the incidence of PONV during the first 48 hours postoperatively. The occurrence of PONV was reported by the patients to the PI or the research assistants through a postoperative interview immediately after surgery in the recovery room, and then at 12 hours, 24 hours, and 48 hours postoperatively. Ward nurses were instructed to call the surgeon for prescription of 4 mg IV therapeutic ondansetron when vomiting occurred, and it could be given up to 4 times in 24 hours, as per the protocol. We examined the incidences of PONV (nausea or vomiting), as well as the incidences of nausea and vomiting separately. Participants were also asked about the exact time they were able to take the first meal or drink postoperatively; this information was used to compute the time to first oral intake (hours). Finally, at discharge from the hospital, the length of hospital stay (days) was recorded and so was the appearance of the surgical incision site to identify any potential sign of wound dehiscence such as an open wound or incisional hernia.

Other Variables Recorded

Other variables collected as potential confounders were age, sex, history of active (firsthand) smoking, and any history of motion sickness or previously experienced PONV in the past surgical history.

Statistical Analysis

Descriptive statistics are presented as means and standard deviations for continuous variables (eg, age) and frequencies and proportions for categorical variables (eg, sex, PONV risk factors, and incidences of PONV). Medians and interquartile ranges (IQRs) were presented for continuous variables that were not normally distributed (eg, length of hospital stay and time to first oral intake). We used linear regression models (for continuous variables) and χ2 tests (for categorical variables) to examine whether patients’ demographic and PONV risk factors differed between preintervention and postintervention.

To address our primary objective, we used an interrupted time series analysis (ITSA) to examine the extent to which the temporal trends of the outcome variables differed between the preintervention and postintervention periods, as suggested post hoc by reviewers. A simple ITSA takes the form of:

Yt=β0+β1Tt+β2Xt+β3TtXt+εt

where Yt is the aggregated outcome variable (eg, rate of nausea incidence) measured at each equally spaced time point t, Tt is the time since the start of the preintervention, Xt represents the preintervention (dummy coded as 0) or intervention (dummy coded as 1) period, and XtTt is the interaction term (0 for the preintervention period, number of time periods from the start of postintervention to the current time period otherwise). β0 is the intercept of the outcome variable, and it represents the initial baseline level (ie, the rate of nausea incidence at time 0). β1 is the slope of the outcome variable during the preintervention period; it represents the amount of change (or lack thereof) in the outcome variable during preintervention. β2 is the change in the level of the outcome variable immediately after the intervention; it represents the amount of change in the outcome variable just after the start of the intervention. β3 represents the difference in the slopes of the outcome variable between postintervention and preintervention periods (ie, postintervention minus preintervention).20

A nonzero estimate of β1 reflects the temporal trend of the outcome variable during the preintervention period; a positive value indicates an increase, whereas a negative value indicates a decrease in the outcome variable during preintervention. β2 reflects the change in the mean of the outcome variable from the end of the preintervention period to the beginning of the postintervention period. A nonzero estimate of β3 reflects a change in the slopes from the preintervention to postintervention periods (postintervention minus preintervention).

Three case-level ITSAs were performed separately for incidences of: (1) nausea or vomiting, (2) nausea only, and (3) vomiting only. As 4 records were available for incidences of nausea and vomiting (immediately and 12 hours, 24 hours, and 48 hours postoperatively) for each patient, ITSA was performed using a generalized estimating equation (GEE) logistic model to adjust for within-patient correlation (assumed first-order autoregressive correlation) on the repeated assessments for patients. We further used ITSA to investigate the extent to which the temporal trends of the secondary outcome variables differ between the preintervention and postintervention periods. We used ITSA logistic regression models to assess the changes in the odds of wound dehiscence (0, no dehiscence and 1, wound dehiscence). ITSA linear regression models were performed to examine the changes in the average time to first oral intake (hours) between preintervention and postintervention. ITSA negative binomial regression models were used to assess the changes in the rate ratios of hospital length of stay (days) between preintervention and postintervention. To control for potential confounding variables, all ITSAs were adjusted for patient age, sex, history of smoking, history of motion sickness, history of PONV, and preoperative and/or anticipated postoperative use of opioids.

