Abstract
Christopher James Ryan, University of Sydney and Department of Psychiatry, Westmead Hospital, Sydney, Australia:
Furrows [1] calls for a review of the practice of intravenous sedation with haloperidol. He is alarmed that this practice is apparently common in Australia's acute admission wards chiefly because of his concern that intravenous haloperidol may precipitate the life-threatening dysrhythmia: Torsades de Pointes. I share Furrows concern about sedation with intravenous haloperidol, but my reasons are different.
Haloperidol has been associated with Torsades de Pointes, as have all other antipsychotics. Despite this, the risk of inducing Torsades with intravenous haloperidol in most psychiatric patients is very small. The numerous reports of the dysrythmia occurring in temporal association with the drug must be seen in context. First, the patients in whom dysrythmias have arisen have almost always been critically ill. Haloperidol appears to be the safest of the antipsychotics from a cardiac point of view and for this reason it is used routinely in intensive care units. Patients in this setting are always very ill, and are often elderly. Second, in most instances the doses of haloperidol used when Torsades has been reported have far exceeded those used by psychiatrists. For example, in the Wilt paper [2] cited by Furrows, the four patients that developed Torsades received the following doses: 580 mg over 4 days, 170 mg in 24 h, 489 mg in 36 h and 10 mg over 4 h. The last case was a 74-year-old woman with cardiac failure, atrial fibrillation, hypothyroidism and renal failure. These four patients were the only four to develop Torsades out of 1100 patients treated with intravenous haloperidol in this unit over 3 years. Hunt and Stern [3], also cited by Furrows, provide an excellent recent review of Torsades and haloperidol. They also describe three cases of their own where patients developed Torsades with cumulative doses of between 44.5 and 1150mg of haloperidol. All three patients had at least one other independent risk factor for developing Torsades.
Despite haloperidol's relative cardiac safety in acute psychosis, Furrows' implication that it should not generally be used for intravenous sedation is well justified. Haloperidol is a poor choice as a sedating agent because haloperidol is not very sedating. Haloperidol's chief effect is as an antipsychotic. There is no evidence that its antipsychotic activity is immediate. When given as a sedating agent in acute psychosis it is almost always given with diazepam and, despite some evidence of augmentation, most of the behavioural control derives from the benzodiazepine [4].
Acute parenteral sedation of the acutely psychotic patient in the emergency department or admission ward is generally best achieved by the sole use of a benzodiazepine. Although diazepam is often used, it too is a poor choice because: (i) it cannot be given intramuscularly, exposing staff to the dangers of finding a vein; and (ii) it has an enormously long half-life so that the patient may remain under its influence for days. Midazolam, in contrast, is rapidly effective given intramuscularly [5]. It is well tolerated, with no risk of Torsades and a relatively limited risk of respiratory depression. Its half-life is short, but having removed the heat from the crisis, its functional effect on the patient usually continues long after its pharmacological effect has ceased. It has no antipsychotic action of course, but as antipsychotics take a considerable time to take effect there is generally little to be lost by a brief delay in their administration. The administration of intramuscular midazolam allows the clinician to gain control of the acute crisis without forfeiting the luxury of calm reflection before deciding whether or not to use an antipsychotic and which one to use.