As the year grows old I’ve been reflecting on the passing of time, and on what it means for us humans to grow older, in sickness and in health. I’ve been remembering the years I spent ill, and the growing number of years I’ve now been well. I’ve been recalling the attempts I made at recovery before the time that worked, and wondering what made the early attempts fail where the last one succeeded. I’ve been thinking about the many readers who write to me as teenagers caught in the confusions of growing up in a world like this, and about the many who have lived longer, with or without illness. Anorexia can take hold at any age, and it can be recovered from at any age, after any number of months or years or even decades.

This post is, above all, for every reader who has been ill for a long time and is losing faith in the idea of life ever being better than this. It will make the controversial suggestion that you shouldn’t necessarily expect recovery from anorexia to be easier the sooner you attempt it. I’d like to stress upfront, though, that this comes with an important caveat: that you should still attempt recovery as soon as you can, any time you can, and that the ‘failures’ are an integral part of the ultimate success.

So, on what basis do I make this suggestion? One plank of the received wisdom about anorexia, and eating disorders more generally (and probably mental and physical ill health more generally) is that recovery is easier, or more likely to be successful, the sooner one attempts it. ‘Catch it early’, we are told; ‘early intervention is key’.

One response to a new UK governmental mental health strategy in 2011, No Health without Mental Health, typifies this tendency in proclaiming:

One of the key themes that emerges clearly from the strategy is that of intervening early. The evidence in almost all areas of mental health is indisputable. Mental health problems are widespread, they affect people of all ages, and the longer they remain hidden, stigmatised and untreated the worse they become and the greater the losses people experience in all areas of their lives.

Early intervention really does have to start early. Parenting support for young families, including before birth, can make a dramatic difference to the emotional health and long-term life chances of our children. (Duggan, 2011)

When we take a closer look at this ‘indisputable evidence’, however, the waters start to look a little murkier. This is certainly true for eating-disorders research; I imagine it is for other forms of mental illness too. Let’s start with a 2015 report by the UK eating disorders charity Beat on the social, health, and economic impacts of eating disorders. The document reports on a survey of 435 respondents suffering from a range of eating disorders, as well as 85 carers, and found amongst other things that

On average 63% of our respondents experienced at least one relapse requiring repeat treatment. With no single treatment regime or type of intervention standing out from our analysis as being substantially superior to other interventions, i.e. capable of improving the chances of recovery for this group, more research is needed to understand what works and why. Nevertheless, the case for earlier intervention appears to be supported given the marked reduction in relapse (of 33%) for those sufferers that recognised their symptoms and sought help quickly. (2015, p. 8)

This seems an unobjectionable statement, one that fits with our intuitions and so serves inconspicuously to strengthen those intuitions. It generates the centrepiece of the message we’re asked to take away from the research: ‘full recovery is possible if intervention is early enough’, ‘the sooner someone gets the help and treatment they need, the better their chances of making a full recovery’ (p. 4).

As ever with research on everything, though, we should take a closer look at what is being claimed on what basis. The crucial thing is, this is a survey. This is a collation of people’s own statements about their experiences, with all the potential for bias and selectivity that entails, both in respondents’ answers and in who responds to the survey in the first place. Maybe those who sought help sooner and got better successfully were more likely to complete the survey, or maybe those who finally got fully better after long illness removed themselves from the eating-disorder world and never saw the survey, or maybe those who sought help sooner were more inclined to overestimate the success of their recovery. Crucially, because this is a survey not an experiment, the possible contributions of these confounding factors and many more can never be determined.

The most a survey, or observational study, can ever establish is correlation, not causation. (Gary Taubes is great on this in the context of nutritional research.) In this case, therefore, the most we can conclude with any confidence at all is that those people who report having sought help early are less likely to report relapse; the two responses correlate. This is not the same as concluding that seeking help early reduces the risk of relapse. Other likely causal factors include severity of illness, strength of personal support network, availability of high-quality healthcare, and similar.

