Most people, me included, probably don’t know a huge amount about human metabolism beyond the fact that it’s a set of chemical reactions by which food is turned into energy, and into building blocks for new cells. What most people care about tends not to be the complex ways in which metabolism ‘allow[s] organisms to grow and reproduce, maintain their structures, and respond to their environments’ (Wikipedia), but mainly the speed at which all these things happen – or, ‘metabolic rate’: when I Google ‘metabolism’, Wikipedia comes up first, but the second hit is the UK NHS page ‘How can I speed up my metabolism to lose weight?’ (spoiler alert: you can’t).

This is where the link with eating disorders comes in. My post back in 2014 on ‘Recovery from anorexia: Why the rules *do* apply to you’ was prompted by many comments and questions from readers that made clear just how prevalent and deep-seated is the anorexic fear that recovery won’t work out as it should for me, even if it might for everyone else. One of the classic examples of this fear is the one I quoted there: ‘because I’ve been so ill for so long, I’ve ruined my metabolism and it’s never going to normalise’. The flipside of the unhealthy but now widespread belief in the possibility of ‘speeding up your metabolism’ to lose weight is the conviction that nothing you do could ever right the past wrongs inflicted on your body. One attitude exaggerates the pliability of the metabolism, the other underestimates it. 

It’s important that we get these facts straight rather than relying on partial or misinformation, because the response of the body’s metabolism to weight restoration after anorexia is of crucial importance when it comes to accepting the necessity of regaining weight, and accepting that its ‘dangers’ are less than the anorexic mind might fear. So, how does the metabolism – or more specifically, metabolic rate – actually change in refeeding? This is a question that’s been bothering me for a while, and which I’ve put off tackling because of the amount of research required to answer it properly. But here goes. This will take a while – indeed, I’ve decided to split the material into two posts to make it a bit more manageable – but bear with me.

Keys, Brozek, and Henschel 1950, p. 329
Figure 1
Source: Keys, Brozek, and Henschel 1950, p. 329

Let’s start with what happens when you starve yourself, and then move on to the recovery process. It’s well established that as an adaptive response to reduced food availability, the body’s basal metabolic rate is reduced, in order to lessen the negative impact of undernutrition, and it’s easy to see the evolutionary reason why this happens. The transition from a response to short-term fasting to the adaptation to extended semi-starvation may not happen immediately (see my post on intermittent fasting, and Wang et al. 2006), but in the longer term this energy-saving mechanism is predictable and robust (see Figure 1). As Keesey and Hirvonen (1997) put it:

when weight loss occurs, resting metabolism declines by an amount significantly in excess of that expected from the loss in metabolically active tissue […]. A decline in resting energy expenditure disproportionately larger than the associated loss in body mass indicates that less energy is required to maintain a gram of tissue in an individual who is weight reduced rather than at the normally maintained body weight.

This metabolic shift has the familiar symptoms we’d expect: lack of energy, low blood pressure, sensitivity to cold, and so on. The question is, what happens when food starts to be more freely available again? How quickly does the body start to ramp up the metabolic rate again in order to restore the body to optimal functioning, and how much does it play things safe in case the energy availability drops off again?

This question is of particular urgency when it comes to the later stages of weight restoration. Everyone accepts that restoration has to proceed to some ‘minimally healthy’ weight, though definitions of that level vary between a BMI of about 18 and 20. But the idea that it may be important to allow restoration to continue beyond this kind of level is more controversial – both amongst sufferers themselves (every fibre of the anorexic mindset screaming that anything more than ‘minimally healthy’ must be avoided at all costs), and, more surprisingly, in the clinical literature.

From my own experience, together with the anecdotal evidence that comes from you, my blog readers, and my reading around some of the anorexia and nutrition research, I concluded some time ago that the common ‘minimally healthy’ target of, say, a BMI of 20 (for Caucasian populations) is not in fact a healthy target at all, because in the vast majority of cases it doesn’t permit a return to ordinary eating habits, a reduction in the hunger that typically accompanies weight restoration, or a full normalisation of metabolic rate. In my post on ‘How and why not to stop halfway’ in recovery, for example, I noted that ‘your metabolism won’t normalise until you reach your natural bodyweight’. I also explained (citing evidence from Dulloo et al. 1997) why the difference in rate of restoration of fat and fat-free mass make a temporary overshoot beyond one’s longterm stable bodyweight likely to be necessary for full recovery. I also drew on guidelines set out by the ED blogger and patient advocate Gwnyeth Olywn, who in her post on ‘Phases of recovery from a restrictive eating disorder’ says similarly that

Once your body reaches its own optimal weight set point (and only your body decides what that is) then it just stops gaining weight and starts maintaining the optimal set point it has reached. It does this seamlessly because the metabolic rate moves back into the optimal range at that same time and biological functions that were on hold are now back on line.

The important consequence of this is that one can maintain a healthy weight on the same energy intake as maintained weight restoration up to that point. In another post, ‘I need how many calories?!’, Olywn adds a bit more detail, outlining three stages of recovery:

The body approaches recovery in the following way:

1 keep everything suppressed and take the energy to deal with the backlog of cellular repair (leading to bloating and water retention initially) and sock the rest away in fat stores (usually disproportionately around the mid-section to insulate vital organs);
2 assuming there is enough energy still coming in, then address longer term repair issues (bone density etc.) and begin to notch up metabolic rates and bring biological functions back on-line;
3 assuming adequate energy continues to come in on a daily basis, then fire up the regular neuroendocrine system back to normal and allow the metabolic rate to go back to normal as well.


Once you hit your body’s optimal weight set point, then the metabolism is normalized and that means that the extra energy you were taking in for weight gain and repair now goes to the usual day-to-day functions that were not happening at all from the moment you first restricted calories (whenever that was).

You gain on the minimum guideline calories+ [a minimum of either 2,500, 3,000, or 3,500 kcal a day, depending on sex, height, and activity levels] and then you maintain on pretty close to that same amount. Shocking, but true.

Here, then, the metabolic rate is assumed to increase somewhat in the early to mid-stages of recovery, before normalising once the ‘optimal set point’ is reached, meaning that dietary intake needn’t change from the weight restoration to the weight maintenance phases. However, no scientific or other evidence is provided to back up these claims. Other helpful posts on this subject at Science of EDs, EDBites, and Barbells and Beakers offer a quick survey of some relevant research, but don’t directly address the question of when full metabolic normalisation happens. So I thought I’d better have a proper look myself.

There are three components of total energy expenditure: resting energy expenditure (REE), which accounts for around 60% of the total; diet induced thermogenesis (DIT), around 10%; and physical activity, on average around 30% (Golden and Meyer 2004). Basal energy expenditure (BEE, also called basal metabolic rate, BMR) and resting energy expenditure are what I’ll be focusing on in this post. The two terms are often used synonymously, but the restrictions are stricter for BMR, which is typically measured lying down, after a night’s fasted sleep, in a temperature-, light-, and humidity-controlled environment, while for REE the value includes estimated daily requirements for light physical activity and digestion. Values are calculated by measuring oxygen consumption and carbon dioxide production (or, in earlier studies, or studies where participants’ diet is known and fixed, just oxygen consumption; see Speakman 2013 on methodological considerations around measurement), with a number of predictive equations used to estimate energy expenditure. Voluntary physical exercise aside, BMR constitutes the single greatest factor determining dietary requirements under semi-starvation conditions (Keys et al. 1950, p. 303)

It’s clear that, just as BMR/REE drop more sharply in semi-starvation than the loss of metabolically active tissue alone would predict, the increase in BMR/REE during refeeding is significantly greater than the increase you’d expect simply from the gain in fat-free mass. In the chronic semi-starvation of anorexia, REE may reduce to as little as 50-70% of predicted levels, but as soon as refeeding begins, energy requirements rapidly increase again. In one study, with 87 participants (Van Wymelbeke et al. 2004), 31% of the total REE increase over 2.5 months of refeeding (initially through tube feeding) happened during the first week, meaning that the metabolic rate of fat-free mass cells can increase within a few days. The rate of increase will be affected by obvious factors like energy intake and physical exercise, but also by things like smoking, mood, and anxiety (e.g. Van Wymelbeke et al. 2004). If energy intake drops again, for example when participants in clinical studies don’t stick to their refeeding diets, REE reduces again within a week; this happened 19 times, in 14 of the 87 participants, in Van Wymelbeke and colleagues’ study, and when participants’ dietary intake dropped below 1.3 x REE, i.e. below the level needed to maintain a stable weight (since REE represents only around 60% of total energy expenditure), REE fell to a level similar to that before refeeding began.

This study is unusual in including a follow-up phase some time after the end of the main study, so we can tell something about people’s outcomes beyond short-term weight gain. At one-year follow-up, 18 participants fulfilled the criteria for recovery (defined as ‘stable and normal BMI (>18.5), normal EI [energy intake] (>1.5× REE), disappearance of fear of eating and of becoming fat, and normal eating behavior at the 1-year visit, without relapse in the previous 2 mo.’), with a ratio of REE to fat-free mass that ‘did not differ significantly from that obtained in the healthy age-matched women’ (134 ± 16 kJ versus 131 ± 15 kJ per kg of fat-free mass per day). (Among those with a strikingly poor outcome after a year, the ratio of REE to fat-free mass was higher than for those who were recovered, perhaps due to factors like continuing elevated anxiety and exercise levels.)

Another earlier study found a similarly rapid increase in metabolic rate in the initial stages of weight restoration. Schebendach and colleagues (1997) measured fasting and post-prandial REE in 50 hospitalised patients with anorexia (with an average age of 16.3, and at an average of 71.6% of an ‘ideal bodyweight’ figure calculated from US medians). Within two weeks, fasting REE increased from 72% to 83.2% of predicted levels, increasing further to 90.1% and then 94.1% in weeks 4 and 6 respectively. However, participants’ bodyweight at this point is not reported, beyond saying that ‘rate of weight gain varied among patients’ (p. 114), with doubts raised about dietary compliance.

Metabolic rate seems not just to return relatively quickly to normal during refeeding, but to overshoot normal levels in a ‘hypermetabolic’ phase in which patients ‘easily lose weight, and need to eat an even larger amount of food to gain weight’ (Marzola et al. 2013; see also Mehler et al. 2010). This is especially true for restricting rather than binge-purge subtype (Weltzin et al. 1991), and for patients beginning treatment at lower weights (Walker et al. 1979). What may seem paradoxical at first glance makes perfect sense on reflection: after prolonged semi-starvation, the body becomes inefficient at dealing with large amounts of energy. It urgently needs to use the energy effectively to replenish fat reserves and repair tissues, but it often can’t do so very well in the early weeks and months of refeeding. Evolutionarily speaking, adapting only slowly to increased availability of food is probably a lesser evil than adapting too readily and finding the food has run out again.

Part of this elevated energy need may be due to conversion of energy into heat, especially at night (hence the frequent phenomenon of night sweats during recovery) (Marzola et al. 2013). Marzola and colleagues (citing Weltzin et al. 1991, above) note that energy needs tend to normalise over the course of 3 to 6 months, which means that ‘To obtain the best chance of long-term weight maintenance recovery, AN patients should persist with an increased caloric intake treatment plan’. Their recommendation is that beyond the initial phase of clinical stabilisation, outpatients will need around 500 kcal over the amount needed for maintenance, and that this amount will need to be increased periodically to carry on sustaining weight gain, with some individuals needing 4,000 or 5,000 kcal a day. Physical exercise can add a huge additional burden, capable of increasing energy requirements almost threefold – for example, from 4,000 to 12,000 kcal to regain 1 kg (Kaye et al. 1988; see also Zipfel et al. 2013).

