Service user and eating disorder therapist views on anorexia nervosa recovery criteria

Abstract

Objectives

Recovery from anorexia nervosa (AN) is difficult to define, and efforts to establish recovery criteria have led to several versions being proposed. Using the perspectives of people with histories of AN and therapists working in the field, we sought to explore the face validity of Khalsa et al (2017) as one of the most recent examples of proposed systematic recovery criteria.

Design

We interviewed 11 health service users (SUs) with histories of AN who had previously received treatment alongside 8 eating disorder therapists (EDTs), exploring their views on the proposed AN recovery criteria.

Methods

Data from verbal and written interviews were analysed thematically. Separate thematic analyses of SU and EDT interviews highlighted where concerns converged and diverged across participants.

Results

Both groups saw some merits of having universally recognized recovery criteria, and the multidimensional approach was welcomed, but EDTs were uncomfortable with considering their use in therapy and SUs felt key components were missing around emotional coping and life quality. SUs disliked the prominence of body mass index (BMI) in the criteria, and all struggled with the proposed duration for recovery. Conceptually, the notion of recovery as an endpoint rather than a journey was contested.

Conclusions

Our findings indicate disparities between academically derived recovery criteria and lived experiences and indicate perceived challenges in using such criteria in therapeutic settings. Including SUs and EDTs in the development of criteria may improve the likelihood of consolidating AN recovery criteria, but conceptual challenges remain.

Practitioner points

• AN recovery is complex, and the use of research‐based AN recovery criteria in therapeutic settings could have a detrimental effect on SUs’ outcomes.

• EDTs should be aware of efforts to define AN recovery criteria.

• EDTs should engage with debates on defining AN recovery and seek to promote participation in such debates to SUs.

Background

There are currently no set criteria for eating disorder (ED) recovery that are universally agreed upon and deemed acceptable by people with EDs, therapists, and researchers. Recognized recovery criteria would be practically useful for researchers and clinicians, as well as helpful for those with EDs trying to understand their own progress. Being able to define a point of recovery is important for researchers examining the effectiveness of treatment approaches for anorexia nervosa (AN), whilst clinically, creating a standardized recovery point would allow a person to have clear aims and support clinical practitioners to consider the process of recovery across patients. Definitions of recovery remain heavily focused on the absence of symptoms and physical aspects of EDs (Bardone‐Cone, Hunt, & Watson, 2018; Khalsa, Portnoff, McCurdy‐McKinnon, & Feusner, 2017), but studies with people with lived experiences of EDs highlight the critical role of psychological aspects of recovery, such as quality of life, and conceptualize recovery as a process rather than a state (Bohrer, Foye, & Jewell, 2020; Dawson, Rhodes, & Touyz, 2014; Kenny, Boyle, & Lewis, 2019; Stockford, Stenfert Kroese, Beesley, & Leung, 2019; Wade & Lock, 2019; Whitley & Drake, 2010).

Nonetheless, efforts to identify an agreed set of recovery criteria have continued (Wade & Lock, 2019). Despite the fact that previously proposed recovery criteria had demonstrable face validity, practitioners have not adopted them (Bardone‐Cone et al., 2018). Whilst De Vos et al., and’s (2017) systematic review revealed the importance of greater focus on psychological factors – such as self‐acceptance, positive relationships, and personal growth – their findings also highlighted two key barriers to reaching consensus on ED recovery criteria. First, recovery is frequently conceptualized in literature both as a state and as a process, making it challenging to define a set of criteria to capture a specific point at which recovery is achieved. Second, whilst eating disorders do overlap to some degree in aetiologies and symptomatologies, they each have different diagnostic criteria. Thus, there is a logic to developing and testing recovery criteria specific to each diagnostic category (e.g., AN) rather than trying to derive a set of recovery criteria that apply across EDs (Kordy et al., 2002). With regard to that second point in particular, we have focused in this project on exploring the validity of criteria for people diagnosed with AN.

