What Predicts Mothers’ Completion of a PTSD-Parenting Intervention in Child Welfare?

Mothers caught between untreated trauma and the child welfare system are among the least likely patients in mental health care to finish treatment, yet a new pilot study reports that nearly two-thirds of them carried a demanding dual-focus therapy through to its end. The study, published online on June 11, 2026, in the Journal of Child and Family Studies, was led by first author Whitney Wortham of New York University’s Silver School of Social Work, with senior author Michael A. Lindsey of NYU’s McSilver Institute for Poverty Policy and Research. It followed 112 mothers enrolled in child welfare family preservation services, all of whom met diagnostic criteria for post-traumatic stress disorder at baseline. Many also carried heavy social burdens: nearly a third were homeless at enrollment, more than three-quarters were unemployed, and almost three-quarters were raising their children without a partner. Despite that pile-up of risk, 63 percent of the mothers — 71 of the 112 — completed the full Parenting-STAIR program, an intervention built to treat maternal PTSD and parenting difficulties within a single course of care. The question was deceptively simple: which mothers stay in treatment, and which fall away? The answers, drawn from standardized baseline assessments and demographic records, point to predictors that could reshape how child welfare agencies deliver trauma-focused mental health care.

The stakes behind that completion statistic are considerable. Post-traumatic stress disorder is strikingly common among birth parents involved with child protective services; a systematic review and meta-analysis cited by the team places its prevalence in this population well above general-population rates, and the disorder is far more than private suffering. PTSD’s core symptom clusters of re-experiencing, avoidance, negative alterations in mood and cognition, and hyperarousal can erode the emotional availability that sensitive parenting requires. A mother flooded by intrusive memories or perpetually braced for threat may struggle to tolerate a toddler’s tantrum, to respond consistently to a child’s distress, or to regulate her own anger before discipline crosses into harm. Previous work by co-author Claude Chemtob and colleagues, published in JAMA Pediatrics, linked maternal PTSD and depression in pediatric primary care with higher rates of child maltreatment and with children’s own exposure to traumatic events. The intergenerational logic is stark: untreated maternal trauma raises the probability of new maltreatment, which in turn seeds the next generation’s trauma burden. Breaking that cycle is the explicit ambition of Parenting-STAIR, which is why who completes it matters so much.

Parenting-STAIR — often abbreviated PSTAIR by the researchers — is an adaptation of STAIR, or Skills Training in Affective and Interpersonal Regulation, a phase-based psychotherapy developed by co-author Marylene Cloitre, of NYU Langone and Stanford University’s National Center for PTSD, for PTSD rooted in childhood abuse. Classical STAIR unfolds in deliberate stages: an early phase of skills training that targets emotion regulation and interpersonal functioning, followed by structured work that processes the traumatic memories themselves. The sequencing is the treatment’s engine; patients first build the capacity to tolerate and modulate intense emotion before confronting trauma material capable of triggering it. Parenting-STAIR extends that architecture into the parenting domain, pairing the emotion-regulation and social-skills curriculum with parenting skills training designed to strengthen the mother–child relationship and interrupt the coercive cycles that feed child behavior problems. The design reflects a practical insight about the child welfare population: mothers referred by the system often arrive seeking help with their parenting rather than their trauma, so folding trauma care into a parenting frame meets clinical need and mandated service requirements simultaneously. An earlier pilot of the approach, reported in the same journal in 2023, established feasibility and preliminary benefit; the new study asks the follow-on question: who stays the course.

To disentangle who completes and who drops out, the team organized the analysis around Andersen’s behavioral model of health services use, a longstanding framework in health services research that sorts predictors into predisposing characteristics, enabling resources, and need factors. The 112 participants were mother–child dyads drawn from family preservation services, meaning the families were receiving in-home supports meant to keep children safely with their parents rather than removed. At baseline, each mother completed a standardized battery: the Posttraumatic Diagnostic Scale for DSM-5, a validated self-report instrument for diagnosing and quantifying PTSD symptoms; the Center for Epidemiologic Studies Depression Scale for depressive symptoms; the Parenting Stress Index–Short Form for stress specific to the parenting role; the Difficulties in Emotion Regulation Scale for emotion dysregulation; and the Eyberg Child Behavior Inventory for the identified child’s behavior problems. Completion was defined as finishing the full course of the intervention. The researchers then entered these baseline variables into logistic regression models — a statistical technique suited to binary outcomes such as completed versus not completed — and reported each predictor’s odds ratio with a 95 percent confidence interval, conveying each effect’s size and precision.

