A Simple Eating Disorder Treatment No One Ever Talks About

Treating eating disorders by normalizing eating habits.

Posted May 11, 2018

An alternative approach to eating disorder treatment with radically improved recovery rates.

My last post painted a fairly sobering picture of the current state of cognitive behaviour therapy (CBT) research for eating disorders. CBT is often now the go-to treatment, especially for bulimia, and it does work very well for some people. My own experience of CBT for anorexia was extremely positive. But with maximum 45% remission rates, roughly 30% relapse rates, remission and recovery defined with despicable laxity, and relapse rates sometimes misleadingly concealed, there’s a lot of room for improvement.

So, what if I told you that there’s a treatment programme which has achieved 75% remission rates, 10% relapse, and zero mortalities after five years in a mixed cohort of 1,428 patients, of whom 40% had anorexia? (Remission rates were the same across the eating disorders, but took longer to achieve for anorexia.) What if I told you that here remission and recovery are defined in ways which actually make sense: not just rattling off the tick boxes of BMI, EDE-Q score, and absence of bingeing and purging for a few weeks, but declaring people in remission ‘when they no longer meet the criteria for an eating disorder, when their body weight, eating behavior, feelings of satiety, physiological status, level of depression, anxiety, and obsession are normal, when they are able to state that food and body weight are no longer a problem, and when they are back at school or work’ (Bergh et al., 2013)? What if these researchers even measured ‘full recovery’, where all these criteria are met at five-year follow-up (Bergh et al., 2002)? What if 90% of those who achieved remission got to full recovery? And what if I said that the essence of the treatment is bafflingly simple…

I guess you might ask: why the hell have I not heard about this? And my answer would turn into a story about the gulf between the theory and the practice of scientific research. Back in 2006 the researchers themselves suggested that ‘Perhaps because this model represents a paradigm-shift it has been slow to catch the attention of clinicians and scientists working within the conventional framework’ (Södersten et al., 2006, p. 577) – the conventional framework being the one that treats eating disorders as mental disorders. I suggested in my previous post that CBT already goes very much against the simplistic disembodied ‘mental illness’  hypothesis, but the Mando view would be that it doesn’t go far enough. Challenges to the dominant paradigm always take time to gain momentum and generate a new paradigm: it was thus for the rise of CBT against the backdrop of psychoanalysis. And this inertia exists for a reason: new theories need a decent weight of accumulated evidence behind them before they deserve acceptance. But in this case we might ask whether the resistance has now outlived its welcome.

Before I go on, I should make clear my own perspective on all this. Last autumn, I had an email from Michael Leon, Professor in the Neurobiology and Behavior School of Biological Sciences at the University of California, Irvine, who specialises in research and treatment of autism. Michael said that he’d appreciated my blog post on the Minnesota study, ‘Anorexia is a physical illness of starvation', and that I might be interested to hear about clinical research on eating disorders which he was involved in with colleagues in Sweden. I was in LA at the time, and I drove down to visit him; we had a fascinating chat about the work they’d been doing, including developing an app to support weight loss for people with obesity, perhaps also to be rolled out for those with mild restrictive or other eating disorders in due course. I meant to write a post about it at the time, but still had lots of unresolved questions about the details, and other projects got in the way of making the time to answer them.

Then last month I went to Sweden to give some talks at Uppsala University, and I thought I’d take the chance to get in touch with those Swedish colleagues and suggest a meeting. They said they’d be happy to meet, so I visited the Mando clinic and we (Cecilia Bergh, Per Södersten, and others on their team) talked for a good 2.5 hours before a current patient showed me round the clinic. Mando have three clinics in Sweden, one in New York, and one in Melbourne. The treatment they offer in Sweden is supported by the Swedish national healthcare programme, and they also accept patients from abroad, who pay unsubsidised rates. They offer acute medical care, plus inpatient, partial hospitalisation, daypatient, and follow-up care, with patients progressing through the stages as necessary. They have published 30 peer-reviewed journal articles over the past 20 or so years. I have no affiliation with or vested interest in Mando, and what I write here is based on our conversations in California and Sweden, on the reading I’ve done of the Mando team’s research publications, and of course on all my other research and experience of CBT and other treatments.

