LSR3_taar1_H.Rmd

---
title: "Trace amine-associated receptor 1 (TAAR1) agonists for psychosis - Results from the human studies <span style='font-size:25px'></style>"
author: "First draft by Spyridon Siafis, commented by Virginia Chiocchia, Charlotte Austin, Claire Friedrich, Robert R McCutcheon, Georgia Salanti, Stefan Leucht <span style='font-size: 15px'></style>"
date: "`r format(Sys.Date(), '%e %b %Y')`"
output:
  html_document:
    toc: true
    df_print: paged
bibliography: ["lsr3_references.bib", "grateful-refs.bib"]
biblio-style: "apalike"
link_citations: true
mainfont: Arialh
---

```{r setup,message=F,echo=F,include=F}
### libraries
library(meta)
library(tidyverse)
library(tibble)
library(stringr)
library(readxl)
library(grid)
library(dplyr)
library(knitr)
library(kableExtra)
library(devtools)
#library(metafor)
library(robvis) #Need to install from devtools::install_github("mcguinlu/robvis")
library(PRISMA2020)
library(grateful)
#devtools::install_github("haozhu233/kableExtra") I need to remove this, otherwise the R markdown didnt work


#Empty the enviroment 
rm(list=ls())

# Set option defaults
knitr::opts_chunk$set(echo = T, warning = F, message = F)

#Data cleaning routines
source("data/clean_data.R")

#Data analysis routines and functions 
source("util/analysis.R")

# Loading 
studies<-read_xlsx('data/studies_2024-01-22_LSR3_H.xlsx', na = "NA") #Descriptive data set with included studies
data <- read_xlsx('data/data_2024-01-22_LSR3_H.xlsx') #Clean data with the RCTs with available data
data_doses<-read_xlsx("data/data_doses_2024-01-22_LSR3_H.xlsx") #Effect sizes for the individual doses of TAAR1 (for a descriptive figure)
ae_categories<-read_xlsx("data/ae_categories.xlsx") #Supporting file with the names of adverse events for the summary plot for the side-effects

#prisma data
prisma<-read.csv("data/LSR3_H_PRISMA_2024.01.23.csv") #csv file for the prisma2020 
prisma<-PRISMA_data(prisma)

#risk of bias data
rob<-read_xlsx("data/rob_secondary_2024.01.24.xlsx")
#results matrix table
results_matrix<-read_xlsx("data/reporting_bias table_2024.01.26.xlsx")
#SoE tables
soe_sch_ta1_pl_eff<-read_xlsx("data/soe_tables_2024.01.30.xlsx", sheet = 1)
soe_sch_ta1_apd_eff<-read_xlsx("data/soe_tables_2024.01.30.xlsx", sheet = 2)
soe_pd_ta1_pl_eff<-read_xlsx("data/soe_tables_2024.01.30.xlsx", sheet = 3)
soe_all_ta1_pl_saf<-read_xlsx("data/soe_tables_2024.01.30.xlsx", sheet = 4)
soe_all_ta1_apd_saf<-read_xlsx("data/soe_tables_2024.01.30.xlsx", sheet = 5)
soe_mechanistic<-read_xlsx("data/soe_tables_2024.01.30.xlsx", sheet = 6)

#Round off results to two digits
options(scipen=1, digits=2)

```

# 1. Description of the methodological approach

1.  We searched for all controlled and uncontrolled studies examining TAAR1 agonists. In this first iteration of the review, we analyzed usable data from all randomized controlled studies comparing any TAAR1 agonist to placebo or current antipsychotics in people with psychosis, other mental health conditions, or healthy volunteers.

2.  We examined efficacy outcomes, including overall symptoms of psychosis (primary outcome) and symptom domains, dropouts due to any reason, and adverse events. Safety outcomes included death, serious adverse events, and specific adverse events. We also sought mechanistic outcomes, specifically neurobiological measures of dopaminergic, glutamatergic, and serotonergic signalling. We examined these outcomes at three time points: immediately after a single dose up to 2 weeks of treatment, 3-13 weeks of treatment (preferably at 6 weeks), and longer-term outcomes.

