Tuesday, July 31, 2007

More shoddy reefer madness reporting of cannabis risks

When you are in the midst of a fully fledged media cannabis panic, as we currently seem to be, you can be quite sure that any new research on the drug:

  • will be pounced upon by lazy journalists
  • will be trawled for any vaguely shocking sounding statistics (by non-scientists and non-statisticians)
  • will have these statistics spun into sexy 'shock' headlines
  • will have any negative findings, statistical ambiguity, commentary on confounding factors/ context / significance etc conveniently glossed over
  • will see politicians responding to the media coverage of the research rather than the research itself *insert daft quote from David Davis*

This is what happened last week with the Lancet meta-study on cannabis and psychosis published last week. The main report headline, that cannabis smoking increases the risk of schizophrenia by 40%, achieved saturation media coverage for a science story, being covered by over 500 news outlets. Putting this headline figure in some sort of statistical context, along with anything that could reasonably be described as intelligent discussion the policy relevance of the paper (remember the endlessly tedious reclassification debate is still rumbling on) was almost completely lacking. (The Lancet, it has to be said, clearly courted this coverage. They played up the 40% figure in there press release at a key juncture on the policy debate (on which the paper does not comment), putting the cannabis story at the top of the issue's press release to maximise its exposure).

The conclusions of the paper, that cannabis use involves a small risk of mental health problems is not really news. This has been known for over a century, and it is acknowledged by everyone that all drugs present potential health dangers, and all drug use involves risk. Here, just for the record is what the Indian Hemp Commission concluded 113 years ago, back in 1894:

“Mental Effects

In respect to the alleged mental effects of the drugs, the Commission have come to the conclusion that the moderate use of hemp drugs produces no injurious effects on the mind. It may indeed be accepted that in the case of specially marked neurotic diathesis, even the moderate use may produce mental injury. For the slightest mental stimulation or excitement may have that effect in such cases. But putting aside these quite exceptional cases, the moderate use of these drugs produces no mental injury. It is otherwise with the excessive use. Excessive use indicates and intensifies mental instability (1:264).”

The Lancet paper tries to quantify the risk and suggests, not unreasonably, it is very real for a small proportion of heavy users, noting that there is a dose response in the phenomenon (the more you use the greater the risk - also hardly a surprise), suggesting that 800 cases of schizophrenia would not have occured if none of the UKs 6.2 million cannabis users had ever tried it. The estimate of a 0.00125% risk of schizophrenia for cannabis users (or you could equally well claim, 99.99875% likelihood of not getting schizophrenia) was mentioned, as far as I can tell, in none of the media coverage for whom the 40% increased risk proved far more panic inducing.

Whilst horribly out-gunned in terms of coverage, there has been some more responsible reporting of this paper:

Will One Joint Really Make You Schizoid?
from Stats.org at George mason University, which notes that whilst cannabis use has leapt since the 50s, schizophrenia diagnosis have remained constant....

Blah cannabis blah blah blah blah Ben Goldacre's bad science column in the Guardian considers the poor media coverage of the lancet paper

Transform's press release on the Lancet paper titled 'Cannabis health risks should not lead to knee jerk policy making', got a bit of coverage, including the Sun (and an honourable mention to the Gallway Evening Echo). More importantly the quote we included from Martin Blakeborough landed him on the Today programme on radio 4 and all subsequent BBC coverage.

Now, just three days after the Lancet brouhaha, we have another almost identical one that has followed almost exactly the same media trajectory, but this time concerning another not terribly surprising and already well known fact that smoking stuff is bad for your lungs. This paper, which unlike the Lancet is based on original research, appears in the journal Thorax, and considers the relative impacts of cannabis and tobacco on a range of lung function measures. The Reuters press release (again some blame must go to the journal publisher) ran with a headline that 'one cannabis joint as bad as five cigarettes' (that has been picked up, again, by 100s of media outlets). With reefer madness alarm bells ringing I looked up the complete study and found the conclusions to be somewhat different. Rather than spell it all out myself, below is a brief comparative review of the paper and the Reuters coverage provided by Jamie Chen, a student at Columbia University in NYC, and an intern at CJPF (with thanks to Eric Stirling):

This is my cursory review of a study found in the journal Thorax, and my response to the Reuters story that smoking one joint is as damaging to your lungs as smoking 5 cigarettes.

The Reuters article appears to have come from the journal publisher

The study was by Sarah Aldington, Mathew Williams, Mike Nowitz, Mark Weatherall, Alison Pritchard, Amanda McNaughton, Geoffrey Robinson, and Richard Beasley. It is entitled The Effects of Cannabis on Pulmonary Structure, Function, and Symptoms. It is published in the July 2007 issue (note: you need an academic id to read it but i have a pdf if anyone is really keen)

Here are some direct quotes from the study, under the subtitle "Lung Function tests":

"Both cannabis and tobacco smoking were associated with a reduction in the FEV1/FVC ratio, but the effect of cannabis was only of marginal statistical significance." (FEV1 being forced expiratory volume in 1 second, FVC being forced vital capacity)

"Cannabis smoking had no effect on FEV1 but tobacco smoking reduced it."

"Cannabis had no statistically significant effect on MMEF and tobacco smoking reduced it." (MMEF is maximum mid-expiratory flow)

"Cannabis increased TLC with marginal statistical significance but tobacco had no effect on TLC." (TLC being total lung capacity)

"For TLCO/VA (adjusted), cannabis had no effect while tobacco smoking reduced this measurement." (TLCO/VA being carbon monoxid transfer factor)

The only test in which cannabis caused worse results than tobacco was sGaw (specific airways conductance).

"Cannabis and tobacco use reduced sGaw, although the effect was of marginal significance for tobacco."

Also, the study said that cannabis smoking had a negative effect on these tests if the subject was also a tobacco smoker.

A "novel finding" was that those who used cannabis regularly had an "increased percentage of low density lung tissue."

So, Reuters says one joint is as bad as 5 cigarettes on your lungs, when really, in every test but one, tobacco had negative effects while cannabis effects were statistically insignificant. Cannabis seems to negatively effect only one aspect of lung function, air flow. But when it comes to hyping the dangers, that single finding is sufficient to ignore the other findings and claim cannabis to be 5 times more damaging than tobacco. Notice that Reuters' headline says 5 times even though the study says 2.5 to 5.

Methodological issues:

Of 3500 randomly sampled adults, the researchers could only find 19 who smoked a joint a day for five years. They had to specifically advertise for people to come to them, which they admit may have attracted people concerned about their respiratory health (since the study focused on the health effects of smoking cannabis). Also, since joints are usually shared and packed at home, the amount of cannabis varies from joint to joint and the amount smoked by each person is different; numbers of joints smoked per person were based on the rough estimate of the individual himself - subjects were told to guess what percentage of a joint they smoked each time and then add up these numbers.

Other issues with the study:

They studied only those who smoked a joint a day for the past five years - these are long-term effects that would not pertain to the average marijuana user (especially if marijuana is not smoked but eaten or vaporised). And the reason there is no filter is because they are made at home, a problem easily remedied by use of a filter in the joint making process. Eliminating the risks of smoking on lungs through alternative methods of administration is not discussed as an obvious harm reduction policy implication, in the paper or any media coverage.

And the papers conclusion: marijuana smoking causes wheezing, coughing, mucus, and chest tightness. Tobacco smoking causes emphysema.

Another lovely example of biased media and badly communicated research.

No comments: