Defining Recovery: What do researchers define as recovery?

Like I mentioned in my previous post, I’m going to be doing a series on my blog about definitions of recovery, what they are, and why they matter. For my first post, I’m going to start by discussing how researchers measure recovery.

Considering that most of my readers are (judging by the email I receive and the comments and other interactions that I have) sufferers and their loved ones, it seems odd that I would lead you all into a fairly esoteric discussion into the minutiae of what researchers would define as recovery. Some of it seems like splitting hairs: should a sufferer be without symptoms for two weeks? Two months? Two years? What would happened if you changed those to three years? Or one?

I know some of you are hard-core nerds like me and find this discussion interesting in and of itself. Most of you- those with outside interests that don’t, say, involve yarn and cats and lengthy discussions on disease outbreaks and eating disorder neurobiology- are probably more concerned with figuring out a way to manage, reduce, and/or eliminate ED symptoms than on technical definitions of recovery. To some extent, rightly so. Yet having a good grasp of what recovery is provides a crucial compass for guiding us towards more effective therapies.

Take body image issues, one of my favorite things to talk about with respect to EDs. My own personal body image (let’s face it) isn’t all that great. Would my recovery be stronger or more durable if I worked on these issues specifically with my therapist? We don’t really know. Most measures of body image look at whether you’re satisfied with your body. Sometimes I am, sometimes I’m not. Often I can’t tell you because I have no real idea what size I actually am. That being said, I generally don’t get hung up on it, and I’ve managed to separate out the pieces of what I look like from who I am. I don’t think about body image that much, aside from a very intellectual sense, which might be odd but is hardly diagnostically significant (except for Extreme Nerdiness- Not Otherwise Specified).

It would be nice if everyone would emerge from an eating disorder loving food and loving their bodies, but that’s not reality. The reality is that the people who develop eating disorders are a fairly heterogeneous group, and so are the people who recover. The goals of defining recovery in a scientific sense is figuring out what, if any, constants appear from people who do get better (and also those who don’t), and if any particular factors are crucial in getting better. If other things seem to be more optional, why?

With that said, let’s dive into the data. Snorkel masks optional.

Morgan-Russell Criteria

The first attempt to define a recovery from an eating disorder (at that point, anorexia nervosa was the only ED that actually existed in the DSM) was in 1975 by British psychologists Morgan and Russell, published in Psychological Medicine. The resulting criteria–creatively termed the Morgan-Russell criteria–looked at  14 different factors, from employment to psychosocial functioning to  body weight and menstrual status. The Morgan-Russell scale also provided  a general outcome measure based on weight and menstruation, which is  the most commonly used criterion in most research studies. Patients  with a good outcome have a body weight that is at least 85 percent of  ideal body weight (IBW) and menstruate regularly. Fair or intermediate  outcome consists of patients with 75 to 85 percent IBW or irregular  menstruation. Poor outcome means the patient’s body weight is less  than 75 percent of ideal. The data, consisting of a 4-year-long follow-up of 41 patients hospitalized for AN, created what many researchers call the “Rule of Thirds.” Roughly one-third of patients had a good outcome, one-third had a fair outcome, and one-third remained chronically ill.

The problem with the general outcome Morgan-Russell Criteria are several. For one, they don’t give you much data. They tell you only two things about a patient, which is their weight and menstrual status. For men, women who are post-menopausal, pre-pubertal, or on birth control, the business about menstruation is irrelevant. Secondly, 85% ideal body weight seems to be setting the bar fairly low in terms of what someone is expecting for recovery. That simply means “no longer meeting the DSM weight criteria for anorexia,” which doesn’t mean that you’re actually well and healthy. This is reflected when the researchers published a second study after following these individuals for twenty years (Ratnasuriya et al., 1991):

Even  among those whose outcome was generally good, a third were regularly  restricting their diet and a third reported eating irregularly. Among  them some reported, for example, that they avoided eating regular meals  unless pressured by their husbands, often restricting themselves to  irregular “snacks.” Others found that they were only able to maintain  their weight if they kept to a very rigid diet. Excessive preoccupations  with thoughts of food and weight also tended to persist, being reported  by over half of the patients who were otherwise doing well.

Which doesn’t sound like a “good” outcome to me.

Newer definitions

Researchers weren’t unaware of these problems, of course, but the fact is that without a lot of outcome data, there wasn’t a whole lot more to work with. Slowly, researchers began to transition more to measuring symptom-free periods as a proxy for recovery. Except how long does a person need to be abstinent from symptoms to indicate that they are in recovery? The researchers don’t agree. There is also a huge variation on the length of follow-up for the study. Studies have classified “recovery” as consisting of everything from being symptom free for several weeks (for example, Herzog et al., 1999) to several years (Von Holle et al., 2008)

The authors of the last study write that:

Recovery rates vary substantially, depending on the definition of recovery used and the length of observation. The present definition of recovery was more conservative than that of other published studies. The length of observation in previous longitudinal studies with survival analysis also varied widely, ranging from 1 to 15 years. Studies with shorter length of follow-up could not require longer symptom-free periods and might have reported higher recovery rates because they failed to capture the cyclicity and relapsing patterns of eating disorder symptoms.