3.  In terms of efficacy outcomes, we pooled the data for each specific subgroup, considering that differences can be expected among different diagnoses (e.g., schizophrenia and Parkinson's disease psychosis) and patient subgroups (e.g., patients with an acute episode of schizophrenia or with predominant negative symptoms). In terms of other outcomes, we combined data from all subgroups but also presented data separately for each subgroup.

4.  Summary of data extraction: For continuous outcomes, we preferred change over endpoint scores and data from appropriate models or imputations considering missing outcome data. For dichotomous outcomes, we again preferred data from all participants, but in case of participants with missing outcome data, we assumed that they did not experience the event; an assumption usually used in meta-analysis in schizophrenia (e.g., @RN2). In terms of crossover studies, we preferred data from appropriate models considering period or carry-over effects. If not available, we assumed a correlation (i.e., 0.2 for adverse events) to correct the variance of the estimates according to @elbourne2002 . If an outcome was measured multiple times after a single dose of the drug, we calculated the area under the curve according to the linear trapezoid rule [@RN120] and divided by the time of assessment to remove the time component.

5.  Risk of bias was assessed using the RoB2 tool [@RN64] for parallel and crossover studies. When only one domain had some concerns, we assigned an overall low risk of bias. Otherwise, we used the worst judgment across domains.

6.  The effect sizes were as follows: 1) standardized mean differences (SMD) for efficacy outcomes, given that various scales have been used in the literature to allow comparability of the findings with other studies of antipsychotic drugs; 2) mean differences (MD) for laboratory measures, i.e., weight (kg), QTc interval (msec), and prolactin levels (ng/ml); 3) odds ratios (OR) for dichotomous outcomes, which were also converted to absolute risks using a control event rate [@RN71]. 95% confidence intervals (95% CI) are presented, but 95% prediction intervals are not presented due to the small number of studies. Other effect sizes were not used in the first iteration of the review, e.g., those used in meta-analysis of variance due to the paucity of data.

7.  Meta-analysis was conducted primarily using a random-effects model with inverse variance and REML estimator of tau-squared. Due to the small number of studies (\<5) in all analyses, we corrected the confidence intervals using the Hartung-Knapp method [@RN82], and we also presented the findings from fixed-effects models using the Mantel-Haenszel method throughout. Meta-analysis of proportions was conducted to pool the control groups to estimate the control event rate using logit transformation of the proportions and inverse variance method [@schwarzer2019]. Moreover, continuity corrections of 0.5 were applied in case of 0 events in one of the groups.

8.  Sensitivity analysis of the primary outcome included using fixed-effect methods (*post-hoc* due to the small number of studies), restricting the analysis to studies with an overall low risk of bias, and excluding studies with imputed values.

9.  Subgroup analyses and meta-regressions were not conducted due to the paucity of data. However, we explored potential dose-effects for the primary outcome by plotting the effect sizes for different fixed doses of TAAR1 agonists compared to placebo (while correcting for the use of common comparators [@higgins2019]).

10. Summary of evidence tables were constructed for all outcomes, including a summary of the association, within-study biases, across-study biases, biases due to indirectness, and other biases. We used the RoB2 tool [@RN64] to assess within-study biases, as mentioned above. We used the ROB-ME approach [@page2023] for evaluating biases in missing evidence and across-study biases. Indirectness was considered especially relevant for non-efficacy outcomes, where we pooled data across diagnoses and subgroups, and considered the percentage of studies and participants with schizophrenia, the main group for which this evidence would be applicable.

11. Finally, as there were no usable data from controlled experimental studies on the effects of TAAR1 agonists on neurobiological measures relevant to dopaminergic, serotonergic, or glutamatergic signalling, we could not analyze them as secondary outcomes in the current review. Nevertheless, to provide mechanistic insights, we narratively described the findings of any available mechanistic data from controlled and uncontrolled studies, irrespective of their inclusion in this review, given that we had comprehensively searched for them [@siafis2023]. We also presented the findings of these studies narratively in summary of evidence tables and discussed potential issues of risk of bias, reporting bias, indirectness, and other biases, but without formally using risk of bias tools.

