The Parenting Therapy Worked in a Research Clinic. In a Real One, It Mostly Didn't.

The paper that landed in BMC Pediatrics this month is a protocol, not a result. It describes a trial that has not finished enrolling, reports no outcomes, and will not answer its own questions for years. It is still one of the more honest documents to appear in addiction treatment research this fall, because it is trying to solve a problem the field usually states and then walks around.

The intervention is called Mothering from the Inside Out, or MIO. The problem is what happens to therapies when they leave the building where they were invented.

What the therapy asks of a mother

MIO is a twelve-week, manualized individual psychotherapy built for mothers in substance use disorder treatment. Its target is not parenting knowledge or child-development facts. It is parental reflective functioning — the capacity to notice what you are feeling in a hard moment with a small child, hold it long enough to think about it, and then respond to what the child needs rather than to what the moment is doing to you.

That sounds soft until you consider the setting. A mother three months into recovery, dealing with a toddler's tantrum at 6 a.m., is running a stress test on the same regulatory machinery her substance use disordered. MIO was designed by Nancy Suchman and colleagues at Yale around the idea that this machinery can be trained through explicit reflection on specific moments, and that improving it changes both caregiving and, downstream, relapse risk.

Two developer-led randomized trials supported that story. Compared with a psychoeducation comparison group, mothers who received MIO showed higher parental reflective functioning, greater caregiving sensitivity, and children whose cues were clearer to them; the trials also reported reductions in depression and psychiatric distress, and in one follow-up a lower proportion of heroin relapses. The cascade looked plausible and the mechanism studies matched it: therapist adherence predicted reflective functioning, which predicted caregiving sensitivity, which predicted attachment security.

Then addiction counselors delivered it

The obvious next step was to hand MIO to the workforce that would actually have to deliver it. A Stage III community-based trial did exactly that: 22 addiction counselors at a nonprofit treatment center in a small northeastern city, four outpatient clinics, 94 mothers caring for a child between 11 and 60 months, randomized to twelve sessions of MIO or twelve sessions of psychoeducation. Enrollment ran from January 2015 to December 2019, and the counselors had already been shown to deliver both treatments with fidelity after training and supervision.

The results, published in Development and Psychopathology, did not reproduce the research-clinic findings. Mothers in the MIO arm did show lower depression, but the caregiving and reflective-functioning advantages that had carried the earlier trials were not there at the same magnitude. The authors put it bluntly in their own abstract: participation in MIO "was not associated with the same degree of improvement that was observed in prior trials where MIO was delivered by research clinicians."

They also offered a reading that has sat in the literature since. MIO, they suggested, may have been protective — holding caregiving steady where it would otherwise have deteriorated over the follow-up period, rather than lifting it. And they named the unresolved question as intervention-intervenor fit: not whether counselors can follow the manual, but whether the therapy's active ingredient survives the conditions counselors work under.

A bigger trial, built around that question

The new protocol, led by Elizabeth Peacock-Chambers of Baystate Health and UMass Chan Medical School with a long author list that includes Suchman-lineage attachment researchers and Peter Friedmann, is designed as a type II hybrid effectiveness-implementation trial. That designation is the whole point. A conventional effectiveness trial asks whether the treatment works in the field and evaluates implementation later, in a separate study that often never happens. A type II hybrid puts fidelity, feasibility and clinical outcomes on equal footing from the start.

Two Massachusetts substance use and behavioral health agencies will supply both the mothers and the counselors — 200 women aged 18 or older with a substance use disorder who are caring for a child aged five or younger, either as guardian or through regular contact while working toward reunification. Randomization runs 2:1: roughly 133 mothers to twelve weekly sessions of 45 to 60 minutes of MIO added to treatment as usual, and about 67 to treatment as usual alone. The imbalance is a deliberate trade — more families get the therapy, a smaller group still provides the comparison.

