Psychopharmacological Treatment in the RAISE-ETP Study: Outcomes of a Manual and Computer Decision Support System Based Intervention

Research data and treatment guidelines (13–17) support distinctive medication strategies for first-episode and multiepisode patients. Our approach to assisting busy clinicians at our nonacademic “real-world” sites to incorporate specialized first-episode treatment strategies into their work started by developing first-episode medication guidelines based on review of the treatment literature. Medication recommendations were limited to marketed agents, given the community facilities setting. NAVIGATE treatment used a shared decision making model (18). For medication selection, patients and prescribers chose from among medications with equivalent evidence based on patient factors and preferences. The shared decision making framework plus the failure of any antipsychotic to demonstrate superior efficacy for initial treatment of psychosis led to the decision to group recommended medications into treatment stages instead of a single medication algorithm. In selecting antipsychotic medications for the treatment stage groups, preference was given to medications with data from studies with first-episode or adolescent patients with psychotic disorders. Side effect and treatment efficacy data from these studies were the primary considerations for assigning medications into initial or subsequent treatment groups. Symptom remission rather than symptom improvement was the treatment goal. If a satisfactory initial response was not obtained with an initial stage medication, medications were chosen from subsequent stage groups. The antipsychotics that were available in the United States during guideline development and for which data were available from contemporary studies with first-episode or adolescent populations were aripiprazole, chlorpromazine, clozapine, haloperidol, olanzapine, quetiapine, risperidone, and ziprasidone. Because of concerns about side effects with chlorpromazine, clozapine, haloperidol, and olanzapine and concerns about lower efficacy for maintenance treatment with haloperidol (19, 20), these medications were excluded from the stage 1 group, which thus consisted of the remaining studied agents: aripiprazole, quetiapine, risperidone, and ziprasidone. Stage 2 agents were the stage 1 agents plus chlorpromazine, haloperidol, and olanzapine. Clozapine was a stage 3 agent. For each medication, first-episode dosing guidelines were developed (e.g., for risperidone, a starting dosage of 1–2 mg/day, a target dosage of 3–4 mg/day, and a maximum dosage of 8 mg/day). Over the 2-year RAISE-ETP treatment duration, continuous antipsychotic treatment was recommended. Patients and prescribers evaluated the potential benefits and disadvantages of switching antipsychotics for participants who entered RAISE-ETP with prescriptions that did not conform to NAVIGATE stage 1 principles. Participants who agreed to take antipsychotic medication but not a NAVIGATE-preferred medication were treated with their preferred agent. Participants who declined to take any antipsychotic had ongoing prescriber monitoring visits. Strategies were also provided for managing side effects (dosage reduction being the usual initial step) and for monitoring and treatment of cardiometabolic abnormalities. Since depressive symptoms in first-episode patients often remit with antipsychotic treatment alone (21), prescription of adjunctive antidepressants for all first-episode patients with depressive symptoms was not advised. Instead, consideration of the persistence and severity of depression was suggested when making decisions about adjunctive antidepressants. The detailed NAVIGATE medication manual is available online (22).

Participants and prescribers used COMPASS, a NAVIGATE-developed computerized clinical decision making tool accessed via a secure web-based platform. COMPASS was designed to facilitate patient-prescriber communication. Participants entered information about symptoms, side effects, treatment preferences, medication adherence and attitudes, and substance use into COMPASS before meeting with their prescribing clinicians. Vital signs data and laboratory test results were also entered. These data were summarized by the COMPASS program for review by the prescriber at the beginning of each medication visit. NAVIGATE medication treatment used a measurement-based approach to guide treatment decisions. The standardized assessments done by the prescriber at each visit were informed by the previously entered participant data (e.g., the COMPASS-suggested probe questions for prescribers were modified at each visit based on the responses on the participant questionnaire, and the prescriber assessment screens displayed the participant’s responses to the corresponding item on the participant self-report screens). Integrating participant treatment priorities and the prescriber’s assessments, COMPASS provided suggested guideline treatments. Prescribers and participants then made medication decisions informed by these recommendations. NAVIGATE guidelines recommended a prescriber visit at least monthly for the first 2 years of treatment.

NAVIGATE prescriber training included a 2-day group session on NAVIGATE principles, followed by individual training via teleconferencing on technical aspects of COMPASS. Monthly group prescriber teleconferences with the NAVIGATE central team included group feedback about challenging clinical cases and NAVIGATE treatment options for them, along with review of relevant psychosis literature.

