Journal of Clinical Anesthesia (2013) 25, 551–559

Original Contribution

Incidence and predictors of postdischarge nausea andvomiting in a 7-day population☆

Jan Odom-Forren PhD, RN, CPAN, FAAN (Assistant Professor)a,⁎,Leena Jalota MD (Visiting Research Associate)a,b,Debra K. Moser DNSc, RN, FAAN (Professor and Gill Endowed Chair)a,Terry A. Lennie PhD, RN, FAAN (Professor and Associate Dean)a,Lynne A. Hall DrPH, RN (Professor and Associate Dean of Research)a,Joseph Holtman MD, PhD (Professor of Anesthesiology)c,Vallire Hooper PhD, RN, CPAN, FAAN (Assistant Professor and Assistant Director)d,Christian C. Apfel PhD, MD (Associate Professor of Anesthesia)b

aCollege of Nursing, University of Kentucky, Lexington, KY 40536, USAbDepartment of Anesthesia and Perioperative Care, University of California San Francisco, San Francisco,CA, 94143–0648; Resident Physician, Department of Internal Medicine, Reading Health System,West Reading PA 19611, USAcDepartment of Anesthesiology, University of Kentucky, Lexington, KY 40536, USAdSchool of Nursing, Medical College of Georgia, Augusta, GA 30901, USA

Received 5 September 2012; revised 3 May 2013; accepted 11 May 2013

8

0h

Keywords:Ambulatory surgery;Emesis control;Ondansetron;Outpatient anesthesia;Postdischarge nauseaand vomiting

AbstractStudy Objective: 1) To quantify the incidence and severity of postdischarge nausea and vomiting (PDNV)for 7 days in adults undergoing outpatient surgeries with general anesthesia; 2) to evaluate whether a riskmodel previously developed for the first two postoperative days may be used to predict the patient’s risk ofPDNV for 7 days; and 3) to verify whether the same risk factors are applicable in the 3 to 7 day period.Design: Prospective study.Setting: Two university-affiliated centers.Patients: 248 adult (N18 years) surgical outpatients undergoing ambulatory surgical procedures withgeneral anesthesia between 2007 and 2008.Measurements: The incidence and severity of PDNV and a simplified risk score for PDNV was assessedprospectively from discharge up to 7 postoperative days.MainResults:The overall incidence of nauseawas 56.9% andof emesiswas 19.4%.The incidence of PDNVwas highest on the day of surgery (DOS), with PDNV of 44.8% and decreasing over time to 6.0% on day 7.Using the simplified risk score for PDNV the area under the receiver operating characteristic (ROC) curve

☆ Supported by Dr. Christian Apfel’s Perioperative Research Core, University of California, San Francisco.⁎ Correspondence: Jan Odom-Forren, PhD, RN, CPAN, FAAN, 535 CON, 760 Rose St., Lexington, KY 40536, USA. Tel.: +1 859 323 6292; fax: +1

57 323 1057.E-mail address: Jlforr2@uky.edu (J. Odom-Forren).

952-8180/$ – see front matter © 2013 Elsevier Inc. All rights reserved.ttp://dx.doi.org/10.1016/j.jclinane.2013.05.008

552 J. Odom-Forren et al.

was 0.766 (0.707, 0.825). A previous history of postoperative nausea and vomiting (PONV; OR 3.51, CI1.70 - 7.27), operating room time (odds ratio [OR] 2.19, 95% CI 1.34 - 3.60), use of ondansetron in thePostanesthesia Care Unit (PACU; OR 6.39, CI 1.65-24.79), and pain during days 3–7 (OR 1.67, CI 1.30 -2.14) were the strongest predictors of PDNV on days 3–7.Conclusions: PDNV affects a significant number of patients after ambulatory surgery, and our simplifiedPDNV score may be applied to a 7-day population. Pain appears to be a factor in late PDNV. It is possiblethat the presence of PDNV during days 3–7 has different origins from the PDNV that resolved over the first48 hours.© 2013 Elsevier Inc. All rights reserved.

1. Introduction

Over 34 million patients undergo ambulatory surgeryannually in the United States [1] and as many as 35% to 50%experience nausea and/or vomiting after discharge [2–6]. Withmore than 60% of all surgeries performed in the ambulatorysetting and millions of patients experiencing postdischargenausea and vomiting (PDNV) every year, it is imperative thatwestudy the incidence and severity of PDNV and identify a modelto predict those patients who aremost likely to suffer from it [7].The impact of PDNV requires that treatment of this complica-tion extend well beyond discharge. An effective predictivemodel would allow preplanning for the care of patients at risk.

Researchers have focused on patient risk for postoperativenausea and vomiting (PONV), nausea and/or vomiting thatoccurs within the first 24-hour period after inpatient surgery[2,8,9]. PONV has been well-described, risk factors have beendetermined, and it is far better studied than PDNV [8,9]. Littleresearch is available in which the incidence and severity ofPDNV are described. PDNV is an underreported condition thatmay affect quality of recovery, has the potential for morbidity,and impacts patient satisfaction and healthcare costs [10–16].