Because there is no direct causal evidence, many other proclamations about early intervention paper over the cracks between causation and correlation. One way of doing this is to use the easily misleading term predict instead of correlate:

The outcome of anorexia nervosa is predicted by body mass index (BMI), physical risk, age and illness duration. Recovery from anorexia nervosa becomes much less likely the longer the illness has persisted. Thus treatment for anorexia nervosa is more likely to be successful if the illness is recognised early – before weight loss becomes too severe and protracted. (Treasure and Russell, 2011)

This is a good example, because prediction (i.e. correlation) is used as the basis of a claim that looks awfully like a causal one (more likely to beif…).

Even if the term correlate is used, the discussion often proceeds in a way that only a causal link would justify:

Early detection and treatment of eating disorders is strongly correlated with better outcomes. The literature reviewed constantly stresses the need to increase capacity to identify and intervene early [and] point to a number of relatively straightforward ways to work toward this goal. [...] (Manitoba Health, 2006)

Even more rarely, we find acknowledgement that the belief is much an article of faith as a conclusion from evidence:

Almost all clinicians would agree that ‘early intervention’ is important in the illness trajectory and outcome of eating disorders. However, the literature is sorely inadequate and there is no evidence to substantiate this belief, nor even consensus as to what should constitute early intervention. (Marks et al., 2003, p. 513)

So, a good sceptical practice is to wonder, every time you read a statement about connections of these kinds, whether correlation or causation is being claimed, and whether the evidence justifies a causal claim, and if not, what the alternative explanations might be. Most often, the most plausible alternative explanation is the possibility that a third, unstated factor is responsible for the association between the two that are mentioned: again, illness severity might be a good candidate here (one possibility is that those with less acute illness are more able to intervene and initiate recovery earlier).

Illness severity is an interesting candidate, because often the assumption seems to be that severity must increase with duration. The core argument for early intervention can perhaps be summarised as follows. Recovery is a demanding process requiring full and active participation of the patient. The longer the patient has been ill, the greater the severity of the illness and therefore the less the likelihood of their being able to participate fully and actively in recovery (for physiological and/or psychological reasons). And in many cases this may be true. But the relationship between duration and severity in anorexia is often more complex than this. The numerous responses to my post on ‘How and why not to stop halfway in recovery’ make clear that many people go through a period of severe illness followed by partial recovery followed by long subsistence in an in-between state of neither health nor severe illness. This phenomenon may well not be picked up on by clinical trials which set irresponsibly low thresholds for ‘recovery’, but its reality challenges a central plank of the early-intervention argument as commonly presented. Someone could easily have been ill a long time and yet not, right now, be severely ill.

The best way to begin to disentangle these complex relationships of correlation and causation would be to conduct a randomised controlled trial (RCT) in which the single difference between groups is the time which elapses between onset of illness (or identification of illness) and seeking (or starting) treatment. This experiment is unlikely ever to be done in this form, because the ethics seem so dubious: wouldn’t we almost certainly be condemning the ‘wait longer’ group to the likelihood of more severe and longer-lasting illness? It's worth noting, though, that many RCTs testing other questions (like whether a new treatment method works better than an old one) do include a 'waitlist' control group (whose duration of illness should average out similarly to the test group) to whom no treatment is given; it's just that the next step is never taken: the control group being given the same treatment with a timed delay.

The closest I’ve found to an RCT on timing of intervention is a 1987 experiment by Russell and colleagues involving 80 participants suffering from anorexia or bulimia and allocated to four subgroups on the basis of age of illness onset, illness duration, and eating-disorder type. The purpose of the study is to assess the efficacy of family therapy versus individual supportive psychotherapy. The main finding was that family therapy is more effective for those whose illness begins at a relatively early age (under 19) and has not become chronic (less than 3 years); a more tentative finding was that individual therapy may be more effective for those in whom it starts later (19 or over). (These results were confirmed at 5-year follow-up; see Eisler et al., 1997.) No differences between efficacy of therapy types were found for those with an illness that had become chronic (over 3 years) despite younger age of onset. So although this research tells us something about the interaction between onset age, duration, and therapy type, it is far from answering the central question: holding all other factors constant, does greater illness duration reduce likelihood of full recovery? At such, it does not justify the conclusion claimed by Treasure and Russell (2011) that ‘Recovery from anorexia nervosa becomes much less likely the longer the illness has persisted’.

