Lunar cycle does affect behaviour: Australian study

December 14, 2009 3:52 PM

A study of patients who presented to an Australian hospital with acute disturbances and violent behaviour appears to show a correlation with lunar cycles.
posted by AmbroseChapel (24 comments total)

Lunacy?
posted by Blazecock Pileon at 3:56 PM on December 14, 2009

If this is a viral for Werewolf I swear to god I'll rip someone's throat out.
posted by The Whelk at 3:56 PM on December 14, 2009

If this is a viral for Werewolf I swear to god I'll rip someone's throat out.
posted by The Whelk at 6:56 PM on December 14

That right there sounds like werewolf talk to me.
posted by MidAtlantic at 3:59 PM on December 14, 2009 [5 favorites]

grar
posted by The Whelk at 4:00 PM on December 14, 2009 [2 favorites]

I imagine a full moon's light allows people to be out without being blind in the dark and makes it easier for cops to find them when they get into trouble.
posted by stavrogin at 4:00 PM on December 14, 2009 [2 favorites]

This is just a stunt to save the failing Trader Vic's chain.
posted by Inspector.Gadget at 4:03 PM on December 14, 2009 [1 favorite]

hmmm. dubious article behind a paywall? i say nay.
posted by Lutoslawski at 4:04 PM on December 14, 2009

Some day I'm going to do a study to see if lunacy studies are more likely to be published during a full moon.
posted by Humanzee at 4:06 PM on December 14, 2009

Somewhere in Utah, Stephanie Meyer is furiously typing out the first chapter of her next set of books.
posted by dnesan at 4:14 PM on December 14, 2009

This is from the Medical Journal of Australia's Christmas competition: "our humour and inspiration stimulus package."

I don't think a paywalled article is great for MetaFilter.
posted by grouse at 4:25 PM on December 14, 2009

From the Australian and New Zealand Journal of Psychiatry, Volume 32, Issue 4 August 1998 (abstract only):

Objective: It is commonly believed that the full moon exerts an influence on violence and aggression in psychiatric settings. This study used a robust methodology to examine the hypothesis that there was an increased frequency of violent and aggressive behaviour among hospitalised psychiatric clients at the time of the full moon.

Method: Prospective data were collected in five inpatient psychiatric settings across the Northern Sydney Area Health Service. Morrison's hierarchy of violence and aggression was used to rate behaviour. Lunar phases were clearly defined and Poisson regression used to examine relationships between lunar phase and violence. Extraneous temporal variation was considered.

Results: No significant relationship was found between total violence and aggression or level of violence and aggression and any phase of the moon. Conclusion: Future research could profitably examine the implications of a belief in the lunar effect among health workers in the face of evidence that no relationship exists between violence, aggression and the lunar cycle.

This is contrary to the above-linked study, which noted [t]here were 91 emergency patients rated as having violent and acute behavioural disturbance at the Calvary Mater Newcastle hospital from August 2008 to July 2009, and almost a quarter of the cases (23 per cent) occurred on a night of full moon and this was double the number for other lunar phases. It looks like this article used a significantly smaller sample size.

It's not only during full moons, but also paydays that bring out the worst in people, though it may just be a localized issue, as such results were not reliably reported elsewhere.

hmmm. dubious article behind a paywall? i say nay.

That site is The Medical Journal of Australia, which has been published since 1914. The publicly viewable section is an abstract of the full article, a very common practice for scientific journals.
posted by filthy light thief at 4:26 PM on December 14, 2009 [2 favorites]

Thanks for taking the thing seriously, light theif. It is, as far as I can see, a reputable publication and a properly-conducted study.

I imagine that most MetaFilter readers would be skeptical about the phenomenon, and put it down to things like confirmation bias -- which is why, like a good skeptic, I'm interested in this study, which does appear to show some evidence the other way.

> I imagine a full moon's light allows people to be out without being blind in the dark and makes it easier for cops to find them when they get into trouble.