Sample Size

Sample size estimation was initially performed for crude comparison between preintervention and postintervention, ignoring time trends. We used OpenEpi Version 3.01, updated in April 2013, for sample size calculation.19 We assumed that the incidence of PONV with routine care seen at CHUK was 50%. We expected that if Apfel score-based prophylaxis was implemented, this incidence would decrease by 50%, corresponding to an absolute risk reduction of 25%. We found that a sample size of at least 116 patients (58 patients in the preintervention group and 58 patients in the postintervention group) would detect this difference with 80% power, a 2-sided type 1 error of 5%. However, our study was not powered to assess differences this small, even if they would be clinically important; larger studies are needed. Our best inference for this and other reported associations is given by the confidence interval (CI) for the odds ratio (OR). All statistical analyses were performed in R version 4.1.2.21 The GEE and Stats packages were used to perform the ITSA analysis.

RESULTS

Fifty-eight patients were enrolled during preintervention and postintervention. Descriptive statistics of patients’ demographic and clinical characteristics are presented in Table 2. There were no statistically significant differences in terms of demographic characteristics and PONV risk factors between the preintervention and postintervention groups, except for smoking, for which the proportion of smokers was higher in the preintervention group (41.4%) compared to the postintervention group (6.9%). Dexamethasone alone was the most commonly prescribed prophylactic regime during preintervention (46%). During the postintervention period, antiemetic prophylaxis combining metoclopramide plus dexamethasone and dexamethasone plus propofol was used in 48% and 21% of patients, respectively (Table 3).

Table 2. - Demographic Data and Apfel Score of Patients in the Current Study
Preintervention, N = 58 Postintervention, N = 58 P
M (SD)/n (%) M (SD)/n (%)
Age 48.0 (15.2) 47.5 (15.7) .86
Sex
 Male 22 (37.9) 14 (24.1) .16
 Female 36 (62.1) 44 (75.9)
PONV risk factors
 Smoking (Yes) 24 (41.4) 4 (6.9) <.001
 History of motion sickness 22 (37.9) 17 (29.3) .43
 History of PONV (if operated before) 10 (17.2) 6 (10.3) .42
 Perioperative and/or anticipated postoperative use of opioids 52 (89.7) 54 (93.1) .74
Apfel score
 1 9 (15.5) 2 (3.4) .06
 2 18 (31.0) 13 (22.4)
 3 23 (39.7) 31 (53.4)
 4 8 (13.8) 12 (20.7)
Abbreviations: M, mean; PONV, postoperative nausea and vomiting; SD, standard deviation.

Table 3. - Rate of PONV Prophylactic Medications Between the 2 Groups
PONV prophylaxis received Preintervention (N = 58), n (%) Postintervention (N = 58), n (%)
None 10 (17) 1 (2)
Dexamethasone alone 27 (46) 0
Ondansetron alone 1 (2) 0
Propofol alone 5 (9) 0
Metoclopramide + dexamethasone 0 28 (48)
Dexamethasone + propofol 15 (26) 12 (21)
Metoclopramide + propofol 0 0
Metoclopramide + dexamethasone + propofol 0 11 (19)
Ondansetron + dexamethasone + propofol 0 6 (10)
Abbreviation: PONV, postoperative nausea and vomiting.