Meanwhile, the mantra of early intervention continues to dominate research, clinical practice, and eating-disorder policy. Statements like the following are often made with no effort to specify the nature of the ‘strong evidence’ referred to:

There is strong evidence that the longer the duration of illness, the harder it is to achieve recovery. […] The earlier the intervention, the more likely a patient will recover. (Rome et al., 2003)

Quite often specific numbers of years are mentioned – most often ten, the neat decade-long span. So we find the observation that ‘recovery is most likely within the first 10 years after symptom onset’, or that ‘10 years represents a meaningful point at which the probability of recovery begins to decrease’ (Von Holle et al., 2008, p. 114). This conclusion too is reached on the basis of retrospective interview data, i.e. on correlational data. The supposed benefits of early intervention are all the harder to argue with because they are so often bound up with those of prevention (indeed, ‘early recognition and intervention’ are sometimes referred to as ‘secondary prevention’ [e.g. Marks et al., 2003]). Prevention is pretty hard to argue with on any grounds, and folding early intervention into it gives the latter an extra layer of incontrovertibility, even though by definition early intervention can happen only when prevention has failed.

Going back to our Beat survey example, it so happens that the wording of the relevant survey questions is also not quite aligned with the conclusion drawn. Respondents were asked

How long after becoming aware of the symptoms of an eating disorder did you first seek help, assistance or advice?


How long after seeking help were you diagnosed with an eating disorder?

as well as

Have there been any subsequent episodes of treatment for a recurrence of the eating disorder?


How long have you suffered from the eating disorder?

None of these questions gives us any insight into the time that may elapse between becoming ill and becoming aware of or accepting the illness – a gap which by definition may not be accessible to self-report. That is, the gap between onset of symptoms and the seeking of help is not determinable by these means. In addition, none of the questions allows for the report of a recurrence of illness for which treatment was not received. This means it’s possible, if unlikely, that seeking help early correlates not with no future relapse but with future relapse not followed by treatment. One could speculate, if in the spirit of devil’s advocate one were looking for support for this possibility, that having tried to ‘nip it in the bud’, and not lastingly succeeded, all faith evaporates in the possibility of succeeding later, and shame grows at the fact of not having managed it before. So a vicious circle would ensue.

However likely or unlikely this possible explanation seems to you, formulating it points to the main reason why I’m writing this post: that the received wisdom may not only turn out to be incorrect, but also do damage in its propagation. From a charity’s point of view, it may seem obvious that making the case (even on slim evidence) for early intervention can only be a good thing: that even if it isn’t actually true that it’s easier to recover if you try sooner, it certainly can’t be harder to recover if you try too soon, and therefore no harm is being done by spreading the meme. (A related point is that achieving higher levels of early detection and intervention require changes in public healthcare provision – awareness-raising amongst medics and the wider population, methods of treatment suited to less acute and/or chronic stages of illness, etc. – that give a mandate to the kind of work such charities do, perhaps especially the work of volunteers like Beat’s ‘Young Ambassadors’, who are recovered former sufferers under 25. This potential source of bias can be interpreted more or less cynically.)

But propagating the idea that recovering sooner must be easier might, as I’ve begun to suggest, come with dangers of its own. Broadly speaking, our prediction might be that the more prevalent this belief is, the more feelings of impotence and/or shame will grow the longer an illness continues. And this is indeed what we see happening to many people. They fall ill, perhaps quite early in life; they make one or two unsuccessful attempts at recovery, and then they subside into a weary conviction that it’s too late to try again, because why would it ever work now when it didn’t before, when the chances were meant to be so much higher back then? They’ve failed at this, as at everything else except being ill. Why fight it?

We can’t blame the earlier-is-easier assumption entirely for this, of course. The nature of anorexia is that it wears you down, wears away your ability to think clearly and flexibly and to imagine that your life could be otherwise, wears down your capacity to even want to be able to take joy in more than eating and not-eating and maintaining your small circle of safety. The feedback loops between the behaviours, the physical state, the emotional and social capacities, the thought patterns, and the behaviours become rapidly more entrenched (see e.g. Treasure and Russell, 2011 – and my post on how and why to start recovery). So the belief in the possibility of recovery, and specifically your own capacity to achieve it, is bound to diminish with time.