Per Södersten and Cecilia Bergh, used with permission
Source: Per Södersten and Cecilia Bergh, used with permission

Normalisation of eating behaviours: the Mandometer device.

The basic claim underpinning the Mando treatment is that anorexia is not an emotional disorder. It is not a psychological disorder. It is, quite literally, a disorder of eating. So the treatment is a treatment of the eating. Their eating treatment has two planks: normalisation of eating speed, and normalisation of hunger and satiety signalling. They’ve developed a simple device to assist with both: the Mandometer. This is basically a weighing scale that talks to an app. You put your plate on the scale, and put the right amount of food on your plate (each patient is given a personal meal plan, but the goal is that everyone should be able to eat ‘normal Swedish food’ of the meat-and-two-veg variety). Before they start treatment, patients use the Mandometer simply to track how much they eat and at what rate. These data are used to determine the patient’s starting meal size and duration, to be adjusted as treatment progresses. Then, once treatment begins, the app displays a graph showing a hypothetical curve for normal eating rate (established on the basis of 10 healthy volunteers [see Bergh et al., 2002], and adjusted to your current state). If you deviate too far from the curve you get a prompt saying ‘please eat a little quicker’, or slower. Every minute you’re also prompted to tap a vertical line to indicate how full you feel, from ‘not at all’ to ‘extremely’, with a training curve provided here too. In essence, the Mandometer is designed to help people tackle that horrible feeling which many months or years of an eating disorder can leave us with: that ‘I don’t know how to eat’. Patients begin by using the device for every meal, and normalise their eating typically over 4–5 months. Thereafter they gradually introduce more meals without the Mandometer, including eating in restaurants and other social settings, until at some point, the team told me, the patient realises they don’t need it any more.

Mando Group AB, used with permission
Source: Mando Group AB, used with permission

I tried the system out in Sweden with a cheese and ham roll I had with me, and it was surprisingly intuitive to use; I tend to eat more quickly than most other people, but at least on this occasion, I found I was sticking to the line pretty naturally (the interesting conversation doubtless helped!), and the fullness ratings were also easy to decide on and felt like they were slowly but surely rising. The treatment for anorexia involves gradually increasing the eating rate, while treatment for bulimia involves decreasing it, both resulting in a 20% change in intake (and with an end goal of eating 350 grams of food in 15 minutes). This is achieved modifying the values of the eating-rate curve between one and four times for each patient, with gaps of on average 35 days between changes. Based on my brief experience of it, I can imagine feeling what the team said many people report: being able to trust the Mandometer, feeling less threatened by it than by a human. Taking comfort in the sense that it can’t lie to you.

The rationale behind the centrality of eating speed is that the gastrointestinal response to a meal is affected by how quickly you eat it. People with anorexia tend to eat too slowly; obese people tend to eat too quickly. Retraining obese adolescents to eat more slowly reduces levels of ghrelin (the ‘hunger hormone’) in fasting conditions and after eating (Galhardo et al., 2012), and since ghrelin levels are chronically elevated in people with anorexia (Prince et al., 2009), one can expect the reverse to hold for those retrained to eat more quickly in anorexia. An interesting twist on this story is that these hormonal actions relate not just to appetite but also to the behaviours associated with the eating: the neuropeptide NPY and the hormone leptin act differentially on responses involved in obtaining food (originally foraging behaviours) and in consuming it (Ammar et al., 2000) depending on the context of food availability, from abundant to scarce. This confirms the centrality of eating-related behaviours in physiological recalibration.

The centrality of the fullness tracking is obvious: the ability to judge when one’s hungry and when full is one of the things which makes eating more again feel so frightening. It seems to justify the grand fear that one could eat and eat and never stop, ending up switching anorexia for obesity. Instead, here you’re gently guided back, bite by bite, to confidence in what fullness means.