Good point. Other factors might be that people just can't sleep with the moon shining through their windows; that the phases of the moon got synchronised with paydays, as noted above; that law-enforcement people believe this urban myth and are therefore more likely to over-react when faced with emotionally disturbed people, and so on and so on.
posted by AmbroseChapel at 4:32 PM on December 14, 2009

I have problems with them using only a single year of data, n=91, from a single hospital. But that shit is pretty common in epidemiological studies, I guess.
posted by Jimbob at 4:33 PM on December 14, 2009

Here are some other studies from the MJA Christmas competitions of previous years:

• Effect of postgraduate exams on putting performance. "Conclusion: Examination stress adversely affects putting performance. Postgraduate examinations should be scheduled to avoid significant golfing fixtures, such as club championships."
  
  
• Hospitals are dangerous places. "Results: The odds of being discharged alive were lowest on the weekend and literally climbed during the week." This paper has an awesome plot that looks like a spider (which satirizes inappropriate normalization techniques) and is full of oblique reference to the same: "The causes are most likely a complex web of specialist availability and patient requests to be home for the weekend. During the week, discharges run as smooth as silk, but things spin out of control at the start of the weekend. This problem has crept undetected into the St Thomas’ Hospital system; removing it may require sweeping changes to staff rosters."
  
  
• Weather patients will come? "Conclusions: As there is a clear relationship between weather conditions and [emergency department] attendance, incorporating meteorological forecasting into emergency medicine training may improve ED scheduling. To improve the morale of ED staff coping with an onslaught of patients on good weather days, the ED environment should simulate sunny weather, with swimming pools, sun lamps, palm trees and Beach Boys music."

posted by grouse at 4:48 PM on December 14, 2009 [1 favorite]

Everything filthy light thief said.

But this study is printed in the "Christmas Offerings" section of the most recent issue of the journal. Other studies from this issue include studies on "how performances by the Staff Christmas Choir of the Peter MacCallum Cancer Centre (“Peter Mac”) affected inpatients, outpatients and visitors in 2008" and on "whether slow or fast bandaid removal is less painful."

I'd bet that they're using real data, but also that they're sort of twisting the data to come up with a few tongue-in-cheek articles for a laugh before the holidays.

On preview, what grouse said.
posted by SpringAquifer at 4:50 PM on December 14, 2009 [2 favorites]

I can confirm the belief among health care staff that there is a lunar cycle effect. My mother worked for many years in a VA psych ward and to this day is convinced that full moons cause behavior problems in the mentally ill.
posted by mullingitover at 5:19 PM on December 14, 2009

Jimbob: "I have problems with them using only a single year of data, n=91, from a single hospital. But that shit is pretty common in epidemiological studies, I guess."

n wasn't really 91, though, it was 603. The violent patients were matched 5.6:1 (why on earth they didn't stop at four is sort of beyond me, but I digress). Case-control studies are quite good when it comes to rare cases. My handy-dandy epi table suggests that even for a control group violence prevalence of 0.05, that sample size should be able to distinguish as little as a 4.0 difference in relative risk ratio at 5% significance and 90% power. Being unable to actually read the damn article despite all my credentials (Ovid only has MJA through this September), I can't say for sure, but I think this study's on the up-and-up statistically speaking.

This is not to say that I believe for a second that this study is anything more than confounding and type-I error.
posted by The White Hat at 5:23 PM on December 14, 2009 [1 favorite]

Ahhh... so lunacy studies are more likely to be published at *Christmas*.

> I imagine a full moon's light allows people to be out without being blind in the dark and makes it easier for cops to find them when they get into trouble.
I have it on good authority (okay, only hokey 80s movies) that everyone in Australia lives in the outback, fighting crocodiles by moonlight, but 'round these here parts we have electrical lighty-thingies that are so much brighter than the moon, you wouldn't believe it until you've seen one. They're becoming so common that even criminals and crazy people can get their hands on them, which is turning the biting of orderlies into a 24/7 business.

