RANDOMIZED CONTROLLED TRIAL
Efficacy of a Single Preoperative Dexamethasone Dose to Prevent Nausea and Vomiting After Thyroidectomy (the tPONV Study) A Randomized, Double-blind, Placebo-controlled Clinical Trial Ignazio Tarantino, MD, MSc, Rene Warschkow, MD,yz Ulrich Beutner, PhD,z Walter Kolb, MD,z Andreas Lu¨thi, MD,§ Cornelia Lu¨thi, BScN,z Bruno M. Schmied, MD,z and Thomas Clerici, MDz
Objective: Does dexamethasone given before thyroidectomy reduce postoperative nausea and vomiting (PONV) in a randomized controlled trial? Background: PONV is an unsettling problem that commonly occurs in patients after thyroidectomy. Various preventive measures have been studied; however, many of these studies have been criticized for their biases (eg, use of opioids, sex selection) or were even retracted. Methods: This single-institution, randomized, double-blind, placebo-controlled, superiority study was performed between January 1, 2011, and May 30, 2013. Patients undergoing thyroidectomy for benign disease were allocated by a block randomized list to receive a preoperative single dose of dexamethasone (8 mg) or placebo. Patients and staff were blinded to the treatment assignment. The primary endpoint was the incidence of PONV assessed at 4, 8, 16, 24, 32, and 48 hours after surgery. To observe an incidence reduction of 50%, a total of 152 patients were required for the study. Results: The total incidence of PONV was reported in 65 of 152 patients (43%; 95% confidence interval [CI], 35–51). In the intention-to-treat analysis, PONV occurred in 22 of 76 patients (29%; 95% CI, 20–40) in the treatment arm and in 43 of 76 patients (57%; 95% CI, 45–67) in the control arm (P ¼ 0.001; odds ratio ¼ 0.31; 95% CI, 0.16–0.61; absolute risk reduction ¼ 28%; 95% CI, 12–42). The number needed to treat was 4. No severe dexamethasone-related adverse events were observed during the study. Conclusions: A single dose of preoperative dexamethasone administration is an effective, safe, and economical measure to reduce PONV incidence after thyroidectomy. Keywords: dexamethasone, PONV, postoperative nausea and vomiting, thyroid, thyroidectomy
(Ann Surg 2015;262:934–940)
A
quarter of surgical patients suffer from postoperative nausea and vomiting (PONV) after general anesthesia.1 The incidence of PONV is higher after thyroidectomy, reaching up to 80%.2 PONV affects patient comfort and health care costs by significantly prolonging hospital stays3 and may lead to serious postsurgical complications such as bleeding, one of the most feared complications after thyroidectomy.4 A study comparing the efficacy of 6 established antiemetic strategies after elective surgery indicated that From the Department of General, Visceral and Transplant Surgery, University of Heidelberg, Heidelberg, Germany; yInstitute of Medical Biometry and Informatics, University of Heidelberg, Heidelberg, Germany; zDepartment of Surgery, Kantonsspital St. Gallen, St. Gallen, Switzerland; and §Department of Anesthesiology, Kantonsspital St. Gallen, St. Gallen, Switzerland. Disclosure: Supported by a grant from the Clinical Trial Unit of the Kantonsspital St. Gallen. The authors declare no conflicts of interest. Reprints: Ignazio Tarantino, MD, MSc, Department of General, Abdominal and Transplant Surgery, University of Heidelberg, Im Neuenheimer Feld 110, 69120 Heidelberg, Germany. E-mail: [email protected]. Copyright ß 2015 Wolters Kluwer Health, Inc. All rights reserved. ISSN: 0003-4932/14/26105-0821 DOI: 10.1097/SLA.0000000000001112
934 | www.annalsofsurgery.com
dexamethasone may be a reasonable first-line treatment option to prevent PONV.5 A decade after the publication of this large study, dexamethasone is still not well established for the prophylaxis and treatment of PONV. Specifically, there is limited evidence for the efficacy of dexamethasone in preventing PONV after thyroidectomy (tPONV). The few studies published to date did not control for confounding factors that may influence PONV, such as postoperative opioid administration, a common cause of nausea and vomiting,6 or the use of various anesthetics within the study arms known to differentially provoke PONV.5 The inherent bias of previous studies, along with concerns about increased infection rates, may explain why dexamethasone has not been established as the standard treatment option for PONV prevention. Recently, our group demonstrated that a bilateral superficial cervical plexus block during thyroid surgery under general anesthesia significantly reduced pain. Of even greater clinical importance, this block prevented the need for postoperative opioids.7,8 Therefore, patients undergoing thyroidectomy along with a bilateral superficial cervical plexus block are an ideal, homogenous cohort for the investigation of the efficacy of dexamethasone for PONV prevention. These patients exhibit a high incidence of PONV and seldom require opioids. Combined with the highly standardized anesthesia protocol at our institution, this setting allows all known biases to be controlled.
METHODS Study Design and Participants This study was a single-institution, randomized, double-blind, placebo-controlled, superiority trial with 2 parallel arms comparing a single preoperative dose of 8 mg of dexamethasone (treatment arm) with saline (placebo arm). Two-dose finding studies have shown that 8 mg of dexamethasone is the minimum effective dose for the reduction of PONV.9,10 Patients older than 18 years scheduled for a primary partial or total thyroidectomy due to benign disease at our institution (Kantonsspital St. Gallen, Department of Surgery) were recruited between January 1, 2011, and May 30, 2013. The exclusion criteria included a history of adverse reactions to dexamethasone, polypropylene glycol, or disodium-EDTA (medication ingredients); patients suffering from chronic pain; the necessity for sternotomy; inability to administer a cervical block; diabetes mellitus type 1 or 2 requiring medication; pregnancy (tested in all women capable of childbearing); glaucoma; administration of antiemetic therapy within 48 hours before the surgery; acute pain requiring analgesics other than acetaminophen or more than a single dose of nonsteroidal antiinflammatory drugs within 48 hours before the surgery; use of antipsychotic drugs; noncompliance; proven malignancy, necessity of lateral neck dissection; and reoperations. All participants provided written informed consent before study participation. The study protocol was approved by the Ethics Committee of the Canton St. Gallen (EKSG 10/082) and by Swissmedic (2011DR3005) before Annals of Surgery Volume 262, Number 6, December 2015
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
Annals of Surgery Volume 262, Number 6, December 2015
enrolling any patients in this trial. The study protocol was previously published (http://www.biomedcentral.com/1471–2253/13/ 19),11 and the trial is registered with ClinicalTrials.gov: NCT01189292.
Randomization and Masking Patients were allocated to the study arms in a 1:1 ratio based on a predefined block randomization list with blocks randomly sized between 4 and 10. The randomization list was generated by the data manager (U.B.), with the hard copy stored in a secure location and the electronic version stored within the trial management software in a password-protected, encrypted fashion. During the study, only the data manager had access to the randomization list. Two to 3 days before surgery, the data manager prepared a computer-generated study medication request form based on the patient’s study number. The request contained the patient’s name and the treatment allocation. The form was placed in an opaque, sealed, and signed envelope and was directly delivered to the pharmacy by study personnel. Study medication was prepared according to the request form by the hospital pharmacy. Medication was delivered in a syringe labeled with the patient’s name and study number and containing either 2 mL of saline (placebo) or 2 mL of saline with 8 mg of dexamethasone (Mephameson; Mepha Pharma, Basel, Switzerland). Syringes containing saline or dexamethasone were indistinguishable. The patients and all study personnel except for the data manager were blinded to the treatment allocation.
