What’s Working? Comparing Partial Hospitalization and Residential Treatment for Eating Disorders

I can’t help but think I often write more about the issues surrounding the general lack of treatment options for eating disorders that I sometimes neglect to comment on what is available. A part of this is that I would refer to myself as somewhat of a treatment modality atheist – I have the luxury of being someone who does eating disorder research but is not involved in directly treating those with eating disorders, and so I don’t need to specialize in one type of treatment. My bottom line tends to be that no one-size-fits-all, and that the type of treatment that works for someone will depend on so many factors (like their gender, ethnicity, socioeconomic status, body size, ability, even their politics to a certain extent) that I wouldn’t want to proclaim one type of treatment as king.

In spite of this treatment modality atheism (or perhaps because of it), I do follow the outcomes literature. As you might expect, I’m interested in what outcomes are beyond weight restoration – that “after” period that some refer to as recovery. And, when I’m writing policy documents or speaking to people about their options, I like to try to give a sense of what the literature says in terms of outcomes. So, I was excited to see a narrative review in the European Eating Disorders Review this summer in which the authors explore outcome studies for residential and partial hospitalization programs – two common settings in which eating disorders are treated.

The Study

Friedman and colleagues (2016) identified 29 studies of treatment outcomes for adults treated for eating disorders (primarily anorexia nervosa and bulimia nervosa) in day hospital/partial hospitalization or residential programs. 22 of these were about outcomes from PHP programs, and 7 were about outcomes of residential programs.

Partial hospitalization (PHP) or day hospital programs are defined as those in which patients spend 6-10 hours, 3-7 days per week, in a setting where they receive a combination of meal support and group/individual therapies. Residential programs, on the other hand, are non-hospital based programs that include a variety of therapies and meal support 7 days per week, including overnight.

Part of the issue in defining and measuring outcomes of these kinds of treatment is the wide variability in programs, even within the same country. While the average length of stay in residential programs is 83 days (Frisch, Franko & Herzog, 2006), things like severity of illness, personal motivation, insurance, and other constraints can impact the length of stay in a significant way. The differences between PHPs are equally challenging to navigate: you can imagine that a program offering 10 hours of support per day, 7 days a week, might differ significantly from one offering 6 hours of support, 3 days per week. Part of the discrepancy is due to a desire to match the level of care to the patient’s level of need, but sometimes it is more of a functional “this is all we have funding for” scenario.

The treatment approach in these settings are generally described as “multidisciplinary,” meaning that they include teams with specialities ranging from psychiatry to occupational therapy to social work and so on, and multi-modal, meaning they offer several different types of therapy (like cognitive behavioural therapy, dialectical behavioural therapy, etc.) offered in both individual and group settings. Again, this makes it challenging to compare across treatments – it can be very hard to know what is making the difference.

To a certain extent, you could make the argument that it doesn’t matter which part is working, just that it’s working. Then again, we might even question what “working” means, in this context. A third wrinkle to add to the complexity of making comparisons between programs is that we still (yes, still!) lack a solid agreement on what constitutes “recovery” from eating disorders. So, it can be hard to gauge which programs are most helpful in getting people there.

Results

That said, this study offers us some insight into the kinds of results that we might typically see from PHP and residential programs. They also tells us a bit more about those variants in program design that I mentioned make it hard to directly compare studies.

Of the 22 studies about PHPs, all shared the following goals:

• Normalize weight
• Normalize eating behaviour
• Resolve ED behaviour

These are very “black box” goals, in my mind – we have to consider that people may hold different ideas of normal eating in society in general and within particular cultures; this is something I rarely see unpack in these studies, and so I like to always flag this. Similarly, if we start to blow up the notion of there being a universal “normal” weight range, that becomes tricky as well. For the purposes of not ranting on for ages, however, let’s assume that they mean allowing people’s bodies to flourish in a way that is appropriate for them.