These methods draw on evidence that the normalisation of eating behaviours, rather than of bodyweight per se, is the key driver of recovery: the ‘semi-starvation neurosis’ seen in anorexia is also present not only in obese people whose BMI is reduced but in people with obesity and binge-eating disorder with no bodyweight change, or people with bulimia, who usually have a ‘normal’ bodyweight (Södersten et al., 2008, p. 458). So, they conclude, the problem isn’t weight loss, but disordered eating behaviours. (Eating behaviours are very unlikely to be normal if you’re severely underweight, and arguably people with bulimia might be underweight for their own bodies if not by population standards; but on the Mando theory, the main driver of improvement is the behavioural normalisation.) This is why tube feeding won’t on its own resolve the problem: the normal behaviours of eating haven’t been re-established.

All this has a whiff of tautology about it: becoming less disordered in one’s eating is what will drive recovery from an eating disorder. But perhaps their point is that we’ve got so led astray down dead ends of complex psychological intervention that simple truths feel like tautologies.

Beyond the eating: The ingredients of efficacy.

The Mandometer itself isn’t the entirety of the treatment, however: patients spend months in the clinic, having numerous aspects of their lives reconfigured. So which elements of the treatment really account for its striking success rates? If it’s ever to be rolled out on a wider scale, or indeed converted into a more universally accessible self-help regimen centred on the Mandometer device, or enhanced to achieve remission and recovery rates over 75%, detailed enumeration and characterisation of the ‘active ingredients’ will be required.

Most sceptically, we need to start by asking whether the success is due to something about the treatment or some other factor. Let’s start with possible confounding factors. The obvious candidate is the type of patient treated. Given the large numbers, systematic biases in patient readiness for treatment seem unlikely. It’s possible that there’s some broad difference between Swedish and e.g. US or UK patients (a large majority of the 1,428 total are Swedish nationals treated in Sweden), but there are plenty of papers reporting on different treatments for different nationalities, and there’s no particular reason to think Scandinavians are more easily treatable. We may also wonder whether the Mando patients are less severely ill than others, but actually the opposite seems to be the case: Sodersten et al. 2017 (p. 186) report lower average BMIs than the majority of other studies.

So could it be that those who end up at the Mando clinic have self-selected to be especially committed to treatment? This possibility might be supported by the team’s estimate that around 70% are motivated to start eating when they arrive; that seems a lower proportion of ambivalence than I would expect, but then I’m not sure what any other clinic would say about typical motivational states on admittance. And the team also remark that of course ‘motivation’ is a complex phenomenon, and often or usually coexists in their new arrivals with profound scepticism: ‘it will never work for me’. This is probably the same everywhere: a degree of hope, a dollop of despair; some energy for change, lots of paralysis hindering it.

The relation to CBT – and to common sense.

Then there’s the question of what the treatment consists of beyond the Mandometer. There are two other major physical and behavioural elements: provision of warmth and restriction of physical activity. Patients are given their own small warm room as well as a shared bedroom, and they rest in the warm for an hour after every meal, with the option to increase the temperature to up to 40 degrees Celsius. This is a method which goes back to William Gull (1874), who provided the first clinical description of anorexia, and it’s intended to reduce anxiety and prevent compensatory activity (Södersten et al., 2006). Physical activity is also monitored and gradually reduced to no more than slow walking around the clinic, and then the restrictions are gradually lifted again as treatment progresses. No psychoactive medications are used, and patients are withdrawn from any previously prescribed.

How about the more cognitive or psychological aspects of treatment? Well, this is where things become a little less clear. In a recent paper the Mandometer team describe their treatment as ‘restor[ing] normal eating behavior using mealtime feedback’ (Södersten et al., 2017, abstract), and state that ‘when eating behavior was normalized, cognitive and emotional abnormalities were resolved at remission without cognitive therapy’. In person, too, they are keen to insist that psychological therapy is not provided: that what they do beyond normalising eating habits using the Mandometer is ‘just common sense’.