*Please see the manuscript, extended data (*<https://osf.io/tdmau/>*) and the protocol [@siafis2023] for more details.*

*A list of abbreviations can be find towards the end of the document.*

# 2. Results

**We present below the results, including the PRISMA flow diagram [@page2021] , table of included studies, risk of bias assessment, and forest plots, to facilitate a better understanding of the summary of evidence tables and the manuscript. Please refer to the detailed description of the findings in the latter part of the manuscript for more information.**

## 2.1 Flow of study selection

```{r echo=F, fig.fullwidth=T, fig.width=12, fig.height=8}
PRISMA_flowdiagram(prisma, previous = FALSE, other=TRUE)
```

**Figure 1** PRISMA 2020 flow diagram. \*It should be noted that due to the limited data on mechanistic insights, we also presented narratively the findings from one of the excluded uncontrolled studies, since this was the only molecular neuroimaging study @RN132; this study is listed along with the 26 excluded studies due to wrong study design in the PRISMA flow diagram.

## 2.2 Table of included studies

```{r echo=FALSE, fig.width=12, fig.height=8}
studies_to_plot<-studies[order(studies$Name),]
knitr::kable(studies_to_plot[, 1:14])
```

**Table 1** Table of included studies. \*It should be noted that due to the limited data on mechanistic insights, we also presented narratively the findings from one of the excluded uncontrolled studies, since this was the only molecular neuroimaging study [@RN132], which characteristics are also presented here as an exception along with the included RCTs. SB: Single-Blind, DB: Double-Blind, DSM: Diagnostic and Statistical Manual for Mental Disorders, RCT: Randomized controlled trial, MDD: Major Depressive Disorder, GAD: Generalzied Anxiety Disorder, fMRI: functional neuroimaging, PET: Positron emission tomography.

## 2.3 Description of included studies

There were 25 eligible studies, which characteristics can be found in **Table 1**, with 20 studies (2147 participants) being completed and 5 studies being still ongoing (according to the status reported in clinicaltrials.gov up to 20.01.2024). There were usable data for this analysis from 9 studies (1683 participants) [@RN116; @RN13; @RN110; @RN111; @RN112; @RN113; @RN115; @RN123; @RN124], which enrolled adult populations across multiple countries and were funded by the industry manufacturing the TAAR1 agonists.

The usable studies compared single or multiple doses of two TAAR1 agonists (i.e., ulotaront or ralmitaront) with placebo or current antipsychotics (i.e., risperidone or amisulpride) in people with schizophrenia, Parkinson's disease psychosis, other mental health conditions (narcolepsy/cateplexy as sleep disorder) and healthy volunteers.

*Five studies* used a parallel design to examine the efficacy and safety of multiple doses of TAAR1 agonists in psychosis within 3-13 weeks: *Four studies* examining daily doses of TAAR1 agonists in adults with an acute episode of schizophrenia spectrum conditions for 4-6 weeks, 3 examined ulotaront [@RN110; @RN111; @RN112] and 1 examined ralmitaront [@RN113]. These studies used placebo as a control group, and the 1 study on ralmitaront also used the antipsychotic risperidone as an active comparator. *One study* comparing daily doses of ulotaront in Parkinson's disease psychosis to placebo for 4 weeks [@RN13].

*Four studies* examined the pharmacokinetics, safety and/or neuroimaging of single doses of TAAR1 agonists or multiple doses for up to 2 weeks: *One study* examining the effects of a single dose of 150mg ulotaront in clinically stable adults with schizophrenia on QTc interval within a day [@RN124]. The study utilized a three-period crossover period comparing ulotaront with placebo and moxifloxacin (an antibiotic used as active comparator, not relevant for our analysis). *Two studies* examined the effects of single doses of ulotaront in healthy volunteers on sleep parameters [@RN116] or fMRI tasks [@RN115]. @RN116 included two cohorts of crossover studies comparing two single doses of ulotaront with placebo. [@RN115] was a parallel trial comparing ulotaront to placebo or the antipsychotic amisulpride on fMRI tasks relevant to reward, working memory and emotion processing in volunteers with high or low schizotypy traits. *One study* examined two single doses of ulotaront in adults with narcolepsy-cateplexy compared to placebo using a three-period crossover design [@RN123].

