A Sequential Adaptive Intervention Strategy Targeting Remission and Functional Recovery in Young People at Ultrahigh Risk of Psychosis

Key Points Question What are the optimal type, timing, and sequence of interventions for individuals at ultrahigh risk of psychosis? Findings In this sequential multiple assignment randomized trial including 342 individuals, a specialized psychological intervention (cognitive-behavioral case management [CBCM]) and a psychopharmacological intervention (CBCM and antidepressant medication) were not more efficacious than control conditions in improving remission and functional recovery. Relapse rates among individuals who remitted were high. Meaning The findings of this study show that addition of sequentially more specialized psychosocial and antidepressant treatment for individuals who did not remit did not lead to superior outcomes, underscoring the need for further adaptive trials, treatment innovation, and an extended duration of care for relapse prevention.


eMethods 2. Information regarding the application of multiple imputation in this study
Multiple imputation is an approach in which imputation of the missing data are carried out a number of times to create a number of complete data sets.These data sets are then analyzed separately and the results from these data sets are then combined in a statistically appropriate manner.
To apply multiple imputation, the usual assumption is that the missing data are missing at random (MAR).To investigate the validity of this assumption, we examined the reasons for the missing data at 6 months, which was the time-point for the study's primary objective.36.3% of the cases had missing 6-month data, of which 9.7% were due to known reasons which did not appear to cause bias in the analysis (moved out of area, withdrew due to time commitment, no longer wanting support and did not want to take medication).2.0% were due to transition to psychosis or deterioration of mental state.These missing data could be 'explained' or predicted in the imputation process using baseline data as well as the non-missing data of other transitioned cases.The remaining 24.6% withdrew due to unknown reasons.These cases were fairly evenly distributed among the 4 treatment regimes at Step 2, i.e., before 6 months (25.0%, 23.3%, 28.9% and 21.2%, respectively for the 4 regimes) and there was no reason to believe that these cases would cause bias in the analysis.Based on the above reasoning, our judgement was that it was reasonable to assume that the missing data were MAR, i.e., the likelihood of observing a value is independent of the value itself, given the data that one has observed.There were missing data at other time-points apart from 6 months.We made the pragmatic assumption that MAR also applied to these missing data.
We applied multiple imputation by using the R package mice 4 and miceadds 5 which conducts multivariate imputation by chained equations.The predictors used in the imputation included age, gender, child trauma score and the longitudinal measures of functioning, general psychopathology, negative symptoms, positive symptoms, depression and quality of life.The longitudinal structure of the data was taken into account in the imputation.The number of imputations used was 100.
An additional point to note is that treatment allocations in this study were dependent on remission status.For many of the drop-outs, their remission status after a particular step could not be determined because the data relevant to the remission criteria were included in the missing data.After data imputation, the missing remission status became determinable.In other words, the missing remission status were also imputed in the process.In turn, treatment allocations for the drop-outs could then proceed in accordance with the imputed remission status in the intentionto-treat analysis.

Pearson correlation between baseline DACOBS total and 6-month outcomes and between change in DACOBS total and change in outcomes (Step 1 non-remitters) Baseline DACOBS total and 6-month outcome Change in DACOBS total and in outcomes 1
1 Change = 6-month score minus baseline score.AQoL, Assessment of Quality of Life; BPRS, Brief Psychiatric Rating Scale; GF: Global Functioning; MADRS, Montgomery-Åsberg Depression Rating Scale; SANS, Scale for the Assessment of Negative Symptoms; SOFAS, Social and Occupational Functioning Assessment Scale.eTable 3.

General linear model analysis comparing SPS and monitoring (Step 1 remitters) at 6 and 12 months with missing data handled using multiple imputation Baseline Month 6 (mean n 1 : 15 SPS; 14 monitoring) Month 12 (mean n 1 : 14 SPS; 15 monitoring) Mean (SE) Mean (SE) Effect size P value Mean (SE) Effect size P value GF: Social Monitoring
AQoL, Assessment of Quality of Life; BPRS, Brief Psychiatric Rating Scale; CAARMS, Comprehensive Assessment of At-Risk Mental States; DS, Disorganized speech; GF: Global Functioning; MADRS, Montgomery-Åsberg Depression Rating Scale; NBI, Non-bizarre ideas; PA, Perceptual abnormalities; SANS, Scale for the Assessment of Negative Symptoms; SPS, support and problem solving; SOFAS, Social and Occupational Functioning Assessment Scale; UTC, Unusual thought content.P value comparing monitoring and SPS with baseline score as a covariate.
1The mean sample sizes for each group over the multiple imputations.eTable 5.

Complete-case and per-protocol analysis comparing CBCM and SPS at 6 months Complete-case analysis Per-protocol analysis
© 2023 McGorry PD et al.JAMA Psychiatry.AQoL, Assessment of Quality of Life; BPRS, Brief Psychiatric Rating Scale; CAARMS, Comprehensive Assessment of At-Risk Mental States; CBCM, cognitive-behavioral case management; DACOBS, Davos Assessment of Cognitive Biases Scale; DS, Disorganized speech; GF: Global Functioning; MADRS, Montgomery-Åsberg Depression Rating Scale; NBI, Non-bizarre ideas; PA, Perceptual abnormalities; SANS, Scale for the Assessment of Negative Symptoms; SPS, support and problem solving; SOFAS, Social and Occupational Functioning Assessment Scale P value comparing SPS and CBCM using general linear model analysis with baseline score as a covariate; UTC, Unusual thought content.©2023 McGorry PD et al.JAMA Psychiatry.eTable 8.

