This piece is co-written with Sam Clifford. We would like to preface this post by stating that we don’t purport to be experts in science communication nor general reporting standards and practice, but we do care about these things. We are also not, in any way, experts on neurochemistry. Caveat lector.
On Monday ABC News posted an article, derived from a report on the 7:30 program, suggesting that the common antidepressant Effexor (venlafaxine) might cause pathological gambling. What prompted this? Did the AMA release a damning report? Did a team of scientists perform some new trials that proved a causal link? No, the ABC spoke to a few people who had become problem gamblers after starting a round of treatment of Effexor, and thought that this was enough to qualify as science journalism.
The article is problematic for a few reasons. Principally, it’s an article masquerading as investigative science reporting, with no science behind it. It’s not even that it’s bad science reporting; the writers have not reported any actual science. They have simply spoken to a few people (some problem gamblers, a psychiatrist, someone from a church counselling service, someone from Pfizer). It’s good that they’re talking to a range of people but these people don’t point to any research that links anti-depressants to gambling (and there are thousands of research papers on the side effects of venlafaxine). At best, the psychiatrist briefly talks about generic neuroscience and the counselling service administrator talks about second-hand anecdotes about people who are seeking counselling for gambling being on Effexor.
The article also points out that one of the problem gamblers seems astonished that Pfizer does not list problem gambling as a side effect of Effexor. There do not seem to be any studies on a link between Effexor and gambling. It would be beyond irresponsible, then, for Pfizer to claim problem gambling as a side effect in their literature. The closest thing to information a cursory Google search can find on a link is this page. Note the numbers. Of nearly 48,000 reports of side effects (and no way of knowing the full sample size) 0.16% are engaged in problem gambling.
Another concern is that the ABC article links Effexor to “similar medications” Permax (pergolide) and Cabaser (cabergoline) which can apparently cause pathological gambling as a side effect, and were the subject of a 2011 lawsuit. The ABC doesn’t mention how these medications are similar to Effexor, and it’s not even clear whether or not they are. Permax and Cabaser are both treatments for Parkinson’s disease. They are agonists which stimulate dopamine receptors in the nervous system and are suggested for Parkinson’s disease because it seems to be linked with a deficiency of dopamine. Effexor is a reuptake inhibitor which aims to lessen the reabsorption of serotonin and norepinephrine, leaving a higher extracellular (outside the cells) concentration of these neurotransmitters. They are “similar” in that they both attempt to effect a higher level of the relevant neurotransmitters in the nervous system, but agonists and reuptake inhibitors have very different mechanisms of action. One stimulates production, the other inhibits reabsorption.
It’s also amazing that the writers would fail to mention that the symptoms reported by the individuals they spoke to are also symptoms of depression. Perhaps then we should be considering not just the population percentage of problem gamblers, but the percentage of those with depression who suffer from the condition.
The way forward is to use conditional probability, a simple piece of statistics that is apparently often ignored in news reports. Statements like “people on Effexor develop problem gambling” have a probability of gambling which is conditional on taking Effexor. What we would like to know is what proportion of people on Effexor are problem gamblers and whether this is “significantly” different to the proportion of people not on Effexor who are problem gamblers. Using the notation of conditional probability, p(A|B) = the probability of event A given that event B has happened, we’re interested in comparing p(problem gambling | Effexor) versus p(problem gambling | no Effexor). It might also be interesting to look at p(problem gambling | anti-depressants which are not Effexor) but this data’s probably difficult to come by.
We don’t have the data to estimate the proportion of people who are problem gamblers but not on Effexor, but if p(problem gambling | Effexor) and p(problem gambling | no Effexor) are the same then they’ll be the same as in the general population, p(problem gambling). Including people on Effexor in the general population won’t skew the numbers too much given that even in the USA, which has a reputation as a country that tends to overprescribe medication, about 6% of the population is prescribed venlafaxine, the active component of Effexor). So it should be safe to look at p(problem gambling | Effexor) versus p(problem gambling), i.e. whether problem gambling is independent of Effexor use.
The Australian government’s problem gambling website cites a gambling participation rate of 70% in 2009. One in six pokie players are problem gamblers according to the same data. Pokies accounted for 62.7% of the $19b total spending on gambling in the 2008-09 financial year. We can’t really infer from this what the number of pokie players is but the Productivity Commission’s report estimates a population of 115,000 problem gamblers, 0.7% of the adult population of Australia (page 11). So while we wouldn’t (or at least shouldn’t) conclude that Effexor reduces the risk of problem gambling (because it hasn’t been studied) it doesn’t sound, prima facie, like Effexor use increases problem gambling. Is it worth studying? Probably. The Commission may at least be interested in it as a possible link between two significant social issues, problem gambling and depression.
The ABC is one of Australia’s most trusted news sources and, as Spider-man’s Aunt May reminded us all, with great power comes great responsibility. The consequences of leading people to believe that their medications may be causing them harm could be dire. If seeing this article caused even one person to come off their anti-depressants without medical supervision it would be a very high price to pay, given the high potential for emotional and physical well-being costs to that person. If the ABC wished to run an article or segment asking the scientific community to once again investigate the side effects of venlafaxine, this time with a specific focus on pathological gambling, they could have done so. Instead, they offered up some anecdotal evidence with no scientific study behind it and spoke to a church counsellor and a psychiatrist for “balance” without actually addressing the science. This is irresponsible at best, and reminds us that science reporting is still not taken seriously by many media outlets.
Accuracy in science reporting. High five, bro!