Procedures The anesthesiologist administered the study medication intravenously 30 to 60 minutes before skin incision. All patients underwent standardized anesthesia, thyroid surgery, and routine postoperative care. During the surgery, recurrent laryngeal nerve monitoring was performed. A bilateral superficial cervical plexus block was performed just before the skin incision as previously described.8,11 The standardized anesthetic protocol consisted of propofol (Disoprivan 1%; AstraZeneca, London, United Kingdom) for hypnosis, fentanyl (Sintenyl; Sintetica, Mendrisio, Switzerland) and remifentanil (Ultiva; GlaxoSmithKline, Mu¨nchenbuchsee, Switzerland) for analgesia, and rocuronium (Esmeron; MSD, Lucerne, Switzerland) for muscle relaxation (for further details, see protocol).11 The postoperative first-line therapy for PONV was intravenous droperidol (Droperidol; Sintetica), with a maximum dose of 3 0.5 mg/24 h if the patient’s blood pressure exceeded 120 mm Hg. The second-line therapy was 4 mg of ondansetron given intravenously (Zofran; Glaxo SmithKline). All surgical procedures were performed by 1 of 2 surgeons (T.C. and W.K.) or under their close supervision during teaching surgical procedures. The postoperative first-line therapy for PONV was intravenous droperidol (Droperidol; Sintetica), with a maximum dose of 3 0.5 mg/24 h if the patient’s blood pressure exceeded 120 mm Hg. The second-line therapy was 4 mg of ondansetron given intravenously (Zofran; Glaxo SmithKline). All surgical procedures were performed by 1 of 2 surgeons (T.C. and W.K.) or under their close supervision during teaching surgical procedures. No analgesia was routinely applied. The first-line reserve for pain relief upon patient demand was 1 g of oral paracetamol (Dafalgan; Bristol-Myers Squibb, Baar, Switzerland), with a maximum dose of 3 g/24 h. The second-line reserve was 1 g of metamizole (Novalgin; Sanofi-Aventis, Meyrin, Switzerland), with a maximum dose of 4 g/24 h. The third-line reserve was intravenous morphine (Morphin HCI; Sintetica); however, this administration was considered a protocol violation, and patients receiving morphine were excluded from per protocol (PP) analysis. ß
2015 Wolters Kluwer Health, Inc. All rights reserved.
Dexamethasone for PONV Prevention
Outcomes The primary endpoint was the incidence of PONV assessed at 4, 8, 16, 24, 32, and 48 hours after surgery (unless the patient was released earlier). If nausea or vomiting was documented at any of these time points, it was considered an incidence of PONV. The nursing staff on the surgical wards assessed PONV. Before accrual of the first patient, nursing staff was instructed and trained about the trial, its objectives, and the outcome assessment. In addition, the assessment was monitored daily by a designated study nurse who also confirmed the outcome by patient interviews. The secondary endpoints were the severity of PONV, pain, length of stay (hours after intervention), wound healing (at discharge and after 6 weeks), the amount of required anesthetic medication, and postoperative inhospital morbidity (wound infection, reoperation, bleeding, prolonged intubation, laryngeal nerve lesion, urinary retention, urinary tract infection, pulmonary infection, gastrointestinal hemorrhage, or newly occurring insulin dependence in diabetic patients). The severity of PONV was rated from 0 to 3 (0 for no nausea; 1 for mild nausea, defined as nausea requiring a single administration of an antiemetic drug; 2 for severe nausea, defined as nausea requiring the repeated administration of antiemetic drugs; and 3 for vomiting). Pain was evaluated at rest and after physical stress (rotating the neck 90 degrees to each side) using the Verbal Rating Scale ranging from 0 (no pain) to 10 (worst pain imaginable). PONV and pain severity were assessed at the time points mentioned earlier. The severity of adverse events was assessed using the DindoClavien classification if related to surgery and with the Common Terminology Criteria for Adverse Events v4.03 if related to medical events.12,13 Adverse events with a grade greater than 3 were considered serious. Adverse events were reported irrespective of causality.
Statistical Analysis To achieve a clinically relevant effect, we assumed a 50% reduction in the incidence of PONV for this study (48% vs 24%). To demonstrate this effect with a 2-sided a level of 5% and a power of 80%, 76 patients per arm were required (including a drop-out rate of 18% and an interim analysis).11 After 70 patients had been allocated, a blinded interim analysis was performed (P ¼ 0.004 for early termination due to superiority, O’Brian-Fleming p-spending). Interim and final analyses were performed according to the intention-to-treat principle. In addition, a PP analysis was performed, excluding patients with protocol violations. Missing values were replaced with the last available value (the last observation carried forward approach). The primary endpoint was analyzed for superiority using x2 statistics (without corrections). Mixed models assessing the mean ranks of the scores with adjustments for time and treatment versus time interactions were applied to analyze the following secondary endpoints: severity of PONV, intensity of pain at rest, and intensity of pain after physical stress. The Mann-Whitney U statistic was used for the following secondary endpoints: amount of anesthetic medication, amount of postoperative analgesics, and length of hospital stay. The 95% confidence intervals (95% CIs) of ratios were calculated using the Wilson score method. An auxiliary, nonconfirmatory post hoc analysis was performed to detect possible risk factors for PONV in a binary logistic regression model (SAS version 9.3; SAS Institute, Inc., Cary, NC).
Role of the Funding Source This trial was supported by a grant from the Clinical Trial Unit of the Kantonsspital St. Gallen. The Clinical Trial Unit had no role in the study design, data collection and analysis, or decision to submit for publication. The final decision on content was made exclusively by the authors. www.annalsofsurgery.com | 935
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
Annals of Surgery Volume 262, Number 6, December 2015
Tarantino et al
RESULTS During the study period, 393 patients were eligible for inclusion. A total of 152 patients were included in the study and randomly allocated to the 2 study arms (Fig. 1). For the PP analysis, 27 patients (18%) were excluded because of protocol violations. In the treatment arm, 11 patients were excluded because of perioperative morphine administration (n ¼ 10) or failure to provide study treatment in time (n ¼ 1). In the control arm, 16 patients were excluded because of perioperative morphine (n ¼ 14) or preoperative use of steroids (n ¼ 2) (Fig. 1). The patient characteristics were well balanced between the study arms (Table 1). In a blinded interim analysis after 70 patients had been allocated, the incidence of PONV was reported in 15 of 35 patients (43%) and 8 of 33 patients (24%) in each arm (P ¼ 0.105). In one arm, 2 patients lacked the outcome measurement at the time of interim analysis. Neither serious nor drug-related adverse events occurred. The trial was continued without modification. The mean time standard deviation of the outcome measurements was 4.0 0.2, 8.0 0.2, 16.1 0.3, 24.1 0.3, 32.0 0.2, and 47.9 0.9 hours postoperatively. The total incidence of PONV was reported in 65 of 152 patients (43%; 95% CI, 35–51). In the intention-to-treat analysis, the incidence of PONV in the treatment arm was reported in 22 of 76 patients (29%; 95% CI, 20–40) compared with 43 of 76 patients (57%; 95% CI, 45–67) in the
control arm (P ¼ 0.001). The odds ratio (OR) was 0.31 (95% CI, 0.16–0.61), the absolute risk reduction was 28% (95% CI, 12–42), and the number needed to treat was 4. In the PP analysis, the overall incidence of PONV was reported in 49 of 125 patients (39%; 95% CI, 31–48). The incidence of PONV in the treatment arm was reported in 18 of 65 patients (28%; 95% CI, 18–40) compared with 31 of 60 patients (52%; 95% CI, 39–64) in the control arm (P ¼ 0.006). The OR was 0.36 (95% CI, 0.17–0.75), the absolute risk reduction was 24% (95% CI, 7–40), and the number needed to treat was 5. In the multivariate analysis, the severity of PONV was significantly reduced by dexamethasone (P ¼ 0.024), with a significant time interaction (P ¼ 0.008) (Fig. 2). Postoperative pain at rest and after physical stress, the amount of anesthetic medication, the amount of postoperatively administered analgesics, and the length of hospital stay did not differ significantly between the study arms (Table 2). No deaths or serious adverse events were observed during the study. No incidence of wound infection or other drug-related adverse events (including newly developed diabetes mellitus) was reported. The majority of adverse events were related to surgery, and the rate did not differ between the study arms (Table 3). Two patients in the treatment arm and none in the placebo arm had to undergo reoperations (P ¼ 0.155). Reoperations were necessary due to postoperative bleeding (n ¼ 1) or a diagnosis of malignancy (completion surgery, n ¼ 1). In an intention-to-treat multivariable logistic regression model with variable selection, a reduced rate of PONV was significantly
Enrollment
393 evaluated for elegibility
241 not randomized • 188 did not meet inclusion criteri • 52 declined to participate • 1 organizational failure
Allocation
Allocated to placebo (n = 76): • 76 received allocated intervention • 0 did not receive allocated intervention
• 0 lost to follow-up • 0 discontinued intervention
• 0 lost to follow-up • 0 discontinued intervention
76 analyzed intention to treat: • 0 excluded from analysis
76 analyzed intention to treat: • 0 excluded from analysis
65 analyzed per protocol: • 11 excluded from analysis (10 patients received morphin, 1 organizational failure)
60 analyzed per protocol: • 16 excluded from analysis (14 patients received morphin, 2 patients received steroid medication preoperatively)
Analysis
Allocated to dexamethason (n = 76): • 75 received allocated intervention • 1 did not receive allocated intervention
Follow-Up
Randomized (n = 152)
936 | www.annalsofsurgery.com
ß
FIGURE 1. CONSORT flow diagram.
