The Antidepressant Wars in the Post-Truth Era

Sometimes there are advantages to being outside the mainstream media. Australia dodged a bullet when the UK press announced claims about antidepressants. There was some coverage, but it’s comforting to hope that when the press releases came this way, our astute journalists were aware that during the previous months our press had already just run a competing big medical story: “Prescription pills killing more people than street drugs.”¹

But the February 2018 UK press releases were all the more potent because they were attributed to such hallowed sources as Lancet, Oxford University and the BBC.² British tabloid press was even more florid with interpretations such as “GPs should prescribe anti-depressants to over a million more Brits, experts claim,” citing an “Oxford University study” that supposedly “found pills were up to 113 per cent more likely to tackle depression than no treatment at all — despite previous claims that they did not work.”³ Even at face value it seemed unlikely The Lancet would have used the words ‘pills’ and ‘tackle.’

At a New Zealand conference, overseas visitors went even further. One headline virtually accused critics of mass-scale negligent homicide: “Claims that antidepressants don’t work are dangerous and akin to climate change denial, mental health professionals and experts say,” citing supposed “’overwhelming evidence’ supporting the view that mental health problems were caused by chemical imbalances in the brain.”⁴

Yet at the same time across the Tasman Sea a Professor of Psychiatry, former Australian of the Year, was assuring ABC listeners in response to a question about “chemical imbalance” that this was attributable to the “drug companies, the DSM, the FDA … American psychiatrists … 40 years ago … now no one believes that any more. If you talk to any modern researcher in neuroscience or psychiatry no one would say that is the explanation. But when you go to a GP or even some psychiatrists they will still trot out that very simplistic explanation to people.”⁵

The big stakes in this are not being exaggerated. As Dr. AJ Frances put it: “We have a medical system that couldn’t’ve been conceived more brilliantly by an enemy of the United States.”⁶ Biopharmaceuticals are regularly cited as leading industries in the US economy. Any threat to them is a threat to the economy.

The Lancet authors note these drugs are now off patent “available worldwide and ready to use also in developing countries,” presumably for the “350 million people affected globally.” This would take numbers beyond the scale of a World War. Hundreds of millions dwarf current concerns about the numbers in school shootings, tsunamis, plagues, or even a nuclear bomb. It is more on the scale of an asteroid hit. Depending on which side of the drug story we pursue, we could be ‘saving’ or ‘killing’ all these millions of lives — or both! The numbers sound apocalyptic but highlight the human impact of traditional epidemiological percentages applied to a populous, troubled world. It could depress some that the Eve of Destruction hit record from 50 years ago still sounds like a current newscast.

The most straightforward empirical perspective for the entire saga lies not in the clinical or statistical sciences but in the social sciences. There is in fact a trove of incontrovertible evidence, i.e. language: the evidence of who said what, where, when.

Despite media attributions to ‘the authors,’ ‘scientists’ and ‘experts,’ the source paper in Lancet made no such direct claims about either millions of exclusively drug treatments, vast improvements, dangerous views of critics, or superiority of drug treatments. The authors went to some lengths to use neutral language. They specifically mentioned “modest effect sizes.”

Nor does it appear that the authors made extreme or even controversial claims in subsequent briefings and interviews. A review of the news coverage shows that quotes mixed in as though from the actual article authors are commonly from the media itself or third party ‘talking heads’ they interviewed for comment. The Lancet paper even leaves open the possibility of other treatments being preferable: “because of inadequate resources, antidepressants are used more frequently than psychological interventions.” This could even support some other journalist interpreting this as a call for more psychological services, rather than more drugs.

Some of the greatest concerns have been about the long-term use of medications. The article concedes “a longlasting debate and concern about their efficacy and effectiveness, because short-term benefits are, on average, modest; and because long-term balance of benefits and harms is often understudied.” This does not indicate the debate has ended. The language is present tense — “are” modest — not that some new breakthrough discovery has changed all this and they are no longer merely modest. Clearly, they remain “modest.”

