A structured narrative review of therapy duration for complex developmental trauma
Population: Adults with complex developmental trauma (CPTSD, personality disorders, structural dissociation) Method: Critical interpretive synthesis, 89 sources, 6 epistemological layers Finding: Duration depends on endpoint — symptom reduction in weeks, structural change in years Framework: Safety prerequisite + endpoint level + structural depth + elapsed time (sleeper effect) License: CC-BY-4.0 | AI-assisted: Claude Opus 4.6 — see
AI methodology
How long does therapy take when the injury is not a single event but a childhood? The clinical literature says weeks to months. The psychoanalytic literature says years. Memoirs describe decades. Client forums describe lifetimes. The field appears to disagree fundamentally about whether healing from complex developmental trauma is a brief intervention or a life's work.
This review examines 89 sources across six epistemological layers — peer-reviewed research, naturalistic outcome studies, the strongest counterpoint evidence for brief treatment, practitioner essays, client narratives from online forums, and published memoirs — and finds that the disagreement is not genuine. It is an artefact of unstated endpoint definitions.
The duration of healing depends almost entirely on what is meant by "healing."
Symptom reduction — fewer flashbacks, scores below clinical thresholds — takes weeks to months. The evidence for this is strong: EMDR intensive formats achieve 88% CPTSD remission in eight days; MDMA-assisted therapy eliminates PTSD diagnoses in 67% of participants after eighteen weeks; concentrated delivery formats produce large effects (d = 1.57) with dramatically lower dropout than standard weekly therapy.
Structural change — personality reorganisation, mentalizing capacity, earned attachment security, integration of dissociated parts — takes years. Perry's dose-response data show 50% personality disorder recovery at 92 sessions and 75% at 216 sessions. The Helsinki Psychotherapy Study found personality organisation still changing at ten-year follow-up. Phase-oriented treatment for dissociative disorders is projected at three to five years, with most participants still symptomatic at study endpoints.
These are not competing claims. They are nested levels of the same process. Both "it takes eight sessions" and "it takes eight years" can be correct simultaneously, because they measure different things. The field's confusion arises from a failure to specify which level of change is being claimed.
The review proposes that four factors determine the timeline:
Factor 0: The safety prerequisite. The therapeutic clock does not start at the first session. It starts when environmental, relational, and autonomic safety is sufficiently established for the developmental process to begin. Pre-safety therapy time — which may span months or years — inflates all duration estimates in the literature. The same therapist, the same modality, can produce nothing during active danger and full healing once safety is achieved.
Factor 1: Endpoint level. Five nested levels emerge from the data, from symptom reduction (weeks) through functional improvement (months), structural change (years), growth and reconnection (years to decades), to recovery as an ongoing direction rather than a destination.
Factor 2: Structural depth. Not all complex developmental trauma produces the same degree of disruption. Patients with primary structural dissociation and preserved adaptive capacity may genuinely respond in months. Patients with tertiary dissociation and pervasive developmental deficits may require years for the same endpoint. The diagnosis is the same; the structural depth is not.
Factor 3: Elapsed time. Evidence from multiple sources suggests that calendar time contributes to healing independently of therapeutic dose. The sleeper effect — therapeutic gains that emerge or grow after treatment has ended — is documented across psychodynamic, mentalization-based, and experiential dynamic therapies. Therapy installs capacity; life develops it.
The strongest evidence for brief treatment — rapid EMDR remission, pharmacologically enhanced protocols, intensive delivery formats — consistently measures symptom-level endpoints. The strongest evidence for long-term treatment — the Helsinki study, MBT eight-year follow-up, Perry's dose-response data — measures structural endpoints. The two literatures are answering different questions. When the endpoint is specified, the contradiction dissolves.
Two findings genuinely complicate the thesis and are acknowledged honestly: the Helsinki convergence (tripling the dose did not improve ten-year outcomes, suggesting diminishing returns beyond a threshold) and natural course data showing BPD remitting gradually over decades regardless of treatment.
The review's evidence base is survivorship-biased: all six layers select for people who stayed in therapy long enough to report. The 18–39% who drop out of standard delivery are invisible. The patients most relevant to the review — those with the deepest structural injuries — are the patients most likely to leave treatment. The review is more confident in the claim that for people who stay in therapy, structural healing takes years than in the broader claim that healing from complex developmental trauma takes years.
The dose-versus-elapsed-time confound cannot be fully resolved: no study has compared the same therapeutic dose delivered at different frequencies over different calendar durations in a complex trauma population. The neuroscientific rationale for why structural change is slow (polyvagal theory, triune brain model) is contested; the clinical observation that it is slow is robust across nine independent theoretical traditions.
git clone https://github.com/sjhogerzeil/narrative-review-therapy-duration.git
cd narrative-review-therapy-duration
./_scripts/tests.sh # verify structural integrity (89 sources, cross-refs, synthesis quality)
./_scripts/sync.sh check # check referential consistencyThen read 4_discussion/1_synthesis.md for the central argument, or browse 3_results/sources/ to inspect individual source annotations.
Requirements: bash, Python 3, and standard Unix tools. No external packages — all scripts use the standard library only. To build the PDF: pandoc and a LaTeX distribution (e.g., TeX Live).
./_scripts/build_pdf.sh # build PDF from source chaptersThis is a structured narrative review using critical interpretive synthesis (CIS; Dixon-Woods et al., 2006) — a methodology designed for questions that span heterogeneous evidence types and require interpretive argument rather than statistical aggregation. The review draws on 89 sources organised into six evidence layers, each representing a different epistemological perspective on the same question. Findings are triangulated across layers: claims supported by multiple independent evidence types are stronger than claims resting on a single layer.
The review was conducted with substantial AI assistance (Claude Opus 4.6, Anthropic). The complete division of labor — which tasks were performed by the human author, which by the AI, and how the AI's work was verified — is documented in 2_methods/13_ai_methodology.md. The human author bears sole responsibility for all claims, interpretations, and conclusions.
The complete audit trail — source annotations, extraction tables, synthesis evolution, process notes, and git history — is preserved in this repository. See Reproducing or adapting this review for instructions on reproducing, verifying, or forking the review.
| If you want to... | Start here |
|---|---|
| Read the full synthesis | 4_discussion/1_synthesis.md |
| Check a specific source | 3_results/sources/ — one file per source, organised by layer |
| See how data was extracted | 3_results/primary/ and 3_results/secondary/ |
| Read the counterpoint response | 4_discussion/5_counterpoint_response.md |
| Understand the methodology | 2_methods/1_design.md |
| See how AI was used | 2_methods/13_ai_methodology.md |
| Verify consistency | Run ./_scripts/sync.sh check |
| Fork for a new question | 2_methods/15_reproducibility.md |
Hogerzeil, S. J. (2026). Eight sessions or eight years? A structured narrative review of therapy duration for complex developmental trauma. Zenodo. https://doi.org/10.5281/zenodo.19171538
See CITATION.cff for machine-readable citation metadata.
This work is licensed under CC-BY-4.0. You are free to share and adapt the material for any purpose, provided you give appropriate credit.
For questions, feedback, or correspondence: s.j.hogerzeil@gmail.com
For errors, missing sources, or methodological discussion: open an issue (how to open an issue).