2015 Wolters Kluwer Health, Inc. All rights reserved.
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
Annals of Surgery Volume 262, Number 6, December 2015
Dexamethasone for PONV Prevention
TABLE 1. Patient Characteristics, Surgical Procedures, and Perioperative Characteristics
No serious adverse events were observed. However, dexamethasone had no influence on postoperative pain, the amount of administered anesthetic medication, the amount of analgesics required postoperatively, or the length of hospital stay. This is the first study demonstrating the efficacy of dexamethasone for the prevention of PONV in a Western population. Previous studies reported biases introduced by postoperative opioid administration, anesthetic technique, or sex.10,14–21 These studies could not conclusively show whether dexamethasone reduced the PONV primarily caused by surgery or the PONV caused by confounding factors such as opioid administration. In our PP analysis excluding patients receiving opioids, we definitively demonstrated that dexamethasone has an effect on the PONV primarily caused by thyroid surgery. This result was confirmed by a multivariable analysis adjusting for known biases, that is, sex or opioid administration. However, the mechanism by which dexamethasone exerts its antiemetic activity is unknown. One can only speculate whether it is the immunomodulatory effect of corticosteroids or the dexamethasone-triggered release of endorphins leading to mood elevation, a sense of well-being, and increased appetite.22 Most patients consider PONV to be worse than postoperative pain23; therefore, reducing PONV by 50% is of high clinical relevance and important for the postoperative well-being of patients. Nevertheless, in our study, dexamethasone treatment had no effect on the length of hospital stay. However, evaluating only patients in the placebo arm, PONV alone had no effect on the length of stay. Thus, it is not surprising that the prevention of PONV would not result in a shorter hospitalization of our patients. Furthermore, the length of stay in our study population was significantly shorter than the average length of stay permitted by Swiss DRG (86.4 hours) (Swiss DRG actually discourages hospital stays of 24 hours for thyroidectomies). The PONV rate was highest 4 and 8 hours after surgery. Whether the beneficial effect of dexamethasone is restricted to the first 8 hours or might even act at later times cannot be derived from our data, because the PONV rate after 8 hours (even in the control arm) was too low to determine any meaningful differences. Ten years ago, a large factorial clinical trial by Apfel et al5 compared the efficacy of 6 antiemetic strategies and their combination for the prevention of PONV. The antiemetic interventions were similarly effective and acted independently. The authors recommended using the safest or least expensive medication first. Therefore, a total intravenous anesthetic medication in combination with dexamethasone was advocated as the first-line treatment method for the prevention of PONV. The authors also recommended that serotonin antagonists, among other antiemetic agents, should be used as a rescue treatment option. This recommendation supports our results, although the mechanism leading to thyroidectomy-related PONV might be different from PONV induced during other surgical intervention. In contrast to previous studies, we did not observe any effect of dexamethasone on postoperative pain.14 We assume that this result is due to the bilateral superficial cervical plexus block we applied to all patients in the study. Our patients, even in the placebo arm, reported mean pain scores at rest of 1 and lower at any time, rendering the measurement of pain reduction meaningless. For comparison, in previous studies, pain scores in the placebo group ranged between 3 and 5. Interestingly, in our multivariable analysis, length of surgery was significantly associated with a lower, not a higher, rate of PONV, whereas more extended surgical procedures (total vs hemithyroidectomy) were associated with a higher PONV rate as one would expect (although without reaching significance). This seemingly contradictory result is difficult to explain. An explanation could be that
Age, yr Sex Male Female Body mass index, kg/m2 Body surface, m2 Nationality Swiss German Other Diagnosis Graves disease Goiter, multinodular Goiter, uninodular Laterality of disease Bilateral Left Right Operation Hemithyroidectomy Near total thyroidectomy Total thyroidectomy Lymph node dissection (C1/level VI) No Yes Operation time, min Incision length, cm Specimen weight, g Histology of specimen PTC FTC Benign disease
Total (N ¼ 152)
Dexamethasone (n ¼ 76)
Placebo (n ¼ 76)
49.0 12.7 (21–81)
49.5 12.0 (21–76)
48.4 13.4 (22–81)
32 (21.1) 120 (78.9) 25.1 4.4 (17.2–37.8) 1.81 0.20 (1.44–2.55)
14 (18.4) 62 (81.6) 24.8 4.1 (18.0–37.8) 1.78 0.16 (1.45–2.24)
18 (23.7) 58 (76.3) 25.4 4.6 (17.2–36.0) 1.84 0.23 (1.44–2.55)
129 (84.9) 10 (6.6) 13 (8.6)
69 (90.8) 4 (5.3) 3 (3.9)
60 (78.9) 6 (7.9) 10 (13.2)
25 (16.4) 68 (44.7) 59 (38.8)
13 (17.1) 35 (46.1) 28 (36.8)
12 (15.8) 33 (43.4) 31 (40.8)
81 (53.3) 32 (21.1) 39 (25.7)
42 (55.3) 15 (19.7) 19 (25.0)
39 (51.3) 17 (22.4) 20 (26.3)
75 (49.3) 32 (21.1)
37 (48.7) 19 (25.0)
38 (50.0) 13 (17.1)
45 (29.6)
20 (26.3)
25 (32.9)
128 (84.2) 24 (15.8) 126.8 43.2 (39–305) 5.7 1.5 (2–10) 48.3 49.9 (4.8–390.0) 10 (6.6) 6 (3.9) 136 (89.5)
63 (82.9) 13 (17.1) 126.2 40.9 (39–266) 5.6 1.4 (2–10) 45.3 40.1 (4.8–230.0) 3 (3.9) 4 (5.3) 69 (90.8)
65 (85.5) 11 (14.5) 127.5 45.7 (60–305) 5.7 1.5 (3–9) 51.2 58.2 (6.9–390.0) 7 (9.2) 2 (2.6) 67 (88.2)
The values given are mean standard deviation (range) or number (percentage). FTC indicates follicular thyroid carcinoma; PTC, papillary thyroid carcinoma.
associated with the study treatment (dexamethasone), male sex, and duration of surgery. There was a tendency for increased rates of PONV to be associated with more extended operations, postoperative morphine use, and the amount of intraoperative fentanyl (Table 4); however, these associations were not statistically significant. The PONV rate was also significantly reduced after dexamethasone treatment when the multivariable model was used according to the PP analysis (OR ¼ 0.29; 95% CI, 0.13–0.63; P ¼ 0.002).