On the other hand, this relentless erosion of life’s meaning can eventually create a backlash against itself. If you half-live like this for long enough, something will eventually come along to shock you into realising what you’ve been reduced to – whether it’s as tiny as the thousandth time you have to say no to the invitation to go out for a drink, or as great as the day someone you love dies having for years and years only known you ill. And the passage of time is not an incidental factor here.

Time matters because it takes time for these kinds of reminders to happen (poignantly or miserably or downright viciously) often enough for them to take root in you as a new feeling that may, in time, become the motivation to embark on something new. Time matters because it provides the undeniable confirmation that yes, things really have got this bad; there is no pretending any more that your life is not trapped in a merciless tapering to nothingness.

Sheer weight of time spent ill is of course not the only thing that can offer this confirmation; an acute crisis like hospitalisation or a shock like losing your job could provide it too. There were no objective crises of that kind for me – moments of heightened emotional turmoil with family, yes, and threats of (what I interpreted as) hospital incarceration, but no point at which ordinary life gave way completely to the inroads of anorexia. So maybe I’m inclined to overestimate the power of the simple passage of time. But I think there are kinds of acceptance that only long intimate love-hate acquaintance can bring; some kinds of questions that only relentless re-posing can really answer.

In the beginning, anorexia poses as an answer to other questions. Later it starts to pose those difficult questions (would I be better off without it?) that are rarely answerable the first or even the tenth time - especially as cognitive functioning is progressively impaired by malnutrition. But anorexia is able to take hold because it seems to offer solutions. It promises control (and self-control is an incontrovertible virtue, no?); it promises simplicity (what more appealing quality in a world as bewildering as this one?); it promises, as an incidental benefit, thinness (who’d say no?). The early days are the honeymoon days, and the violence becomes clear only later. It’s not for me to risk the sweeping generalisation of how much later the clarity is likely to come – we all have to find that out for ourselves. But if there were none of those seductions to suck you in to begin with, anorexia would be a different beast entirely.

Perhaps, then, to expect early intervention always to be preferable is to ignore the reality of anorexia’s time course. Step in too early, and perhaps you risk stepping in at the point where the apparent benefits outweigh the costs, leaving a space of gently corrupted ideals that other life experiences will later rush to fill: remember the good old days before they forced me to ‘get better’…? Wouldn’t everything be better if I were a bit more like that again?

The other dynamic shape to bear in mind is that of recovery. Recovery from any kind of illness, especially one with a pronounced psychological component, is not a linear affair. There are phases in which enormous effort bears little apparent fruit, phases where the progress starts to feel self-sustaining; periods of stumbling, falling, relapsing. Recovery from anorexia often requires several attempts, sometimes separated by years. Perhaps if the first one came sooner, this would be less often true; perhaps not. I’ve written before (here and here), though, about how different I felt the last time from all the other times: I sensed a steel thread of certainty which had never been there before. Other people tell me the same thing; one hears it, too, about giving up smoking or bidding farewell to other addictions. At some point you are ready to declare: enough is enough. There is much that can be done to hasten the arrival of this moment – by yourself and by others around you – but you can’t force or predict the moment of its appearance, nor can you fake it.

And when it comes and you start recovery in earnest, for the final time, you draw on all the wisdom you’ve accumulated in those previous efforts where your heart wasn’t fully in it: your hard-won knowledge of the things that make change harder or easier for you, whether they relate to food types, mealtimes, exercise, stress, other people, or anything else. From one perspective those past attempts are one long series of failures. From another they are the important, maybe even the necessary, stepping stones to success. The idea of multiple relapses is never a pretty one. Even now, the time aged 21 when recovery was going well and a single identifiable factor brought it to a standstill and me back apparently to square one still makes me wonder, a little wistfully, what if… But it taught me something. Everything we live through does, especially relapse.

