• •

  To estimate the effectiveness of haloperidol for the prophylaxis of PONV in adults undergoing surgery, diagnostic or therapeutic procedures under general or regional anesthesia or under monitored anesthetic care.

• •

  To estimate the frequency of adverse effects associated with the administration of haloperidol.

• •

  Estimate the need for therapeutic antiemetic agents following the prophylactic administration of haloperidol for PONV.

• •

  To assess whether the risk of PONV changes depending on the route of administration of haloperidol.

• •

  To evaluate whether the risk of PONV changes depending to the time of administration of haloperidol.

• •

  To evaluate whether the risk of PONV changes according to the dose of haloperidol administered.

Systematic literature review and met-analysis of clinical trials.

• •

  Type of studies: Randomized clinical trials, both published and unpublished, evaluating the effectiveness of haloperidol in the prevention of PONV.

• •

  Type of participants: Adult patients undergoing diagnostic or therapeutic procedures under general/regional anesthesia or sedation.

• •

  Types of interventions: Haloperidol, at any dose or route of administration, any moment prior to the occurrence of PONV versus placebo, another drug or no treatment. The drug could be administered in the preoperative period, during the induction of anesthesia, in the intraoperative or postoperative period (prior to the occurrence of nausea and/or vomiting).

• •

  Types of outcomes: Incidence of nausea, vomiting, PONV, rescue remedies requirement, Q-T segment disorders and any haloperidol-related adverse effect.

• •

  Studies evaluating combined therapies in the haloperidol group, without specifying individual antiemetic effects.

• •

  Trials with less than 20 participants.

The e-search of the literature was performed in the following databases: Medline, EMBASE, CINAHL and The Cochrane Controlled Trials Register. The search used a combination of key words (as described in the search terms) and a filter recommended in Medline's Clinical Queries section with 93% sensitivity and 97% specificity to identify randomized clinical trials. The search had no restrictions with regard to language or year of publication.

• 1.

  (MESH) AND haloperidol (MESH) or postoperative (ALL FIELDS) AND nausea (ALL FIELDS) AND vomiting (ALL FIELDS) AND Haloperidol (ALL FIELDS) MeSH-NAUSEA OR NAUSEA* OR INAPPETENCE

• 2.

  MeSH-VOMITING OR VOMIT* OR EMESIS OR EMET*

• 3.

  MeSH-POSTOPERATIVE NAUSEA AND VOMITING OR POSTOPERATIVE NAUSEA AND VOMITING

• 4.

  Randomized Controlled Trials

• 5.

  #1 OR #2 OR #3

• 6.

  MeSH-POSTOPERATIVE OR POST-OPERATIVE

• 7.

  MeSH-ANESTHESIA OR ANAESTHESIA OR ANESTHET* OR ANAESTHET*

• 8.

  MeSH-HALOPERIDOL

Filter: (randomized controlled trial [Publication Type] OR (randomized[Title/Abstract AND controlled[Title/Abstract]AND trial[Title/Abstract]).

A search of the unpublished literature was accomplished by contacting the most representative authors and the haloperidol pharmaceutical manufacturers. The search was done on the references of the selected articles, PONV clinical practice guidelines, editorials and relevant review articles.

Two independent researches carried out the search and any discrepancies were settled by consensus.

The titles of the articles found in the search were reviewed to identify any relevant articles. Then a selection by abstract was done, and an attempt was made to get the full text of the selected article. The inclusion and exclusion criteria by two independent authors were used and any conflicts were settled by consensus.

Two independent researchers reviewed the selected articles in order to extract the information about the participants, the interventions and outcomes. Any differences were agreed upon by consensus.

The quality evaluation of the selected trials was performed using the instrument suggested by the GRADE Guidelines (Guyatt G.H.) and the CONSORT 2010 Statement checklist (Murphy J.F.), for the evaluation of clinical controlled trials. Two independent reviewers carried out the evaluation and any differences were settled by consensus.

Initially, a descriptive analysis of the information extracted and the quality of the primary trials was performed. The quantitative analysis was done using the Cochrane collaboration RevMan5 software.

An attempt was made to contact the authors in order to obtain the missing data of the primary studies. If the data could not be obtained, the information was excluded from the analysis. The selected trials for which the complete text was not available (Maggi and Dyberg) were included, despite the inability to complete the planned quality analysis; the data were obtained from the systematic review published by Buttner in 2004. The decision to include the data was made based on the fact that the sample size of the relevant references was representative for the analysis. A sensitivity analysis was done, excluding these articles, in order to measure their impact on the overall effectiveness estimate.

The subgroup analysis was performed on the basis of the outcome measurements times, the type of antiemetic control, time of administration of haloperidol, and the route of administration of haloperidol.

