Marco Sivilotti

Years since first academic appointment: More than 16 years

Title: Haloperidol Versus Ondansetron for Cannabis Hyperemesis Syndrome (Havoc): A Randomized Controlled Trial

Full author list: Aaron J. Ruberto, Marco L.A. Sivilotti, Savannah Forrester, Andrew G. Day

Institution: Queen’s University


Despite intense interest in the legalization of cannabis for medicinal and recreational use, little is known about the adverse effects of chronic use. Emergency physicians have increasingly encountered and recognized so-called “cyclic vomiting,” or cannabis hyperemesis syndrome, a debilitating condition characterized by paroxysms of sustained vomiting associated with chronic heavy use of cannabis. First described in a 14-patient case series in 2004 yet still poorly understood, this dramatic emergency presentation is remarkable in part for its resistance to standard anti-emetic treatment, and for a high rate of recidivism. Recently, investigators in Australia have proposed, and our own experience corroborates, that haloperidol may be the anti-emetic of choice in these patients.

Research Question

In non-pregnant adults presenting to the emergency department with recurrent hyperemesis attributed to chronic cannabis use, is intravenous haloperidol (at either 0.05 mg/kg or 0.1 mg/kg) superior to 8 mg ondansetron in reducing nausea and abdominal pain by 2 hours?

Trial Design

We propose a blinded, randomized, cross-over clinical trial in which patients will serve as their own control on subsequent presentations. Specifically, this three-period, balanced trial with >7-day washout will primarily aim to compare haloperidol regardless of dose to ondansetron. Eligible subjects will provide explicit written consent on their index visit, and encouraged to return to the same study site for any subsequent bout of hyperemesis to allow cross-over. We anticipate moderate attrition in which some subjects return only once, or never. To maximize the number of patients who will receive both ondansetron and at least one dose of haloperidol, participating patients will be randomized in equal numbers within site to one of the following 4 treatment sequences: 1-OHh, 2-OhH, 3-HOh, 4-hOH (O=Ondansetron, h=low dose of, and H=high dose of haloperidol).

Inclusion/Exclusion (summary):

We hope to enroll enroll 80 adults with recurrent bouts of witnessed vomiting and self-reported frequent use of cannabis (i.e. ~daily x 6 months), and a working emergency diagnosis of hyperemesis cannabis. Daily use of opioids, pregnancy, recent intravenous antiemetics or antipsychotics, severe mental health or language barriers, and intolerance to study medications will preclude enrollment.

Statistical Analysis

The primary goal is to compare haloperidol at either dose with ondansetron. The primary outcome is the average change in abdominal pain and nausea scores at 2 hours vs. baseline using a visual analog scale.  Our primary analysis will only utilize the first two periods and will include all patients regardless of whether they return for a second period.  We will estimate the difference in outcome (with 95% confidence interval) between haloperidol and ondansetron using a linear mixed effect model with a random patient effect and fixed effects of treatment, period, sequence and site. This model makes efficient use of data when a considerable number of patients do not complete both periods. If the primary analysis identifies a significant benefit of haloperidol (one-sided alpha of 0.025), then we will compare the two doses of haloperidol using a mixed effects model based on all three periods. If <25% of the patients return for the second period, however, we will switch our primary analysis to an analysis of variance using only the first period.


  1. Do you see this condition and are you interested in participating as a site?
  1. Will you be able to identify potentially eligible subjects when already given antiemetics on index visit, yet consent in anticipation of recidivism in the near future?
  1. Is it too much to ask to answer more than one question with an RCT?