DISCUSSION This investigation provides compelling evidence that a single preoperative dose of dexamethasone administration reduces the rate of tPONV by half. Even after the exclusion of patients treated with opioids (which may induce PONV) in the PP analysis, the effect was still both statistically significant and clinically relevant. Every fourth patient medicated with dexamethasone benefited from the treatment. ß
2015 Wolters Kluwer Health, Inc. All rights reserved.
www.annalsofsurgery.com | 937
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
Annals of Surgery Volume 262, Number 6, December 2015
40
Tarantino et al
n=12
20
Percentage
30
PONV: Vomiting Severe nausea Mild nausea
n=3
n=12
10
n=3 n=1
n=3 n=8
n=2
n=1
n=15
0
n=9
4h Time: Arm: Dex Plc
n=5
n=7
n=3 n=1 n=1
Plc
Dex
8h Dex
n=1 n=2
n=3
Plc
Dex
16h
n=1
n=2
n=1
Plc
Dex
24h
n=2 n=1
n=2
Plc
Dex
32h
n=3
48h Plc
FIGURE 2. Time course of PONV in the intention-to-treat analysis. Bars represent the incidence of PONV at 6 time points after thyroidectomy in the dexamethasone and placebo arm. Distribution of severity is indicated by different shades of grey. The number of patients affected is indicated within the bars (n ¼). Severity of PONV was significantly reduced by dexamethasone (P ¼ 0.024), with a significant time interaction (P ¼ 0.008) (mixed model over the mean ranks of the PONV score). Dex indicates dexamethasone; Plc, placebo. TABLE 2. Secondary Endpoints (Intention-to-treat Analysis) Endpoint
Time/Unit
Pain (at rest) (VRS 0–10)
4h 8h 16 h 24 h 32 h 48 h 4h 8h 16 h 24 h 32 h 48 h
Pain (after physical stress) (VRS 0–10)
Length of postoperative stay, h Propofol, mg Remifentanil, mg Fentanyl, mg Rocuronium, mg Paracetamol total, g Paracetamol day 1, g Paracetamol day 2, g Metamizole total, g Metamizole day 1, g Metamizole day 2, g Morphine total, mg Morphine day 1, mg Morphine day 2, mg
Total (N ¼ 152)
Dexamethasone (n ¼ 76)
Placebo (n ¼ 76)
P
0.86 1.28 (0–8) 0.77 1.16 (0–6) 0.66 1.14 (0–7) 0.51 1.08 (0–8) 0.41 0.81 (0–4) 0.36 0.69 (0–4) 1.99 1.63 (0–8) 1.95 1.64 (0–7) 1.77 1.49 (0–7) 1.56 1.37 (0–8) 1.28 1.40 (0–6) 1.20 1.31 (0–5) 48.2 14.2 (22.2–120.0) 1132 422 (0–2846) 1961 1151 (18–9192) 436 109 (200–1000) 39.2 7.8 (0.0–70.0) 5.1 1.5 (0–8) 2.2 0.8 (0–4) 2.9 1.1 (0–4) 3.0 2.4 (0–8) 1.8 1.1 (0–5) 1.2 1.6 (0–5) 1.5 3.7 (0–17) 1.4 3.5 (0–15) 0.1 0.8 (0–10)
0.72 1.09 (0–4) 0.82 1.22 (0–6) 0.63 1.00 (0–4) 0.53 0.96 (0–5) 0.37 0.63 (0–2) 0.36 0.67 (0–3) 1.88 1.52 (0–5) 1.88 1.54 (0–6) 1.72 1.40 (0–5) 1.50 1.22 (0–5) 1.25 1.39 (0–5) 1.18 1.36 (0–5) 48.3 15.7 (22.2–120.0) 1097 422 (0–2257) 1944 1197 (18–9192) 427 111 (200–1000) 38.7 7.3 (0.0–50.0) 5.2 1.6 (0–8) 2.2 0.9 (0–4) 2.9 1.1 (0–4) 3.3 2.5 (0–8) 2.0 1.2 (0–5) 1.3 1.6 (0–4) 1.2 3.3 (0–15) 1.2 3.3 (0–15) 0.0 0.0 (0–0)
1.00 1.44 (0–8) 0.73 1.11 (0–5) 0.69 1.26 (0–7) 0.50 1.19 (0–8) 0.46 0.96 (0–4) 0.36 0.72 (0–4) 2.09 1.73 (0–8) 2.01 1.75 (0–7) 1.83 1.58 (0–7) 1.62 1.51 (0–8) 1.32 1.43 (0–6) 1.22 1.26 (0–5) 48.1 12.6 (25.8–92.8) 1168 422 (469–2846) 1978 1111 (573–5460) 444 108 (300–800) 39.7 8.3 (3.5–70.0) 5.1 1.4 (2–8) 2.2 0.8 (1–4) 2.9 1.1 (0–4) 2.8 2.3 (0–8) 1.7 1.0 (0–4) 1.1 1.6 (0–5) 1.7 4.0 (0–17) 1.6 3.7 (0–15) 0.2 1.2 (0–10)
0.947
0.674y
0.835z 0.374z 0.930z 0.360z 0.596z 0.474z 0.727z 0.629z 0.260z 0.156z 0.429z 0.391z 0.417z 0.156z
The values given are mean standard deviation (range). Pain measured on a VRS ranging from 0 (no pain) to 10 (worst pain imaginable). Mixed model over the mean ranks (time interaction: P ¼ 0.547). yMixed model over the mean ranks (time interaction: P ¼ 0.995). zMann-Whitney U test. VRS indicates Verbal Rating Scale.
938 | www.annalsofsurgery.com
ß
2015 Wolters Kluwer Health, Inc. All rights reserved.
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
Annals of Surgery Volume 262, Number 6, December 2015
Dexamethasone for PONV Prevention
TABLE 3. Adverse Events Grade
Adverse Event Surgical adverse events Recurrent laryngeal nerve palsy Hoarseness Bleeding Hypoparathyroidism Bleeding requiring reoperation Medical adverse events Coughing Sinus tachycardia Temporary compensated renal insufficiency Allergic reaction (to metamizole) Hypocalcemia Total adverse events
Total (N ¼ 152)
Dexamethasone (n ¼ 76)
I I I II IIIb
4 1 1 4 1
(2.6) (0.7) (0.7) (2.6) (0.7)
2 1 1 2 1
1 1 1 2 2
1 1 1 1 2 17
(0.7) (0.7) (0.7) (0.7) (1.3) (11.1)
1 (1.3)
Placebo (n ¼ 76)
Py
2 (2.6)
1.00 0.32 0.32 1.00 0.32
(2.6) (1.3) (1.3) (2.6) (1.3)
8 (10.5)
2 (2.6)
1 (1.3) 1 (1.3) 1 (1.3) 2 (2.6) 9 (11.8)
0.32 0.32 0.32 0.32 0.16 0.80
The values given are number (percentage). All adverse events were unrelated to the study treatment. Surgical adverse events: Dindo-Clavien classification (I–V)12; medical adverse events: CTCAE grade (1–5).13 2 y x test. CTCAE indicates Common Terminology Criteria for Adverse Events.
more and extended surgical manipulations lead to a higher PONV rate that correlates with the assumption that tPONV is caused by edema and inflammation around the neck tissues, leading to evoked parasympathetic impulses through the vagus, recurrent laryngeal, and glossopharyngeal nerves to the vomiting center.2,24,25 In our study, we applied bilateral superficial cervical plexus block in all patients. Is it possible that this not commonly applied procedure might influence the PONV rate or the efficacy of
dexamethasone? In a randomized controlled study, very similar to ours but without application of a cervical plexus block, the PONV rate in the control arm was 60% compared with 57% in our study.15 Thus, an influence of the cervical block on PONV seems unlikely. The efficacy of dexamethasone was somewhat less in this study (risk reduction of 36% compared with 50%), but this difference is within the limits of normal interstudy variation and probably not due to the lack of a cervical block.