A sensitivity analysis was carried out to assess the impact of the data included from sources of articles not available in their complete text and based on the methodological quality of primary studies.

10 trials published since 1962 to 2010 were analyzed and these included 2711 participants. The age of the participants ranged from 15 to 75 years. One of the trials was performed in patients who had undergone laparoscopic cholecystectomy (Chu), two trials on gynecological laparoscopy (Aouad and Wang), another article in open gynecological surgery (Chu) and the rest included mixed surgical population. The doses of haloperidol used ranged from 0.25mg to 5mg.

The time of administration of prophylactic treatment for PONV was following the induction of anesthesia in 4 trials and during the intraoperative period in another 4 trials. Two trials did not specify the time of administration and it was impossible to obtain the information when trying to contact the authors.

The control drugs used were ondansetron, droperidol, dexametasone and placebo.

The outcomes included in the trials were incidence of nausea, vomiting, PONV, evaluated at 2h, 4h, 6h and 24h, in addition to the need for rescue antiemetic agents and prolongation of the QT interval.

We were unable to obtain data on the quality of the experiments published by Dyberg in 1962 and Maggi 1964. The outcome data were obtained from the meta analysis published by Buttner in 2004 and the appendixes to this review, published over the Internet.22–25

All the trials included are randomized clinical trials. Six of the trials included reported patients lost to follow up, but none of these losses impacted the validity of the results. Two of the trials had no patients lost to follow up and other two trials do not report any related data. Only three trials use the intention-to-treat analysis but the rest do not report any related data.26–29

Eight of the trials included report that the participants were blind; eight trials report that the treating physicians were blind and in six the evaluations were blind. Six of the trials specify the development and blind nature of the randomization process.26–34

The quality of the experiments and the amount of information reported in the articles improved according to the year of publication, probably because of the development of checklists that facilitate the job of the researchers in organizing and assembling the reports of the trials. The articles included in this systematic review are average to high quality. Eight of the articles meet 5 or more of the 7 items evaluated under Guyatt's scale. The two remaining articles (Tornetta and Maggi) do not meet the quality criteria evaluated. The quality evaluation suggested by CONSORT and applied to our meta analysis highlights the fact that 5 articles met at least 11 of the items; 1 article meets 9 of the items and 2 articles (Dyberg and Maggi) have no data available to do the quality evaluation.

In view of the clinical heterogeneity accounted for by diversity of the control interventions, we decided to undertake a subgroup analysis by type of control.

Six of the studies included compare haloperidol against placebo for the prophylaxis of PONV. As a whole, the reports favor the use of haloperidol, but it was not possible to do the quantitative synthesis due to the broad clinical heterogeneity resulting from the broad range of doses used; hence, we decided to do the per dose analysis.

Just Tornetta in 1972 evaluated the prophylactic effect of I.M. haloperidol 0.25mg, 45min before the induction of anesthesia (Fig. 1).

Fig. 1.

Haloperidol 0.25mg.

(0.07MB).

There is no statistically significant difference in the incidence of PONV in the first 6 postoperative hours, which was the only time period evaluated. However, Buttner reported that there is no significant difference in PONV at 24h: RR 0.57 95% CI (0.18–1.28) for nausea and RR 0.85 (0.4–1.79) for vomiting.

Only Tornetta used I.M. haloperidol 0.5mg, 45minutes prior to surgery (Fig. 2).

Fig. 2.

Haloperidol 0.5mg.

(0.06MB).

The control group has a larger number of participants. This study was developed in two phases: the initial phase included 4 groups with doses of 0.5, 1 and 2mg, versus placebo; the second phase evaluated 0.25mg and 4mg doses versus placebo.

There is a significant difference in the prevention of PONV at 6h. According to Buttner, the differential analysis of the incidence of nausea and vomiting at 24h indicates a RR 0.48 95 CI % (0.24–0.96) for nausea and RR 0.28 95% CI (0.08–0.63) for vomiting. There is no differential decrease in the incidence of nausea or vomiting (Fig. 3).

Fig. 3.

Haloperidol 1mg.

(0.17MB).

Aouad studied the participants who underwent laparoscopic surgery; Parlow focused on urological and lower limbs surgery under regional anesthesia with intrathecal morphine and Wang focused on gynecological laparoscopy. The total effect shows benefit with the administration of haloperidol 1mg in the prevention of nausea and vomiting at 24h. There is moderate heterogeneity, probably accounted for in Parlow's trial (intrathecal morphine-associated anesthesia) with a baseline risk of PONV different from the other trials included in the comparison; nevertheless, we do not know the PONV risk difference among the various types of surgeries and anesthetic techniques.

Haloperidol's route of administration in Aouad and Wang's trials is intravenous, while in Parlow's it is intramuscular, which could account for the heterogeneity. Tornetta was excluded from this analysis because he only evaluated the outcome after 6h.