Results of the case-level ITSA on nausea or vomiting, nausea only, and vomiting only, adjusted for confounders, are shown in Table 4. There was no change in the odds of nausea or vomiting during the preintervention period (time [months] β1) per month. There was no change in the odds of nausea or vomiting at the beginning of the intervention compared to just before (intervention, β2), and no statistically significant difference in the slope between preintervention and postintervention (time [months] × intervention, β3) per month (Figure, A). The odds of nausea decreased during the preintervention period (β1, −0.52; OR, 0.60; 95% CI, 0.36–0.97) per month. However, the ORs at the start of the intervention (β2) and change in the slope between preintervention and postintervention (β3) were not statistically significant (Figure, B). Regarding the odds of vomiting, there was no statistically significant change during the preintervention period (β1) per month. At the beginning of the intervention, there was a statistically significant decrease in odds of vomiting (β2, −2.32; OR, 0.10; 95% CI, 0.02–0.47; P = .004). The change in slope between preintervention and postintervention (β3) was not statistically significant (Figure, C). Results from the interrupted time series linear regression model showed no significant slope in the average time to first oral intake (hours; Table 5 and Supplemental Digital Content 1, https://links.lww.com/AA/E76). At the beginning of the postintervention period, there was a significant decrease in the average time to first oral intake (β2, −14.1; standard error [SE], 5.7; 95% CI, −25.4 to −2.78), reflecting an average decrease of about 14 hours in the first oral intake. There was no difference in the slopes of time to first oral intake (hours) between the 2 periods. There was an increasing trend in the odds of wound dehiscence during preintervention (β1, 2.3; OR, 10.3; 95% CI, 2.0–82.0) per month (Supplemental Digital Content 2, https://links.lww.com/AA/E76; Supplemental Digital Content 3, https://links.lww.com/AA/E76). However, there was a decrease in the odds of wound dehiscence at the beginning of the intervention (β2, −4.5; OR, 0.01; 95% CI, 0–0.42) compared to just before. The difference in the slope of the odds of wound dehiscence was not statistically significant between preintervention and postintervention (Supplemental Digital Content 3, https://links.lww.com/AA/E76). However, it should be noted that this parameter (β3) was estimated with large SEs. Considering that the rate of wound dehiscence was low (10.3% and 3.4% in preintervention and postintervention, respectively), our study may not be powered to assess a difference this small; future studies with a larger sample size are needed to better assess the potential difference in temporal trends of wound dehiscence.

Table 4. - Case-Level Interrupted Time Series Analysis, Generalized Estimating Equation Logistic Models on Nausea or Vomiting, Nausea Only, and Vomiting Only, Adjusted for Confounders
Parameter Estimate Robust SE OR (95% CI) P
Outcome: nausea or vomiting
 Intercept, 3.17 1.03
 Time (mo), —0.26 0.27 0.77 (0.46––1.31) .34
 Intervention, —1.46 0.75 0.23 (0.05–1.01) .05
 Time (mo) × intervention, 0.11 0.47 1.11 (0.44–2.82) .82
Outcome: nausea
 Intercept, 3.40 1.00
 Time (mo), −0.52 0.25 0.60 (0.36–0.97) .04
 Intervention, −0.04 0.69 0.96 (0.25–3.72) .95
 Time (mo) × intervention, 0.39 0.46 1.48 (0.60–3.65) .40
Outcome: vomiting
 Intercept, 0.40 0.79
 Time (mo), 0.01 0.26 1.01 (0.61–1.67) .97
 Intervention, −2.32 0.80 0.10 (0.02–0.47) .004
 Time (mo) × intervention, −0.27 0.71 0.76 (0.19–3.05) .70
β0
: intercept: log-odds of nausea/vomiting at time 0 with intervention = 0.
β1
: preintervention slope: change in log-odds of nausea/vomiting per month preintervention.
β2
: log-odds ratio of nausea/vomiting at the start of intervention compared to end of preintervention.
β3
: difference between periods in slope of nausea/vomiting over time: postintervention minus preintervention.
Abbreviations: CI, confidence interval; OR, odds ratio; SE, standard error.

Table 5. - Case-Level Interrupted Time Series Analysis, Linear Regression Model on Time to First Oral Intake (Hours), Adjusted for Confounders
Parameter Estimate SE 95% CI P
Intercept, 39.16 7.15
Time (mo), 0.04 2.17 −3.90 to 4.71 .83
Intervention, −14.11 5.71 −25.43 to −2.78 .02
Time (mo) × post, 2.26 2.88 −3.45 to 7.97 .44
β0
: intercept: average time to first oral intake (hours) at time 0 with intervention = 0.
β1
: preintervention slope: change in average time to first oral intake (hours) per month preintervention.
β2
: average time to first oral intake (hours) at the start of intervention compared to end of preintervention.
β3
: difference between periods in slope of time to first oral intake (hours) over time: postintervention minus preintervention.
Abbreviations: CI, confidence interval; SE, standard error.

F1
Figure.:
Rates of nausea or vomiting (A), nausea (B), and vomiting (C) over time. Each point represents the rate of nausea/vomiting for each case. The blue line represents the average rate of nausea/vomiting in each month. Preintervention: April 2019 to June 2019. Postintervention: July 2019 to September 2019.