And when I’m looking back with more realism, my overwhelming feeling is that it couldn’t have been otherwise. There was much I needed to learn about anorexia and how unremittingly shit it is, and as a very young woman I simply couldn’t have learned it all yet. Nonetheless, teenage recovery is often promoted as part of the emphasis on early recovery. In many studies of anorexia, earliness in the course of the illness is conflated with youth (i.e. earliness in the life course). For example, family-based therapy (typically aimed at children and adolescents) often goes hand in hand with ‘early intervention’ (e.g. Treasure and Russell, 2011; Jones, 2012). In general – and understandably, given that onset of anorexia often occurs during adolescence – early intervention often in practice means intervention with teenagers:

Early intervention appears to interrupt the increasingly restrictive dieting and growing social isolation. But in many individuals, dieting and weight loss dominate mental life for years and severely impair physical health and social and occupational development. […] Anorexia nervosa in adults, including those only in their 20s, is generally much more refractory to intervention. […] How is it that children and adolescents in generally good health develop an illness that is often chronic, severe, and possibly deadly but that may, especially if treated early, fully remit without any long-term consequences? […] Once they have become established as habitual, the behaviors are highly resistant to change and are a critical element in the persistence of the disorder. (Walsh, 2013, my italics)

Here the notions of short duration and young age are seamlessly intertwined. And this is another important confounding factor that makes it hard to be confident about the standard message: perhaps the key difference is between the cognitive and/or physical malleability of adolescence versus the greater entrenchment of adulthood, as well as or rather than length of illness. But the life-course factor has tended to serve as another element that helps the case for early intervention make itself. Given how profoundly the direction of one’s life is shaped by those early years, it makes complete sense to predict that disruption caused by illness to education, initial employment, the development of social and sexual relationships, and so on would have profound negative consequences that could easily contribute to a lastingly vicious circle of ill health. Again, it’s worth remembering that making complete sense isn’t the same as being true, and for some people the awfulness of that disruption may, again, be the most real motivation for change. But in any case, it’s even more important to remember that if you don’t happen to develop anorexia in adolescence, that correlation between age and length of illness won’t operate in the same way for you: early intervention will mean something different for you.

And this is why I’m writing: not to perform a perverse picking-apart of the research literature for its own sake, but in the spirit of a reassuring reminder. This post is for those of you who have suffered from anorexia for years and years, to the point where you can’t imagine life, or yourself, without it. It’s for those of you who are no longer young, maybe no longer middle-aged, and feel ignored by everything that’s ever written about anorexia. It’s for those of you who have tried countless times to get better and have barely any energy left to try again. It’s for you if the labels of ‘chronic’ or ‘severe and enduring’ translate bleakly into ‘lost cause’ and ‘past helping’.

Until you’re dead, it’s never too late. Every failure has something to teach us. Not everything is easier when we are younger. Not everything is easier when we are less ill. We’re asked to take on trust that fatalism means: not believing that the first possible opportunity is the best you’ll ever have. But in fact perhaps fatalism comes also from precisely that message: from believing that once the early days are behind you, there is little hope left.

One of the problems with eating-disorder research which I’ve written about before is that the threshold for ‘full recovery’ is set far too low. Another is that follow-ups after more than a year or two are rare. Combined with the particular limitations specific to research on early intervention, we’re left not knowing what kind of success is really being measured in these findings of early intervention’s success. For me, although it’s easy to wish I’d managed to do what my child psychiatrist hoped I had, and kicked the anorexic habit in time to be well throughout my 20s, I suspect that being that ill for that long was one thing that helped me bid anorexia farewell with the conviction I ultimately did.

As I said at the beginning, none of this is to say that any effort at escape from anorexia should ever be deferred. Some of the potency of the passing of the years comes from the repeated efforts towards recovery that feel like failures at the time but make ultimate success more likely. If you have any conviction at all that you don’t want to be ill for the rest of your life, act on it whenever and however you can, whether that flash of steel certainty is there or not. But by the same token, if you feel the light is fading and winter descending on the story of your illness, take comfort in that, not despair, for it means that anorexia’s days, not yours, are now numbered.


Beat. (2015). The costs of eating disorders: Social, health and economic impacts. Full text here

Duggan, S. (2011). Early intervention is rightly at the heart of new mental health strategy. Full text here.

Eisler, I., Dare, C., Russell, G.F., Szmukler, G., le Grange, D., and Dodge, E. (1997). Family and individual therapy in anorexia nervosa: A 5-year follow-up. Archives of General Psychiatry, 54(1), 1025-1030. Abstract here.

You are reading

A Hunger Artist

An App for Recovery from Anorexia

Baby steps towards a new way of helping full recovery happen.


A thought

How Much Does a Blog Title Matter?

Why this blog is called A Hunger Artist