In conclusion, there is a 30% PONV risk reduction with haloperidol versus placebo 95% CI (0.3–49%, P: 0.03).

Need for rescue antiemetic agents

In Aouad's trial, patients initially received intravenous promethazine 12.5mg and ondansetron 4mg when symptoms persisted. In Parlow's trial, patients received I.V. diphenhydrinate 50mg and if symptoms persisted they were administered I.V. prochloperazine 10mg and ondansetron 4mg. Wang's trial does not report the specific rescue antiemetic agents (Fig. 4).

Fig. 4.

Need of rescue antiemetics.

(0.16MB).

Haloperidol decreases the need for rescue antiemetic agents by 60%. The major effect is evidenced in Wang's trial with a higher weight of the analysis because of a larger sample size. The uncertainty in the other trials results from the small sample size; however, the specific estimates favor the use of haloperidol. There is no heterogeneity among the trials included in this analysis

Sedation

Aouad and Wang's trials measured sedation up to 2h after surgery, using the 0–10 numerical rating scale; there were no differences in the sedation rating scores between the groups (Fig. 5).

Fig. 5.

Haloperidol 2mg.

(0.15MB).

This analysis included the trials comparing haloperidol's 2mg effectiveness versus placebo in the prophylaxis of PONV at 24h. The overall estimate favors the haloperidol group, with statistical significance, reduced incidence of PONV by 34% at 24h and 95% CI (18–49%). The effect is the same for nausea 0.71 95% CI (0.51–0.99) and vomiting 0.61 95% CI (0.41–0.89) (Fig. 6).

Fig. 6.

Haloperidol 2mg vs. placebo in the first 12h.

(0.1MB).

In this analysis we included the mean outcomes at 2h by Chu, at 12h by Parlow and at 6h by Tornetta. In the light of the heterogeneity, it was not possible to do a meta-analytical summary; however, two of the trials do evidence the protective effect of haloperidol at 6 and 12h. There is no statistically significant difference in the incidence of PONV in the first 2h (Fig. 7).

Fig. 7.

Need for rescue antiemetic agents.

(0.15MB).

Chu used ondansetron 4mg as rescue medication, while Parlow used I.V. diphenhydrinate 50mg; when symptoms persisted, I.V. prochloperazine 10mg and ondansetrón 4mg were administered. There is evidence of a reduced need for the use of rescue agents in the patients receiving haloperidol versus placebo: 53% CI (18–73%).

Chu measured the QTc interval before and 10min after the administration of haloperidol. No changes were found in the length of the interval following the administration of the agents.

Haloperidol 4mg

The studies comparing haloperidol 4mg against placebo are Tornetta 1972 and Maggi 1964. The data are taken from Buttner's 2004 meta-analysis (Fig. 8).

Fig. 8.

Haloperidol 4mg.

(0.15MB).

Maggi randomized 140 patients to receive either haloperidol 4mg or placebo at the end of surgery. Tornetta administered 4mg of I.M. haloperidol prior to the induction of anesthesia. A 38% reduction in the incidence of postoperative nausea and vomiting was observed 95% CI (36–78%). Despite the difference in the routes of administration, the effect is maintained.

Dyberg evaluated the effectiveness of I.V. haloperidol 5mg, administered during the course of surgery for prophylaxis of PONV in a mixed surgical population. Follow up was for 24h. The data are taken from Buttner's systematic review and show a reduction in the risk of nausea of 63% 95% CI (54–70% P<0.00001) and a 70% reduction CI (59–78% P<0.00001).

Chu and Wang compared the effectiveness of haloperidol and droperidol for the prophylaxis of PONV. Chu included women over 18 years of age, undergoing laparoscopy-assisted vaginal hysterectomy to compare I.V. haloperidol 2mg versus droperidol 1.25mg, 15min after the induction of anesthesia. Wang randomized adult women undergoing gynecological laparoscopy to either I.V. haloperidol 1mg or droperidol 0.625, 15min prior to the induction of anesthesia (Fig. 9).

Fig. 9.

Incidence of PONV at 24h.

(0.15MB).

There is no difference in the incidence of PONV at 4h, RR 0.97 95% CI 0.52–1.79 P: 0.91, more between 4 and 24h RR 0.99 95% CI (0.64–1.53 P: 0.96).

Need for rescue agents

I.V. Ondansetrón 4mg was administered in both studies as rescue medicine to patients reporting intolerable nausea and vomiting (Fig. 10).

Fig. 10.

Need for rescue medicines.

(0.14MB).

There is no statistically significant difference in the need for rescue medicines between groups in both trials. The measurements for this outcome were taken at 4h in Wong's trial and at 24 in Chu's.