Studies with individuals who have experiences of AN characterize recovery as a highly individualized experience that is only partly explained by the restoration of physical health and the cessation of AN behaviours (Kenny et al., 2019; Stockford et al., 2019; Whitley & Drake, 2010). Individuals with AN have conceptualized recovery as a process involving identity change, self‐acceptance, motivation to change, and reconnecting with social support (Bowlby, Anderson, Hall, & Willingham, 2015; Conti, 2018; Duncan, Sebar, & Lee, 2015; Dawson, Rhodes, & Touyz, 2014b; Hay & Cho, 2013; Jenkins & Ogden, 2012; Stockford et al., 2019; Lewke‐Bandara, Thapliya, Conti, & Hay, 2020; Piot et al., 2019; Romano & Ebener, 2019; Stockford et al., 2019). However, this personalized conceptualization of recovery poses a challenge for researchers who wish to define AN recovery in a universally applicable way.

In 2017, Khalsa et al reviewed 27 studies addressing criteria for AN recovery, remission, and relapse, with the intention of developing universally applicable criteria. The authors suggested using BMI, AN symptoms, Eating Disorder Examination (Cooper & Fairburn, 1987) scores, presence of AN behaviours, and duration to identify full and partial states of recovery, remission, and relapse (see Table 1). We were interested in reviewing these recovery criteria firstly because their relative simplicity compared to other suggested criteria makes it more likely that they could be utilized by researchers and used in clinical settings. Secondly, as Khalsa et al., and’s (2017) structured review of previous criteria was published open access in an influential journal in the field, and has demonstrated academic traction in the relatively short time since publication (with nearly 100 citations at the time of writing), these criteria have a good chance of continued impact. If so, the criteria must have validity within the world of health care practice, as they may be applied into therapeutic practice and could therefore have a direct impact upon the process of recovery for service users (SUs) (Romano & Ebener, 2019) and the practice of eating disorder therapists (EDTs). Moreover, evidence suggests that if SUs believe they are at a particular stage of recovery whilst a set of criterion being used in their treatment indicates otherwise, this could cause distress and may inhibit self‐efficacy (Bardone‐Cone, 2012; Callard, 2012; Dawson et al., 2014b; Slof Op’t Landt, Dingemans, de la Torre Y Rivas, & van Furth, 2019). Therefore, the views of these individuals on the recovery criteria proposed by Khalsa et al are important and informative.

Table 1.

Proposed definitions for recovery, remission, and relapse from Khalsa et al., (2017)

Illness status	BMI criteria	Symptoms	Behaviours	Scales	Duration
Full recovery	BMI ≥ 20 or ≥ 90% ideal body weight	No significant fear of gaining weight or disturbance in body image	No restricting, bingeing, or purging	EDE within 1 SD of normal	12 months
Partial recovery	BMI ≥ 19 or ≥ 85% ideal body weight	No significant fear of gaining weight or disturbance in body image	No restricting, bingeing, or purging	EDE within 1.5 SD of normal	6 months
Full remission	BMI ≥ 19 or ≥ 85% ideal body weight	Fear of gaining weight or disturbance in body image present	No restricting, bingeing, or purging	EDE within 2 SD of normal	3 months
Partial remission	BMI of ≥ 18.5 or ≥ 85% ideal body weight	Fear of gaining weight or disturbance in body image present	No restricting, bingeing, or purging	EDE within 2 SD of normal	1 month
Partial relapse	BMI ≤ 18.5 or ≤ 85% ideal body weight	Significant fear of gaining weight or disturbance in body image	Restricting, bingeing, or purging present	EDE ≥ 2 SD of normal	1 month
Full relapse	BMI ≤ 18.5 or ≤ 85% ideal body weight	Significant fear of gaining weight or disturbance in body image	Significant restricting, bingeing or purging	EDE ≥ 2 SD of normal	3 months

EDE = Eating Disorder Examination and BMI = body mass index.

