Scientific Papers

Antiemetic prophylaxis with droperidol in morphine-based intravenous patient-controlled analgesia: a propensity score matched cohort study | BMC Anesthesiology

Clinical setting and patient selection criteria

This study was approved by the Taipei Medical University – Joint Institutional Review Board, Taipei, Taiwan (approval number: TMU-JIRB-N202205095; date of approval: 9 June 2022). The need for written informed consent was waived by the Institutional Review Board due to the retrospective nature of this research. All methods were performed following the standards of the Helsinki Declaration and relevant study guidelines [23].

We used the electronic medical record database of Shuang Ho Hospital, Taipei Medical University to collect retrospective data from 1,512 consecutive patients who underwent surgery with general or neuraxial anesthesia and received opioid-based IV-PCA for postoperative pain control from January 1, 2020 to November 30, 2022. The exclusion criteria were: duplicate cases, missing data of IV-PCA dosage, age < 20 years, using non-morphine analgesics for IV-PCA, and switching to a droperidol regimen during IV-PCA. The included patients were classified into the droperidol group and control group based on whether or not they received the addition of droperidol to IV-PCA. Data were extracted by two independent resident anesthesiologists, who were not involved in the data analysis. The quality of the datasets was validated using random sampling by other authors.

Intravenous patient-controlled analgesia protocol

Contraindications to IV-PCA were the inability to maintain consciousness, cognitive impairment, and postoperative mechanical ventilation support or intensive care beyond 24 h. IV-PCA was typically initiated at the post-anesthesia care unit after surgery and administered using an ambulatory infusion pump (CADD®-Solis Infusion System, Smiths Medical, Inc., Minneapolis, MN, USA), programmed to deliver morphine sulfate 1 mg/mL in normal saline [24, 25]. The infusion settings were a loading dose of 0–5.0 mL, a demand dose of 0.5–2.0 mL, a basal infusion rate of 0–1.5 mL/hr, and a lockout time of 5–10 min. As an antiemetic prophylaxis, we used a droperidol regimen of 0.025–0.075 mg/mL added to the IV-PCA infusate based on previous literature [15,16,17,18,19,20,21,22, 26, 27]. In the non-droperidol regimen, no antiemetic was added to the morphine solution. The pain service team evaluated the patients’ response at 12-hourly intervals, and more frequently in patients with inadequate analgesia or relevant adverse events (e.g., nausea, vomiting, and sedation). The severity of PONV was rated using a 4-point verbal descriptive scale: no PONV: no complaint of nausea or vomiting; mild PONV: patients complained of nausea but refused antiemetic drugs; moderate PONV: patients complained of nausea and requested antiemetic drugs; severe PONV: patients complained of nausea and had episodes of vomiting requiring antiemetic treatments [20, 28]. For patients with mild PONV, the IV-PCA infusion parameter was adjusted to reduce morphine dosage. When patients had moderate-to-severe PONV, antiemetic medications were administered, and IV-PCA infusion rates were reduced to relieve the symptom. The pharmacologic treatment included dopamine antagonists (e.g., metoclopramide and prochlorperazine), 5-HT3 antagonists (e.g., ondansetron), corticosteroids (e.g., dexamethasone), and histamine antagonists (e.g., diphenhydramine). In most patients, IV-PCA was used for 48 to 72 h and switched to oral acetaminophen or nonsteroidal anti-inflammatory drugs thereafter.

Study outcomes

The primary outcome was the rate of any episodes of nausea and/or vomiting within 72 h after surgery. The secondary outcomes were the rate of PONV which needed rescue antiemetic medications, the severity of PONV, unintentional sedation within 72 h after surgery, postoperative opioid consumption, and daily maximum pain scores within 72 h after surgery. The occurrence of PONV, level of sedation, and pain intensity were evaluated regularly by certified nurse anesthetists of the pain service team at 12-hourly intervals at the institution. The medical records were reviewed to determine whether the patients received antiemetic medications for PONV during IV-PCA. The University of Michigan Sedation Scale (UMSS) grading system was used to evaluate the sedation level during IV-PCA, as follows: 0, fully awake; 1, drowsy with closed eyes; 2, easily aroused with light tactile stimulation or simple verbal commands; 3, arousable only by strong physical stimulation; 4, unarousable [29]. Unintentional sedation was defined as a maximum UMSS score ≥ 1 after excluding planned sedation. Postoperative pain intensity was evaluated both at rest and during movement using a self-reported 11-point numeric rating scale (NRS) with response options from “no pain” (0) to “the worst pain” (10).

