Analysis based on the August 12 Frontiers in Psychiatry systematic review and its August 27 PubMed record, alongside current NIMH, WHO, and AOTA guidance on schizophrenia, rehabilitation, and occupational therapy in mental health.
A forest plot can give a scattered literature the appearance of a verdict. In a new schizophrenia review, the pooled estimates point toward occupational therapy: lower overall symptom scores, better social-function scores, and better activity-of-daily-living scores. The direction is encouraging. The fine print is doing heavy work. Ten small trials supplied the evidence, all from China or Japan. Several outcomes rested on two to four studies, provider credentials were often unreported, and heterogeneity ran as high as 96%. For mental-health OT, this is a signal worth reading and a prescription for better research. It is too fragile to set a protocol, promise an effect, or stand alone in a service-line proposal.
In brief
- The review pooled 10 randomized trials with 958 adults. Seven trials involving 722 participants favored OT on total Positive and Negative Syndrome Scale scores by a mean 5.76 points.
- Secondary findings favored OT on social-function, behavioral, and ADL scales, though each estimate drew from only two to four trials and heterogeneity ranged from 73% to 96%.
- All trials were conducted in China or Japan. Provider credentials were frequently absent, participant and therapist blinding was impossible, outcome-assessor blinding was often unclear, and the funnel plot suggested possible publication bias.
Ten trials lean in one direction
Researchers at Jiamusi University and Daqing Third Hospital searched seven English- and Chinese-language databases for randomized trials published from 2015 through February 2026. They found 10 studies from China and Japan, with samples ranging from 40 to 200 participants and 958 participants in all. The review was registered in PROSPERO and followed the PRISMA reporting framework.
Seven trials involving 722 participants reported total scores on the Positive and Negative Syndrome Scale. The pooled occupational therapy group scored 5.76 points lower than the comparison group, with a 95% confidence interval from 3.79 to 7.72 points lower. Lower PANSS totals indicate fewer overall symptoms. The analysis had moderate heterogeneity, with an I-squared value of 54%.
That result held its direction when the researchers removed individual studies one at a time. The review cannot say whether positive symptoms, negative symptoms, or both changed because the included studies did not report those subscales separately. A total score compresses several clinical questions into one number.
The secondary numbers carry wider margins
The function results followed the same general direction and rested on thinner evidence. Two trials with 293 participants favored OT on the Social Disability Screening Schedule. Four trials with 311 participants favored OT on the Nurses' Observation Scale for Inpatient Evaluation. Three trials with 229 participants favored OT on an activities-of-daily-living scale.
The studies disagreed substantially about the size of those effects. Heterogeneity reached 88% for social disability, 96% for the nursing-observation scale, and 73% for ADLs. Differences in participants, intervention content, staff, dose, timing, settings, and measurement can all widen that spread. A pooled mean supplies a summary; it does not make unlike programs interchangeable.
The duration analysis offers no shortcut. Trials lasting more than 10 weeks produced a statistically significant estimate, while the shorter group did not. The formal test found no significant difference between duration subgroups. The review therefore provides no reliable minimum dose and no evidence that a longer program works better.
Occupational therapy was loosely defined
The most important methods sentence appears before the forest plots. Provider credentials were not required for inclusion because most original studies did not report them. The review accepted interventions connected to the eight occupational therapy domains, whether delivered individually, in groups, or through a combination. Only one trial is described in the discussion as being delivered exclusively by a full-time occupational therapist; three others used uniformly trained health-care personnel.
That boundary matters in a profession-specific claim. Structured activity, life-skills practice, social training, cognitive work, leisure, and vocational preparation can belong inside occupational therapy. Their presence alone does not establish that an occupational therapist evaluated occupational performance, built an occupational profile, analyzed context, selected an intervention through clinical reasoning, or adjusted the plan with the person receiving care.
The review supports an intervention family whose members shared occupation-related content. It gives limited evidence about which elements were distinctively OT, who was qualified to deliver them, how fidelity was maintained, or which clinical decisions drove the effect. Program leaders should keep those uncertainties attached to every presentation of the pooled result.
Bias and geography set the boundary
Seven trials adequately described random sequence generation. Three were unclear. Blinding participants and personnel was infeasible, leaving all studies at high risk of performance bias. Most reports did not clearly say whether outcome assessors were blinded. The funnel plot suggested possible publication bias, and several outcome groups were too small to support a confident assessment of missing negative studies.
The trials were mostly small, single-center, and short term. All came from China or Japan. Service systems, hospital use, professional roles, intervention labels, family expectations, community supports, and outcome measurement can travel differently across countries. The review includes no U.S. trial and offers little long-term evidence about housing, school, employment, relationships, community participation, readmission, or quality of life.
Those limits do not erase the consistent direction. They define the distance between an encouraging synthesis and a portable clinical standard. The authors describe the strength of evidence as limited and call for larger multicenter trials, standardized interventions, rigorous methods, and longer follow-up.
Use the review as a program audit
A mental-health OT service can put this paper to work without borrowing its effect size. Start by mapping what the service actually delivers: the occupations addressed, assessment process, intervention theory, staff credentials, supervision, frequency, session length, duration, setting, and handoff to home and community. If a program cannot describe those elements, it will struggle to reproduce an intervention or explain an outcome.
Keep symptoms and function in separate columns. PANSS change can matter while leaving the person's chosen life goals unresolved. Track the outcomes that occupational therapy is positioned to influence: starting and sustaining daily routines, managing health, completing ADLs and IADLs, using transportation, returning to school or work, maintaining housing, building social participation, and carrying strategies into real environments. Record whose goal each measure represents.
NIMH places psychosocial treatment beside medication and describes coordinated specialty care as collaborative, recovery-focused team care that includes employment and education support, case management, psychotherapy, medication, and family education. WHO similarly includes ADL training, self-management, social skills, supported housing, supported education, and supported employment in rehabilitation for schizophrenia. OT should enter that team with a clear functional contribution and shared measures, rather than claiming the entire recovery package.
Build the trial the service needs
For clinics and universities, the research agenda is unusually concrete. Define the intervention in enough detail for another team to deliver it. Report provider qualifications and treatment fidelity. Blind outcome assessors where feasible. Register the protocol and publish null results. Recruit across multiple sites and communities. Follow people long enough to see whether gains reach ordinary life and last.
The outcome set should include the person's priorities alongside symptom measures: participation, occupational performance, quality of life, education, employment, housing stability, social connection, readmission, and adverse effects. Researchers should report positive and negative symptom subscales separately and examine whether intervention response varies with stage of illness, cognitive needs, culture, setting, and access to community supports.
AOTA describes mental-health OT as comprehensive assessment of client needs and environmental context followed by client-centered intervention for participation in meaningful daily life. That description can become a fidelity question. Did the study measure the life the participant wanted to resume, and did the intervention change the conditions that made participation difficult?
An encouraging line, drawn in pencil
This review gathers a fragmented literature and finds a recurring signal: people receiving occupation-related rehabilitation often left the study period with better symptom and function scores than comparison groups. The direction deserves a larger, cleaner test.
The current evidence cannot identify a standard OT protocol, establish who must deliver its active ingredients, or promise the same effect in a U.S. community program. It does give practitioners a disciplined place to begin: name the intervention, protect the person's agency, measure participation as carefully as symptoms, and make the next result easier to trust.
The diamonds on the forest plot are an invitation. The field still has to draw the map.