The aim of the current study is to examine the face validity of Khalsa et al., and’s (2017) recovery criteria with SUs and EDTs. To contextualize this study, we aimed to explore how SUs and EDTs conceptualize recovery themselves, before examining how they respond to and perceive Khalsa et al., and’s (2017) AN recovery criteria in the contexts of their own experiences, their thoughts on the potential use of these criteria in therapeutic settings, and the ways in which the views of these two groups converge and diverge.

Method

Materials and methods

The semi‐structured interview schedule was developed to investigate beliefs about concepts of recovery, remission, and relapse through open questions, and to investigate views on Khalsa et al., and’s (2017) criteria (Appendix S1). A structured written interview schedule was adapted from this for participants responding in writing (see Appendix S2). Interviews began with participants exploring their own conceptualizations of recovery before discussing their views on the proposed criteria.

For semi‐structured interviews, participants were able to choose whether to participate in an interview in‐person, via Skype, or through a written interview email. In‐person interviews were conducted at the researchers’ institution or at participants’ homes and lasted between 23 and 62 min. With participant consent, verbal interviews were digitally recorded and were transcribed verbatim. Participant names were replaced with pseudonyms.

For written interviews, participants received an email explaining the process. Once they gave written consent, they were sent the structured written interview schedule. This document stated all information which would be given at the beginning of a semi‐structured interview, and to contact the researchers for further information, along with support options if the participant felt distressed (no queries were received in this regard, other than to clarify how long they could take to complete the interview). Participants replied in their own time, having been informed that if the written interview had not been received within 2 weeks, a reminder email would be sent. This process required a single response in which all questions were covered by the participants, rather than any further interactions or clarifications by the research team. The quality of data received by this format was sufficient to be included within the analytic framework used, although lower word counts than semi‐structured interview transcripts were typical.

Author 1 and Author 2 were primary analysts. Author 1 has personal experience of eating disorders including AN. Author 1 finds it challenging to conceive of recovery as a state for herself but does not oppose the idea of recovery being a state experienced by others.

Ontologically, our position was one of relativism and our epistemological approach was based on social constructionism (Burr, 2015) and interpretivism. Our aims are largely deductive in nature and, as the study aim is to produce practically applicable findings, an overarching pragmatism has been applied. However, our work was not entirely deductive, and our analytic approach, influenced by Fletcher (2017), allowed inductive insights to be coded and incorporated. We have drawn on Hammarberg, Kirkman, and De Lacey (2016), using their guidance to critically challenge our credibility, reliability, and trustworthiness throughout the analytic process.

Participants and data collection

Recruitment and data collection occurred between April and July 2018. Ethical approval was granted by the authors’ institution. SUs were recruited purposively through adverts in public spaces and online. EDTs were recruited through ED service managers who agreed to distribute participant information within their services. All participants received an information sheet and a consent form to complete and were fully informed about the purpose of the study and how their data would be used.

11 SUs (two men and nine women, age range 18‐55, all British, Irish, or Canadian, with self‐reported experience of AN) participated. Four interviews were in‐person, three via Skype and four written. 8 EDTs (two men and six women) from two UK‐based ED services (one NHS adult ED service and one charitable adult ED service) were interviewed in‐person. All participants were sent a copy of the Khalsa et al., (2017) paper prior to interview.

Analytic procedure

Transcripts were thematically analysed based on the process described by Braun and Clarke (2006), with additional guidance from Swain (2018) for developing inductive codes alongside deductive codes. This allowed for an inductive exploration of SUs’ and EDTs’ experiences, with a deductive exploration of viewpoints regarding Khalsa et al., and’s (2017) recovery criteria.

Author 2 was the primary coder for SU data and Author 1 for EDT data. The primary coder of each data set read each transcript multiple times, identifying key information, developing codes reflecting personal experiences, and identifying all information relating to recovery as defined in the Khalsa et al., (2017) criteria. Given our aims and partly deductive approach, we determined in advance two overarching themes (individual views on recovery, and individual views on the Khalsa et al., (2017) definition), into which coded data were allocated before an inductive approach allowed us to establish the subthemes which characterized their responses to these themes. Authors 1 and 2 independently coded transcripts from each other’s data set to check for reliability and validity of codes. Authors 3 and 4 independently provided validity and reliability checks by coding four transcripts, with reflections discussed following coding. Following this coding process, we reviewed our codes and built these into the subthemes presented below that capture the essential perceptions and messages conveyed in the interviews. Our final themes and subthemes are given in Table 2.