In the longer term, it seems that all metabolic changes are fully reversible: Dellava and colleagues (2009) found no significant differences between a group of 16 women fully recovered (for an average of 6.4 years) from anorexia (full recovery defined as defined as having a BMI over 18.5, absence of binging and purging or another eating disorder, and exercise not exceeding US guidelines) and healthy controls. Measures were made of both body composition (e.g. percentage and location of bodyfat) and REE, and the only factor that made any difference to the latter was the amount of fat-free bodymass. The only difference was a higher rate of fat oxidation (breaking down fat molecules for fuel) in the recovered women, perhaps due to past over-exercising or to unexplored dietary differences.

So, we know that BMR or REE reduces in semi-starvation, we know that it increases again during refeeding, and we know that in fully recovered individuals it’s back more or less to normal, but we haven’t yet seen any detailed evidence about the nature of the metabolic changes that occur during later weight-restoration phases, or whether there’s anything to be learned about metabolic rate that might be relevant to getting into that ‘fully recovered’ bracket in the first place.

One of the main problems here is methodological. Many studies break off long before anything we could realistically think of as full recovery. A 1993 study by Krahn and colleagues, for instance, found a REE of 123% of expected normal levels in the final weight maintenance phase. But no figures are given for participants’ final BMI, with the third and final refeeding phase ending once a target weight is reached (within 10% of an ideal weight again calculated using US medians), and then followed by a drastic reduction in energy intake, from 3,600 kcal/day to as little as 1,800. Anyone who did this in the real world would be setting themselves up perfectly for relapse, which is, I imagine, exactly what happened to many of these participants. But nothing beyond discharge day is reported here, even for the 6 out of 10 patients who made it to the end of the study.

Even in the few studies that do cover ‘full recovery’ (Van Wymelbeke et al. 2004, Dellava et al. 2009), the qualifying BMI is usually strikingly low (in both these studies, the minimum was 18.5, though the actual means were 20.3 ± 1.6 and 21.9 ± 2.2 respectively). In Van Wymelbeke and colleagues’ study, as is often the case, details about how eating behaviours and attitudes were assessed are not provided, nor are exercise levels mentioned; while in Dellava and colleagues’, exercise was briefly assessed, but no assessment at all was made of attitudes towards food or body shape or weight – in either the recovered or the control group. It’s therefore hard to tell whether and to what extent disordered attitudes and behaviours might be exerting effects on physiology, activity, and intake in these women, which makes the metabolic results less meaningful than they could be.

In short, none of the clinical studies I’ve been able to find reports in enough detail, and also continues the intervention long enough, for us to be able to judge with confidence what the effects are of full weight restoration brought about by a dietary intake that goes beyond more or less arbitrarily imposed clinical limits. This really matters, because the rates of dropout (Fassino et al. 2009) and relapse (Steinhausen 2002) amongst participants in clinical studies on eating disorders are high, and one of the blindingly obvious reasons for this, which a baffling number of clinicians and researchers seem to neglect or choose to ignore, is that weight ‘restoration’ is halted at a much lower level than makes any sense. (This will perhaps be the subject of a future post.) What happens if we set aside our fear of overweight, and our target-driven efforts to declare as many people as possible discharged and dealt with? What happens if we allow recovery to proceed beyond clinical minimums and calorie-counted meals to something that more persuasively resembles full health?

Tune back in for my next post to find out!

You are reading

A Hunger Artist

Early Intervention as Panacea: Reality or Myth?

Questioning the evidence base for the efficacy of early intervention in anorexia

Christmas in Recovery

Tackling a few of the trickiest questions about anorexia recovery

Lose Your Phone, Find Your Body

Or: Turn on, tune in, drop out