In the midst of the twittersphere furore surrounding Johann Hari’s high-profile book Lost Connections, stating that there are better alternatives to antidepressants, patients and practitioners around the country are unheard as they wrestle with difficult decisions. Sadly, as clinicians, we often don’t share our worries with patients, and vice versa. Prescription rates continue to rise. Most GPs and many patients would love there to be ready alternatives, based on the social causes of low mood, such as at the Rushey Green Time Bank in Lewisham. Meantime with long waits for therapy, antidepressants are such an easy answer.
NICE guidance advocates medication for those not responding to psycho-social interventions – it does suggest discussing side effects and withdrawal, but does not advocate detailing the poor evidence for effectiveness or lack of long term studies ruling out harm. A positive take on the evidence suggests they only work in a minority of patients (number needed to treat = 7 for SSRIs) with a small effect, while analyses based on unpublished trials, suggest that average effects are smaller than what is considered to be clinically significant. It’s been an ongoing debate for years. Compared to the benefits of antibiotics in pneumonia and cataract surgery antidepressant medication cannot be described as very effective. Yet we only demand a small mean difference in a meta-analysis for it to get into the NICE guideline as standard treatment.
As a clinician, I have started being more honest about the evidence, explaining that “while most people get better while taking on antidepressants we won’t know if, when you are better, this is due to placebo, other positive things you are doing, the natural course of mood changes or least likely a positive effect of the drug.” I know I am reducing the placebo response, not just for the benefits of being more honest, but because I now share Gotzsche’s serious worries  about long term harms. It’s a much tougher conversation and has to be delivered with sensitivity, compassion and if at all possible with some alternative solutions.
Perhaps what is most bizarre, is that, as doctors, we find it hard to believe that chemicals which are designed to get into the brain and affect synapse function, might just have the potential to do harm, if infused continuously over months or years. Initially they cause a chemical imbalance which might feel good or bad (or more often neutral), the brain then reacts to compensate and create a new steady state (feeling better or worse) – stopping means this transition goes into some kind of reverse (sometimes with side effects), whereas continuing appears to result in some kind of neuronal structural changes, the benefits or harms of which are uncertain.
I tend to put significant physical and psychological effects reported by patients starting down to real drug side effects. My convoluted logic then thinks that if some people get worse then there probably are some people who do benefit significantly in the short term (balancing out as a marginal net average improvement). This takes us back to the quandary of whether the patient who is better has actually benefited from the drug. It’s not quite as bad as a needle in the haystack, but we should not simply interpret improvements as being down to the medication. Quite the reverse. But should we just hope that if the average effect in short term trials is slightly positive we should just continue prescribing? Guidelines suggest we should. When patients are quite clear of benefit, particularly when sustained or repeated, and nothing else good has happened in their lives which might have helped, I am more likely to believe the drug has had some positive effect, and continue to prescribe. Without such clear-cut temporal evidence of benefit, I now worry a lot about continuing to prescribe, but will normally continue if the patient requests despite sharing my concerns.
It’s a different dilemma if the patient’s mood remains low despite several weeks of treatment. This is common. Do we stop, increase the dose or switch? While the US based STAR-D study showed that assertive changes to medication could lead to increased remission rates – 47% is not a very impressive proportion of those in the study, which saw more dropping out of their (free) care than achieving remission in each step2. Once antidepressants have failed the chance of the them working reduces significantly.
As even supporters of antidepressants have argued, the problem with antidepressants is that often the wrong people are getting them. This mismatch is because of an interaction of three aspects of human nature – doctors’ desire to help people in distress, doctors’ complicity with the norms of professional practice, and patients’ desire to take chemical substances to take away distress. This leads to us prescribing, not just to those meeting the criteria for depression (most of whom won’t benefit), but also to those about whom we wonder ‘creatively’ if they might benefit or because waiting times for substantive therapy remain long.