This report focuses on the first 2 years of patient participation, the minimum by design for all participants. Patients 15–40 years of age receiving treatment for a first episode of psychosis due to schizophrenia, schizoaffective disorder, schizophreniform disorder, brief psychotic disorder, or psychotic disorder not otherwise specified and who had taken antipsychotics for ≤6 months during their lifetime were recruited from 34 community mental health treatment facilities nationwide that did not have preexisting first-episode specialty care programs. Written informed consent was obtained from adult participants; for participants under 18, written consent was obtained from parents or guardians and written assent from participants. RAISE-ETP was conducted under the guidance of the institutional review boards at the coordinating center and the various sites, as well as the National Institute of Mental Health’s Data and Safety Monitoring Board.

RAISE-ETP employed cluster randomization. The 17 randomly assigned NAVIGATE sites recruited 223 participants, and the 17 community care sites recruited 181 participants. Community care clinicians were trained on recruitment, informed consent, and study assessment procedures but received no guidance on treatment approaches. The study design and assessments have been described previously (9). Monthly patient self-report data on prescriptions (medications and dosages) and on number of medication management visits were obtained with the Service Use and Resource Form (23). At baseline and at 3, 6, 12, 18, and 24 months, participants reported in a yes/no format whether they had experienced, during the past 30 days, any of 21 common side effects of antipsychotic medications (dizziness, blurred vision, dry mouth, excess saliva, nausea, constipation, increased appetite, weight gain, weight loss, restlessness, shaking, rigidity, fatigue, drowsiness, excess sleep, insomnia, decreased libido, other sexual problems, breast swelling or discharge, impaired sexual performance, and amenorrhea). Concurrently, vital signs were obtained and fasting blood samples were collected. Participants taking medications also completed the Adherence Estimator, a self-report scale (24) measuring beliefs related to intentional nonadherence that has been validated against pharmacy claims (25).

As RAISE-ETP participants had psychotic disorders, our paramount medication question was whether NAVIGATE, compared with community care treatment, was associated with a greater likelihood of antipsychotic prescription. We also compared the likelihood of participants receiving a prescription that conformed to NAVIGATE stage 1 (“first-line”). We determined, by study month, whether participants received prescriptions for antipsychotic monotherapy with a NAVIGATE stage 1 antipsychotic. We allowed a broad range of antipsychotic dosages (instead of our targeted dosage ranges) to qualify as first-line to allow for low dosages for antipsychotic initiation and higher dosages for management of treatment resistance (for example, the qualifying dosage range for risperidone was 1–8 mg/day, based on 1 mg/day being the NAVIGATE recommended lowest starting dosage and 8 mg/day the highest dosage). Participants receiving concurrent stimulants were classified as not receiving prescriptions for first-line medications; participants receiving antipsychotic monotherapy with paliperidone at approved dosages were classified as receiving a first-line medication prescription (NAVIGATE training included review of the administration advantages of paliperidone palmitate over risperidone microspheres). Given that depression outcomes were better with NAVIGATE than with community care, we also compared the likelihood of receiving an antidepressant prescription in the two conditions. In addition to these explorations of medication groups, we wished to characterize choice of specific antipsychotic agents and dosages prescribed. For this, we examined the likelihood of the most commonly used agents being prescribed (irrespective of dosage or other medications prescribed) and the mean modal dosage for the oral formulations of each agent.

The primary measure for general side effects was the total number of side effects (excluding amenorrhea, as it is not applicable to male participants). Secondary measures were amenorrhea and a priori side effect groupings (sedation, extrapyramidal symptoms, anticholinergic side effects, increased appetite or weight gain, and sexual problems). A side effect group was considered present if any side effect within that group was present.

Longitudinal analyses of the any-antipsychotic-use (yes/no) outcome and other binary outcomes were performed using a generalized linear mixed-models analysis with a logit link. PROC GLIMMIX in SAS, version 9.4, was used. Each subject’s modal dosage was calculated, and the mean modal dosage was compared between conditions using a mixed-models analysis with a random intercept for site. Longitudinal analysis of the cardiometabolic outcomes and total number of side effects was performed with mixed models using the PROC MIXED procedure in SAS. The mixed-models approach takes into account the within-subject correlation of the repeated measurements. The difference in the trajectories between the two treatment groups was assessed by including a time-by-treatment interaction term in the mixed models. Least square means, which estimate the population marginal means for a balanced design, were computed. In all longitudinal analyses, cluster correlation within site was addressed by including a random intercept for subjects nested within sites. The limited number of clusters in clustered randomized trials can cause an imbalance between treatment groups on baseline measures, potentially confounding the relationship between treatments and outcomes. In accordance with the overall RAISE-ETP statistical analysis plan (5, 26), variables with significant baseline group imbalance were included as covariates in our analyses if they were correlated with the outcome of interest at a level ≥0.30.