TABLE 4. Logistic Regression Analysis for Incidence of PONV (Intention-to-treat Analysis) Univariate
Treatment Placebo Dexamethasone Age, yr Sex Male Female Body mass index, kg/m2 Nationality Swiss German Other Diagnosis Goiter Graves disease Operation Hemithyroidectomy (Near) total thyroidectomy Lymph node dissection No Yes Operation time, h Postoperative morphine No Yes Propofol (100 mg) Remifentanil (100 mg) Fentanyl (100 mg) Rocuronium (10 mg)
Full Multivariate Model
Optimized Modely
OR (95% CI)
P
OR (95% CI)
P
OR (95% CI)
P
Reference 0.31 (0.16–0.61) 1.01 (0.99–1.04)
0.001
Reference 0.27 (0.12–0.57) 1.01 (0.98–1.05)
0.001
Reference 0.26 (0.12–0.54) —
Efficacy of a Single Preoperative Dexamethasone Dose to Prevent Nausea and Vomiting After Thyroidectomy (the tPONV Study) A Randomized, Double-blind, Placebo-controlled Clinical Trial Ignazio Tarantino, MD, MSc, Rene Warschkow, MD,yz Ulrich Beutner, PhD,z Walter Kolb, MD,z Andreas Lu¨thi, MD,§ Cornelia Lu¨thi, BScN,z Bruno M. Schmied, MD,z and Thomas Clerici, MDz
Objective: Does dexamethasone given before thyroidectomy reduce postoperative nausea and vomiting (PONV) in a randomized controlled trial? Background: PONV is an unsettling problem that commonly occurs in patients after thyroidectomy. Various preventive measures have been studied; however, many of these studies have been criticized for their biases (eg, use of opioids, sex selection) or were even retracted. Methods: This single-institution, randomized, double-blind, placebo-controlled, superiority study was performed between January 1, 2011, and May 30, 2013. Patients undergoing thyroidectomy for benign disease were allocated by a block randomized list to receive a preoperative single dose of dexamethasone (8 mg) or placebo. Patients and staff were blinded to the treatment assignment. The primary endpoint was the incidence of PONV assessed at 4, 8, 16, 24, 32, and 48 hours after surgery. To observe an incidence reduction of 50%, a total of 152 patients were required for the study. Results: The total incidence of PONV was reported in 65 of 152 patients (43%; 95% confidence interval [CI], 35–51). In the intention-to-treat analysis, PONV occurred in 22 of 76 patients (29%; 95% CI, 20–40) in the treatment arm and in 43 of 76 patients (57%; 95% CI, 45–67) in the control arm (P ¼ 0.001; odds ratio ¼ 0.31; 95% CI, 0.16–0.61; absolute risk reduction ¼ 28%; 95% CI, 12–42). The number needed to treat was 4. No severe dexamethasone-related adverse events were observed during the study. Conclusions: A single dose of preoperative dexamethasone administration is an effective, safe, and economical measure to reduce PONV incidence after thyroidectomy. Keywords: dexamethasone, PONV, postoperative nausea and vomiting, thyroid, thyroidectomy
(Ann Surg 2015;262:934–940)
A
quarter of surgical patients suffer from postoperative nausea and vomiting (PONV) after general anesthesia.1 The incidence of PONV is higher after thyroidectomy, reaching up to 80%.2 PONV affects patient comfort and health care costs by significantly prolonging hospital stays3 and may lead to serious postsurgical complications such as bleeding, one of the most feared complications after thyroidectomy.4 A study comparing the efficacy of 6 established antiemetic strategies after elective surgery indicated that From the Department of General, Visceral and Transplant Surgery, University of Heidelberg, Heidelberg, Germany; yInstitute of Medical Biometry and Informatics, University of Heidelberg, Heidelberg, Germany; zDepartment of Surgery, Kantonsspital St. Gallen, St. Gallen, Switzerland; and §Department of Anesthesiology, Kantonsspital St. Gallen, St. Gallen, Switzerland. Disclosure: Supported by a grant from the Clinical Trial Unit of the Kantonsspital St. Gallen. The authors declare no conflicts of interest. Reprints: Ignazio Tarantino, MD, MSc, Department of General, Abdominal and Transplant Surgery, University of Heidelberg, Im Neuenheimer Feld 110, 69120 Heidelberg, Germany. E-mail: [email protected]. Copyright ß 2015 Wolters Kluwer Health, Inc. All rights reserved. ISSN: 0003-4932/14/26105-0821 DOI: 10.1097/SLA.0000000000001112
934 | www.annalsofsurgery.com
dexamethasone may be a reasonable first-line treatment option to prevent PONV.5 A decade after the publication of this large study, dexamethasone is still not well established for the prophylaxis and treatment of PONV. Specifically, there is limited evidence for the efficacy of dexamethasone in preventing PONV after thyroidectomy (tPONV). The few studies published to date did not control for confounding factors that may influence PONV, such as postoperative opioid administration, a common cause of nausea and vomiting,6 or the use of various anesthetics within the study arms known to differentially provoke PONV.5 The inherent bias of previous studies, along with concerns about increased infection rates, may explain why dexamethasone has not been established as the standard treatment option for PONV prevention. Recently, our group demonstrated that a bilateral superficial cervical plexus block during thyroid surgery under general anesthesia significantly reduced pain. Of even greater clinical importance, this block prevented the need for postoperative opioids.7,8 Therefore, patients undergoing thyroidectomy along with a bilateral superficial cervical plexus block are an ideal, homogenous cohort for the investigation of the efficacy of dexamethasone for PONV prevention. These patients exhibit a high incidence of PONV and seldom require opioids. Combined with the highly standardized anesthesia protocol at our institution, this setting allows all known biases to be controlled.
METHODS Study Design and Participants This study was a single-institution, randomized, double-blind, placebo-controlled, superiority trial with 2 parallel arms comparing a single preoperative dose of 8 mg of dexamethasone (treatment arm) with saline (placebo arm). Two-dose finding studies have shown that 8 mg of dexamethasone is the minimum effective dose for the reduction of PONV.9,10 Patients older than 18 years scheduled for a primary partial or total thyroidectomy due to benign disease at our institution (Kantonsspital St. Gallen, Department of Surgery) were recruited between January 1, 2011, and May 30, 2013. The exclusion criteria included a history of adverse reactions to dexamethasone, polypropylene glycol, or disodium-EDTA (medication ingredients); patients suffering from chronic pain; the necessity for sternotomy; inability to administer a cervical block; diabetes mellitus type 1 or 2 requiring medication; pregnancy (tested in all women capable of childbearing); glaucoma; administration of antiemetic therapy within 48 hours before the surgery; acute pain requiring analgesics other than acetaminophen or more than a single dose of nonsteroidal antiinflammatory drugs within 48 hours before the surgery; use of antipsychotic drugs; noncompliance; proven malignancy, necessity of lateral neck dissection; and reoperations. All participants provided written informed consent before study participation. The study protocol was approved by the Ethics Committee of the Canton St. Gallen (EKSG 10/082) and by Swissmedic (2011DR3005) before Annals of Surgery Volume 262, Number 6, December 2015
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
Annals of Surgery Volume 262, Number 6, December 2015
enrolling any patients in this trial. The study protocol was previously published (http://www.biomedcentral.com/1471–2253/13/ 19),11 and the trial is registered with ClinicalTrials.gov: NCT01189292.
Randomization and Masking Patients were allocated to the study arms in a 1:1 ratio based on a predefined block randomization list with blocks randomly sized between 4 and 10. The randomization list was generated by the data manager (U.B.), with the hard copy stored in a secure location and the electronic version stored within the trial management software in a password-protected, encrypted fashion. During the study, only the data manager had access to the randomization list. Two to 3 days before surgery, the data manager prepared a computer-generated study medication request form based on the patient’s study number. The request contained the patient’s name and the treatment allocation. The form was placed in an opaque, sealed, and signed envelope and was directly delivered to the pharmacy by study personnel. Study medication was prepared according to the request form by the hospital pharmacy. Medication was delivered in a syringe labeled with the patient’s name and study number and containing either 2 mL of saline (placebo) or 2 mL of saline with 8 mg of dexamethasone (Mephameson; Mepha Pharma, Basel, Switzerland). Syringes containing saline or dexamethasone were indistinguishable. The patients and all study personnel except for the data manager were blinded to the treatment allocation.