Table 2.

Table of themes and subthemes

Theme	Subtheme
(1) What is recovery	(1.1) Recovery as a journey, not just an end‐state
	(1.2) Experience‐informed recovery definitions
(2) Views on Khalsa et al., and’s (2017) definition	(2.1) BMI as problematic
	(2.2) Duration
	(2.3) Pros and cons of a multidimensional approach
(3) The use of Khalsa et al., and’s (2017) criteria in therapeutic settings

Results

For ease of reading, our overarching findings are presented here with minimal exemplar quotes from participants; more comprehensive data can be seen in Appendix S3.

1) What is recovery?

Given that we were expecting participants to give their comments on Khalsa et al.,’ (2017) definitions of recovery, it was appropriate to start from a point of understanding their own existing definitions. This enabled them to have an anchor point in their existing perspectives when considering Khalsa et al, and allows us to analytically contrast these perspectives.

1.1) Recovery as a journey, not just an end‐state

All participants recognized recovery as an important term but ranged in their degree of belief in a recovered state. Some SUs reported being comfortable with the term recovery because they considered themselves recovered or had experienced a sustained period of recovery, but most were more tentative, finding the idea of recovery as an ‘end‐state’ difficult to imagine in light of their experiences:

I don't believe there is such a thing as complete recovery, I think it's like an addiction, it’s with you for life (Alex, SU)

Where they accepted the concept of a recovery end‐state, the possibility of recovery was important to several SUs in terms of their motivation to take care of themselves:

Recovery being possible is vital to me every day because it stops me day‐dreaming about the potential positives of going back to anorexia. (Jodie, SU)

Similarly, most of the EDTs in the sample framed recovery in some way as being a journey rather than a destination, with some such as Chris also believing in a recovered state:

I see it as a direction of travel… I would judge someone as being recovered if that person was feeling confident that they had got to a place where they felt that they could maintain their gains (Chris, EDT)

Most EDTs did not believe that individuals ever achieved a fixed state of recovery, but that this was a fluid concept representing a state of living and a process of learning.

1.2) Experience‐informed recovery definitions

All SUs had developed their own personal criteria for AN recovery, which usually featured managing AN thoughts and behaviours sufficiently to engage in life.

I think it’s being more flexible as well and letting other things in your life take over, but I don’t think it’s necessarily it disappearing altogether (Leslie, SU)

In keeping with the views of SUs, several of the EDTs reported exploring with their clients what the term “recovery” meant to them, rather than defining it for them.

I was saying earlier that recovery I don’t think you can define yourself. I think it comes from the person you’re seeing. That’s why I’d always ask them… (Carol, EDT)

However, most EDTs directly or indirectly talked of how their own views influenced their understanding of recovery. As seen above, Carol said that she did not have a definition of recovery, but she did provide a description of recovery during the interview:

recovery is somebody being able to have a healthier relationship with food so that it doesn’t completely preoccupy their thoughts all of the time, which equates to having a healthy BMI as well. (Carol, EDT)

What Carol’s quotes capture is something that was voiced by all the EDTs in the sample to some degree; namely, that ideas about recovery are held by EDTs but these tend to be individual and are driven by how each EDT ‘sees’ things, developed from their own clinical experiences. Each EDT viewed their ideas about recovery as being a version rather than a ‘true’ definition of recovery and of being less meaningful than the version of recovery held by their client. For these EDTs, the question of who ‘owns’ the definition of recovery is one that by its nature contains questions of power balances in therapy.

2) Views on Khalsa et al., and’s (2017) definition

Although presented here jointly, it is worth noting that SUs tended to focus on each individual criterion in turn, whereas EDTs spoke more about their general impression of Khalsa et al., and’s (2017) complete criteria set.