I find this keenness to demarcate their practices from therapy intriguing. It makes sense in various ways. First, it creates a powerfully distinctive message: you don’t need therapy, you just need to be taught how to eat. Second, Per and Cecilia told me that they developed their method before they learned about CBT, meaning that as far as the genesis of what they do is concerned, formal psychotherapeutic methods weren’t needed. This kind of personal aspect can be a potent driver of how we understand what we do, even if there may be many different routes to a single idea. Third, a recent paper (Gutiérrez and Carrera, 2018) observed that a non-specific treatment protocol (Specialist Supportive Clinical Management) used as a placebo in five recent anorexia trials performed as well as the specialised eating-disorder treatments (e.g. CBT-E and the Maudsley method for adults, MANTRA), suggesting that current thinking on anorexia would do as well to return to the common-sense basics of a strong therapeutic relationship, plenty of ‘praise, reassurance and advice’, and a focus on eating normalisation and weight restoration (McIntosh et al., 2005). 

Last but not least, the response to their findings from the clinical mainstream (not least from CBT practitioners) has, by all accounts, been frosty at best, with clinicians and researchers following the time-honoured pattern of ignoring it, then trying to disprove it, then saying they knew it all along (Archie Roy, in Knight and Butler, 2004).

The rejection has in some cases been embarrassingly incoherent, with a renowned eating disorder specialist resorting to calling the treatment ‘bullshit’ (in a 2006 Australian news story here). The team told me that their invitations to conduct randomised controlled trials comparing the Mando treatment with standard treatments have been repeatedly declined, and a single attempt to compare the Mando treatment with ‘treatment as usual’ (van Elburg et al., 2012) was riddled with methodological problems, as made clear in Bergh and colleagues’ 2013 response (which, incidentally, was rejected by the journal that published the original study).

It would be entirely understandable if the hostility from those who practise more psychological forms of therapy had over time encouraged a forceful demarcation of their methods – though the strong claim that ‘“psychopathology” is an effect, not a cause, of starvation’ was present in the very first paper on the treatment (Bergh and Södersten, 1996, p. 612). The lack of others’ willingness to collaborate must feel especially galling since the team have also been recognised with 12 healthcare and entrepreneurship awards since 1998 (some of them for the obesity-focused dimension of their work; see here). Södersten tells me that many researchers invited to visit the Stockholm clinic have declined the invitation; that one who did visit was actively hostile to Bergh; and that another who visited subsequently dismissed the data he had reviewed during his visit as insignificant, saying they shouldn’t be published. (They were later published in the prestigious Proceedings of the National Academy of Sciences [Bergh et al., 2002]).

Whether there’s really such a stark divide between CBT and the Mando treatment’s accompanying support, though, is unclear. The team remark in one paper that a ‘great deal of time is spent convincing and coaxing the patients to start resuming their normal social interactions’ (2013, p. 881). More extensively, the Mando clinic website outlines the ‘four cornerstones of treatment’: normalising eating behaviour, heat and rest, decreasing physical activity, and social reconstruction. Under social reconstruction they say:

In addition to attending to our patients’ nutritional needs and eating behaviour we help them accept and appreciate their bodies, understand the mechanisms underlying eating disorders and recognise warning signals, develop emotional regulation, and eventually, return to school or work. The treatment also focuses on improving self esteem and self awareness, building confidence and enjoyment, and managing social situations and interpersonal relationships.

Each patient works with their case manager to set up short term goals, such as visiting a friend or reading a book, and long term goals, such as going on a holiday or learning to drive. A structured plan, for returning to school or work, is arranged with the patient. We encourage our patients to resume social activities, such as meeting with friends, going to parties, or taking a summer job or volunteer role.

As patients reach their goals their confidence increases, and in many cases a patient’s confidence is better after Mandometer treatment than it was prior to becoming ill.

This sounds a lot like the description of a course of CBT: helping the patient to understand the mechanisms maintaining their disorder, to practise more acceptant thoughts and attitudes, regulate emotions, enhance interpersonal skills, and so forth. And yes, both sound a lot like common sense because to some extent they are: cognitive-behavioural principles resonate because they have a logical simplicity to them that often makes us feel: how did I not realise that all along?