The authors note the eclectic and varied approaches to treatment amongst PHPs. All of the programs had group therapy as a key component of their approach. Several offered group meals, as well as groups for skill development. Additionally:

• 14 offered cognitive behaviour therapy (CBT) and/or dialectical behaviour therapy (DBT)
• 3 offered primarily psychodynamic therapy
• 12 offered dietitian consultations and meal planning
• 4 included family-based treatment (FBT), and 8 others included consultations with families

There was no consensus around which intensity of PHP was most effective; some suggest that programs of at least 5 days in duration are more effective at reducing symptoms like binge/purge episodes (Olmstead et al., 2013). Similarly, the wide variation in length of stay precluded a solid answer about how long patients might stay to receive the most benefit; duration ranged from 22 days to 182 days.

For residential programs, all of the programs reviewed included 24-hour care but non-medical monitoring; the main treatment modality offered in these settings is CBT and/or DBT. Again, the programs offered several different types of treatment, including:

• Individual therapy
• Family therapy
• Group therapy
• Nutrition counselling
• Nursing support
• Medication management

Treatment duration for residential treatment ranged from 28.5 days to 105 days.

The authors grouped studies into several types: open studies (non-randomized) reporting outcomes immediately at discharge, open studies reporting outcomes at an interval after discharge, and randomized-controlled trials. Most open trials that reported outcomes at discharge (for both PHPs and residential) reported that patients had improved in weight and/or symptoms; then again, as the authors note, this makes sense – generally, these are criteria for discharge. There is mounting evidence that there is little link between discharge outcome and more long-term outcomes; many people relapse following discharge (Lock et al., 2013). Even the types of treatment we tend to refer to as “evidence-based” face this issue – recently, LeGrange et al. (2014) reported that 2/3 of patients who were weight restored at discharge from FBT remain so four years later.

In longer-term follow-up studies, relatively few patients were followed into their post-discharge outcomes: an average of 66% for PHPs and 37% for residential. This means that authors lack data about outcomes for a significant number of people – and likely those doing less well. So, we need to take the favorable outcomes of these studies with a grain of salt; those doing well are more likely to want to continue to take part in follow up research, as the authors indicate.

There were 2 randomized controlled trials included in the bunch that did demonstrate comparable outcomes for PHPs versus earlier RCTs on inpatient treatment. In both, participants were largely weight-restored at 1 year follow up.

Outcomes and Recovery

While the authors note the importance of looking beyond weight restoration and symptom remission, they are limited in how much they can say about deeper recovery by the data in the studies they reviewed. As ever, studies don’t assess recovery in any kind of consistent way or comment on the nuances of recovery – like how people feel in their social worlds.

I understand the need to focus primarily on the measurable – we want evidence for what is working, so we go with things that are more tangible. There is so much that is intangible, however, about recovery. I’ve said it many times before but it bears repeating: outcomes are not the same as recovery. You could make the argument that they are equally important, but they are not the same.

As always, the studies included in this review, and the review overall, does not point to one particularly promising intervention for eating disorders. As the authors note, it is hard to secure funding for studies of treatment outcomes – and even when they happen, there is the possibility for corporate interests to sneak their way in to the results. I would argue it is also hard to get money to fund studies that look at recovery beyond the checks and balances required to discharge patients; those providing funding simply don’t tend to be as interested in the “after” or the more nuanced aspects of recovery, which is a shame, in my opinion!

We also need to better understand how diverse people respond to treatment, of course. This study didn’t note the demographic profiles of those involved in the studies, but we need to remember that those who are able to access PHPs and residential treatment likely largely fit the stereotype of eating disorders. If we want to be sure that treatment is really effective, we need to consider people’s experiences of how that treatment fits with their lives – otherwise, we’re only reproducing more of the same.

Reference

Friedman, K., Ramirez, A., Murray, S., Anderson, L., Cusack, A., Boutelle, K., & Kaye, W. (2016). A Narrative Review of Outcome Studies for Residential and Partial Hospital-based Treatment of Eating Disorders European Eating Disorders Review, 24 (4), 263-276 DOI: 10.1002/erv.2449