Treating Overcontrol with Overcontrol: The Hidden Harm

How clinician rigidity replicates the very dynamic it’s trying to treat — and what that costs our patients

There is a word in behavioral health that does a great deal of work while asking very little of us.

Noncompliant.

The patient declined residential. The family resisted recommendations. ROIs weren’t signed. The treatment team lost alignment. And somewhere in the gap between what we recommended and what the patient did, disagreement became pathology.

This framing is comfortable. It locates the problem outside of us. And in eating disorder treatment especially, where patients can be medically fragile, ambivalent, and genuinely resistant to care, it’s easy to see why it persists.

But I want to make a different argument. One that is, frankly, harder to sit with.

When we treat overcontrolled patients with rigid, compliance-based systems, we are not treating the disorder. We are replicating it.

The field has a word for what many of our most treatment-resistant patients are experiencing. Overcontrol. Maladaptive perfectionism. Cognitive rigidity. An insistence on sameness, an inability to tolerate uncertainty, a deep need to manage outcomes from the inside out. These are not personality flaws; this is the architecture of the illness.

And yet, our response is often to bring our own version of the same architecture to the table: rigid level-of-care algorithms, compliance-contingent treatment, polarized team dynamics, and a frame in which ambivalence is pathology rather than information.

We need to be honest about what that does.

The Control Paradox

Research on eating disorders has a name for this problem. It’s called the control paradox.

Eating disorders, particularly anorexia nervosa, have been conceptualized as dyscontrol syndromes: the patient’s overcontrol is the symptom, not the defense against it. The goal of treatment, therefore, is to develop internal agency and flexibility. An internal locus of control. Not compliance with external demands, but genuine, self-directed engagement with recovery.

What happens, then, when we introduce coercive or rigidly compliance-based structures?

Therapeutic strategies aim to develop internal control in the patient, rather than employing external coercive tools. Control and compulsion have a simultaneous and complementary nature in the treatment of eating disorders — and this is precisely what makes them so dangerous.

Túry et al., Journal of Clinical Psychology, 2019

This isn’t just theoretical. The data bears it out. A systematic review and meta-analysis comparing compulsory and voluntary treatment in eating disorder patients found that forced hospitalization resulted in stays that averaged three weeks longer, without producing significantly better weight outcomes. Longer control. Same results. The external pressure didn’t accelerate recovery; it prolonged the fight.

Compulsory treatments in eating disorders: a systematic review and meta-analysis, 2021

We keep reaching for control as the intervention. And we keep getting the same outcome.

What the Dropout Numbers Are Actually Telling Us

Dropout rates in eating disorder treatment range from 20–51% in inpatient settings and 29–73% in outpatient settings. Nearly half of patients who enter our most intensive programs leave before completing them.

Fassino et al., BMC Psychiatry meta-analysis, 2009

We tend to read these numbers as a patient problem. Resistance. Noncompliance. The illness talking.

But there is another interpretation- one the research also supports.

Studies have found that the strongest predictor of treatment dropout is not illness severity, not psychiatric comorbidity, not even readiness to change. It is the patient’s experience of the therapeutic relationship. Patient ratings of the alliance (whether they felt heard, respected, and safe) were significantly associated with whether they stayed in treatment. Readiness to change, by itself, was not.

Finding et al., 2015; Helping Alliance Questionnaire research in outpatient ED settings

Think about what that means. A patient can walk in ambivalent and still stay in treatment, if the relationship holds. A patient can be highly motivated and still leave- if it doesn’t.

We are not losing patients because they are too sick. We are losing patients because we are not making treatment safe enough to stay in.

The best predictor of patient outcomes was neither how sick they were nor how much hospitalization they required. It was their baseline interest in making change — and whether the treatment environment honored that.

Geller, as reported in Filter Magazine, 2023

What RO-DBT Teaches Us About Our Own Rigidity

Radically Open DBT- the treatment model developed specifically for overcontrolled presentations, including restrictive eating disorders, is built on a central clinical premise: that overcontrolled patients need openness modeled for them, not demanded of them.