Data from the following studies were unavailable: *One study* examined the efficacy and safety of ulotaront in comparison to placebo in adolescents with an acute episode schizophrenia (NCT04072354). This study had two parts with adults and adolescents. This study has been recently completed, and although some usable data in the adult population are available (see above), these are not available for the adolescents. *One study* examined the long-term efficacy and safety of ulotaront in comparison to the antipsychotic quetiapine in adults with clinically stable schizophrenia (NCT04115319). The study should have been completed in 12.2022, but there are currently no usable data available. *One study* examined monotherapy or add-on treatment with ralmitaront in comparison to placebo in adults with schizophrenia or schizoaffective disorder and predominant negative symptoms (NCT03669640). The study has been recently completed in 05.2023, and it has been reported to be failed, yet no usable data were available. *Nine studies* examined pharmacokinetics and safety of single or few multiple doses of ulotaront or ralmitaront in healthy volunteers (Fowler 2015, JapicCTI-194581, NCT02699372, SEP361-101, DA801002, NCT01940159, Koblan 2016, NCT04325737, NCT04865835).

The following studies are ongoing: - *One study* is examining the efficacy and safety of two ulotaront doses of 50mg/d and 75mg/d in comparison to placebo in adults with an acute episode of schizophrenia (NCT04825860). The study is expected to be completed in 06.2025. - *Two studies* are examining the efficacy and safety of ulotaront compared to placebo in other mental health conditions, i.e., adults with major depression (NCT05593029) and generalized anxiety disorder (NCT05729373). These trials are expected to be completed in 2025, and their safety data are relevant for this review.

The studies used to describe mechanistic insights are provided in the relevant section below. It should be noted that the findings from an additional single-arm molecular neuroimaging are described in this section [@RN132], which was uncontrolled and formally excluded from this iteration of the review. The characteristics of this study and its findings are described in the "mechanistic insights" section.

## 2.4 Risk of bias assessment

The risk of bias of the studies was examined using the RoB2 tool [@RN64] for parallel and crossover studies for all outcomes. Here, we present the risk of bias of using the worst judgement of a domain across outcomes. It should be noted that when there were some concerns in only one of the domains, we assigned an overall low risk of bias according to our protocol.

```{r echo =F, fig.fullwidth=T, fig.width=7.2, fig.height=6}
rob_traffic_light(rob_all, tool="Generic")
```

**Figure 2** Risk of bias of studies with usable data

All the studies were double-blind randomized controlled studies, with a low risk of bias in most cases in their respective domains of biases due to randomization, deviations from intended interventions, and outcome measurement. There were also no indications of selectively reported results; thus, the studies had a low risk of bias in their respective domains. Most of the studies had missing outcome data, but there was no clear indication that the missingness depended on the true value of the effects for most of the outcomes. Therefore, there were some concerns, except for @RN13, which had a high risk of bias for efficacy outcomes but not for adverse events and other outcomes. Additionally, there were few differences in this domain across outcomes, such as a low risk for the outcome of dropouts. In terms of the three crossover studies, there was a low risk of bias in period and carry-over effects, except for some concerns in the three-period crossover study of @RN124 . Few differences across outcomes can be found in the overall judgment of the risk of bias for each study per outcome in the respective forest plots below.

**It should be noted that the x-axis in the figures below, which present dichotomous outcomes using odds ratios, is logarithmically scaled.** **When the forest plots present only the effect sizes, there were crossover studies in which variances were corrected assuming a correlation of 0.2.**

## 2.5 Comparison 1: *TAAR1 agonist* vs *placebo*

### 2.5.1 Primary outcome: Overall symptoms

#### Timepoint 1: *1 day - 2 weeks*

No usable data available from RCT.