Patients have expressed their aversion to nausea andvomiting after surgery [17,18]. When asked to rankpostoperative and postanesthesia outcomes from the mostundesirable to least undesirable, patients rated vomiting asthe most undesirable outcome, ranking it as more undesirablethan pain or shivering [18]. Patients who had experiencednausea after surgery were willing to pay $73 (USD) out-of-pocket for a preventive drug, and those who experiencedvomiting were willing to pay $100 (USD) [19].

The purposes of this study were to: 1) quantify theincidence and severity of PDNV for 7 days in adultsundergoing outpatient surgeries with general anesthesia; 2)evaluate whether a risk model previously developed for thefirst two postoperative days may be used to predict thepatient’s risk of PDNV for 7 days; and 3) verify whetherthe same risk factors are at work in the 3–7 day period.This study was part of a multi-site study whose primaryobjective was the development of a simplified risk modelfor predicting those patients who were most likely to sufferfrom PDNV in the first 48 hours postdischarge [6]. Thisstudy differed by following a subsample of patients fromtwo sites included in the original 48-hour study over a 7-

day period to better describe the incidence and severity ofPDNV over a long time frame.

2. Materials and methods

Institutional Review Board approval for the descriptive,prospective 7-day study was obtained from the University ofKentucky and the Medical College of Georgia for the 7-daystudy. Informed consentwas obtained from all participants. Thisreport was based on 248 (of 260) patients from the twoambulatory surgery siteswho returned completed 7-day records.

Eligible patients were adults (N 18 yrs of age) undergoingan outpatient surgical procedurewith general anesthesia and anendotracheal tube or Laryngeal Mask Airway. Excluded fromthe studywere those patients whowere unable to communicatein English, individuals whose surgery resulted in a planned orunplanned inpatient stay, current pregnancy, persistent orrecurrent nausea and/or vomiting before anesthesia, andpatients who required regional anesthesia only. Eligibleparticipants were recruited consecutively either in preadmis-sion testing or in the preoperative unit on the day of surgery bythe primary investigator or trained research assistant.

2.1. Variables and measurements

Participants were asked to self-report demographiccharacteristics and medical and surgical history. Other data(eg, use of volatile anesthetics, type and dose of opioids,duration of anesthesia, type of surgery, antiemetics includinguse of glucocorticoids, postoperative type and dose ofopioids, experience of nausea and/or vomiting in the PACU,time and type of first drink and first food intake) wereobtained from the clinical record. Incidence and severity ofnausea were recorded during the PACU stay at admission,15, 30, 60, 120, and 240 minutes and at discharge; during theride home; and for 7 days following discharge. The incidenceof PONV or PDNV was measured as any number greaterthan 0 on the numerical rating scale, and severe nausea wasdefined as nausea of 7 or greater on the numerical ratingscale. Severity of nausea was measured using an 11-pointnumerical rating scale, where 0 represented “no nausea” and10 represented “worst nausea imaginable.”

553PDNV incidence and predictors

Vomiting was measured as the number of episodes thatoccurred at least one minute apart and severe vomiting as threeor more emetic episodes. Because there is a lower incidence ofretching in these patients [20,21], it was combinedwith vomitingfor this study. At the time of discharge, participants were given aself-report PatientDiary consisting of questions about nausea andvomiting for each day in the comingweek. The PDNVdatawerecollected from the clinical record and patient interview on days 1,2, and 7, and from the patient diary.

2.2. Procedure

At the time of consent, participants were informed that aninvestigatorwould call to ask questions about nausea, vomiting,and other symptoms that occurred after discharge onDays 1 and2, and again on day 7. Patients were instructed on use of thePatient Diary. The Patient Diary contained a space for patientsto record medication use each day. Participants were remindedwith a telephone call on Day 7 to mail the Patient Diary to theinvestigator in the self-addressed, stamped envelope provided.Patient medical treatment remained fully at the discretion of thetreating anesthesia provider and surgeon in order to reflect real-life conditions and current clinical practice in the U.S.

2.3. Sample size estimation

Power estimates were obtained using nQuery Advisor(Statistical Solutions, Saugus, MA, USA) [22]. With a samplesize of 120 and a PDNV prevalence of 30%, approximately 36participants were expected to have this adverse event; theremaining 84 would not. With these group sizes and an alphalevel of 0.05, the power of a two-sample t-test to detect asignificant group difference was approximately 84% if the ratioof the difference in means to the standard deviation was assmall as 0.6. Cohen considers a difference of this magnitude tobe slightly larger than a medium effect size [23].With a total of120 subjects and a significance level of 0.05, the power of thechi-square test of association to detect an odds ratio (OR) assmall as 4 was approximately 84%. More subjects wereavailable due to the requirements of the multi-site study, so thedata analyses are at least as powerful as determined prior to datacollection [24].