Procedures The anesthesiologist administered the study medication intravenously 30 to 60 minutes before skin incision. All patients underwent standardized anesthesia, thyroid surgery, and routine postoperative care. During the surgery, recurrent laryngeal nerve monitoring was performed. A bilateral superficial cervical plexus block was performed just before the skin incision as previously described.8,11 The standardized anesthetic protocol consisted of propofol (Disoprivan 1%; AstraZeneca, London, United Kingdom) for hypnosis, fentanyl (Sintenyl; Sintetica, Mendrisio, Switzerland) and remifentanil (Ultiva; GlaxoSmithKline, Mu¨nchenbuchsee, Switzerland) for analgesia, and rocuronium (Esmeron; MSD, Lucerne, Switzerland) for muscle relaxation (for further details, see protocol).11 The postoperative first-line therapy for PONV was intravenous droperidol (Droperidol; Sintetica), with a maximum dose of 3 0.5 mg/24 h if the patient’s blood pressure exceeded 120 mm Hg. The second-line therapy was 4 mg of ondansetron given intravenously (Zofran; Glaxo SmithKline). All surgical procedures were performed by 1 of 2 surgeons (T.C. and W.K.) or under their close supervision during teaching surgical procedures. The postoperative first-line therapy for PONV was intravenous droperidol (Droperidol; Sintetica), with a maximum dose of 3 0.5 mg/24 h if the patient’s blood pressure exceeded 120 mm Hg. The second-line therapy was 4 mg of ondansetron given intravenously (Zofran; Glaxo SmithKline). All surgical procedures were performed by 1 of 2 surgeons (T.C. and W.K.) or under their close supervision during teaching surgical procedures. No analgesia was routinely applied. The first-line reserve for pain relief upon patient demand was 1 g of oral paracetamol (Dafalgan; Bristol-Myers Squibb, Baar, Switzerland), with a maximum dose of 3 g/24 h. The second-line reserve was 1 g of metamizole (Novalgin; Sanofi-Aventis, Meyrin, Switzerland), with a maximum dose of 4 g/24 h. The third-line reserve was intravenous morphine (Morphin HCI; Sintetica); however, this administration was considered a protocol violation, and patients receiving morphine were excluded from per protocol (PP) analysis. ß
2015 Wolters Kluwer Health, Inc. All rights reserved.
Dexamethasone for PONV Prevention
Outcomes The primary endpoint was the incidence of PONV assessed at 4, 8, 16, 24, 32, and 48 hours after surgery (unless the patient was released earlier). If nausea or vomiting was documented at any of these time points, it was considered an incidence of PONV. The nursing staff on the surgical wards assessed PONV. Before accrual of the first patient, nursing staff was instructed and trained about the trial, its objectives, and the outcome assessment. In addition, the assessment was monitored daily by a designated study nurse who also confirmed the outcome by patient interviews. The secondary endpoints were the severity of PONV, pain, length of stay (hours after intervention), wound healing (at discharge and after 6 weeks), the amount of required anesthetic medication, and postoperative inhospital morbidity (wound infection, reoperation, bleeding, prolonged intubation, laryngeal nerve lesion, urinary retention, urinary tract infection, pulmonary infection, gastrointestinal hemorrhage, or newly occurring insulin dependence in diabetic patients). The severity of PONV was rated from 0 to 3 (0 for no nausea; 1 for mild nausea, defined as nausea requiring a single administration of an antiemetic drug; 2 for severe nausea, defined as nausea requiring the repeated administration of antiemetic drugs; and 3 for vomiting). Pain was evaluated at rest and after physical stress (rotating the neck 90 degrees to each side) using the Verbal Rating Scale ranging from 0 (no pain) to 10 (worst pain imaginable). PONV and pain severity were assessed at the time points mentioned earlier. The severity of adverse events was assessed using the DindoClavien classification if related to surgery and with the Common Terminology Criteria for Adverse Events v4.03 if related to medical events.12,13 Adverse events with a grade greater than 3 were considered serious. Adverse events were reported irrespective of causality.
Statistical Analysis To achieve a clinically relevant effect, we assumed a 50% reduction in the incidence of PONV for this study (48% vs 24%). To demonstrate this effect with a 2-sided a level of 5% and a power of 80%, 76 patients per arm were required (including a drop-out rate of 18% and an interim analysis).11 After 70 patients had been allocated, a blinded interim analysis was performed (P ¼ 0.004 for early termination due to superiority, O’Brian-Fleming p-spending). Interim and final analyses were performed according to the intention-to-treat principle. In addition, a PP analysis was performed, excluding patients with protocol violations. Missing values were replaced with the last available value (the last observation carried forward approach). The primary endpoint was analyzed for superiority using x2 statistics (without corrections). Mixed models assessing the mean ranks of the scores with adjustments for time and treatment versus time interactions were applied to analyze the following secondary endpoints: severity of PONV, intensity of pain at rest, and intensity of pain after physical stress. The Mann-Whitney U statistic was used for the following secondary endpoints: amount of anesthetic medication, amount of postoperative analgesics, and length of hospital stay. The 95% confidence intervals (95% CIs) of ratios were calculated using the Wilson score method. An auxiliary, nonconfirmatory post hoc analysis was performed to detect possible risk factors for PONV in a binary logistic regression model (SAS version 9.3; SAS Institute, Inc., Cary, NC).
Role of the Funding Source This trial was supported by a grant from the Clinical Trial Unit of the Kantonsspital St. Gallen. The Clinical Trial Unit had no role in the study design, data collection and analysis, or decision to submit for publication. The final decision on content was made exclusively by the authors. www.annalsofsurgery.com | 935
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
Annals of Surgery Volume 262, Number 6, December 2015
Tarantino et al
RESULTS During the study period, 393 patients were eligible for inclusion. A total of 152 patients were included in the study and randomly allocated to the 2 study arms (Fig. 1). For the PP analysis, 27 patients (18%) were excluded because of protocol violations. In the treatment arm, 11 patients were excluded because of perioperative morphine administration (n ¼ 10) or failure to provide study treatment in time (n ¼ 1). In the control arm, 16 patients were excluded because of perioperative morphine (n ¼ 14) or preoperative use of steroids (n ¼ 2) (Fig. 1). The patient characteristics were well balanced between the study arms (Table 1). In a blinded interim analysis after 70 patients had been allocated, the incidence of PONV was reported in 15 of 35 patients (43%) and 8 of 33 patients (24%) in each arm (P ¼ 0.105). In one arm, 2 patients lacked the outcome measurement at the time of interim analysis. Neither serious nor drug-related adverse events occurred. The trial was continued without modification. The mean time standard deviation of the outcome measurements was 4.0 0.2, 8.0 0.2, 16.1 0.3, 24.1 0.3, 32.0 0.2, and 47.9 0.9 hours postoperatively. The total incidence of PONV was reported in 65 of 152 patients (43%; 95% CI, 35–51). In the intention-to-treat analysis, the incidence of PONV in the treatment arm was reported in 22 of 76 patients (29%; 95% CI, 20–40) compared with 43 of 76 patients (57%; 95% CI, 45–67) in the
control arm (P ¼ 0.001). The odds ratio (OR) was 0.31 (95% CI, 0.16–0.61), the absolute risk reduction was 28% (95% CI, 12–42), and the number needed to treat was 4. In the PP analysis, the overall incidence of PONV was reported in 49 of 125 patients (39%; 95% CI, 31–48). The incidence of PONV in the treatment arm was reported in 18 of 65 patients (28%; 95% CI, 18–40) compared with 31 of 60 patients (52%; 95% CI, 39–64) in the control arm (P ¼ 0.006). The OR was 0.36 (95% CI, 0.17–0.75), the absolute risk reduction was 24% (95% CI, 7–40), and the number needed to treat was 5. In the multivariate analysis, the severity of PONV was significantly reduced by dexamethasone (P ¼ 0.024), with a significant time interaction (P ¼ 0.008) (Fig. 2). Postoperative pain at rest and after physical stress, the amount of anesthetic medication, the amount of postoperatively administered analgesics, and the length of hospital stay did not differ significantly between the study arms (Table 2). No deaths or serious adverse events were observed during the study. No incidence of wound infection or other drug-related adverse events (including newly developed diabetes mellitus) was reported. The majority of adverse events were related to surgery, and the rate did not differ between the study arms (Table 3). Two patients in the treatment arm and none in the placebo arm had to undergo reoperations (P ¼ 0.155). Reoperations were necessary due to postoperative bleeding (n ¼ 1) or a diagnosis of malignancy (completion surgery, n ¼ 1). In an intention-to-treat multivariable logistic regression model with variable selection, a reduced rate of PONV was significantly
Enrollment
393 evaluated for elegibility
241 not randomized • 188 did not meet inclusion criteri • 52 declined to participate • 1 organizational failure
Allocation
Allocated to placebo (n = 76): • 76 received allocated intervention • 0 did not receive allocated intervention
• 0 lost to follow-up • 0 discontinued intervention
• 0 lost to follow-up • 0 discontinued intervention
76 analyzed intention to treat: • 0 excluded from analysis
76 analyzed intention to treat: • 0 excluded from analysis
65 analyzed per protocol: • 11 excluded from analysis (10 patients received morphin, 1 organizational failure)
60 analyzed per protocol: • 16 excluded from analysis (14 patients received morphin, 2 patients received steroid medication preoperatively)
Analysis
Allocated to dexamethason (n = 76): • 75 received allocated intervention • 1 did not receive allocated intervention
Follow-Up
Randomized (n = 152)
936 | www.annalsofsurgery.com
ß
FIGURE 1. CONSORT flow diagram.