Clinics are seeing cohorts of people who have been on a range of psychotropic medications for decades, and the article specifically cautions: “Some of the adverse effects of antidepressants occur over a prolonged period, meaning that positive results need to be taken with great caution, because the trials in this network meta-analysis were of short duration.”

The Lancet article itself relies on statistics not widely understood by the general public or even most health professionals. Meta-analysis is specialized and a career in its own right. Perhaps a modest effect size might save lives. A claim in The Sun UK stated that the drugs were 113% more likely to ‘tackle’ depression; this might mean anything, e.g. someone “113%” taller than myself might either mean a head taller (100% of my height plus 13%) or the height of a giraffe (me standing on top of me plus 13% = 213%).

And 113% of what? There is no direct text string of “up to 113 percent” apparent in the Lancet article. So unless the journalist was experienced in meta-analysis, this would likely be an interpretation given them by someone unattributed. Perhaps the tabloid was being modest. It might have said “213%” as one of the few places in the article where the text-string ‘13’ appears in the source article is “ORs ranging between 2·13.” The authors define this as “summary odds ratios (ORs) for dichotomous outcomes.” ORs are usually shown as a decimal format, 2.13, rather than the 2·13 in the article, which at first glance might have led me to think it could denote a range from 2 to 13, since they used the phrase “ranging between 2·13.” Since the article uses a hyphen (1·89–2·41) to denote range, we might assume it to be a 2.13 figure, presumably referring to the probability of someone moving from one classification to another across a threshold criterion.

The Lancet article states this threshold to be “a reduction of ≥50% of the total score on a standardised observer-rating scale for depression.” That sounds impressive — less than half as depressed? But to someone untrained in meta-analytic argot, the word ‘standardised’ rings a bell as this term commonly refers to placement on a normal curve rather than raw score. Assuming they mean reduction of raw scores (and ignoring whether some of the item oddities in the scales may not directly assess depression), it raises the question of the starting level of those raw scores. An extremely depressed person might lower their scores and remain depressed. A low-scoring person might merely halve an already low score. It literally seems to mean that among those with much lowered symptoms more of them had received a pill that had some noticeable effect than from the group who had drawn an inactive placebo.

The OR refers not to the actual level of change in raw score but to the relative probability of a person who has a 50% reduction being in the drug group rather than the placebo group. But we’re only told the relative probabilities, not how many people this might be. Could there be a very small number who reduce scores by 50%? Would they be high or low scorers? Would those in the placebo group necessarily have far smaller reductions or nearly the same? The authors mention “the high percentage of placebo responders in antidepressant trials,” so how strongly do they respond? A cynic might ask if the threshold had been set to maximize the difference. If the figures are merely the ratio between numbers in drug and placebo groups, we are reliant on the authors to describe what this effect size might mean in raw score, let alone clinical terms, which they describe three times in the Lancet article as “modest.” This could be serving dual purposes as a statistical term relating to probabilities (more commonly termed “moderate”) or as a clinical benefits term, as when they say “short-term benefits are, on average, modest.”

The intricacies of meta-analysis are for experts. Each dot on a plot has meaning, e.g. 2·13 or 2.13? But evidence anyone with a computer can check at home is — where in the Lancet article does it specifically spell out the “113 per cent” figure? Who interpreted this to the tabloids? And, if deciphered, what would such a figure mean in clinical practice? If an odds ratio, then does it mean in practice a few more people checked a few more items on a depression rating scale? People are much happier or less gloomy? Bill is 113% less depressed than Jenny or than he was last visit? Fewer suicide attempts? The Lancet article states, “We were not able to quantify some outcomes, such as global functioning.”