Take the ‘feeling fat’ example: in a cognitive-behavioural context you’ll be invited to ask, does this feeling of fat mean I am fat?, and to identify all the other reasons you might ‘feel fat’, from having just seen a photo of a very slim woman to having food in your stomach because you’ve just eaten. Understanding that physical states and behaviours and thoughts and moods/emotions all interact with each other, and that the interactions can be changed, is commonsensical: of course what I’ve just eaten affects how my body feels, and of course reminding myself of this fact makes a difference to how I interpret that sensation. (The next step, that this interpretation affects subsequent behaviour, or that thought can affect action, also seems obvious but is contested in the Mando theory; see my Appendix below.) The retrospective obviousness of these interactions makes it all the more valuable to have our attention directed to them by someone who has expertise in how they tend to manifest and how they can most efficiently be tweaked into healthier patterns.

The Mando team say more on the common-sense theme in their 2017 paper:

As a matter of fact, ‘cognitive’ support is also used in the treatment that aims at normalization of eating behavior, e.g., setting short-term social goals, modifying these goal as they are reached, and informing the patients that normal eating will facilitate reaching these goals. However, these goals, including getting a hair-cut and starting school, are not specific for patients with eating disorders but are rather more like good advice for anyone in need of a haircut or schooling. Negotiating goal setting in CBT for eating disorders is done using the Socratic approach, i.e., eliminating contradictions, but it is difficult to question the advice, considering that the aims are self-evident, e.g., establishing a good relationship with the patient, her friends and relatives, informing her of the adverse effects of dieting and the physical consequences of starvation, improving her problem solving abilities, etc. No one would question such reasonable suggestions, which are more like common-sense than based on scientific considerations. (p. 186)

It’s interesting that goals and techniques which are reasonable, self-evident, and generalisable beyond a clinical population are considered to be unscientific. Indeed, we might not need science to come to them, but if the scientific method is used to confirm their benefits to people with eating disorders, and to progressively refine and enhance them, then they turn into scientifically based methods. Sometimes scientific findings strikingly refute common sense (the world is flat) or folk psychology (change blindness blindness [Levin et al., 2000]); sometimes the science and the intuitions align (smells and flavours are indeed potent cues of emotional memories). We can’t assume that common sense and truth always agree or disagree.

Meanwhile, there’s good reason to expect cognitive support to be helpful in starting and maintaining recovery from anorexia, in everything from the decision to embark on recovery, to continuing despite discomfort, to developing resistance to sociocultural pressures that would push us back to illness when we’re vulnerable for other reasons. The Mando team offer some relevant observations on the first of these, the starting eating. They describe in an early paper (Bergh et al., 2002) how two patients who were not yet ready to start eating were trained to eat using the behaviourist principle of ‘successive approximations’, or taking baby steps towards the ultimate goal. Thus, ‘food was placed on the plate, patients placed empty forks in their mouths, food was placed on the fork, patients were encouraged to smell the food, and so forth. After three and six daily training sessions [respectively], patients started to eat in front of the monitor’ (this was before the days of the mobile app). Rewards for each iteration include verbal reinforcement, small gifts, and the promise of doing something nice later. This reconfiguration of the reward system gradually disrupts the reward associated with dietary restriction and exercise mediated by dopamine release (Södersten et al., 2008; Södersten et al., 2016). You can kind of call this common sense (one step at a time, the carrot instead of the stick, etc.), but it was also formalised, in B.F. Skinner’s animal experiments, into a distinct method of differential reinforcement in behaviourist conditioning.

I wonder how broadly successful such a method might be for encouraging eating in anorexia. Does everyone who is subject to a persistent enough version of this behavioural progression towards eating end up eating, and eating reliably and adequately, or does resistance sometimes persist, or eating never increase to nutritionally viable levels? Whether foolproof or not, it’s not clear to me that there’s a point where science has to stop being commonsensical.

Where does behavioural stop and cognitive start?

This method also raises the question of where exactly the dividing line is between the cognitive and the behavioural. An early study comparing behavioural and cognitive-behavioural therapies for bulimia found that BT was just as effective as CBT but worked quicker and had lower drop-out rates, among other advantages (Freeman et al., 1988): getting the disordered behaviours under control was apparently what really improved patients’ self-esteem and self-perception. But BT included ‘self monitoring, systematic modification of eating behaviour using graded tasks, and the teaching of alternative coping strategies, including relaxation training’ (p. 522). In these examples, as in the successive approximations towards eating, thought (and sensation, and emotion) is involved as well as action.