Finally, in the adjusted ITSA negative binomial model on hospital length of stay, there was no change in the temporal trend of hospital length of stay (days) in the preintervention period and no change in hospital length of stay (days) at the beginning of the postintervention period. However, there was an improvement in the slope of hospital length of stay (ie, decreased length of stay) during the postintervention period (β3, −0.35; OR, 0.70; 95% CI, 0.5–1.0) per month, reflecting about one-third day less (Supplemental Digital Content 4, https://links.lww.com/AA/E76; Supplemental Digital Content 5, https://links.lww.com/AA/E76).

DISCUSSION

This study demonstrates that in settings of a low-income country like Rwanda, preoperative risk factor screening based on Apfel score and the use of a structured PONV prevention protocol is feasible and is associated with a reduced incidence of vomiting and time to first oral intake immediately after implementation compared to just before the intervention. There was no substantial difference in the trends of PONV incidence, nausea and vomiting incidence, and time to first oral intake before and after implementation, although observations over a longer period are needed to assess the extent to which these effects are sustained over time. Interestingly, although we did not observe impact on hospital length of stay immediately after protocol implementation, there was a slight decrease in hospital length of stay over time in the postintervention period.

Before implementing this management approach, the overall incidence of PONV was 85%. This incidence was consistent with Apfel score prediction, as the preintervention phase of our study showed that >80% of patients had 2 or more risk factors. PONV prophylaxis was underutilized and combination therapy was not used, explaining the high incidence of PONV. Demonstrating the effectiveness of this approach in variable resource settings is critical to validate risk assessment scoring, as the setting of this study is very different from the settings in which the risk score was developed and treatment approaches were validated.

Differences from high-income settings include surgical indications, demographics (younger age group), coexisting diseases and nutritional status, and anesthetic drugs utilized, such as thiopental, halothane, and ketamine, which are common in low-resource settings and rarely used in high-income countries. Therefore, risk stratification and prevention strategies developed in high-income settings need revalidation in low-income countries.

This study was designed as a pragmatic operational research trial, with associated strength and limitations. Study observations were obtained in the regular clinical setting in a low-resource location, not in a highly controlled research environment that limits the variabilities inherent in patient care due to resource availability. Therefore, results have direct and practical implications for clinical management and can be immediately applied to practice. As the hospital site for this study has many similarities to referral hospitals in other low-income settings, our results may be applicable in other sites. Our sample size was sufficient to assess the primary outcome. Adequate assessment of the impact of the protocol on secondary outcomes such as the incidence of postoperative wound dehiscence will require larger studies. The study was also single blinded. This is an appropriate approach for operational research studies but may have increased the likelihood of bias in our findings.

The prophylactic drugs used in our protocol were selected based on studies indicating their efficacy in patients at risk of PONV.17,23–25 We followed existing guidelines by providing no prophylaxis in low-risk patients, while 1 or 2 medications were administered to medium-risk patients, and 3 medications were administered to high-risk patients.18 By using a similar protocol in another “before-after” study, Sigaut et al5 demonstrated that similar educational strategies aiming at improving medical care by systematic recording of simplified items (such as scores) were efficient.

Guidelines should provide fallback options in settings where drug availability cannot be consistently assured. Therefore, we included alternatives to be used when the primary choice of drug was not available. During the postintervention period, most of our patients received metoclopramide instead of ondansetron due to unavailability of the latter drug. Although metoclopramide is considered less effective than ondansetron,26 we still observed a significant reduction of PONV.

PONV may increase time spent in PACU as well as hospital length of stay.27 In this study, we found a median (IQR) length of hospital stay of 5 (3–7) days during preintervention compared to 4 (2–6) days during postintervention (P = .007). Although we cannot assign causality to this association, it is in agreement with previous studies linking duration of hospitalization to PONV.10–16 This would be a particularly important benefit of PONV reduction, as hospital capacity is frequently a major challenge in low-income settings.

Serious complications may arise from untreated PONV. PONV may delay recovery, induce wound dehiscence, and cause pulmonary aspiration of gastric contents, leading to aspiration pneumonia.17,24,28 In our study, we observed signs of wound dehiscence in about 10% of patients during the preintervention period compared to 4% in the postintervention period. However, our study may not be powered to assess a difference this small; future studies with a larger sample size will be needed to assess the association between PONV and wound dehiscence.