#### Timepoint 2: *3 weeks - 13 weeks* (primary timepoint)

```{r echo=F, fig.fullwidth=T, fig.width=12, fig.height=4}
forest(meta_comp_overall_sch, 
       col.diamond = "royalblue4",
       col.square =   "royalblue1",
                 rightcols = c("effect", "ci","Overall_overall"), 
                 rightlabs = c("SMD", "95%-CI", "Risk of bias"),  
       xlim=c(-0.8,0.8),
                 label.e = unique(meta_comp_overall_sch$data$treat1_new), 
       label.c=unique(meta_comp_overall_sch$data$treat2_new),
                 subgroup.hetstat = FALSE, overall.hetstat =TRUE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = TRUE, prediction.subgroup = FALSE,
                 subgroup = FALSE, fixed=TRUE,
                 label.left = "Placebo better", label.right="TAAR1 better")
```

**Figure 3a** Forest plot for overall symptoms (primary outcome, measured with PANSS total) for the comparison of TAAR1 agonist vs. placebo in people with acute episode of schizophrenia at 3-13 weeks (primary timepoint). The overall risk of bias of the studies for this outcome is presented. SMD:Standardized mean difference, 95%CI: 95% confidence intervals, SD: Standard deviation.

```{r echo=F, fig.fullwidth=T, fig.width=12, fig.height=4}
forest(meta_comp_overall_pdp, 
       col.diamond = "royalblue4",
       col.square =   "royalblue1",
                 rightcols = c("effect", "ci","Overall_overall"), 
                 rightlabs = c("SMD", "95%-CI", "Risk of bias"),  
       xlim=c(-0.8,0.8),
                 label.e = unique(meta_comp_overall_pdp$data$treat1_new), 
       label.c=unique(meta_comp_overall_pdp$data$treat2_new),
                 subgroup.hetstat = FALSE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = FALSE, prediction.subgroup = FALSE,
                 subgroup = FALSE, fixed=TRUE,
                 label.left = "Placebo better", label.right="TAAR1 better")
```

**Figure 3b** Forest plot for overall symptoms (primary outcome, measured with the NPI total score) for the comparison of TAAR1 agonist vs. placebo in people with Parkinson's disease psychosis at 3-13 weeks (primary timepoint). The overall risk of bias of the studies for this outcome is presented. SMD:Standardized mean difference, 95%CI: 95% confidence intervals, SD: Standard deviation.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.5.2 Secondary outcome: Positive symptoms

#### Timepoint 1: *1 day - 2 weeks*

No usable data available from RCT.

#### Timepoint 2: *3 weeks - 13 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}
forest(meta_comp_positive_sch, 
       col.diamond = "royalblue4",
       col.square =   "royalblue1",
                 rightcols = c("effect", "ci", "Overall_positive"), 
                 rightlabs = c("SMD", "95%-CI", "Risk of bias"),  xlim=c(-0.8,0.8),
                 label.e = unique(meta_comp_positive_sch$data$treat1_new),
       label.c=unique(meta_comp_positive_sch$data$treat2_new),
                 subgroup.hetstat = TRUE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = FALSE, prediction.subgroup = FALSE,
                 subgroup = TRUE, fixed=TRUE,
                 label.left = "Placebo better", label.right="TAAR1 better")
```

**Figure 4a** Forest plot for positive symptoms for the comparison of TAAR1 agonist vs. placebo in people with acute episode of schizophrenia (measured with PANSS positive) at 3-13 weeks (primary timepoint). The overall risk of bias of the studies for this outcome is presented. SMD:Standardized mean difference, 95%CI: 95% confidence intervals, SD: Standard deviation.

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}
forest(meta_comp_positive_pdp, 
       col.diamond = "royalblue4",
       col.square =   "royalblue1",
                 rightcols = c("effect", "ci", "Overall_positive"), 
                 rightlabs = c("SMD", "95%-CI", "Risk of bias"),  xlim=c(-0.8,0.8),
                 label.e = unique(meta_comp_positive_pdp$data$treat1_new),
       label.c=unique(meta_comp_positive_pdp$data$treat2_new),
                 subgroup.hetstat = TRUE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = FALSE, prediction.subgroup = FALSE,
                 subgroup = TRUE, fixed=TRUE,
                 label.left = "Placebo better", label.right="TAAR1 better")
```

**Figure 4b** Forest plot for positive symptoms for the comparison of TAAR1 agonist vs. placebo in people with Parkinson's disease psychosis (measured with SAPS-PD) at 3-13 weeks (primary timepoint). The overall risk of bias of the studies for this outcome is presented. SMD:Standardized mean difference, 95%CI: 95% confidence intervals, SD: Standard deviation.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.5.3 Secondary outcome: Negative symptoms

#### Timepoint 1: *1 day - 2 weeks*

No usable data available from RCT.