The primary effectiveness endpoints are parental reflective functioning, parenting stress and substance use at the end of the intervention. The implementation endpoints get the same billing: counselor fidelity to core components on the MIO Treatment Fidelity Scales–Revised and the MIO Fidelity Checklist, and feasibility defined narrowly and usefully as delivering MIO at the intended treatment intensity. An adaptation of the Implementation Research Logic Model structures a mixed-methods read of how the two sites differ in the way they implement, and contextual determinants at the level of site, counselor and mother are modeled as predictors of both kinds of outcome.

The counselors are the experiment

Sixteen counselors will be trained. They must be at least 21, employed by a participating agency, and hold a bachelor's degree or higher in a counseling discipline — no infant mental health fellowship, no psychodynamic training, no research background. Before they deliver a single MIO session they have to demonstrate adequate fidelity, therapeutic alliance and clinical reflective functioning on standardized assessments, after which they receive supervision both from their own agency and from designated MIO trainers.

Read that requirement as a filter and it changes the interpretation of everything downstream. The trial is not asking whether an average counselor can do this. It is asking whether a counselor who can pass a reflective-functioning gate, working inside an agency that agreed to host a research study, can then sustain the work at full dose across twelve weeks of somebody else's chaotic life. Feasibility, in this protocol, is not a courtesy metric. It is the load-bearing one, because the earlier community trial suggests the clinical effect is the fragile part.

The quiet expansion nobody has flagged

The population has moved, too. The 2015–2019 trial enrolled mothers already in outpatient treatment, caring for a child aged 11 to 60 months at home. The new trial keeps the child under five but adds mothers who have regular contact with a child while working toward reunification — mothers whose children are, in the language of child welfare, placed elsewhere.

That is a substantially harder group to treat. Contact is supervised or scheduled, the parenting moments MIO works on are rarer and observed, and the mother's motivation is entangled with a court or agency timeline she does not control. It also happens to be the group where a parenting intervention has the most to offer, since reunification decisions turn partly on whether a caseworker believes the parent can manage a child's distress. Alaska's child-removal data, published this month, made the surrounding context plain: two-thirds of the state's removals between 2020 and 2025 involved a parent's alcohol or drug use, against a national figure near 38 percent.

Whether a therapy built around reflective functioning can move a reunification timeline is not a question this trial is powered to answer. It is, however, the question every family court in the country is already answering without evidence.

What the protocol cannot tell you

Three limits deserve to be stated before anyone quotes this study as good news.

The endpoints are planned, not observed. No mother has been reported as improving on anything. A protocol paper is a description of an intention, and the fact that the trial was registered on ClinicalTrials.gov in August 2025 while the protocol appeared more than a year later means the design was fixed well before publication — useful for credibility, irrelevant to whether the therapy works.

The 2:1 allocation buys access at the cost of statistical power in the comparison. If MIO's true effect on parenting stress is modest — and the community trial suggests the effect is at minimum smaller than the original trials implied — a 67-mother usual-care arm may not separate from noise. A null result from this trial will be genuinely ambiguous.

And the trial cannot repair the finding it is responding to. It can show that MIO is deliverable at dose with adequate fidelity in two willing Massachusetts agencies. It cannot show that reflective functioning is the mechanism by which caregiving improved in the developer trials. The community replication weakened that causal chain, and nothing in the new design re-tests it.

Why this matters beyond one therapy

For families living with co-occurring substance use and mental health conditions, parenting is rarely the thing treatment addresses. Addiction programs treat use. Child welfare systems treat risk. Mental health systems treat symptoms. The mother who is trying to do all three at once is usually the person in the room with the least institutional support and the most consequences.

MIO is one of a small number of interventions built specifically for that intersection, which is why its replication problem matters more than its sample size suggests. The science-to-service gap is usually described as a problem of dissemination — get the evidence-based treatment into more agencies. What the Connecticut trial showed is closer to the reverse: the treatment arrived, the counselors were trained to fidelity, and the effect still thinned out. That is not a marketing failure. It is a question about what the therapy needs in order to work, and whether an ordinary addiction clinic can supply it.

The trial now enrolling in Massachusetts is, in effect, the field's attempt to find out. It will report fidelity numbers before it reports outcomes, and those numbers will be the more interesting half.