So after trying to work out what was said, the easier questions are who originally said it where and when. The most sensational interpretations and quotes are from the media and third parties interviewed about the paper. The lead Lancet author, Oxford’s Dr. Andrea Cipriani, told the BBC it was “the final answer … anti-depressants work …” but hedged rather unadventurously: “Medication should always be considered alongside other options, such as psychological therapies, where these are available,“ and the BBC cited the ‘authors’ as saying “it did not mean that anti-depressants should always be the first form of treatment.”⁷ Much of this is consistent with existing guidelines. The argument would be whether it was “the final answer” or any new answer at all.

It is no surprise that some of the press omitted the ‘other options’ and drew a straight inference that all this supposed life-saving treatment could now be unleashed because of the alleged new scientific findings about the drugs. This is a British literary trope, memorable from HG Wells’ War of the Worlds in which the Deputation waved a futile white flag, hoping to talk to the monsters, only to be turned to fire by the death ray.

The spin has created an epic science fiction classic, a ‘boys’ own’ world of tomorrow conspiracy adventure; scientific advances which had presumably been held back because of ‘dangerous’ cowardly views of extremist Luddite critics and timid or old-fashioned GPs. It conjures up the image of the timid softies, vainly hoping to communicate with the repulsive creature (in this case the Black Dog, the brain disease, chemical imbalance depression), rather than being action heroes saving hundreds of millions of Earth lives by courageously manning the professor’s magic bullet ray-gun invention (the wonder drugs).

A “suicide attempter” stands atop a building. Clint Eastwood asks: “Anyone try and talk him down?” The reply:”Yeah, but no luck. But now that you’re here…” After bravely climbing a ladder, confronting, tackling and saving the man, Mr. Eastwood says: “Now you know why they call me ‘Dirty Harry’ — every dirty job that comes along.” Mere talk therapy compared to a man of action willing to risk all and get hands dirty for life-saving results. Enough talk! The only thing that stops a bad guy with a chemical weapon is a good guy with a chemical weapon. Re-arm the doctors! Let them get on with it, run boldly into the building and .44-Magnum-blast that chemical-armed Black Dog.

By contrast with the jingoistic media, the Lancet article does not demean talk therapy, merely noting “inadequate resources” and “where these are available.” However, lest high-priced talk therapists take heart from this, the authors noted that no treatment at all can get results: “Depressive symptoms tend to spontaneously improve over time.” And even if the drugs beat placebos in trials, that doesn’t necessarily mean placebo (or doing nothing) is ineffective, merely less so. Would it be ‘modest’ to ‘spontaneously improve’ by doing nothing?

The article was explicit that the authors did not review or compare alternative treatments, all of which have their own supporters and critics, for example the well-publicized comparisons of physical exercise versus anti-depressants. The article specifically states, “We did not cover important clinical issues that might inform treatment decision making in routine clinical practice (eg, specific adverse events, withdrawal symptoms, or combination with non-pharmacological treatments).” These are the very “important clinical issues” that have been most questioned over decades, not whether these drugs were different from placebos.

It must be remembered that the Lancet article is not a report of a new treatment. It is in itself a secondary source, analysing and citing previous articles. So today’s public has been conditioned to ask which level of repartee is now the ‘fake news’? For example, the New Zealand article claimed that “mental health professionals and experts” viewed criticism as “incredibly dangerous,” but the only directly-attributed use of that term was from someone who said, “Speaking both as a mental health nurse and someone who’s [on] anti-depressants, … antidepressants saved my life…” Yet even this participant “…agreed that a conversation was needed on the issue of over diagnosis and overprescribing.”

Aren’t the latter the very issues that were the core of that conference titled: “Mental Health in Crisis”?⁸ Other media reported a completely different spin on the same conference: “Antidepressants don’t work and mental health system is failing, experts say.”⁹

Where else do we see such florid discourse and counter-accusations? The U.S. school gun debate? Nuclear disarmament? And in what sector other than mental health would someone cite being a patient as among their qualifications as an expert? I can’t see putting my hands up at a Dental Conference claiming expertise by saying, “It saved my tooth.”