The treatment uses dialectical principles to encourage cognitively rigid patients to think and behave more flexibly. It works precisely because the therapeutic stance itself is the opposite of the patient’s pathology. The clinician brings flexibility, openness to uncertainty, and genuine relational warmth.

ABCT Fact Sheet; Hatoum & Burton systematic review, 2024

What RO-DBT makes explicit is something we know but resist: you cannot use the same mechanism to treat and to cure. You cannot treat rigidity with rigidity. You cannot treat control with control. The relational stance of the treatment has to be the antidote to the symptom, not its mirror.

And yet the field’s default response to resistance- escalate the level of care, enforce compliance, make treatment contingent on submission, is precisely that mirror.

This is worth sitting with. Especially because many of us who are drawn to this work are ourselves conscientious, perfectionistic, overcontrolled. These are the traits that make us good clinicians. They also make us vulnerable to replicating, under stress, the very relational dynamics our patients are trapped inside.

Conditional acceptance. Compliance-based worthiness. Rupture in response to autonomy. Fear-based engagement.

When a patient declines a recommendation and we respond by threatening to withdraw care, we have just recreated, in the treatment relationship, the exact structure the patient lives inside their illness.

The Level-of-Care Trap

I want to be precise here, because this argument is sometimes misread.

Residential treatment can be lifesaving. For some patients, it is unequivocally necessary. The question is not whether RTC helps. The question is what we do when a patient declines it.

There is a meaningful clinical and ethical difference between recommending a higher level of care, and making continued treatment contingent on submission to that recommendation.

The first is our job. The second is a power move that, in the literature, has a name: perceived coercion. And perceived coercion has measurable clinical consequences.

Patients who experience treatment as coercive report higher distress, lower therapeutic alliance, and greater difficulty engaging with recovery. Involuntary hospital treatment takes longer and patients struggle significantly more with weight restoration compared to those who enter voluntarily.

Narrative Review, MDPI Healthcare, 2023

This does not mean we never push. It means we are honest about what pushing costs, and whether the account can afford it.

Because the “ideal” level of care on paper is clinically meaningless if the patient will not engage in it. Patients do not recover from ASAM grids or treatment algorithms. They recover through sustained participation, alliance, safety, repetition, and time.

In mental health treatment, continuity is often the intervention. Sometimes staying in relationship with a struggling patient — imperfectly, incrementally, pragmatically — is the most clinically defensible thing we can do.

Not because it’s ideal. Because it’s what the patient can actually engage in right now. And engagement is the only variable that actually predicts recovery.

A Note on “Splitting”

No clinical concept in eating disorder treatment is deployed more loosely, or does more damage when misused, than splitting.

Splitting exists. Manipulation exists. Eating disorders are ego-syntonic and can absolutely exploit systems and relationships. This is not a naive argument.

But splitting is increasingly invoked whenever a patient seeks autonomy, a family hesitates, providers disagree, or a treatment team loses alignment. In these moments, the term stops being a clinical observation and becomes a defensive explanation: a way of locating discomfort in the patient rather than examining what’s happening in the system.

A patient refusing to sign an ROI is not automatically splitting. It may reflect fear, relational distrust, previous coercive experiences, attachment ruptures, or a completely reasonable assessment that the treatment environment does not feel safe.

Before we call it splitting, we might ask a more useful question:

What made collaboration feel unsafe here?

That question opens a treatment. The other one closes it.

What We Owe Our Patients

Recovery from a restrictive eating disorder is not a single event. It is long, nonlinear, relational, and deeply personal. Many patients disengage and return. Many make progress only after years of partial participation. The research on therapeutic alliance is unambiguous: that alliance, not level of care, not treatment intensity, is associated with treatment completion, symptom reduction, and faster recovery, particularly in anorexia nervosa.

Zaitsoff et al., 2015; Graves et al., 2017; Frontiers in Psychology, 2024

Our ethical obligation is not to recommend the theoretically optimal pathway and then withdraw when the patient doesn’t take it. Our ethical obligation is to remain engaged: reassessing risk, preserving alliance, building motivation over time, and holding the door open.

The most dangerous thing we can do for an overcontrolled patient is confirm, through our own rigidity, that the world only accepts them when they comply.

That is the message their illness is already delivering. We do not have to deliver it too.