2.4. Statistical analysis

Data analyses were performed with SPSS (version 18.0software; SAS, Inc., Chicago, IL, USA). Descriptive statistics,including means and standard deviations or frequencydistributions, were used to describe the sample and summarizethe data. Bivariate analyses, including two-sample t-tests andchi-square tests of association, were used to identify factorsassociated with PDNV. One-way analysis of variance(ANOVA) with Tukey’s HSD post-hoc tests were used todetermine significant differences in means of nausea betweengroups. Alpha was set at b 0.05.

To validate the simplified PDNV score developed byApfel et al [6], the PDNV risk was calculated and correlatedwith the actual 7-day incidence by generating calibration andreceiver operating characteristic (ROC) curves. The actualincidence of PDNV was plotted against the mean of thepredicted incidence and compared using weighted linearregression analysis in this sample. The slope and intercept ofthe fitted regression line demonstrate whether the scoreaccurately predicts PDNV. The actual incidence of PDNV inthe overall study (n = 2170) was plotted against the mean ofthe predicted incidence and compared using weighted linearregression analysis. The area under the ROC curve was usedto estimate the discriminating power of the risk scores (AUC).An AUC of 0.5 means that the score cannot discriminatebetween patients with and without PDNV, while an AUC of 1denotes a perfect discrimination.

Finally, a logistic regression analysis was conducted toexamine relevant factors for PDNV for days 3 to 7.Significance was set at the P b 0.05 value. Tested predictorswere: patient-specific predictors (eg, female gender, age,nonsmoking for at least two weeks before surgery, history ofPONV, history of motion sickness, concomitant medica-tions, preexisting diseases); intraoperative predictors (eg, useof volatile anesthetics, type and dose of opioids, duration ofanesthesia, type of surgery, antiemetics including use ofglucocorticoids); and postoperative predictors (eg, postop-erative type and dose of opioids, experience of nausea and/orvomiting in the PACU, and postoperative pain).

3. Results

3.1. Incidence

The overall incidence of nausea for the study was 56.9%and emesis was 19.4%, with an overall incidence of PDNV of56.9%. In the PACU, the incidence of nausea was 22.6% and19.4% experienced vomiting. During the ride home afterdischarge, 34.7% of patients had nausea, with 8.1% of patientsexperiencing emesis. After arrival home on the day of surgery,nausea increased to the highest of any single day (44.8%) andemesis increased to 13.3%. The next day after surgery (day 1),the incidence of nausea and vomiting decreased to 25% and5.2%, respectively. Nausea decreased to 6.0%on day 7; emesiswas stable on days 6–7 at 0.8% (Fig. 1). There was asignificant difference in the presence of PDNV based ongender during the PACU stay, day of surgery, postoperativeday 1, and postoperative days 2–4. Overall, gender played asignificant role in the presence of PDNV until day 4 (Fig. 2).

3.2. Severity of nausea

Severe nausea (N 7 on scale of 0–10) affected 19.4% ofpatients on the day of surgery and continued to affect 2% ofpatients on day 7. Of those who were nauseous, 30%

Fig. 1 Incidence of nausea and/or vomiting in the PostanesthesiaCare Unit (PACU), and from the day of surgery (DOS) topostoperative day 7. Solid line=nausea; dotted line=vomiting.

554 J. Odom-Forren et al.

experienced severe nausea during the drive home, increasingto 34% of those with nausea during the day of surgery. Of the6.0% with nausea on day 7, a third (33%) continued to havesevere nausea.

3.3. Differences in demographic and clinicalcharacteristics

Demographic and clinical characteristics of patients whoexperienced and did not experience PDNV are shown inTable 1. The presence of PDNV was significantly higher forwomen than men. Younger patients were more likely thanolder patients to experience PDNV. Those with and withoutnausea did not differ on BMI, nor was the presence of nauseaassociated with educational level, ethnicity, or income.

A history of motion sickness, previous PONV, migraineheadaches, and pain on activity postdischarge were allsignificantly related to the presence of PDNV. The onlyopioid associated with PDNV was the use of fentanyl in thePACU. Patients who had a longer OR time were significantlymore likely to experience PDNV than those who had less ORtime. Patients who experienced nausea and vomiting in thePACU were significantly more likely to experience PDNV.

Fig. 2 Nausea mean by gender and day. Solid line=male, dottedline=female.

Nitrous oxide, use of a regional block as an adjunct toinhalation anesthesia, and ASA physical status were notassociated with PDNV.