Yes, the discomfort induced by sitting in front of a plate of food is overcome by repeatedly sitting, smelling, starting to taste. But the changes that occur are cognitive as well as behavioural: learning to sit with the fear, discount it in favour of the verbal encouragement or the promised reward, work through the instinctive inner screaming that you shouldn’t be eating or can’t eat, cope with the panic that sets in after eating. The learning process initiated and sustained by behavioural prompts (getting you to sit at a table, to pick up the fork, etc.) is a cognitive and emotional learning process too. And this must be equally the case with ‘self-monitoring’ and ‘the teaching of alternative coping strategies’, or even with the heat treatment, which seems squarely physical, but is explicitly described as anxiety-reducing, i.e. effective in a specifically cognitive domain.

Overall, it seems to me that an unnecessary gulf has emerged between the Mando team, who see themselves as using behavioural techniques, and CBT practitioners, who call what they do cognitive-behavioural. Part of the problem is perhaps that the practice of CBT for eating disorders may have shifted since its inception to a more cognitive and less behavioural version. Recent studies which place more emphasis on eating normalisation (Dalle Grave et al., 2014; Calugi et al., 2017) seem, correspondingly, to have met with better success rates, but even here patients seem to lose weight during the follow-up period. (With Dalle Grave et al. weight was lost by 6 months and then mostly regained by 12 months – and of course we know nothing beyond that point. Relapse data are, as Lampard and Sharbanee [2015] point out, worryingly scarce across the field, and can be subject to the kinds of reporting problems I discussed in my previous post.) So perhaps the Mando team’s innovations should be seen less as a rejection of CBT and more as a call to return CBT to its origins: to a profound insight into the inseparability of thought, emotion, mood, behaviour, and physical state.

The Mando team admit this essential proximity at times: ‘The cognitive procedures that are part of CBT probably play a minor role in producing positive outcomes. Indeed, the cognitive therapy can be considered to be good advice similar to that given to patients whose eating behavior is being normalized’ (Södersten et al., 2017, p. 187). The hypothesis that the cognitive procedures play only a minor role is only a hypothesis, though, and a good way to test it would be to administer the Mandometer device as a self-help intervention alongside instructions about heat and no exercise, and see whether the remission rates remained as high. If you found that people don’t comply with the no-exercise rule or persist with the use of the device, this would indicate that some additional support is needed to ensure compliance. These other mechanisms might take various forms, and the Mando team refer to one obvious candidate: a change of place.

Because learning depends on place, the cues that maintain anorexic eating are eliminated when anorexics eat in a new place. As patients do, body weight increases. Because learning also depends on state, anorexics will enter a new state as they gain weight and eventually, when they reach their normal weight, the patients are no longer bothered by their previous condition. Their serious psychological problems resolve completely.

One hypothesis is therefore that the Mandometer-centred intervention is less likely to work when implemented in one’s everyday environment. This would mean that a self-help version of the Mando treatment at home might not be feasible. But it’s worth asking what degree of environmental change is really needed: could simply eating one’s meals in a different room be enough, and could other contextual changes like presence or absence of others while eating, use of different utensils and crockery, use of music or other accompaniments, play a role? These are all empirical questions which demand investigation as part of the broader question about what the active ingredients of this method really are. This investigation will be best furthered, in my opinion, by recognising that whether an intervention starts with something that looks bodily, behavioural, or cognitive, the changes it elicits will necessarily cascade throughout the entire embodied and environmentally embedded system that is a human being.

The potential for breaking down the current barriers to collaboration seems great, and important to seize. What I learnt at the Mando clinic suggested that the cognitive aspects of treatment are probably both quite significant and also currently a little opaque. For example, the ‘case managers’ who oversee both the day-to-day details of patients’ care and their progression between different phases appear to have a good deal of autonomy to adopt principles and methods of their own choosing. Because all are clinically trained elsewhere, what they choose may often be shaped by therapeutic methods that have, in one way or another, formalised ‘common sense’ into a psychological framework. So we might well find that the ‘common-sense’ cognitive support in this paradigm actually ends up looking surprisingly (or unsurprisingly) similar to kinds of therapy practised elsewhere.