#### Timepoint 2: *3 weeks - 13 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}
forest(`meta_comp_taar1_vs_placebo_3-13 weeks_negative`, 
       col.diamond = "royalblue4",
       col.square =   "royalblue1",
                 rightcols = c("effect", "ci", "Overall"), 
                 rightlabs = c("SMD", "95%-CI", "Risk of bias"),  
       xlim=c(-0.8,0.8),
                 label.e = unique(`meta_comp_taar1_vs_placebo_3-13 weeks_negative`$data$treat1_new), 
       label.c=unique(`meta_comp_taar1_vs_placebo_3-13 weeks_negative`$data$treat2_new),
                 subgroup.hetstat = TRUE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = FALSE, prediction.subgroup = FALSE,
                 subgroup = TRUE, fixed=TRUE,
                 label.left = "Placebo better", label.right="TAAR1 better")

```

**Figure 5** Forest plot for negative symptoms for the comparison of TAAR1 agonist vs. placebo (measured with PANSS negative) at 3-13 weeks (primary timepoint). The overall risk of bias of the studies for this outcome is presented. SMD:Standardized mean difference, 95%CI: 95% confidence intervals, SD: Standard deviation.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.5.4 Secondary outcome: Depressive symptoms

#### Timepoint 1: *1 day - 2 weeks*

No usable data available from RCT.

#### Timepoint 2: *3 weeks - 13 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}

forest(`meta_comp_taar1_vs_placebo_3-13 weeks_depression`, 
       col.diamond = "royalblue4",
       col.square =   "royalblue1",
                 rightcols = c("effect", "ci", "Overall"), 
                 rightlabs = c("SMD", "95%-CI", "Risk of bias"),  xlim=c(-0.5,0.5),
                 label.e = unique(`meta_comp_taar1_vs_placebo_3-13 weeks_depression`$data$treat1_new),  label.c=unique(`meta_comp_taar1_vs_placebo_3-13 weeks_depression`$data$treat2_new),
                 subgroup.hetstat = TRUE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = FALSE, prediction.subgroup = FALSE,
                 subgroup = TRUE, fixed=TRUE,
                 label.left = "Placebo better", label.right="TAAR1 better")
```

**Figure 6** Forest plot for depressive symptoms for the comparison of TAAR1 agonist vs. placebo (measured with PANSS anxiety/depression factor) at 3-13 weeks (primary timepoint). The overall risk of bias of the studies for this outcome is presented. SMD:Standardized mean difference, 95%CI: 95% confidence intervals, SD: Standard deviation.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.5.5 Secondary outcome: Cognitive symptoms

#### Timepoint 1: *1 day - 2 weeks*

No usable data available from RCT.

#### Timepoint 2: *3 weeks - 13 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}

forest(`meta_comp_taar1_vs_placebo_3-13 weeks_cognition`, 
       col.diamond = "royalblue4",
       col.square =   "royalblue1",
                 rightcols = c("effect", "ci", "Overall"), 
                 rightlabs = c("SMD", "95%-CI", "Risk of bias"),  xlim=c(-0.8,0.8),
                 label.e = unique(`meta_comp_taar1_vs_placebo_3-13 weeks_cognition`$data$treat1_new),  label.c=unique(`meta_comp_taar1_vs_placebo_3-13 weeks_cognition`$data$treat2_new),
                 subgroup.hetstat = TRUE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = FALSE, prediction.subgroup = FALSE,
                 subgroup = TRUE, fixed=TRUE,
                 label.left = "Placebo better", label.right="TAAR1 better")
```

**Figure 7** Forest plot for cognitive impairment for the comparison of TAAR1 agonist vs. placebo (measured with MMSE) at 3-13 weeks (primary timepoint). The overall risk of bias of the studies for this outcome is presented. SMD:Standardized mean difference, 95%CI: 95% confidence intervals, SD: Standard deviation.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.5.6 Secondary outcome: Quality of life

#### Timepoint 1: *1 day - 2 weeks*

No usable data available from RCT.

#### Timepoint 2: *3 weeks - 13 weeks*

No usable data available from RCT.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.5.7 Secondary outcome: Functioning

#### Timepoint 1: *1 day - 2 weeks*

No usable data available from RCT.

#### Timepoint 2: *3 weeks - 13 weeks*

No usable data available from RCT.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.5.8 Secondary outcome: Response to treatment

#### Timepoint 1: *1 day - 2 weeks*

No usable data available from RCT.

#### Timepoint 2: *3 weeks - 13 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}
forest(meta_comp_response_sch, 
       col.diamond = "royalblue4",
       col.square =   "royalblue1",
                 rightcols = c("effect", "ci", "Overall_response"), 
                 rightlabs = c("OR", "95%-CI", "Risk of bias"),  xlim=c(0.2,5),
                 label.e = unique(meta_comp_response_sch$data$treat1_new),
       label.c="placebo",
                 subgroup.hetstat = TRUE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = FALSE, prediction.subgroup = FALSE,
                 subgroup = TRUE, fixed=TRUE,
                 label.left = "Placebo better", label.right="TAAR1 better")