3.4. Prediction of PDNV in a 7-day sample

We used the simplified PDNV scoring system developedby Apfel et al [8] to validate the effectiveness of predictingPDNV in a 7-day sample. The 5 factors in the scoring systemare female gender, age less than 50 years, history of PONV,opioids administered in the PACU, and nausea in the PACU.Using the simplified risk score for PDNV, in which eachfactor counted for one point, led to an ROC-AUC of 0.766(0.707, 0.825). According to this simplified PDNV score,when 0, 1, 2, 3, 4, or 5 of these risk factors were present, theassociated PDNV incidences were 40%, 19.4%, 39.7%,65%, 88.4%, and 93.8%. In a multivariate logistic regressionof preoperative, intraoperative, and postoperative factors;female gender; age less than 50 years; history of PONV; andnausea in the PACU were the strongest predictors (Table 2).

Since the Apfel PDNV predictor scoring system wasdeveloped from the larger dataset of this study based only ondata from 48 hours postdischarge, we wanted to determinehow it would predict later PDNV (days 3–7). When appliedto those who had PDNV on days 3–7, the ROC-AUC was0.668 (0.586, 0.750). The calibration plot of the predictedand actual incidences of PDNV from the original dataset (n =2170) resulted in a calibration line with a slope of 1.04 and anintercept of 0.006. The calibration plot of the predicted andactual incidences of PDNV for days 0–7 in this subset ofpatients (n = 248) resulted in a calibration line with a slope of0.87 and an intercept of 0.10 (Figs. 3, 4, 5).

3.5. Risk factors for days 3–7

To determine why the AUC was lower for patients whoexperienced PDNV during days 3–7, a logistic regressionwas done. A stepwise forward logistic multivariableregression analysis of patient-related anesthesia and intrao-perative and postoperative factors showed that previoushistory of PONV (OR 3.51, CI [1.70 - 7.27], OR time (OR2.19, 95% CI 1.34 - 3.60), use of ondansetron (OR 6.39, CI1.65 - 24.79), and pain during days 3–7 (OR 1.67, CI 1.30 -2.14) were the strongest predictors of PDNVwithin days 3–7.

In univariate analysis, endoscopy had a significant effecton PDNV with OR (0.53, 95% CI 0.20 - 0.53) (Table 2).

4. Discussion

Our findings substantiate the continuing problem ofnausea and vomiting after outpatient surgery, a problem thatis unrecognized and undertreated [14,25]. The overall PDNincidence of 56.9%, PDV incidence of 19.4%, and overallPDNV incidence of 56.9% occurs at the top end of the

Table 1 Demographic and clinical characteristics of patients with and without postdischarge nausea and vomiting (PDNV; n = 248)

PDNV(n = 141)

No PNDV(n = 107)

P-value

DemographicsAge (yrs) (mean±SD) 48.86 ± 14.29 56.84 ± 14.04 b 0.001BMI (kg/m2) (mean±SD) 28.77 ± 6.77 29.36 ± 6.58 0.494

N (%) N (%)Gender women

men114 (46)27 (10.9)

44 (17.7)63 (25.4)

b 0.001

Ethnicity African-AmericanCaucasianother

12 (4.8)120 (48.4)9 (3.6)

9 (3.6)96 (38.7)2 (0.8)

0.321

Clinical characteristicssmoker no

yes117 (47.2)24 (9.7)

94 (37.9)13 (5.2)

0.286

previousPONV

noyes

63 (25.4)78 (31.5)

77 (31)30 (12.1)

b 0.0001

previousmotionsickness

noyes

93 (37.5)48 (19.4)

83 (33.5)24 (9.7)

b 0.046

migraineheadachehistory

noyes

99 (39.9)42 (16.9)

91 (36.7)16 (6.5)

b 0.006

Procedures generalbreastcholecystectomycystoscopyD&CENTGYNherniaknee arthroscopyorthopedicprostateupper extremity

18 (7.3)19 (7.7))25 (10.1)4 (1.6)12 (4.8)11 (4.4)14 (5.6)7 (2.8)20 (8.1)2 (0.8)1 (.4)8 (3.2)

19 (7.7)8 (3.2)9 (3.6)6 (2.4)11 (4.4)4 (1.6)8 (3.2)12 (4.8)18 (7.3)4 (1.6)2 (.8)6 (2.4)

0.152

Surgical approach conventionalarthroscopicendoscopiclaparoscopic

53 (21.4)27 (10.9)19 (7.7)42 (16.9)

38 (15.3)22 (8.9)21 (8.5)26 (10.5)

0.526

ASA physical status 123

26 (10.5)101 (40.7)14 (5.6)

16 (6.5)71 (28.6)20 (8.1)

0.13

OR time (hrs) (mean±SD) 1.47 ± 0.69 1.27 ± 0.62 0.023Pain after discharge(0–10) a

(mean±SD) 3.12 ± 2.15 2.40 ± 2.19 0.01

Regional block noyes

54 (21.8)87 (35.1)

40 (16.1)67 (27)

0.883

Nitrous oxide noyes

133 (53.6)8 (3.2)

102 (41.1)5 (2.0)

0.726

BMI=body mass index, PONV=postoperative nausea and vomiting, D&C=dilatation and curettage, ENT=ear, nose, and throat, GYN=gynecological, OR=operating room.

a Numerical rating scale (0=“none at all” and 10=“worst possible pain”).