Complete-case and per-protocol analysis comparing CBCM+fluoxetine and CBCM+placebo at 12 months Complete-case analysis Per-protocol analysis
© 2023 McGorry PD et al.JAMA Psychiatry.AQoL, Assessment of Quality of Life; BPRS, Brief Psychiatric Rating Scale; CAARMS, Comprehensive Assessment of At-Risk Mental States; DS, Disorganized speech; GF: Global Functioning; MADRS, Montgomery-Åsberg Depression Rating Scale; NBI, Non-bizarre ideas; PA, Perceptual abnormalities; SANS, Scale for the Assessment of Negative Symptoms; SOFAS, Social and Occupational Functioning Assessment Scale; UTC, Unusual thought content. 1 Comparing placebo and fluoxetine using general linear model analysis with baseline score as a covariate. 2nteraction between Step 2 treatment (SPS/CBCM) and Step 3 treatment (placebo/fluoxetine) using general linear model analysis with baseline score as a covariate.©2023 McGorry PD et al.JAMA Psychiatry.eTable 9.

Complete-case and per-protocol analysis of remission rates
CBCM, cognitive-behavioral case management; SPS, support and problem solving.P value calculated using the chi-square test. 1 Refers to sustained remission (i.e., remission criteria were met at weeks 4 and 6 for Step 1, 12 and 24 for Step 2, and 36 and 52 for Step 3).© 2023McGorry PD et al.JAMA Psychiatry.eTable 10.

Complete-case and per-protocol analysis comparing relapse rates between SPS and monitoring (Step 1 remitters) at 6 and 12 months Complete-case analysis Per-protocol analysis
SPS, support and problem solving.P value calculated using the Fisher's exact test.© 2023 McGorry PD et al.JAMA Psychiatry.eTable 11.

Complete-case and per-protocol analysis comparing relapse rates between SPS and monitoring (Step 2 remitters) at 12 months using logistic regression Complete-case analysis Per-protocol analysis
Interaction between the factors SPS vs CBCM and monitoring vs SPS.

eTable 12. Complete-case and per-protocol analysis comparing SPS and monitoring (Step 1 remitters) at 6 months Complete-case analysis Per-protocol analysis
© 2023 McGorry PD et al.JAMA Psychiatry.AQoL, Assessment of Quality of Life; BPRS, Brief Psychiatric Rating Scale; CAARMS, Comprehensive Assessment of At-Risk Mental States; DS, Disorganized speech; GF: Global Functioning; MADRS, Montgomery-Åsberg Depression Rating Scale; NBI, Non-bizarre ideas; PA, Perceptual abnormalities; SANS, Scale for the Assessment of Negative Symptoms; SOFAS, Social and Occupational Functioning Assessment Scale; SPS, support and problem solving; UTC, Unusual thought content.P value: comparing monitoring and SPS using general linear model analysis with baseline score as a covariate.eTable13.

Complete-case and per-protocol analysis comparing SPS and monitoring (Step 1 remitters) at 12 months Complete-case analysis Per-protocol analysis
McGorry PD et al.JAMA Psychiatry.AQoL, Assessment of Quality of Life; BPRS, Brief Psychiatric Rating Scale; CAARMS, Comprehensive Assessment of At-Risk Mental States; DS, Disorganized speech; GF: Global Functioning; MADRS, Montgomery-Åsberg Depression Rating Scale; NBI, Non-bizarre ideas; PA, Perceptual abnormalities; SANS, Scale for the Assessment of Negative Symptoms; SOFAS, Social and Occupational Functioning Assessment Scale; SPS, support and problem solving; UTC, Unusual thought content.P value comparing monitoring and SPS using general linear model analysis with baseline score as a covariate.

eTable 14. Complete-case and per-protocol analysis comparing SPS and monitoring (Step 2 remitters) at 12 months Complete-case analysis Per-protocol analysis Baseline Month 12 Baseline Month 12 n Mean (SE) Mean (SE) Effect size P value 1 P value 2 n Mean (SE) Mean (SE) Effect size P value 1 P value 2
AQoL, Assessment of Quality of Life; BPRS, Brief Psychiatric Rating Scale; CAARMS, Comprehensive Assessment of At-Risk Mental States; DS, Disorganized speech; GF: Global Functioning; MADRS, Montgomery-Åsberg Depression Rating Scale; NBI, Non-bizarre ideas; PA, Perceptual abnormalities; SANS, Scale for the Assessment of Negative Symptoms; SOFAS, Social and Occupational Functioning Assessment Scale; SPS, support and problem solving; UTC, Unusual thought content.
© 2023 McGorry PD et al.JAMA Psychiatry. 1 Comparing monitoring and SPS using general linear model analysis with baseline score as a covariate. 2nteraction between the factors SPS vs CBCM and M vs SPS using general linear model analysis with baseline score as a covariate.eTable16.

Treatment adherence rates
SPS, support and problem solving.P value calculated using the Fisher's exact test.