Undifferentiated Agitation in the ED: A new RCT

When I was preparing an M+M last fall, I came across this notable study called Management of Acute Undifferentiated Agitation in the ED: A Randomized Double-Blind Trial of Droperidol, Ziprasidone, and Midazolam . It’s by Martel et al (including Michelle Biros, who’s editor of Academic Emergency Medicine), and appeared in Academic Emergency Medicine in December 2005 (Vol 12, No 12, pp1167 – coincidentally, right after Dr. Richardson’s EMPATH study).

It’s a good study that we might otherwise overlook, because it came out on the eve of the ACEP guidelines for agitation management and thus, wasn’t included in that extensive lit review.

We all know how important it is to manage agitation — for our safety, and also from a diagnostic perspective to figure out what’s really wrong with the patient. If you’ll recall from my talk, interpreting the agitation management studies isn’t easy — there are different scoring scales, different modes of delivery. And, of course, different practice environments. Some studies exclude alcohol agitation, others populations are dominated by alcoholics. Some studies seem applicable to us, until you find out it’s a study of consenting psych ward patients.

So these authors tried to keep it straightforward to better aid practice — they looked at a convenience sample of adults presenting to a large urban ED with undifferentiated agitation — could be drugs, psych, anything. A “public consent” was used — local facilities and shelters were notified of the study, and therefore individuals could be enrolled in the study, exempt from informed consent (proxies were used when available).

After being selected by an ED doc as having “undifferentiated agitation, the patients (assuming they were not pregnant, not incarcerated, and not allergic) were randomized to receive either 5 mg of midazolam (versed) IM, 5 mg of inapsine(droperidol) IM, or 20 mg of ziprasidone (geodon) IM.

College and medical students then scored the patients on the Altered Mental Status Scale (AMS – a modified version of BARS) at the time of med administration and frequently for two hours. Pulse ox and end-tidal CO2 were also monitored, which I believe is a first for agitation sedation studies.

They enrolled about 50 patients in each arm of the study. Though most characteristics were randomized, the midazolam arm ended up with more head-injured patients. It’s important to note that 90% of their patients were eventually shown to have alcohol on board.

What did they find? Midazolam and droperidol are fastest — adequate sedation (an AMS score less than or equal to 0) was achieved at 15 minutes after the IM dose, compared to 30 minutes for those receiving ziprasidone.

Specifically, 33 of the 48 midazolam patients were sedated at 15 minutes, and 30 of the 50 droperidol patients were sedated. Only 18 of the 46 zirprasidone (geodon) patients were sedated at 15 minutes, though by 30 minutes all groups were statistically similar, and by 45 minutes, the midazolam group was starting to get agitated again (14 of 48 were agitated, compared to droperidol’s 9 of 50, and ziprasidone’s 9 of 46.

Time to discharge / dispo was not affected by drug choice — it was always around 6-7 hours. The droperidol patients needed the least amount of “rescue” sedation — only five patients, for a total of six doses. For geodon, nine patients needed 11 doses, and significantly, 24 midazolam patients — half of them — needed more sedation (30 doses altogether).

A few patients (six altogether) got akathesia. Interestingly, four of those patients were geodon recipients — I thought that was a selling point of these new antipsychotics.

The respiratory issue was of particular concern to the authors, and yet they didn’t really report anything specific about end-tidal CO2 results. About half of all patients had some respiratory depression, no significant differences between arms. Twenty of the 48 midazolam (versed) patients needed O2, but that was not significantly worse than the other groups. No one got intubated. No one had dysrhythmias.

So, I liked this study because they took pains to make it applicable to practice — they didn’t try to separate out the causes of agitation. Nor did they lose patients by trying to obtain individual informed consent. Unfortunately, the vast majority of their study group had alcohol on board (although that may better simulate the EHC experience). As one critic noted, if the undifferentiated agitation was due to cocaine or alcohol withdrawal, you give benzos and don’t want to waste time with other drugs. Furethermore,

The effectiveness of sedation in this study was defined as an AMS score of less than 0; however, we think that the most important measures of effective sedation are the time to establishment of an intravenous line, as well as determination of a fingerstick glucose level and a rectal temperature, thus allowing for administration of medications and identifying rapidly reversible causes of AUA.

It’s a good point, but if you interpret this literally you end up physically restraining everyone until you get a diagnosis, and that puts the staff and patient at risk. And how often do we really need stat labs and tox on our agitation patients? As always, it depends on the center — a place like EHC gets a  lot of drunks, but a patient at MSH might be more likely to have drug interactions causing agitation. Use your judgment.

Another commenter noted that the FDA black-box warning on droperidol mandates EKGs and 2-3 hours of tele monitoring. However, Martel – and the ACEP guidelines — note that most everyone has had a good and safe experience with droperidol.

This study by itself isn’t enough to change practice, but taken together with the ACEP guidelines it makes a pretty strong case for the safety and efficacy of benzos in general and midazolam in particular. Just be ready to give more when it wears off. I’ll excerpt some from page 6 of the ACEP recommendations, which make much of a Class II RCT by Nobay (see the guidelines for the reference) comparing ativan, versed and haldol :

The authors compared IM midazolam (5 mg) to IM lorazepam (2 mg) or IM haloperidol (5 mg). Midazolam had a significantly shorter time to sedation than did lorazepam or haloperidol. The mean time to sedation was 18.3 minutes for midazolam, 28.3 minutes for haloperidol, and 32.2 minutes for lorazepam. The time to arousal (81.9 minutes) in patients given midazolam was also significantly shorter than that of the other therapies.

In addition, several class III studies found midazolam (2.5 to 3 mg IM) to be efficacious in reducing agitation. It produced rapid sedation, within 6 to 8 minutes, in a small series of acutely agitated patients. Midazolam was significantly better than haloperidol in controlling motor agitation in a small study of schizophrenic patients. A large series reported by the TREC Collaborative group found that midazolam (15 mg) was superior to haloperidol (5 mg) plus promethazine (50 mg) in producing rapid sedation at 20- and 40- minute endpoints.

Again, take care to note the different agitation scoring methods, different routes of admin, and the different patient populations. It’s explained well on the ACEP review. And I should note, in summary, ACEP’s final recommendations — all class B — are for benzos (midazolam or lorazepam) or typical antipsycholtics (haldol or droperidol) for effective monotherapy of an undifferentiated agitation. Atypical antipsychotics should be reserved for agitated patients with known psych history — the case against them in undifferentiated agitation has been eloquently stated. EMcrit.org has some other studies cited as well, plus some practice notes for sedating violent patients.