555PDNV incidence and predictors

published range of 30% - 55% [3–5,14,26]. In this study,PDN was a surrogate for PDNV because all patients whoexperienced PDV also had experienced PDN. It is possiblethat this subgroup of two centers had a higher percentage ofPDNV than the larger study group (56.9% vs 37.1%)

because of the significant difference in certain risk factors; ie,a significantly higher number of women (P b 0.03) andlaparoscopic surgeries (P b 0.0001) [6]. Several demograph-ic and clinical characteristics were significantly related to the7-day PDNV in this study, including female gender, a history

Table 2 Incidences with unadjusted and adjusted odds ratios (ORs) and 95% confidence intervals (CIs) for variables potentiallyinfluencing postdischarge nausea and/or vomiting (PDNV) days 3–7 and days 0–7

Variables Unadjusted OR(95% CIs):days 3 - 7

Adjusted OR(95% CIs):days 3 - 7

P-value Adjusted OR(95% CIs) ofmain predictors:days 0 - 7

P-value

Patient characteristic gender (W/M) 1.64 0.80 - 3.33 1.96 1.03 - 3.72 0.039age, b 50 / ≥ 50 yrs 2.01 1.10 - 3.67 3.03 1.69 - 5.44 b 0.001PONV history, yes/no 2.76 1.50 - 5.11 3.51 1.70 - 7.27 0.001 2.78 1.54 - 5.03 0.001nonsmoking, yes/no 1.14 0.50 - 2.57motion sickness history, yes/no 1.23 0.65 - 2.34fentanyl≥ 125 / b 125 μg 1.01 0.99 - 1.035-HT3 (ondansetron)antagonists, yes/no

1.11 0.54 - 2.28

glucocorticoids, yes/no 1.76 0.94 - 3.27Surgical approach arthroscopy 0.87 0.38 - 1.97

endoscopy 0.53 0.20 - 0.53laparoscopy 0.91 0.44 - 1.88

Type of surgery breast 1.08 0.347 - 3.413cholecystectomy 1.120 0.384 - 3.265cystoscopy 1.33 0.284 - 6.263D&C 0.296 0.058 - 1.517ENT 1.414 0.387 - 5.172GYN 0.691 0.185 - 2.584hernia 0.778 0.206 - 2.936knee 0.830 0.281 - 2.450orthopedic 1.556 0.243 - 9.954prostrate 0.999 0.000 - 0.000upper extremity 0.848 0.193 - 3.732OR time 2.244 1.49 - 3.39 2.19 1.34 - 3.59 0.002surgical procedure 0.97 0.89 - 1.06

Postoperative opioids in PACU (yes/no) 1.016 0.991 - 1.04nausea in PACU (yes/no) 1.12 0.99 - 1.26 1.34 1.12 - 1.61 0.002vomiting in PACU (yes/no) 1.48 0.63 - 3.47pain in the PACU 1.12 1.01 - 1.24pain postdischarge 1.69 1.37 - 2.09 1.67 1.30 - 2.14 b 0.001rescue antiemetic inPACU (yes/no)

5.81 2.10 - 16.08 6.39 1.65-24.79 0.007

Postdischarge morphine equivalentsafter discharge

1.02 1.00 - 1.03

PONV=postoperative nausea and vomiting, PDNV=postdischarge nausea and/or vomiting, PACU=Postanesthesia Care Unit, D&C=dilatation and curettage,ENT=ear, nose & throat, GYN=gynecological.

556 J. Odom-Forren et al.

of PONV or motion sickness, and OR time. Somecharacteristics usually associated with PONV were notassociated in this sample, eg, nonsmoking status and use ofopioids (morphine equivalents). We found that severity ofPDNV peaked on the day of surgery and gradually decreasedover the next 7 days.

We attempted to provide a real-world incidence of PDNVin a sample of patients who received general anesthesia foroutpatient surgery by not excluding by procedure, anesthesiaregimen, or risk. The only high-risk inclusion criterion wasuse of inhalation anesthesia. This study described theincidence and severity of PDNV over a 7-day period in asample of ambulatory surgery patients. Without focusingspecifically on patients at demographic or procedural high-

risk, the overall incidence of PDNV was 56.9%. Other workreported an overall PDNV incidence of 35.7% over a 5-dayperiod [3]. Incidences of PDNV across studies are difficult tocompare because of the varying patient samples, eg, high-risk patients, specific procedures performed, or using womenonly. Many of the studies reported PDNV as a secondaryoutcome or assessed PDNV only for the first 24 hoursinstead of the 7 days postdischarge [14].