2015 Wolters Kluwer Health, Inc. All rights reserved.
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
Annals of Surgery Volume 262, Number 6, December 2015
Dexamethasone for PONV Prevention
TABLE 1. Patient Characteristics, Surgical Procedures, and Perioperative Characteristics
No serious adverse events were observed. However, dexamethasone had no influence on postoperative pain, the amount of administered anesthetic medication, the amount of analgesics required postoperatively, or the length of hospital stay. This is the first study demonstrating the efficacy of dexamethasone for the prevention of PONV in a Western population. Previous studies reported biases introduced by postoperative opioid administration, anesthetic technique, or sex.10,14–21 These studies could not conclusively show whether dexamethasone reduced the PONV primarily caused by surgery or the PONV caused by confounding factors such as opioid administration. In our PP analysis excluding patients receiving opioids, we definitively demonstrated that dexamethasone has an effect on the PONV primarily caused by thyroid surgery. This result was confirmed by a multivariable analysis adjusting for known biases, that is, sex or opioid administration. However, the mechanism by which dexamethasone exerts its antiemetic activity is unknown. One can only speculate whether it is the immunomodulatory effect of corticosteroids or the dexamethasone-triggered release of endorphins leading to mood elevation, a sense of well-being, and increased appetite.22 Most patients consider PONV to be worse than postoperative pain23; therefore, reducing PONV by 50% is of high clinical relevance and important for the postoperative well-being of patients. Nevertheless, in our study, dexamethasone treatment had no effect on the length of hospital stay. However, evaluating only patients in the placebo arm, PONV alone had no effect on the length of stay. Thus, it is not surprising that the prevention of PONV would not result in a shorter hospitalization of our patients. Furthermore, the length of stay in our study population was significantly shorter than the average length of stay permitted by Swiss DRG (86.4 hours) (Swiss DRG actually discourages hospital stays of 24 hours for thyroidectomies). The PONV rate was highest 4 and 8 hours after surgery. Whether the beneficial effect of dexamethasone is restricted to the first 8 hours or might even act at later times cannot be derived from our data, because the PONV rate after 8 hours (even in the control arm) was too low to determine any meaningful differences. Ten years ago, a large factorial clinical trial by Apfel et al5 compared the efficacy of 6 antiemetic strategies and their combination for the prevention of PONV. The antiemetic interventions were similarly effective and acted independently. The authors recommended using the safest or least expensive medication first. Therefore, a total intravenous anesthetic medication in combination with dexamethasone was advocated as the first-line treatment method for the prevention of PONV. The authors also recommended that serotonin antagonists, among other antiemetic agents, should be used as a rescue treatment option. This recommendation supports our results, although the mechanism leading to thyroidectomy-related PONV might be different from PONV induced during other surgical intervention. In contrast to previous studies, we did not observe any effect of dexamethasone on postoperative pain.14 We assume that this result is due to the bilateral superficial cervical plexus block we applied to all patients in the study. Our patients, even in the placebo arm, reported mean pain scores at rest of 1 and lower at any time, rendering the measurement of pain reduction meaningless. For comparison, in previous studies, pain scores in the placebo group ranged between 3 and 5. Interestingly, in our multivariable analysis, length of surgery was significantly associated with a lower, not a higher, rate of PONV, whereas more extended surgical procedures (total vs hemithyroidectomy) were associated with a higher PONV rate as one would expect (although without reaching significance). This seemingly contradictory result is difficult to explain. An explanation could be that
Age, yr Sex Male Female Body mass index, kg/m2 Body surface, m2 Nationality Swiss German Other Diagnosis Graves disease Goiter, multinodular Goiter, uninodular Laterality of disease Bilateral Left Right Operation Hemithyroidectomy Near total thyroidectomy Total thyroidectomy Lymph node dissection (C1/level VI) No Yes Operation time, min Incision length, cm Specimen weight, g Histology of specimen PTC FTC Benign disease
Total (N ¼ 152)
Dexamethasone (n ¼ 76)
Placebo (n ¼ 76)
49.0 12.7 (21–81)
49.5 12.0 (21–76)
48.4 13.4 (22–81)
32 (21.1) 120 (78.9) 25.1 4.4 (17.2–37.8) 1.81 0.20 (1.44–2.55)
14 (18.4) 62 (81.6) 24.8 4.1 (18.0–37.8) 1.78 0.16 (1.45–2.24)
18 (23.7) 58 (76.3) 25.4 4.6 (17.2–36.0) 1.84 0.23 (1.44–2.55)
129 (84.9) 10 (6.6) 13 (8.6)
69 (90.8) 4 (5.3) 3 (3.9)
60 (78.9) 6 (7.9) 10 (13.2)
25 (16.4) 68 (44.7) 59 (38.8)
13 (17.1) 35 (46.1) 28 (36.8)
12 (15.8) 33 (43.4) 31 (40.8)
81 (53.3) 32 (21.1) 39 (25.7)
42 (55.3) 15 (19.7) 19 (25.0)
39 (51.3) 17 (22.4) 20 (26.3)
75 (49.3) 32 (21.1)
37 (48.7) 19 (25.0)
38 (50.0) 13 (17.1)
45 (29.6)
20 (26.3)
25 (32.9)
128 (84.2) 24 (15.8) 126.8 43.2 (39–305) 5.7 1.5 (2–10) 48.3 49.9 (4.8–390.0) 10 (6.6) 6 (3.9) 136 (89.5)
63 (82.9) 13 (17.1) 126.2 40.9 (39–266) 5.6 1.4 (2–10) 45.3 40.1 (4.8–230.0) 3 (3.9) 4 (5.3) 69 (90.8)
65 (85.5) 11 (14.5) 127.5 45.7 (60–305) 5.7 1.5 (3–9) 51.2 58.2 (6.9–390.0) 7 (9.2) 2 (2.6) 67 (88.2)
The values given are mean standard deviation (range) or number (percentage). FTC indicates follicular thyroid carcinoma; PTC, papillary thyroid carcinoma.
associated with the study treatment (dexamethasone), male sex, and duration of surgery. There was a tendency for increased rates of PONV to be associated with more extended operations, postoperative morphine use, and the amount of intraoperative fentanyl (Table 4); however, these associations were not statistically significant. The PONV rate was also significantly reduced after dexamethasone treatment when the multivariable model was used according to the PP analysis (OR ¼ 0.29; 95% CI, 0.13–0.63; P ¼ 0.002).