Adjustment for multiple comparisons was done by controlling the false discovery rate using the Benjamini-Hochberg procedure (27, 28). The R multtest package was used. False discovery rate correction was applied to groups of analyses that addressed the same clinical question. The blocks were medication classes prescribed, specific agents, daily dose, vital signs, laboratory findings, number of side effects, and specific side effects. Significance was declared for analyses with false discovery rate–corrected p values <0.05.

Odds ratios from the analyses of medication classes and specific agents were converted to Cohen’s d using the formula . Elsewhere, effect sizes of the difference between least square means were calculated using the formula , where t is t-value and df is degrees of freedom.

The participants’ demographic and clinical characteristics are summarized in Table S1 in the data supplement that accompanies the online edition of this article. Overall, 73% were men, and the mean age was 23 years. The most frequent racial categories were Caucasian (54%) and African American (38%), and the most frequent diagnoses were schizophrenia (53%) and schizophreniform disorder (17%).

Of the 223 NAVIGATE participants, 211 (94.6%) completed one or more COMPASS visits. During their first 2 years of study participation, NAVIGATE participants collectively completed 3,004 COMPASS assessments.

As illustrated in Figure 1, NAVIGATE participants had significantly more medication visits compared with community care participants (treatment-by-time interaction, F=3.78, df=23, 9246, p<0.001; effect of treatment, F=12.80, df=1, 9246, p<0.001). Over the 2 years, the least squares mean estimate of the number of medication visits per month was 0.29 (95% CI=0.23, 0.36) for the community care group and 0.55 (95% CI=0.42, 0.69) for the NAVIGATE group.

As shown in Table 1, NAVIGATE participants were significantly more likely compared with community care participants to receive an antipsychotic prescription (odds ratio=3.73, 95% CI=1.71, 8.16) and less likely to receive an antidepressant prescription (odds ratio=0.39, 95% CI=0.16, 0.94).

Over the trial, NAVIGATE participants were more likely to receive prescriptions conforming to NAVIGATE first-line principles (odds ratio=2.19, 95% CI=1.08, 4.42). Prescriptions at study entry for NAVIGATE and community care participants were equally likely not to conform with NAVIGATE first-line principles (t=−0.49, df=744, p=0.63). In post hoc analyses of participants who were not receiving a NAVIGATE first-line prescription at baseline, 62.7% of the 110 NAVIGATE participants later received a NAVIGATE first-line prescription, compared with 44.4% of the 90 community care participants (odds ratio=2.07, 95% CI=1.02, 4.16; t=2.11, df=31, p=0.043).

As shown in Tables 1 and 2, the specific antipsychotics prescribed and the mean modal daily dose did not significantly differ between conditions for any of the major antipsychotics. NAVIGATE participants were more likely than community care participants to receive prescriptions for aripiprazole and less likely to receive prescriptions for haloperidol, although these differences fell short of statistical significance.

As presented in Table 3, both weight and body mass index (BMI) analyses revealed significant treatment-by-time interactions. The estimated mean increase in BMI from baseline to month 24 was 2.10 (95% CI=1.32, 2.89) for the NAVIGATE group and 2.45 (95% CI=1.90, 2.99) for the community care group; the corresponding estimated weight gain was 6.51 kg (95% CI=4.61, 8.41) kg for the NAVIGATE group and 7.31 kg (95% CI=5.62, 9.00) for the community care group. No significant treatment-by-time interactions or treatment effects were detected in analyses of other vital signs data or of lipid or carbohydrate metabolism measures.

Analysis of the number of medication side effects revealed a significant treatment-by-time interaction. As shown in Table 4, NAVIGATE and community care participants reported equal numbers of side effects at baseline, but NAVIGATE participants reported fewer side effects at subsequent visits (the difference was significant at months 6 and 12 and fell short of significance at months 3, 18, and 24). A secondary analysis controlling for antipsychotic prescription similarly revealed an advantage for NAVIGATE treatment (treatment-by-time interaction, F=2.88, df=5, 1087, p=0.004). Table S2 in the online data supplement presents the analyses of side effect groups. NAVIGATE participants were significantly less likely to have sedation or anticholinergic side effects; they were also less likely to have extrapyramidal symptoms, appetite increase, and sexual dysfunction, although these differences fell short of significance.

Scores on the Adherence Estimator scale did not differ between groups at baseline, and they decreased (indicating fewer beliefs associated with nonadherence) significantly among NAVIGATE but not community care participants (treatment-by-time interaction, F=2.46, df=5, 940, p=0.032). Least squares mean estimates of baseline and 24-month scores were 8.33 (SE=0.86) and 6.087 (SE=0.69) with NAVIGATE (change decrease of 2.24 [SE=1.06]) and 7.12 (SE=0.80) and 7.90 (SE=0.81) with community care (change increase of 0.78 [SE=0.93]).