Design: 12-year follow-up of a randomised controlled trial of extended early intervention, with latent growth mixture modelling
Population: 148 patients with first-episode psychosis; 106 completed 12-year follow-up
Primary outcome: Social-occupational functioning trajectory class over 3 years, and functional outcome at 12 years
Effect: Four trajectory classes identified; over a quarter persistently poor, with worse 12-year function on most measures; baseline negative symptoms distinguished this class

Symptomatic remission in psychosis has never been the same thing as functional recovery, but there has been little to say about when the divergence sets in. This 12-year follow-up of a randomised trial of extended early intervention for first-episode psychosis took 148 patients with Social and Occupational Functioning Assessment Scale scores at three or more timepoints in the first three years, and followed 106 to the 12-year endpoint.

Latent growth mixture modelling produced four trajectory classes over those first three years: persistently good, gradually improved, suboptimal stable, and persistently poor. Most patients fell into the suboptimal or poor classes. More than a quarter were persistently poor, and that class did worse on most functional measures at 12 years. What distinguished them at baseline was severity of negative symptoms, not positive symptom burden.

The clinical reading is about measurement. Services track symptoms closely in the first three years and function loosely or not at all, and the variable with the long shadow is the one not being recorded. A SOFAS score at a few fixed points in the first three years is a cheap way to see which trajectory a patient is on while there is still time to act.

Record SOFAS or an equivalent function score at fixed intervals in the first three years, not just symptom scales
Baseline negative symptoms are the signal — assess them properly rather than inferring from positive symptom control
A patient in symptomatic remission with flat function is the group this identifies
Vocational and social intervention belongs in the first three years, not after symptoms settle
Four classes, not two — 'improving slowly' is a distinct group from 'never improved'

Why it matters

It puts the window for changing functional outcome inside the first three years, where services are currently measuring symptoms instead.

The statistics, in plain English

Latent growth mixture modelling finds groups in data; it does not prove those groups exist as distinct entities. With 148 patients and four classes, some classes are small, and 106 of the original 148 reached 12 years, so attrition could shift the picture. The finding is best read as confirming that early functional course tracks with long-term course, not as a validated classification to apply to an individual.