DISCUSSION This investigation provides compelling evidence that a single preoperative dose of dexamethasone administration reduces the rate of tPONV by half. Even after the exclusion of patients treated with opioids (which may induce PONV) in the PP analysis, the effect was still both statistically significant and clinically relevant. Every fourth patient medicated with dexamethasone benefited from the treatment. ß
2015 Wolters Kluwer Health, Inc. All rights reserved.
www.annalsofsurgery.com | 937
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
Annals of Surgery Volume 262, Number 6, December 2015
40
Tarantino et al
n=12
20
Percentage
30
PONV: Vomiting Severe nausea Mild nausea
n=3
n=12
10
n=3 n=1
n=3 n=8
n=2
n=1
n=15
0
n=9
4h Time: Arm: Dex Plc
n=5
n=7
n=3 n=1 n=1
Plc
Dex
8h Dex
n=1 n=2
n=3
Plc
Dex
16h
n=1
n=2
n=1
Plc
Dex
24h
n=2 n=1
n=2
Plc
Dex
32h
n=3
48h Plc
FIGURE 2. Time course of PONV in the intention-to-treat analysis. Bars represent the incidence of PONV at 6 time points after thyroidectomy in the dexamethasone and placebo arm. Distribution of severity is indicated by different shades of grey. The number of patients affected is indicated within the bars (n ¼). Severity of PONV was significantly reduced by dexamethasone (P ¼ 0.024), with a significant time interaction (P ¼ 0.008) (mixed model over the mean ranks of the PONV score). Dex indicates dexamethasone; Plc, placebo. TABLE 2. Secondary Endpoints (Intention-to-treat Analysis) Endpoint
Time/Unit
Pain (at rest) (VRS 0–10)
4h 8h 16 h 24 h 32 h 48 h 4h 8h 16 h 24 h 32 h 48 h
Pain (after physical stress) (VRS 0–10)
Length of postoperative stay, h Propofol, mg Remifentanil, mg Fentanyl, mg Rocuronium, mg Paracetamol total, g Paracetamol day 1, g Paracetamol day 2, g Metamizole total, g Metamizole day 1, g Metamizole day 2, g Morphine total, mg Morphine day 1, mg Morphine day 2, mg
Total (N ¼ 152)
Dexamethasone (n ¼ 76)
Placebo (n ¼ 76)
P
0.86 1.28 (0–8) 0.77 1.16 (0–6) 0.66 1.14 (0–7) 0.51 1.08 (0–8) 0.41 0.81 (0–4) 0.36 0.69 (0–4) 1.99 1.63 (0–8) 1.95 1.64 (0–7) 1.77 1.49 (0–7) 1.56 1.37 (0–8) 1.28 1.40 (0–6) 1.20 1.31 (0–5) 48.2 14.2 (22.2–120.0) 1132 422 (0–2846) 1961 1151 (18–9192) 436 109 (200–1000) 39.2 7.8 (0.0–70.0) 5.1 1.5 (0–8) 2.2 0.8 (0–4) 2.9 1.1 (0–4) 3.0 2.4 (0–8) 1.8 1.1 (0–5) 1.2 1.6 (0–5) 1.5 3.7 (0–17) 1.4 3.5 (0–15) 0.1 0.8 (0–10)
0.72 1.09 (0–4) 0.82 1.22 (0–6) 0.63 1.00 (0–4) 0.53 0.96 (0–5) 0.37 0.63 (0–2) 0.36 0.67 (0–3) 1.88 1.52 (0–5) 1.88 1.54 (0–6) 1.72 1.40 (0–5) 1.50 1.22 (0–5) 1.25 1.39 (0–5) 1.18 1.36 (0–5) 48.3 15.7 (22.2–120.0) 1097 422 (0–2257) 1944 1197 (18–9192) 427 111 (200–1000) 38.7 7.3 (0.0–50.0) 5.2 1.6 (0–8) 2.2 0.9 (0–4) 2.9 1.1 (0–4) 3.3 2.5 (0–8) 2.0 1.2 (0–5) 1.3 1.6 (0–4) 1.2 3.3 (0–15) 1.2 3.3 (0–15) 0.0 0.0 (0–0)
1.00 1.44 (0–8) 0.73 1.11 (0–5) 0.69 1.26 (0–7) 0.50 1.19 (0–8) 0.46 0.96 (0–4) 0.36 0.72 (0–4) 2.09 1.73 (0–8) 2.01 1.75 (0–7) 1.83 1.58 (0–7) 1.62 1.51 (0–8) 1.32 1.43 (0–6) 1.22 1.26 (0–5) 48.1 12.6 (25.8–92.8) 1168 422 (469–2846) 1978 1111 (573–5460) 444 108 (300–800) 39.7 8.3 (3.5–70.0) 5.1 1.4 (2–8) 2.2 0.8 (1–4) 2.9 1.1 (0–4) 2.8 2.3 (0–8) 1.7 1.0 (0–4) 1.1 1.6 (0–5) 1.7 4.0 (0–17) 1.6 3.7 (0–15) 0.2 1.2 (0–10)
0.947
0.674y
0.835z 0.374z 0.930z 0.360z 0.596z 0.474z 0.727z 0.629z 0.260z 0.156z 0.429z 0.391z 0.417z 0.156z
The values given are mean standard deviation (range). Pain measured on a VRS ranging from 0 (no pain) to 10 (worst pain imaginable). Mixed model over the mean ranks (time interaction: P ¼ 0.547). yMixed model over the mean ranks (time interaction: P ¼ 0.995). zMann-Whitney U test. VRS indicates Verbal Rating Scale.
938 | www.annalsofsurgery.com
ß
2015 Wolters Kluwer Health, Inc. All rights reserved.
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
Annals of Surgery Volume 262, Number 6, December 2015
Dexamethasone for PONV Prevention
TABLE 3. Adverse Events Grade
Adverse Event Surgical adverse events Recurrent laryngeal nerve palsy Hoarseness Bleeding Hypoparathyroidism Bleeding requiring reoperation Medical adverse events Coughing Sinus tachycardia Temporary compensated renal insufficiency Allergic reaction (to metamizole) Hypocalcemia Total adverse events
Total (N ¼ 152)
Dexamethasone (n ¼ 76)
I I I II IIIb
4 1 1 4 1
(2.6) (0.7) (0.7) (2.6) (0.7)
2 1 1 2 1
1 1 1 2 2
1 1 1 1 2 17
(0.7) (0.7) (0.7) (0.7) (1.3) (11.1)
1 (1.3)
Placebo (n ¼ 76)
Py
2 (2.6)
1.00 0.32 0.32 1.00 0.32
(2.6) (1.3) (1.3) (2.6) (1.3)
8 (10.5)
2 (2.6)
1 (1.3) 1 (1.3) 1 (1.3) 2 (2.6) 9 (11.8)
0.32 0.32 0.32 0.32 0.16 0.80
The values given are number (percentage). All adverse events were unrelated to the study treatment. Surgical adverse events: Dindo-Clavien classification (I–V)12; medical adverse events: CTCAE grade (1–5).13 2 y x test. CTCAE indicates Common Terminology Criteria for Adverse Events.
more and extended surgical manipulations lead to a higher PONV rate that correlates with the assumption that tPONV is caused by edema and inflammation around the neck tissues, leading to evoked parasympathetic impulses through the vagus, recurrent laryngeal, and glossopharyngeal nerves to the vomiting center.2,24,25 In our study, we applied bilateral superficial cervical plexus block in all patients. Is it possible that this not commonly applied procedure might influence the PONV rate or the efficacy of
dexamethasone? In a randomized controlled study, very similar to ours but without application of a cervical plexus block, the PONV rate in the control arm was 60% compared with 57% in our study.15 Thus, an influence of the cervical block on PONV seems unlikely. The efficacy of dexamethasone was somewhat less in this study (risk reduction of 36% compared with 50%), but this difference is within the limits of normal interstudy variation and probably not due to the lack of a cervical block.
TABLE 4. Logistic Regression Analysis for Incidence of PONV (Intention-to-treat Analysis) Univariate
Treatment Placebo Dexamethasone Age, yr Sex Male Female Body mass index, kg/m2 Nationality Swiss German Other Diagnosis Goiter Graves disease Operation Hemithyroidectomy (Near) total thyroidectomy Lymph node dissection No Yes Operation time, h Postoperative morphine No Yes Propofol (100 mg) Remifentanil (100 mg) Fentanyl (100 mg) Rocuronium (10 mg)
Full Multivariate Model
Optimized Modely
OR (95% CI)
P
OR (95% CI)
P
OR (95% CI)
P
Reference 0.31 (0.16–0.61) 1.01 (0.99–1.04)
0.001
Reference 0.27 (0.12–0.57) 1.01 (0.98–1.05)
0.001
Reference 0.26 (0.12–0.54) —