The baseline portrait of the sample reads like a catalogue of compounded adversity. Mothers reported exposure to an average of 5.4 categories of traumatic events on the assessment’s trauma checklist, a burden reflecting accumulated adversity rather than a single catastrophic event. Ninety-two percent of the mothers showed at least moderate PTSD symptoms, and 28 percent fell into the severe-to-extreme range. Thirty-nine percent crossed the clinical cutoff for comorbid major depression. Layered over the clinical picture were acute social precarities: 30 percent were homeless at enrollment, 78 percent were unemployed, and 73 percent had no partner. Against that backdrop, the 63 percent completion rate is the study’s quiet headline. In adult psychotherapy generally, meta-analytic work cited by the team indicates that a substantial share of patients terminate treatment prematurely, and dropout in PTSD care specifically has been documented as a persistent threat to delivering an adequate therapeutic dose. That mothers facing homelessness, unemployment, and solitary caregiving stayed at this rate suggests the program struck a chord that many conventional clinical services miss.

The regression results are where the study turns counterintuitive. Mothers with more severe PTSD symptoms at baseline were more likely, not less, to complete the program: each one-point increase on the PTSD symptom measure raised the odds of completion by roughly 8 percent (odds ratio 1.08, 95 percent confidence interval 1.00–1.16, p < .05). Greater parenting stress told the same story, with each unit increase nudging completion odds upward (odds ratio 1.04, 95 percent confidence interval 1.00–1.08). Maternal age pushed in a straightforward direction — each additional year of age was associated with 25 percent higher odds of completing (odds ratio 1.25, 95 percent confidence interval 1.11–1.42, p < .05). The barriers, by contrast, were relational and structural rather than clinical. Mothers without a partner had odds of completion roughly 80 percent lower than partnered mothers (odds ratio 0.20, 95 percent confidence interval 0.06–0.69, p < .05), and each additional child in the household cut the odds by about 45 percent (odds ratio 0.55, 95 percent confidence interval 0.34–0.90, p < .05). Notably, baseline depression scores, difficulties with emotion regulation, and the severity of the identified child’s behavior problems did not significantly predict completion once the other variables were accounted for.

The pattern linking symptom severity to persistence rather than flight has a plausible clinical reading. In the vocabulary of the Andersen model, PTSD severity and parenting stress are need factors, and perceived need is one of the strongest engines of help-seeking and adherence: mothers whose symptoms are loudest may have the most urgent internal reason to keep showing up. That interpretation inverts the intuition many clinicians carry — that sicker patients are the hardest to retain — and it echoes a point from the treatment-burden literature the team cites, which holds that patients weigh the demands of a treatment against the distress it promises to relieve, and when distress runs high the balance can tip toward persistence. Parenting stress may operate through relevance as much as through misery; a program that visibly targets the daily frictions of raising children under the oversight of a welfare agency may feel worth its logistical cost in a way that generic therapy does not. The age effect fits life-course accounts in which accumulated parenting experience, and often greater material stability, make a standing weekly commitment easier to sustain. What the results do not support is the assumption that the most symptomatic mothers are the least reachable.

The barriers the study exposed are less about motivation than logistics, and they concentrate in precisely the mothers the system most needs to reach. An unpartnered mother has no second adult to absorb childcare during sessions, to share school pickups, or to hold the household together when a weekly appointment collides with a shift job or a housing office visit. Each additional child multiplies those demands — more school schedules, more illnesses, more bedtimes — and the odds ratios suggest that the cumulative weight eventually tips the balance against attendance. The team situates these findings within the literature on the social determinants of mental health, which documents how poverty, housing instability, and unemployment constrain not only the onset of mental disorders but also access to care and persistence within it. The practical implications follow directly: completion-supportive designs for this population would build childcare into sessions, offer flexible or wraparound scheduling, provide transportation assistance, and consider shortened or modular formats for mothers carrying the largest caregiving loads. Because child welfare involvement in the United States falls disproportionately on low-income families and on Black children and mothers, engineering these supports is also a matter of equity in service delivery, not merely of convenience.

Caveats attach, as they do to any pilot. The trial was open, with no randomization and no comparison condition, so the completion figures cannot be attributed to the intervention’s design alone; the measures were largely self-reported; and the sample was drawn from a single service system, limiting generalizability. Completion, moreover, is a gateway measure rather than an outcome: finishing treatment does not by itself demonstrate symptom reduction or lower maltreatment risk. All procedures were approved by New York University’s institutional review board, and the work was funded by the Robin Hood Foundation; the authors report no competing interests. The decisive test is already in motion — the team’s larger Safe Mothers, Safe Children study, whose protocol appeared in the journal Trials, is a randomized controlled trial designed to determine whether Parenting-STAIR treats maternal PTSD and reduces maltreatment recidivism. Whatever the trial shows, the pilot’s message to child welfare systems is already actionable: dropout among trauma-exposed mothers is not a fixed property of the population but a modifiable function of support, and the mothers most engulfed by symptoms may be the ones most ready to stay — provided the system clears the practical obstacles standing between them and the clinic door.