The incidences of PDN and PDV during the ride homehave seldom been reported in the few studies available onPDNV [28,27], although Ernst and Thwaites [28] reported20% with nausea and 4% with vomiting during the ridehome. In our study, 34.7% of patients experienced PDNduring the ride home and 8.1% experienced PDV. The

Fig. 3 Receiver operating characteristic (ROC) curve for days 0–7 based on the Apfel PDNV Score. Dotted line=reference line; solidline=receiver operating characteristic (ROC) curve.

ig. 4 Calibration plot of the predicted and actual incidences ofostdischarge nausea and vomiting (PDNV) with 95% CIs in (A)alibration based on the Original Model and (B) Calibration basedn the model derived from this dataset, n = 248. The circle area ofe data points is proportional to the sample size of each risklassification group.

557PDNV incidence and predictors

incidence of PDN and PDV peaks after arrival home on theday of surgery to 44.8% and 13.3%, respectively. After theday of surgery, the incidences of PDN and PDV decreasegradually until day 7 when the overall incidence is 6.0% fornausea and 0.8% for emesis. The results from this study aresimilar to those of Philip et al1, who reported an incidence ofPDN of 46% in high-risk patients during the first 24 hoursafter surgery decreasing to 8% on day 7, and an incidence ofPDV of 12% during the first 24 hours decreasing to 1% onday 7. It is significant that even a week later in this studysome patients were still experiencing the negative symptomsof nausea and vomiting, symptoms described by patients asworse than pain. The experience of PDNV may be verydistressing to the patient in terms of resuming normalfunction [17,18].

Apfel et al [8] determined that female gender, nonsmok-ing status, history of PONV or motion sickness, and use ofperioperative opioids were strong predictors for PONV.Using the Apfel simplified PONV risk assessment tool, therisk of PONV increased from 10% for no risk factors to 79%if all 4 factors were present. In a study of predictors ofPDNV, Apfel et al [6] determined that female gender andhistory of PONV were strong predictors for PDNV for 48hours postdischarge. Patients in our study over a 7-dayperiod who experienced PDNV were also more likely to bewomen and to have a history of PONV.

Interestingly, nonsmoking status, strongly related toPONV [8,29], was not related to PDNV in our sample.

1 Philip BK, Chen YT, Gan TJ, Ma L, Hu HX. Post-operative nausea/vomiting after high-risk ambulatory surgeries [Abstract]. Anesthesiology2007;107:A40.

FpCothc

Neither nitrous oxide nor the use of inhalation agents wererelated to PDNV, contrary to the findings of other PONVresearch [29–33]. This was more than likely due to theshort-acting effects of the volatile anesthetics [31]. In onestudy, patients who experienced PONV in the PACU wereunlikely to experience PDNV [3]. Wright et al [34] foundthat patients who experienced PONV in the PACU weremore likely to have PDNV. We found that patients whoexperienced PONV in the PACU were significantly morelikely to experience PDNV. The use of opioids (morphineequivalents) postdischarge also was significantly related toPDNV. This finding was similar to other studies, in whichresearchers found that postoperative opioids given in thePACU were linked to PONV [8,29]. Patients with higherpain scores had higher nausea scores, likely due to higherconsumption of opioids.

Fig. 5 Receiver operating characteristic (ROC) curve for days 3–7 based on the Apfel PDNV score. Dotted line=reference line; solidline=ROC curve.

558 J. Odom-Forren et al.

Apfel et al [6] recently identified predictors for asimplified assessment tool for early PDNV (48 hrs aftersurgery). The 5 statistically significant independent riskfactors for PDNV for the first 48 hours were female gender,age less than 50 years, history of PONV, opioidsadministered in the PACU, and nausea in the PACU [6].We used this simplified scoring system and determined thatthe ROC-AUC in this subsample of 248 patients was 0.77.To determine any differences in patients experiencing laterPDNV during days 3–7, the AUC was calculated to be 0.68.It is possible that the presence of PDNV later postdischargehas different causes from the PDNV that has resolved overthe first 48 hours. As the risk score was less predictive fordays 3–7, we conducted a logistic regression that resulted inpostdischarge pain as a strong predictor for PDNV on days3–7. Pain was significantly related to PDNV every day for 7days. Interestingly, the amount of opioids consumed was nota significant factor. Opioids given in the PACU or thepresence of PONV in the PACU were not significantpredictors of late PDNV (days 3–7). Opioids given in thePACU may not be significant for days 3–7 since the half-lifeof opioids is only a few hours. If we look at daily opioidconsumption, there may be an acclimation that is occurring,or it is possible that since there was a lower incidence ofpatients consuming opioids on days 3–7 than in 48 hours, wedo not have the power to detect its effect.