We also observed a decreased time interval to first oral intake: approximately 24 (24–36) hours during preintervention compared to 17.5 (12–24) hours during postintervention (P < .001). Bisgaard and Kehlet demonstrated that tolerance to early oral nutrition is enhanced by a multimodal PONV prevention strategy.29 Oral intake will reduce catabolism and the usual postsurgical loss of lean body mass.

Our study did not address the financial implications of our PONV protocol. However, Hirsch30 has shown that PONV can result in additional cost-related consequences not only to the patient but also to the hospital.

In conclusion, even in a low-resource setting, where surgery, comorbidities, as well as anesthetic and surgical management are different compared to high-resource countries, structured PONV prevention protocols were validated and demonstrated to be significant, sustained benefits. Implementing a locally modified multimodal prevention regimen for PONV based on an Apfel score preoperative risk stratification is simple and easily implemented. Our protocol is particularly appropriate in settings where preferred treatment drugs are not always available and substitutions have to be made.

ACKNOWLEDGMENTS

Tania Rodriguez-Carpio, Leah Reichle, and Elizabeth Harvey, medical students from the University of Virginia, Charlottesville, VA, contributed to data collection, and we thank them for their invaluable efforts.

DISCLOSURES

Name: Jean de Dieu H. Tuyishime, MD, MMed.

Contribution: This author helped develop the study concept, design the study, collect and analyze the data, and write the manuscript, and read and accepted the final version of the manuscript.

Name: Joseph Niyitegeka, MD, MMed.

Contribution: This author helped design the study, participated in writing the manuscript, and read and accepted the final version of the manuscript.

Name: Adeyemi J. Olufolabi, MBBS, FRCA.

Contribution: This author helped design the study, assisted with protocol implementation, and read and accepted the final version of the manuscript.

Name: Samuel Powers, MA.

Contribution: This author helped with data analysis, participated in manuscript writing, and read and accepted the final version of the manuscript.

Name: Bhiken I. Naik, MBBCh, MSCR.

Contribution: This author helped with data analysis and rewriting the manuscript, and read and accepted the final version of the manuscript.

Name: Siny Tsang, PhD.

Contribution: This author had primary responsibility for statistical analysis, and read and accepted the final version of the manuscript.

Name: Marcel E. Durieux, MD, PhD.

Contribution: This author helped design the study, participated in writing the manuscript, and read and accepted the final version of the manuscript.

Name: Theogene Twagirumugabe, MD, MMed, FCCM, PhD.

Contribution: This author was involved in study design, data analysis, and writing the manuscript, and read and accepted the final version of the manuscript.

This manuscript was handled by: Angela Enright, MB, FRCPC.