So the most salient point I seek to make is that the original article was not what the general public would see. The news-reading public would likely not locate, let alone read or try to digest the source article. If I wasn’t employed in the sector, neither would I.

It is noteworthy that the Lancet site and the Lancet Facebook page made reference to those secondary and tertiary media sources rather than its own article or any presumed press release, let alone the original meta-analyzed source medical articles. Much medical news in the UK in particular comes through filters such as the UK Science Media Centre, which held briefings and cited “expert reaction.”¹⁰

Oxford, Lancet and BBC are prestigious. Mad in America was quick to pick up on the potential impact of this propaganda. Dr. Moncrief’s MIA article¹¹ warned about the dangers of extrapolating the claims to the much larger population of the United States. It would fulfill the 80s prophesy of a Prozac Nation. Australia has a smaller population but a very high proportion of psychotropic medical prescriptions, so many thousands here might have been potentially influenced by this article and its press support.

The claim that this article “puts to bed the controversy” and has settled a scientific debate is the most portentous. The Lancet article itself, had anyone bothered to read it, clearly pointed out that the authors did not investigate “adverse events, withdrawal symptoms, or non-pharmacological treatments.” These are the typical controversies from within the literature and profession over many years rather than whether these drugs “work.”

There is no scientific or news value in finding that psychotropic drugs are more active than sugar pills. Water, vodka and gin look alike, but the psychoactive properties of alcohol are immediately apparent to anyone sampling them, even wearing a blindfold and blocked nose. Any supermarket substance like alcohol, caffeine and cigarettes, let alone the Laudanum or Medicine Show nostrums of the earlier centuries, will have an effect and be easily distinguished from a placebo.

The more important issue is how they compare with an active placebo, i.e. a substance that gives a buzz or other effect, and even more importantly how they compare with alternative treatments like psychotherapy and activities such as physical exercise, let alone social interventions that might ameliorate whatever was depressing them in the first place. One might well expect placebo to be effective. Depression is a state of mind. It’s not like toothache. It should be no surprise that a placebo response which is a state of mind, “I’m being treated,” might have an impact.

Which straw man ever said psychoactive substances “don’t work”? The problem, more commonly, is that these potions work too well. We’ve known for centuries that they are not placebos. Imagine the furore in substituting the Egyptian pyramid crew’s beer, the Union Soldier’s alcoholic ‘medicinal’ Bitters, or the British Navvy’s rum ration with coloured water! Bayer’s miracle painkiller drug Heroin was removed from the market not because of lack of effect but the opposite. It was recreational. It was addictive. Wars were and are fought over opium and cocaine. There have long been movements to prohibit alcohol because its sedative and disinhibiting properties are so popular that it has led to social problems. Marijuana enthusiasts form political groups.

So to find that substances like antidepressants are more psychoactive than an inactive placebo is no finding. It’s long been known and assumed. What had been surprising was the earlier number of findings that seemed not to be able to detect a treatment difference from placebo. That negative history had not been some mere fear or bias but had been the research history. The Lancet authors specifically mention this history of negative findings: “Depressive symptoms tend to spontaneously improve over time and this phenomenon contributes to the high percentage of placebo responders in antidepressant trials.” That is not the language of ‘the past’ now ‘put to bed.’ The authors are not here referring to the drugs, but rather to the trials, the depression condition itself, and natural human response. They are speaking in present tense, with no indication that human nature had now changed.

The article observed that “drugs tended to show a better efficacy profile when they were novel and used as experimental treatments than when they had become old,” echoing the old medical placebo adage: “Use the new medicines while they still work.”

So before even starting to probe deeper, and without learning the arcane language of meta-analysis, we have at least some solid incontrovertible evidence available to all that the claims in the press do not directly match the text of the source article. This is UK English, not the Dead Sea Scrolls. Even a non-English-speaker with a word processor can comb through the text of the Lancet article and news releases to see precisely who said what, where and when.