2.1) BMI as problematic

Most SUs commented first on the BMI criterion. There were concerns both about the actual BMI value used and about how BMI could give a misleading impression of recovery when other symptoms were still present.

I just don’t think that should be there, necessarily, or should be an optional thing. I think you could have a significant fear of gaining weight or disturbance in body image, you could be restricting, binging or purging and still be in a relapse, even if your BMI wasn’t under 18.5. (Jamie, SU)

Jamie speaks clearly here about her discomfort with BMI being a necessary criterion. Jamie’s comment indicates confusion as to how the complete set of recovery criteria would be applied, such as how fears and beliefs about one’s body would be considered alongside BMI. SUs commonly voiced their view that BMI was problematically overvalued as a marker of AN:

Giving equal importance to the BMI in this table reinforces the eating disorder belief that a person’s weight does indicate how well they are. (Jodie, SU)

Both comments above indicate a sense that BMI should be given a lesser role in defining recovery and that failing to do so perpetuates a weight‐based concept of AN that runs counter to SUs’ experiences.

Where BMI was discussed by EDTs, it was also felt that BMI could not be universally applied as every individual has their own ‘normal weight’:

my belief is that we all have slightly different bodies, and slightly normal weights, so I think some people can be recovered at a BMI of 19. Some people can’t be recovered at BMI 19. (Chris, EDT)

2.2) Duration

SUs struggled with the idea of a fixed timeframe for recovery and felt that a 12‐month period was too short:

I don’t think it can be that clear‐cut… it can take ages for people to recover, it depends a lot on comorbidities and things. (Leslie, SU)

Leslie’s language suggests discomfort with an oversimplification of AN recovery, common to responses to many of the criteria. SUs felt conflicted about stipulating any timeframe, given their belief in the individual nature of recovery and their concerns about the power to do harm to someone’s recovery by having fixed durations:

I think recovery definitely takes at least 12 months. I wouldn’t want to say to someone who considered themselves in full recovery after only a year, ‘You can’t be fully recovered’. (Louise, SU)

As with SUs, EDTs struggled with the idea of having a set duration for recovery. Chris, who believed in the notion of service users reaching a point of being recovered, felt the 12‐month timeframe proposed was too short.

…if someone has made consistent changes which have been sustained for 18 months, 2 years, I think that’s meaningful. At 12 months I think there’s still room for concern. (Chris)

Whilst Carol was broadly in agreement with the proposed recovery criteria, she also struggled with the 12‐month timeframe. In contrast to Chris, however, this struggle appeared to be related to the concept of a service user being ‘fully recovered’, which was an idea Carol did not fully ascribe to:

I would agree with the full recovery, but again I guess, the duration; is 12 months of doing that full recovery or is that remission? It’s so hard to say. Full recovery means that they’re fully recovered and they’ll never go back, but in eating disorders that’s very rare for somebody to never, ever have any eating disorder thoughts ever, ever again. (Carol)

The stipulation of a set duration of recovery was not only seen as conceptually difficult, but was felt by some of the EDTs to create a pressure upon service users:

people sometimes can really get caught in trouble if they are judging themselves that they haven’t recovered, or if others are judging them or criticising them. (Will)

2.3) Pros and cons of a multidimensional approach

Whilst SUs largely expressed negative views about the recovery criteria proposed, the multidimensional approach of Khalsa et al., and’s (2017) criteria was welcomed by some. Sam, who was the most positive SU about having set criteria, captures a sense of necessity about the proposed framework despite the challenges to having set criteria for recovery:

The attempt to standardize definitions of recovery, relapse, and remission for ED is essential, especially within the Khalsa framework which utilizes BMI, observable behaviours, subjective measures, standardized ratings, and specific durations of follow‐up… I would caution on whether they can be universally applied to AN patients. I think factors such as comorbidity, gender, and the duration of the illness prior to diagnosis are also important factors, especially when it comes to the proposed time periods. (Sam, SU)