REFERENCES

1. Kenny GN. Risk factors for postoperative nausea and vomiting. Anaesthesia. 1994;49(suppl):6–10.
2. Bruderer U, Fisler A, Steurer MP, Steurer M, Dullenkopf A. Post-discharge nausea and vomiting after total intravenous anaesthesia and standardised PONV prophylaxis for ambulatory surgery. Acta Anaesthesiol Scand. 2017;61:758–766.
3. Gan TJ, Meyer TA, Apfel CC,, et al.; Society for Ambulatory Anesthesia. Society for Ambulatory Anesthesia guidelines for the management of postoperative nausea and vomiting. Anesth Analg. 2007;105:1615–1628.
4. Chatterjee S, Rudra A, Sengupta S. Current concepts in the management of postoperative nausea and vomiting. Anesthesiol Res Pract. 2011;2011:748031.
5. Sigaut S, Merckx P, Peuch C, Necib S, Pingeon F, Mantz J. Does an educational strategy based on systematic preoperative assessment of simplified Apfel’s score decrease postoperative nausea and vomiting? Ann Fr Anesth Reanim. 2010;29:765–769.
6. Diemunsch P, Gan TJ, Philip BK,, et al.; Aprepitant-PONV Protocol 091 International Study Group. Single-dose aprepitant vs ondansetron for the prevention of postoperative nausea and vomiting: a randomized, double-blind phase III trial in patients undergoing open abdominal surgery. Br J Anaesth. 2007;99:202–211.
7. Pierre S, Benais H, Pouymayou J. Apfel’s simplified score may favourably predict the risk of postoperative nausea and vomiting. Can J Anaesth. 2002;49:237–242.
8. Smith HS, Smith EJ, Smith BR. Postoperative nausea and vomiting. Ann Palliat Med. 2012;1:94–102.
9. Gan TJ. Risk factors for postoperative nausea and vomiting. Anesth Analg. 2006;102:1884–1898.
10. Hambridge K. Assessing the risk of post-operative nausea and vomiting. Nurs Stand. 2013;27:35–43.
11. Gan TJ, Diemunsch P, Habib AS,, et al.; Society for Ambulatory Anesthesia. Consensus guidelines for the management of postoperative nausea and vomiting. Anesth Analg. 2014;118:85–113.
12. Macario A, Claybon L, Pergolizzi JV. Anesthesiologists’ practice patterns for treatment of postoperative nausea and vomiting in the ambulatory post anesthesia care unit. BMC Anesthesiol. 2006;6:6.
13. Chalya P, Mhwewa O, Mabula J. Postoperative nausea and vomiting at a tertiary care hospital in north-western Tanzania. Tanzan J Health Res. 2015;17:9.
14. Ahmed S, Lema G. Incidence and factors associated with postoperative nausea and vomiting among elective adult surgical patients at University of Gondar comprehensive specialized hospital, Northwest Ethiopia, 2019: a cross-sectional study. Int J Surg Open. 2020;22:57–61.
15. Apfel CC, Heidrich FM, Jukar-Rao S,, et al. Evidence-based analysis of risk factors for postoperative nausea and vomiting. Br J Anaesth. 2012;109:742–753.
16. Thomas JS, Maple IK, Norcross W, Muckler VC. Preoperative risk assessment to guide prophylaxis and reduce the incidence of postoperative nausea and vomiting. J Perianesth Nurs. 2019;34:74–85.
17. Brainard A, Gresham C. Prevention and treatment of motion sickness. Am Fam Physician. 2014;90:41–46.
18. Apfel CC, Korttila K, Abdalla M,, et al.; IMPACT Investigators. A factorial trial of six interventions for the prevention of postoperative nausea and vomiting. N Engl J Med. 2004;350:2441–2451.
19. OpenEpi. https://www.openepi.com/Menu/OE_Menu.htm.
20. Mascha EJ, Sessler DI. Segmented regression and difference-in-difference methods: assessing the impact of systemic changes in health care. Anesth Analg. 2019;129:618–633.
21. Core Team R. R: A language and environment for statistical computing. Vienna, Austria: Foundation for Statistical Computing. Available at https://www.R-project.org/.
22. SPSS Statistics for Windows. IBM Corp.
    23. Polati E, Verlato G, Finco G,, et al. Ondansetron versus metoclopramide in the treatment of postoperative nausea and vomiting. Anesth Analg. 1997;85:395–399.
    24. Rodseth RN, Gopalan PD, Cassimjee HM, Goga S. Reduced incidence of postoperative nausea and vomiting in black South Africans and its utility for a modified risk scoring system. Anesth Analg. 2010;110:1591–1594.
    25. Mayeur C, Robin E, Kipnis E,, et al. Impact of a prophylactic strategy on the incidence of nausea and vomiting after general surgery. Ann Fr Anesth Reanim. 2012;31:e53–e57.
    26. Domino KB, Anderson EA, Polissar NL, Posner KL. Comparative efficacy and safety of ondansetron, droperidol, and metoclopramide for preventing postoperative nausea and vomiting: a meta-analysis. Anesth Analg. 1999;88:1370–1379.
    27. Skledar SJ, Williams BA, Vallejo MC,, et al. Eliminating postoperative nausea and vomiting in outpatient surgery with multimodal strategies including low doses of nonsedating, off-patent antiemetics: is “zero tolerance” achievable? ScientificWorldJournal. 2007;7:959–977.
    28. Fortier J, Chung F, Su J. Unanticipated admission after ambulatory surgery–a prospective study. Can J Anaesth. 1998;45:612–619.
    29. Bisgaard T, Kehlet H. Early oral feeding after elective abdominal surgery–what are the issues? Nutrition. 2002;18:944–948.
    30. Hirsch J. Impact of postoperative nausea and vomiting in the surgical setting. Anaesthesia. 1994;49(suppl):30–33.

    Supplemental Digital Content

    Copyright © 2022 International Anesthesia Research Society