These human factors also of course come into play when reviewing progress. Explaining that the small average benefits shown in meta analyses of discontinuation trials may well be due to methodological flaws is just not going to happen in a general practice consultation. And it rarely feels like the time to let on that discontinuation syndromes, and sexual dysfunction are more likely to occur if treatment continues for longer. And certainly not spoil the party by sharing the emerging evidence that “for some people antidepressants may over time actually cause depression”.
The three factors are joined by a fourth and fifth – our tendency to see associations (eg prescribing and getting better) as causal, and our inability to notice change that occurs gradually over long periods. If progress is good after three months it’s much easier to suggest continuing medication than go against guidelines and discuss the option of consolidating further improvement through positive personal actions (creativity, generosity, making connections, being curious, physical exercise) which we have evidence to support but which require complex changes in behaviour and have slow incremental effects. Even easier when presented by a request to continue medication as part of the pile of prescription requests received daily by the practice, to just respond positively to the patient’s implied wish, and send a new script to the pharmacy.
So, what about those discussions that should happen at the end of treatment as prescribed by NICE? Yet again its human factors rather than evidence that wins the day. Fears that withdrawing medication will lead to a relapse dominate some patients’ and doctors’ thinking. All those trials showing continuation helps prevent relapse might apply to my patient, even though it’s their first episode.
So, given these natural tendencies it’s no surprise that prescriptions for antidepressants have doubled every 10 years for the last three decades to over 70 million in the UK, despite a very flimsy evidence base for both starting and continuing treatment. In the continuation phase of the STAR-D study, the weighted mean relapse rates over the year were 37.4% for those who had gone into remission and fully 64.4% of those who went into continuation treatment having only ‘improved’ (rather than remitted). Relapse rates were also markedly higher for those who has remitted following more than one step of antidepressant treatment to get to remission. And while we don’t know the outcomes of those who dropped out, only a startlingly low 5.8% of those continuing treatment remained in the study and stayed in remission. So, as well as there being high rates of relapse when provided with ‘state of the art’ anti-depressant STAR-D care, a significant proportion of patients voted with their feet and discontinued their (free) treatment – sometimes very early on in. We don’t know if they are right or wrong to trust their instincts, but there is precious little evidence to suggest they are wrong.
And all these dilemmas before even starting to deconstruct the edifice of mental health diagnoses, which even the DSM III architects have started to pull down. We must as GPs start to gain and share knowledge about the pharmacology of so called antidepressants. Joanna Moncrieff explains how they have wide ranging effects on most parts of the body and brain; it turns out they are not simply ‘antidepressants’ but drugs which might have unpredictable abilities including inducing drowsiness or numbness. These effects might indirectly improve or worsen mood, depending on the individual. As doctors, we should therefore focus more on interpreting the varied and idiosyncratic effects of drugs on each patient we see. This means using all the evidence presented by the patient, alongside the evidence from research, to try to work out together whether they are one of the minority to benefit, while being mindful of potential long-term harm for others.
- Pigott H.E.aLeventhal A.M.b · Alter G.S.a · Boren J.J.b Psychothreapy and Psychosomatics. Efficacy and Effectiveness of Antidepressants: Current Status of Research.
- Gotzsche, P. (2017). Psychiatry gone astray. [Blog] Dr David Healy. Available at: https://davidhealy.org/psychiatry-gone-astray/
- Whitaker, R. (2010). Anatomy of an epidemic: magic bullets, psychiatric drugs, and the astonishing rise of mental illness in America. New York, Crown Publishers.
- Hengartner, M.P. Methodological Flaws, Conflicts of Interest, and Scientific Fallacies: Implications for the Evaluation of Antidepressants’ Efficacy and Harm. Psychiatry, 07 December 2017 | https://doi.org/10.3389/fpsyt.2017.00275
- Moncrieff, J., Cohen, D., & Porter, S. (2013). The Psychoactive Effects of Psychiatric Medication: The Elephant in the Room. Journal of Psychoactive Drugs, 45(5), 409–415. http://doi.org/10.1080/02791072.2013.84532