For some SUs, having multiple components was seen as a counterbalance to some perceptions of AN as a dieting‐based disorder. However, the presentation of the multidimensional approach raised concerns. SUs did not know whether all Khalsa et al., and’s (2017) criteria needed to be met in order for an individual to be considered in full recovery. They felt that meeting all criteria for 12 months was highly unlikely, with several reflecting that their own symptoms would leave them between recovery and other categories:

I’d have been all over the matrix. I don’t know how scoring would have gone.… I think it would be really remiss of clinicians to expect their clients to fit in a line of boxes. (Louise, SU)

Furthermore, many SUs felt that essential components of AN recovery were missing. SUs felt that coping skills were missing, be that dealing with emotions or dealing with not being in control of situations. Below, Jamie suggests that recovery might be defined by the coping mechanisms that replace the AN behaviours:

I think a big part of recovery for me is – and was – about learning to cope with negative emotions in ways other than turning to food. (Jamie, SU)

Several commented on the importance of wellbeing and a more holistic view, and the following quote from Jenny reflects the notion of gaining a life outside of AN being a marker of recovery.

I still feel there’s something about missing the person… for me, I wanted recovery, is having that release from a complete focus on food, because… you can’t have any kind of life when that’s what your head space is (Jenny, SU)

There was a concern about seeing AN recovery in isolation and not attending to comorbid conditions.

I don’t know how you would bring in comorbid things because it’s often comorbid with a lot of other influences (Leslie, SU)

EDTs shared the sense that the recovery criteria do not tell the full story of AN recovery, as voiced here by Val:

An eating disorder is about much more than this, I suppose, is what I’m getting at. This is one aspect of it and it’s important but it’s not the entirety of it. (Val, EDT)

Most of the EDTs’ comments relating to perceived missing elements within the AN recovery criteria were indirect or were in broad reference to individual recovery being more complicated than the picture presented by the table. In this quote, Joanna speaks of what she sees missing when thinking about the individuals she works with:

The thing that’s coming up for me is that a lot of eating disorders are based in self‐esteem and self‐worth… My fear is that it’s a guide, but people are individuals. (Joanna, EDT)

3) The use of Khalsa et al., and’s (2017) criteria in therapeutic settings

Potential use of these criteria in therapeutic settings was a source of real concern in both groups. Amongst SUs, there was widespread concern that having recovery criteria could lead EDTs to see SUs as needing to meet a certain set of goals, leading to the potential use of the criteria as a ‘ticking the boxes’ method of deciding if SUs were recovered, or could lead to difficulties when SUs did not fit the suggested profile.

it takes a bit of professional discretion being used by a clinician to guide… I think it would be really remiss of clinicians to expect their clients to fit in a line of boxes. (Jenny, SU)

SUs were concerned that having recovery criteria might create negative judgements of themselves, in terms of either their progress relative to themselves or their progress relative to others. According to Poppy, the effects of this might undermine an individual’s work towards recovery:

I would worry that having a chart/definitions such as this available would only incur thoughts like ‘I’m still only in partial remission, I’ve failed in recovery. (Poppy, SU)

There was a strong sense from SUs of their vulnerability when accessing services, and their fear of being judged by EDTs comparing them to the recovery criteria. The potential for EDTs to exacerbate AN symptoms by using recovery criteria was linked by some SUs to perfectionist thinking patterns and to a tendency to see oneself as having ‘failed at recovery’. Nonetheless, most SUs, having often detected anxieties held by EDTs they had encountered, felt the guidelines would help EDTs.