Administration of ondansetron in the PACU predictedlater postdischarge nausea. It is likely that the use ofondansetron is a surrogate for nausea in the PACU, since allpatients who received ondansetron in the PACU presentedwith nausea. History of PONV is a strong predictor of PONV

and both early and late PDNV. OR time was a significantpredictor. It is possible that types of surgeries requiringgreater amounts of time for surgery are more complex andtherefore more likely to contribute to a longer recovery andare associated with pain and nausea. It is possible that agewas not in our final model for late PDNV because itcorrelated with pain, ie, the younger the patient, the higherthe pain score for those with late PDNV.

For days 0–2, our findings were similar to the largermulti-site study with the exception of one predictor (opioidsin the PACU), which may be attributed to differencesbetween the larger sample and our subsample. However,from days 3 to 7, we found different significant predictorsfrom Apfel et al: OR time, postoperative pain with activity,ondansetron in the PACU, and nausea in the PACU. It ispossible that the presence of nausea later postdischarge hasdifferent causes from the PDNV that has resolved over thefirst 48 hours. Postoperative pain is a very important factorand has not been given proper consideration in discussions ofPDNV. Patients who go home and have pain with activity areforced to consume more pain medications, which then resultsin more nausea and vomiting postdischarge, a verydistressing situation for the patient. Our study points to theneed for a larger study to further examine the relationship ofPDNV and pain.

This study was conducted in a sample of patients fromtwo centers in the U.S. It differed in some aspects from thesample of 1910 patients who were followed only for48 hours. It is possible that patients were treated differentlyin the two centers preoperatively and intraoperatively, whichmay have contributed to the differences in samples; however,patients came from two different settings: a university settingin an academic center and the other a not-for-profit privatehospital. The smaller subsample (n = 248), however, waspowered to detect significant group differences betweenthose with PDNV and those without.

While it would have been better to estimate the ROCcurve for days 0-7 without using the data from this study,the redundancy is likely minimal since our data compriseless than 15% of the overall data (n = 2,170) used to createthe curve.

In conclusion, pain seems to be a factor in late PDNV.The Apfel PDNV scoring system worked well in predictingPDNV in ambulatory surgery patients over a 7-day period.Patients at high risk for PDNV should be given prophylactic,long-acting antiemetics before discharge.

Acknowledgments

Nausea Study Group: Mary Kay Rayens, PhD; KurtViele, PhD; Apfel Perioperative Group, UCSF Departmentof Anesthesia: Oliver Radke, PhD, MD; Zohn Centimole,MSN, CRNA; Melissa Thomas, BSN, RN; Susan Westneat,MS; Rachel Whelan, BS; Carrell Rush, MPH.

559PDNV incidence and predictors

References

[1] Cullen KA, Hall MJ, Golosinskiy A. Ambulatory Surgery in theUnited States, 2006. Natl Health Stat Report 2009;11:1-25.

[2] American Society of PeriAnesthesia Nurses PONV/PDNV StrategicWork Team. ASPAN'S evidence-based clinical practice guideline forthe prevention and/or management of PONV/PDNV. J Perianesth Nurs2006;21:230-50.

[3] Carroll NV, Miederhoff P, Cox FM, Hirsch JD. Postoperative nauseaand vomiting after discharge from outpatient surgery centers. AnesthAnalg 1995;80:903-9.

[4] Gupta A, Wu CL, Elkassabany N, Krug CE, Parker SD, Fleisher LA.Does the routine prophylactic use of antiemetics affect the incidenceof postdischarge nausea and vomiting following ambulatory surgery?A systematic review of randomized controlled trials. Anesthesiology2003;99:488-95.

[5] Odom-Forren J, Moser DK. Postdischarge nausea and vomiting:a review of the literature. Ambul Surg 2005;12:99-105.

[6] Apfel CC, Philip B, Cakmakkaya OS, et al. Who is at risk forpostdischarge nausea and vomiting after ambulatory surgery?Anesthesiology 2012;117:475-86.

[7] Joshi GP, Twersky RS. Fast tracking in ambulatory surgery. AmbulSurg 2000;8:185-90.

[8] Apfel CC, Läärä E, Koivuranta M, Greim CA, Roewer N. A simplifiedrisk score for predicting postoperative nausea and vomiting:conclusions from cross-validations between two centers. Anesthesiol-ogy 1999;91:693-700.

[9] Glass PS, White PF. Practice guidelines for the management ofpostoperative nausea and vomiting: past, present, and future. AnesthAnalg 2007;105:1528-9.

[10] Candiotti KA, Kovac AL, Melson TI, Clerici G, Joo Gan T;Palonosetron 04–06 Study Group. A randomized, double-blindstudy to evaluate the efficacy and safety of three different doses ofpalonosetron versus placebo for preventing postoperative nausea andvomiting. Anesth Analg 2008;107:445-51.