The Von Holle et al. study seemed to find fairly low rates of full symptom absence for three years: roughly 10% of people with AN and 15% with BN were totally recovered for three years. This disorders are episodic, so maybe expecting 100% abstinence for such a long period of time is perhaps raising the bar too high. Blips in recovery appear to be the definition, not the exception. Still, the disorders showed slightly different patterns of recovery over time. In the first figure, you can see the diseases mapped on the same graph. The second figure shows how many people reported recovery during that particular time period, which gives you an idea of when most people recovery and the momentum of wellness.

Kaplan-Meier hazard analysis of recovery from AN and BN. It measures the overall proportion of patients that meet the criteria for recovery. Adapted from Von Holle et al., 2008.

Kaplan-Meier hazard analysis of recovery from AN and BN. It measures the overall proportion of patients that meet the criteria for recovery. Adapted from Von Holle et al., 2008.

The proportion of patients with AN or BN that say they have recovered at each time point. When you add up all of these patients, you get the final numbers on the survival ratio above. Adapted from Von Holle et al., 2008.

The proportion of patients with AN or BN that say they have recovered at each time point. When you add up all of these patients, you get the final numbers on the survival ratio above. Adapted from Von Holle et al., 2008.

Definitions of remission/recovery also help determine our definition of relapse (Olmsted, Kaplan, & Rockert, 2005). In this study, researchers are beginning to recognize that, like the illness itself, recovery exists on a spectrum. Figuring out what that spectrum is and where any significant markers are on that spectrum is crucial in understanding what happens to people with eating disorders.

Holistic Recovery

One of the most comprehensive definitions of ED recovery comes from a 2010 paper in Behavior Research and Therapy (Bardone-Cone et al., 2010). In this paper, the researchers lay out and validate three different aspects of ED recovery:

  • Physical recovery: No longer underweight, no other acute physical symptoms of an ED (low electrolytes from purging, etc).
  • Behavioral recovery: No food restriction, binge eating, purging, excessive exercise for three months.
  • Psychological recovery: Within the normal range of responses on the Eating Disorders Examination and other questionnaires.

A person was deemed to have made a full recovery when they met all three criteria and a partial recovery when they met the first two but not the third. The researchers acknowledged that the time period of behavioral recovery was on the shorter side, but they reasoned that when this was combined with signs of psychological recovery, it was a good indicator of true recovery.

As expected, the researchers found that people who made a full recovery (fulfilled all three of the criteria above) were indistinguishable from healthy controls on both measures of ED thoughts and behaviors and significantly improved on aspects of psychosocial functioning compared to the other ED groups. The active ED group looked like an active ED group from other research, and the partially recovered group fell somewhere in between on measures of ED-specific and general life problems.

Most of the assessments used to measure psychological recovery are tilted towards the sociocultural aspects of EDs. While I don’t have a problem with assessing these factors, I would have preferred to see an examination of other factors that drive EDs, such as regulating mood and anxiety. I think there’s the possibility of missing a significant subset of patients who are less-than-fully recovered and still needing help, but because they don’t say they feel fat or want to be thin, they appear normal and healthy.

That being said, I do like the way Bardone-Cone and colleagues break down the three different aspects of ED recovery. I think it’s important to recognize that stopping behaviors is just one aspect of recovery, and that normalizing your own ED-related thoughts is a crucial step to getting better. Like I said above, I would incorporate other measures of ED psychopathology than the researchers used but I like the concept.

So that’s your somewhat brief introduction to various measures of recovery. I’m sure there are others, but these are a reasonable example of the different criteria out there.

Edited to add: small sections of this post were adapted from Decoding Anorexia.

Twitter Digg Delicious Stumbleupon Technorati Facebook Email

4 Responses to “Defining Recovery: What do researchers define as recovery?”

  1. This is what my dissertation is on! So cool to see a post about it!


  1. Defining recovery: What do patients define as recovery? | ED Bites - March 14, 2013

    […] I posted about how researchers define recovery in a variety of clinical situations. I think we have made a lot of progress in our understanding of […]

  2. The treatment variables that affect long-term recovery | ED Bites - September 14, 2013

    […] there’s no standard definition of what constitutes recovery (I’ve blogged about this here, here, here, and here). Second, there’s no understanding of what should be targeted during […]

  3. Unpacking Recovery Part 2: The Multiple Facets of Recovery | Science of Eating Disorders - August 15, 2014

    […] at ED Bites wrote a few posts about recovery on her blog, and the first in the series can be found here. In this post, Carrie looks at the 3 dimensions of recovery that surface in Bardone-Cone’s […]