Anesthesia management

All patients received a 12-lead electrocardiogram before surgery to rule out clinically important QTc prolongation and severe cardiac arrhythmia. General anesthesia was typically induced using fentanyl 1–2 µg/kg and propofol 1–2 mg/kg. Rocuronium 0.6–1.0 mg/kg or cisatracurium 0.1–2.0 mg/kg was given for endotracheal intubation. Inhalational sevoflurane or desflurane was used to maintain general anesthesia. Reversal agents were always administered when neuromuscular blocking agents were used, including sugammadex 2 mg/kg or neostigmine 0.05 mg/kg. For spinal anesthesia, bupivacaine 6–15 mg without opioids was used. In combined general and neuraxial anesthesia, we used a continuous epidural infusion of ropivacaine 5 mg/mL with or without fentanyl 2.5–5 µg/mL. Midazolam 2–5 mg was given intravenously during the induction of general anesthesia or neuraxial anesthesia for anxiolysis on a case-by-case basis. In perioperative fluid management, crystalloid fluids (sodium chloride 0.9% or lactated Ringer’s solution) were administered according to the current practice guidelines [30].

Covariates for adjustment

To adjust for potential confounding factors, the following patient and clinical covariates were collected based on the available data, physiological plausibility, and existing literature [3, 31]. Demographic attributes were age, sex, and body mass index. The recorded preoperative clinical covariates were the American Society of Anesthesiologists physical status, current cigarette smoking (within 30 days before surgery), previous history of PONV, coexisting diseases (hypertension, diabetes mellitus, major depression, and malignancy), and preoperative blood tests (hemoglobin, creatinine, aspartate aminotransferase, alanine aminotransferase, and estimated glomerular filtration rate based on the Cockcroft-Gault formula) [3, 32]. Intraoperative variables included the site of surgery (categorized into extremity, head and neck, breast, upper abdomen, lower abdomen, thorax, spine, and other), uses of laparoscopic or robotic techniques, types of anesthesia, use of volatile anesthetics, total intravenous anesthesia, duration of anesthesia, intraoperative blood loss and fluid volume, intraoperative use of dexamethasone and midazolam, intraoperative use of non-steroidal anti-inflammatory drugs, intraoperative opioid dosage, and type of neuromuscular blockade reversal agent (neostigmine, sugammadex, or nothing) [3, 31]. Postoperative factors included the duration of IV-PCA, postoperative use of non-steroidal anti-inflammatory drugs, and opioid consumption within 72 h after surgery [3]. The dosages of non-morphine opioids were transformed into morphine milligram equivalents for analysis (Supplementary Table S1) [33, 34].

Statistical analysis

Normality of the included variables was checked using the Shapiro-Wilk test and Anderson-Darling test. Normally distributed data were expressed as mean ± standard deviation. Non-normally distributed variables were presented as median with inter-quartile range and log-transformed to reduce distribution skewness, including preoperative blood test results, intraoperative blood loss and fluid volume, duration of anesthesia, duration of IV-PCA, and intraoperative and postoperative opioid consumption. To minimize potential confounding effects, a propensity score matching procedure was performed as follows. First, non-parsimonious multivariable logistic regression analyses were used to estimate a propensity score for the patients who did and did not receive droperidol. The patients who received droperidol were matched to those who did not in a ratio of 1:1 using a greedy matching algorithm within a caliper width of 0.05 standard deviations of the log odds of the calculated propensity score and without replacement to adjust for all of the collected covariates. Baseline patient characteristics were compared between the patients who did and did not receive droperidol using the absolute standardized mean difference (ASMD) [35]. Covariate balance between groups was defined as an ASMD less than 0.1 [36]. Conditional logistic regression analyses were used to evaluate associations between the collected variables and PONV. The significant variables in univariate analysis were incorporated into multivariable models to calculate adjusted odds ratios (aORs) with 95% confidence intervals (CIs) for the independent factors of PONV. For sensitivity analysis, the inverse probability treatment weighting (IPTW) method was used to eliminate potential confounding effects of imbalances in the collected covariates, as previously described [37]. Briefly, the inverse of estimated propensity score was used for weighted logistic regression analyses. We truncated 1% of subjects that were at the end of weighting distribution to decrease the effect of the large weights. Subgroup analyses according to Apfel’s simplified risk score (i.e., female sex, non-smoker, history of PONV, and postoperative opioids) [38], age, sex, current smoker or not, previous PONV or not, type of anesthesia, use of volatile anesthetics or not, use of neostigmine or sugammadex, and intraoperative use of dexamethasone or not were also conducted to examine the association of droperidol with PONV in these strata. The aORs were transformed into relative risks using Zhang and Kai’s method [39]. A meta-analysis by Weibel et al. showed that droperidol reduced the risk of PONV by 39% compared to placebo in adults undergoing general anesthesia [13]. For sample size estimation, at least 347 patients in each group were required to detect a relative risk of 0.61 between the droperidol group and control group, accepting a type I error of 5% and type II error of 20% with an anticipated PONV rate of 20% in the control group [13, 40]. The number of patients enrolled in this cohort substantially exceeded the minimum necessary sample size. Furthermore, Austin et al. indicated that at least 20 events per variable are required in logistic regression analyses [41]. The number of PONV events in the matched cohort was 187, suggesting that a maximum of 9 variables could be considered in the multivariable model without affecting the model performance. Since we included 3 variables in the final model, our analyses met the requirement of this criterion. A two-sided p value of < 0.05 was used to define a statistically significant difference. All statistical analyses were conducted using SAS software, version 9.4 (SAS Institute Inc., Cary, NC, USA).

Source link