Where next from here?

Ultimately, I’d love to see a proper comparison done of the methods involved in the Mando treatment and in some CBT or other cognitive or psychological treatment-as-usual paradigm. And what I think is most important is not a randomised controlled trial of one hermetically sealed treatment method versus another (it’s pretty clear that the Mando works far better than other treatments), but careful conversation and reciprocal observation between practitioners of the distinct methods. What are they actually doing day-to-day, hands-on with individual patients? What descriptive terminology should we use to ensure we all know what we mean when clinicians and researchers talk about what they do with patients? This kind of investigation will require and encourage collaboration, and will allow the significance of the kinds of findings being generated here to be fully explored and translated into widely accessible treatment possibilities. It also aligns with other calls for closer integration between CBT and other therapeutic traditions, recognising the potential for combination of complementary active ingredients in order to enhance efficacy (Lampard and Sharbanee, 2015).

My hunch is that the core of the Mando treatment (the device plus post-meal warmth and minimal exercise) plus something that looks a lot like some version of what is usually called CBT could be a powerful combination. The work needs doing to test this hunch and other people’s, because a treatment apparently this successful needs to be better understood, and this understanding can come only from more intensive scrutiny of its constituent parts. Until that happens, the Mando method and its reception remains an interesting case study in the human difficulties of doing science, and an important development in the treatment of anorexia.

Let’s not let it stop there. We urgently need to challenge the silence of the status quo. We need to get past the politics to do better – more open, more collaborative – science. We need to bring an end to what is not just a bizarre situation, but an unacceptable one: we need to remember that the health of real people is at stake here.


An appendix on mental causation

The deeper question underlying all the discussion of cognitive versus behavioural interventions concerns whether there is mental causation: whether thoughts can affect actions. The Mando answer is no: ‘thoughts or cognitions are driven by behaviour not the reverse’. I’m not sure the answer can be so simple, though. As I emerge from three years’ hard work collaborating on the third edition of a textbook on consciousness, I’m acutely aware that no one really knows what conscious thought even is or how to make sense of its relationship to the material brain or organism or environment or to behaviour. It seems hard to claim that there is no relationship between thinking about, say, how much better you’d feel about yourself if you were five kilos lighter and taking actions to lose five kilos. The relationship may not in fact be a causal one: thought causes action. Indeed, it almost certainly can’t be, since the claim that ‘consciousness itself’ has causal power basically amounts to invoking magic. But a system in which thoughts about self-esteem and weight loss are always sloshing around is a lot more likely also to be a system in which weight-loss-related actions occur. Even if in reality the temporal sequence

have thought (e.g. this picture of a supermodel in a bikini makes me feel rubbish about myself) ➙ take action (e.g. skipping lunch)

doesn’t also represent a cause-and-effect relationship, the state of the organism in which such thoughts and such behaviours occur is not a healthy one, and the state of the system seems to be alterable by intervening in the thought processes themselves: for example, asking oneself where the feeling of rubbishness comes from, challenging its premises, etc. It may be that if both thoughts and actions are caused by some underlying processing, then changing a thought necessitates some new underlying process, which may cause different actions. Thus, I think we can make a case for the importance of cognitive intervention independently of any strong philosophical claims about mental causation.


Ammar, A.A., Sederholm, F., Saito, T.R., Scheurink, A.J.W., Johnson, A.E., and Sodersten, P. (2000). NPY-leptin: Opposing effects on appetitive and consummatory ingestive behavior and sexual behavior. American Journal of Physiology: Regulatory, Integrative and Comparative Physiology278(6), R1627-R1633. Open-access full text here.

Bergh, C., Brodin, U., Leon, M., Kreier, F., Buijs, R., and Södersten, P. (2015). Comparison of Mandometer treatment and treatment as usual for anorexia nervosa: Standards of evidence. International Journal of Eating Disorders. Direct PDF download (preprint) here.

Bergh, C., Brodin, U., Lindberg, G., and Södersten, P. (2002). Randomized controlled trial of a treatment for anorexia and bulimia nervosa. Proceedings of the National Academy of Sciences99(14), 9486-9491. Direct PDF download (final version) here.