It would help the clinicians, because I think…it’s just not necessarily about the weight, and if they are getting terrified because somebody has gone down to a really low BMI on their watch…there’s something about compassion and fear, you know, it’s hard to be compassionate in a fearful situation. (Jenny, SU)

As with SUs, all the EDTs saw the table as unreflective of their clinical experiences:

On paper, these things look good in the sense that I would agree that that’s what I would define. In terms of an individual, I don’t think you can necessarily say that’s what it should look like, because for that person, that might be as good as it can get. (Carol, EDT)

Given this universally held view, it was interesting to explore where and how the criteria might be seen to be useful. There were comments here about the benefits of the criteria both within and outside of therapy settings. For example, in therapy settings, the criteria were thought by some of the therapists to be potentially helpful to certain individuals who conceptualized their AN symptoms and AN experiences in a similar way to Khalsa et al., (2017). However, all the therapists in the sample talked about the limitations of the criteria and the need to use then alongside careful clinical judgement, given the sense of judgement that might come up for SUs.

EDTs were more concerned that recovery cannot be defined in a universal way and, in relation to Theme 1, recovery needs to be considered with each individual recovering from AN on their own terms.

When we start labelling and categorising, are we building on people with that failing, because they don’t fit in a box? ‘I went to my GP and my BMI wasn’t low enough, so I’m not even good at this’. That’s coming up from it. It doesn’t sit quite right. (Joanna, EDT)

Like Joanna, most EDTs felt uneasy and unsure about the implications of using a fixed set of criteria for recovery. Val’s quote below illustrates how she felt she would struggle with using Khalsa et al’s recovery definition in a therapeutic setting.

I don’t think these tables can be used on their own… If someone said to me, right, this is what you’ve got to use, and this is recovery, I would say, I’m not using it, it’s not helpful. (Val, EDT)

What comes through in these comments is a sense of fear of the criteria doing harm, which contrasts with Jenny’s view as an SU that the criteria would give EDTs reassurance.

Discussion

Our analysis indicated the largely shared concerns of SUs and EDTs about Khalsa et al., and’s (2017) AN recovery criteria, which were viewed as problematic by both groups on categorical (can recovery be captured by such criteria?), therapeutic (will AN recovery criteria make AN symptoms worse?), and conceptual (does a state of AN recovery exist?) grounds. These concerns can be understood in the context of the participants’ existing views on AN recovery, as presented above.

When directly considering Khalsas et al.,’s (2017) criteria, SUs were consistently negative about using BMI as a recovery criterion, which appeared related to perpetuating the perception of AN as a weight‐determined illness and to neglecting to attend to emotional, cognitive, and social aspects of AN. Although de Vos et al., and’s (2017) review showed individuals who had recovered from a range of EDs favoured emotional and holistic recovery factors over physiological ones, our finding demonstrates a stronger position, in that some SUs favoured the removal of physiological criteria from AN recovery definitions. EDTs had concerns about the use of one universal BMI for recovery but did not go so far as to favour removing BMI entirely from recovery criteria. In either case, the broader indication that both SUs and EDTs strongly valued the inclusion of facets of social functioning when considering recovery from AN poses a significant challenge to Khalsa et al’s criteria, which objectively fails to consider any aspect of social functioning. Whilst we recognize that the diagnostic criteria for AN do not specify an impairment in social functioning, our participants’ accounts demonstrate the importance of considering the direct lived experience of those who will be using such criteria in practice to ensure that they address aspects that are meaningful in recovery experiences. Failure to consider such holistic aspects of SUs’ lives risks rendering recovery criteria to a technocratic and meaningless exercise.

EDTs and SUs shared concerns about the recovery duration, which was seen as too short and too rigid. As AN recovery is often reported to be a fluctuating journey rather than a stable state (Bowlby et al., 2015), the concerns about the rigidity of the duration are perhaps unsurprising. Furthermore, all participants were unclear if all criteria had to be met for 12 months, and this is not made clear in Khalsa et al., and’s (2017) review. Our findings unveiled many uncertainties about interpreting the criteria, indicating the potential confusion and distress that could be caused if clear guidelines are not provided for use.

Participants had mixed feelings about the components included in Khalsa et al., and’s (2017) criteria. Both EDTs and SUs in our study saw AN recovery as being a unique personal process requiring more than gaining weight and displaying certain behaviours. The sense of recovery being holistic chimes with numerous other studies on recovery (e.g., Dawson et al., 2014b; Duncan et al., 2015), which have demonstrated that the time required for this process was unique for everyone, tying to concerns raised here about fixed durations for recovery.