[11] Kapur PA. The big “little problem". Anesth Analg 1991;73:243-5.[12] White PF, O'Hara JF, Roberson CR, Wender RH, Candiotti KA;

POST-OP Study Group. The impact of current antiemetic practices onpatient outcomes: a prospective study on high-risk patients. AnesthAnalg 2008;107:452-8.

[13] Odom-Forren J, Fetzer SJ, Moser DK. Evidence-based interventionsfor post discharge nausea and vomiting: a review of the literature.J Perianesth Nurs 2006;21:411-30.

[14] Pan PH, Lee SC, Harris LC. Antiemetic prophylaxis for postdischargenausea and vomiting and impact on functional quality of living duringrecovery in patients with high emetic risks: a prospective, randomized,double-blind comparison of two prophylactic antiemetic regimens.Anesth Analg 2008;107:429-38.

[15] Tramèr MR. Strategies for postoperative nausea and vomiting. BestPract Res Clin Anaesthesiol 2004;18:693-701.

[16] Kovac AL, Eberhart LH, Kotarski J, Clerici G, Apfel CC;Palonosetron 04–07 Study Group. A randomized, double-blindstudy to evaluate the efficacy and safety of three different doses of

palonosetron versus placebo in preventing postoperative nausea andvomiting over a 72-hour period. Anesth Analg 2008;107:439-44.

[17] Lee A, Gin T, Lau AS, Ng FF. A comparison of patients' and healthcare professionals' preferences for symptoms during immediatepostoperative recovery and the management of postoperative nauseaand vomiting. Anesth Analg 2005;100:87-93.

[18] Macario A, Weinger M, Carney S, Kim A. Which clinical anesthesiaoutcomes are important to avoid? The perspective of patients. AnesthAnalg 1999;89:652-8.

[19] Gan T, Sloan F, Dear Gde L, El-Moalem HE, Lubarsky DA. Howmuch are patients willing to pay to avoid postoperative nausea andvomiting? Anesth Analg 2001;92:393-400.

[20] Fetzer SJ, Hand MC, Bouchard PA, Smith H, Jenkins MB. Evaluationof the Rhodes Index of Nausea and Vomiting for ambulatory surgerypatients. J Adv Nurs 2004;47:74-80.

[21] Ming JL, Kuo BI, Lin JG, Lin LC. The efficacy of acupressure toprevent nausea and vomiting in post-operative patients. J Adv Nurs2002;39:343-51.

[22] Elashof J. nQuery Advisor. Saugus (MA): Statistical Solutions;1995–2005.

[23] Cohen J. Statistical Power Analysis for the Behavioral Sciences.Hillsdale (NJ): Lawrence Erlbaum Associates; 1988. p. 40.

[24] Hoenig JM, Heisey DM. The abuse of power: the pervasive fallacy ofpower calculations for data analysis. Am Statistician 2001;55:19-24.

[25] Fetzer SJ, Hand MA, Bouchard PA, Smith HB, Jenkins MB. Self-careactivities for postdischarge nausea and vomiting. J Perianesth Nurs2005;20:249-54.

[26] Wu CL, Berenholtz SM, Pronovost PJ, Fleisher LA. Systematic reviewand analysis of postdischarge symptoms after outpatient surgery.Anesthesiology 2002;96:994-1003.

[27] Waterman H, Leatherbarrow B, Slater R, Waterman C. Post-operativepain, nausea and vomiting: qualitative perspectives from telephoneinterviews. J Adv Nurs 1999;29:690-6.

[28] Ernst EM, Thwaites RM. Incidence and impact of pain, nausea andvomiting following discharge from a day surgery unit. J Appl Ther1997;1:257-63.

[29] Gan TJ. Risk factors for postoperative nausea and vomiting. AnesthAnalg 2006;102:1884-98.

[30] Apfel CC, Korttila K, Abdalla M, et al; IMPACT Investigators. Afactorial trial of six interventions for the prevention of postoperativenausea and vomiting. N Engl J Med 2004;350:2441-51.

[31] Apfel CC, Kranke P, Katz MH, et al. Volatile anaesthetics may be themain cause of early but not delayed postoperative vomiting: arandomized controlled trial of factorial design. Br J Anaesth 2002;88:659-68.

[32] Habib AS, White WD, Eubanks S, Pappas TN, Gan TJ. A randomizedcomparison of a multimodal management strategy versus combinationantiemetics for the prevention of postoperative nausea and vomiting.Anesth Analg 2004;99:77-81.

[33] Tramèr M, Moore A, McQuay H. Omitting nitrous oxide in generalanaesthesia: meta-analysis of intraoperative awareness and postoperativeemesis in randomized controlled trials. Br J Anaesth 1996;76:186-93.

[34] Wright CD, Jilka J, Gentry WB. Efficacy of promethazine suppositoriesdispensed to outpatient surgical patients. Yale J BiolMed 1998;71:391-5.