Bergh, C., Callmar, M., Danemar, S., Hölcke, M., Isberg, S., Leon, M., ... and Palmberg, K. (2013). Effective treatment of eating disorders: Results at multiple sites. Behavioral Neuroscience127(6), 878. Direct PDF download (final version) here.

Bergh, C., and Södersten, P. (1996). Anorexia nervosa, self-starvation and the reward of stress. Nature Medicine2(1), 21-22. Direct PDF download (final version) here.

Calugi, S., El Ghoch, M., and Dalle Grave, R. (2017). Intensive enhanced cognitive behavioural therapy for severe and enduring anorexia nervosa: A longitudinal outcome study. Behaviour Research and Therapy89, 41-48. Paywall-protected journal record here.

Dalle Grave, R., Calugi, S., El Ghoch, M., Conti, M., and Fairburn, C.G. (2014). Inpatient cognitive behavior therapy for adolescents with anorexia nervosa: Immediate and longer-term effects. Frontiers in Psychiatry5, 14. Open-access full text here.

Freeman, C.P.L., Barry, F., Dunkeld-Turnbull, J., and Henderson, A. (1988). Controlled trial of psychotherapy for bulimia nervosa. British Medical Journal296(6621), 521. Direct PDF download (final version) here.

Galhardo, J., Hunt, L.P., Lightman, S.L., Sabin, M.A., Bergh, C., Sodersten, P., and Shield, J.P.H. (2012). Normalizing eating behavior reduces body weight and improves gastrointestinal hormonal secretion in obese adolescents. The Journal of Clinical Endocrinology & Metabolism97(2), E193-E201. Direct PDF download (final version) here.

Gull, W.W. (1874). Anorexia nervosa (apepsia hysterica, anorexia hysterica). Transactions of the Clinical Society of London7, 22–28. Reprinted in 1997 in Obesity Research5(5). Open-access full text here

Lampard, A.M., and Sharbanee, J.M. (2015). The cognitive‐behavioural theory and treatment of bulimia nervosa: An examination of treatment mechanisms and future directions. Australian Psychologist50(1), 6-13. Paywall-protected journal record here. Direct PDF download (preprint) here.

Levin, D.T., Momen, N., Drivdahl IV, S.B., and Simons, D.J. (2000). Change blindness blindness: The metacognitive error of overestimating change-detection ability. Visual Cognition7(1-3), 397-412. Direct PDF download (final version) here.

Prince, A.C., Brooks, S.J., Stahl, D., and Treasure, J. (2009). Systematic review and meta-analysis of the baseline concentrations and physiologic responses of gut hormones to food in eating disorders. The American Journal of Clinical Nutrition89(3), 755-765. Open-access full text here.

Södersten, P., Bergh, C., and Leon, M. (2016). Commentary: New insights in anorexia nervosa. Frontiers in Neuroscience10, 483. Direct PDF download (final version) here.

Södersten, P., Bergh, C., and Zandian, M. (2006). Understanding eating disorders. Hormones and Behavior50(4), 572-578. Direct PDF download (final version) here.

Södersten, P., Nergårdh, R., Bergh, C., Zandian, M., and Scheurink, A. (2008). Behavioral neuroendocrinology and treatment of anorexia nervosa. Frontiers in Neuroendocrinology29(4), 445-462. Direct PDF download (final version) here.

van Elburg, A.A., Hillebrand, J.J., Huyser, C., Snoek, M., Kas, M.J., Hoek, H.W., and Adan, R.A. (2012). Mandometer treatment not superior to treatment as usual for anorexia nervosa. International Journal of Eating Disorders45(2), 193-201. Paywall-protected journal record here.

Gutiérrez, E., and Carrera, O. (2018). Anorexia nervosa treatments and Occam's razor. Psychological Medicine, 1-2. Paywall-protected journal record here.

McIntosh, V.V., Jordan, J., Carter, F.A., Luty, S.E., McKenzie, J.M., Bulik, C.M., ... and Joyce, P.R. (2005). Three psychotherapies for anorexia nervosa: a randomized, controlled trial. American Journal of Psychiatry162(4), 741-747. Open-access full text here.