Therapeutically, there was a concern that having recovery‐state criteria would create competition or feelings of failure, compounding key cognitive difficulties associated with AN. This was one of a number of reasons that participants ranged from cautious to being fearful of the use of these criteria in therapeutic settings.

EDTs were uncomfortable with the idea of imposing a definition of recovery on SUs, as they believed that this was at odds with the principles/goals of therapeutic work. EDTs’ views were in keeping with recent work by Wetzler et al. (2020) in emphasizing the person‐centred nature of recovery. What is interesting to note is that despite SUs agreeing with EDTs that recovery is individually defined, there was not a wholescale rejection of the use of recovery criteria in therapy by SUs and some felt that Khalsa et al., and’s (2017) criteria might be reassuring for EDTs. As with EDTs, SUs believed that criteria would need to be used sensitively and with clinical judgement. Once recovery criteria are made public, their use cannot be controlled; concerns about their use in shaping therapy access or practice were prominent and caution is urged in how these or other AN recovery criteria are portrayed. Clear guidelines about their use and limitations may help avoid misuse.

Conceptually, SUs and EDTs largely believed recovery to be a process rather than an end‐state, in line with many other existing studies (e.g., Jenkins & Ogden, 2012). This suggests that, even if categorical issues can be resolved, SUs and EDTs may find it difficult to get past the core concept (that there can be a defined point of AN recovery) upon which AN recovery criteria sets are based. We note that Khalsa et al., (2017) have limited discussion on how their criteria were developed. It may be that a wider discussion about the acceptability of such criteria needs to be facilitated before further recovery criteria are proposed.

Limitations

Our study has focused on one set of proposed criteria, which may have limited the range of points discussed; it would have been interesting to present alternative models to enable EDTs and SUs to consider their relative strengths and weaknesses. Our sample size was adequate for the design and nature of the study, but we cannot know how representative the views of the two groups are of other SUs and EDTs. We recruited EDTs from two services, and a wider diversity of organizational influences may have led to a broader assortment of observations. The multiple data collection methods used aided recruitment and offered choice to our participants. However, written interviews had less flexibility to explore participants’ thoughts and experiences, and therefore yielded slightly less wordy responses than in‐person or Skype interviews. We also made the decision not to collect personal data such as diagnostic history in SUs or therapeutic background in EDTs, partially out of concern for EFFECTING recruitment numbers; however, we are aware that this means that part of the context for these participants’ views is missing, and this might have influenced elements of their responses. As with all qualitative study, there is a strength in the richness of perspective able to be brought by participants, but there will always be critique for the ways in which researcher individuality can affect results; whilst our team approach has sought to address this element to a degree, the future work in this field will benefit from mixed methods that allow for both richness and objective countability.

Conclusion

Through SUs and EDTs reviewing Khalsa et al., and’s (2017) criteria for AN recovery in the light of their own experiences of AN, we have identified several areas of concern pertinent to all researchers developing AN recovery criteria. We must consider whether trying to define an AN recovery state is a valid goal, even when its practical utility for research is evident. Our findings suggest that the balance of opinion in SUs and EDTs may be weighed against attempts to define AN recovery in this way. However, if the goal of producing AN recovery criteria remains, our work leads us to argue for the value of a ‘co‐produced’ definition of AN recovery that is potentially acceptable to those in recovery and EDTs. Our results suggest that SUs and EDTs can aid discussions on how criteria may be perceived, the reactions they may cause and the potential benefits and concerns about using such criteria in therapeutic settings.

Conflicts of interest

All authors declare no conflict of interest.

Supporting information

Appendix S1 Semi‐structured interview schedule.

Appendix S2 Written interview schedule.

Appendix S3 Data supporting analysis reported in main paper.

Data Availability Statement

Data have been retained at NTU, but service users did not consent to open access and so requests for data access should be addressed to the corresponding author.