Seriously, raw p-values mean nothing when taken out of context. The context here being, that for whatever mule-headed reason, people keep on doing these damned moon studies. I'm not sure what the palliative implications are here ---should we blow up the moon or something? That would be totally metal, and I can get behind that. But mostly these studies just analyze a couple correlations, and most of them come back negative, even when they obviously go data-mining. For starters, what constitutes a full moon is somewhat vague, and subject to author choice. And only acute disturbances and violent crime? What do you want to bet they'd be happy with other correlations? Once you're doing multiple comparisons, with multiple studies published, it becomes a guarantee that you're going to see something "statistically significant" pop up randomly (think dead salmon that can read people's emotions).

The agonizing bit is the hypothesized explanations for the connection. The one that really gets my dander up is the "gravity from the moon" one, since the moon's gravity, aside from being ludicrously tiny on Earth, also doesn't correlate with moon phase. (I'm not claiming that this paper mentions this ---I can't read it because I'm at home and therefore don't have the right proxy).
posted by Humanzee at 5:24 PM on December 14, 2009 [3 favorites]

That's just loony.
posted by rmmcclay at 5:46 PM on December 14, 2009

Warewilf
posted by The Whelk at 5:48 PM on December 14, 2009

The bit that jumps out to me is that they have two types of patients, ones that are deemed 'severe' and the rest. Only the 'severe' ones show the trend. But the less severe patients aren't actually specific patients but calls to security, so I'm not even clear how many actual people are involved (we can assume that one call to security equals one patient but it doesn't actually say). Also it's a somewhat subjective call as to which category a patient falls into and there's no detailed data about how the call was made here. For example, it says about the 'less severe' group: "Few of these patients were physically restrained or sedated" so what makes one patient who's sedated extreme and another patient who's sedated not extreme? There's mention of a clinical trial too which is confusing, is that related to how they made the call? You can easily cause a trend by how you stratify your data, just draw the line in the place that gives the result you want. I don't think that's been done here on purpose but again, so little detail how can we tell?

It seems to be a fairly straight forward study (assuming there are objective rules about how they stratify the patients) so I'd like to see it repeated elsewhere. Then we'd begin to have some ecvidence to work with.
posted by shelleycat at 5:49 PM on December 14, 2009

I'm not impressed. Beyond going against other research in this field it is in the journal's Christmas offerings along with:


A prescription for a smile
Paula H Johnson — Med J Aust 2009; 191 (11/12): 594.

The ethical rap
Diann S Eley — Med J Aust 2009; 191 (11/12): 647.

Publication celebration!
Peter W New — Med J Aust 2009; 191 (11/12): 663.

Fast versus slow bandaid removal: a randomised trial
Jeremy S Furyk, Carl J O’Kane, Peter J Aitken, Colin J Banks and David A Kault — Med J Aust 2009; 191 (11/12): 682-683.

“The moment is all we have”: patients and visitors reflect on a staff Christmas choir
Clare C O’Callaghan, Colin J Hornby, Elizabeth J Pearson and David L Ball — Med J Aust 2009; 191 (11/12): 684-687.

This is a dump of smile inducing semi-studies and anecdotes.
posted by dances_with_sneetches at 7:00 PM on December 14, 2009 [1 favorite]

You know, this reminds me a lot of that kid who cited one of Google's April Fool's gags in a university paper.
posted by Jilder at 7:36 PM on December 14, 2009 [1 favorite]

I can only roll my eyes at the full moon thing, it's going to take a damned lot more than one article to change my position on that.

BUT I just wanted to offer that the timing of the article is probably due to this being a "blue moon" month. We get two full moons this month, and the second one is called a "blue moon."

It turns out to not be particularly rare, but rare enough to be noted in folk aphorism.
posted by ErikaB at 7:47 PM on December 14, 2009