```

**Figure 8a** Forest plot for response (\>50% or 20% symptom reduction in PANSS total from baseline) for the comparison of TAAR1 agonist vs. placebo in people with acute episode of schizophrenia at 3-13 weeks (primary timepoint). The overall risk of bias of the studies for this outcome is presented. OR=Odds ratio, 95%CI: 95% confidence intervals.

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}
forest(meta_comp_response_pdp, 
       col.diamond = "royalblue4",
       col.square =   "royalblue1",
                 rightcols = c("effect", "ci", "Overall_response"), 
                 rightlabs = c("OR", "95%-CI", "Risk of bias"),  xlim=c(0.2,5),
                 label.e = unique(meta_comp_response_pdp$data$treat1_new),
       label.c="placebo",
                 subgroup.hetstat = TRUE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = FALSE, prediction.subgroup = FALSE,
                 subgroup = TRUE, fixed=TRUE,
                 label.left = "Placebo better", label.right="TAAR1 better")

```

**Figure 8b** Forest plot for response (\>50% symptom reduction in SAPS-PD from baseline) for the comparison of TAAR1 agonist vs. placebo in people with Parkinson's disease psychosis at 3-13 weeks (primary timepoint). The overall risk of bias of the studies for this outcome is presented. OR=Odds ratio, 95%CI: 95% confidence intervals.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.5.9 Secondary outcome: Relapse

#### Timepoint 1: *1 day - 2 weeks*

No usable data available from RCT.

#### Timepoint 2: *3 weeks - 13 weeks*

No usable data available from RCT.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.5.10 Summary outcome plot: Efficacy outcomes

```{r echo=F, fig.fullwidth=T, fig.width=10, fig.height=5, warning=FALSE}

outcome_categories<-outcome_names %>% 
  select(outcome, event_text) 

efficacy_primary<-meta_outcome %>% left_join(outcome_categories) %>% 
  filter(outcome=="cognition" | outcome=="depression" | outcome=="negative" | outcome=="overall" | outcome=="positive") %>% #efficacy outcome
  filter(!is.na(TE.random)) %>% filter(!is.na(event_text)) %>%
  mutate(point=TE.random,
         lb=TE.random-1.96*seTE.random,
         ub=TE.random+1.96*seTE.random) %>% 
  unique() %>%
  filter(comparison=="taar1_vs_placebo") %>% 
  mutate(n_text=paste0("N=", k,", n=",n))

efficacy_primary$population <- factor(efficacy_primary$population, levels=unique(efficacy_primary$population))
efficacy_primary$event_text <- factor(efficacy_primary$event_tex, levels=unique(efficacy_primary$event_tex))
level_outcomes<- c("Overall symptoms", "Positive symptoms", "Negative symptoms", "Depressive symptoms", "Cognitive impairement")
level_population<-c("Schizophrenia spectrum", "Parkinson Disease Psychosis")

ggplot(data=efficacy_primary, 
       aes(y=factor(event_text, level=rev(level_outcomes)), x=point))+
  geom_text(aes(x=0.65, label=n_text), colour="black" , position=position_dodge(width = 0.9))+
  geom_point(aes(size=n), position=position_dodge(width = 0.5), colour="royalblue4")+ #Size proportional to the number of participants
  coord_cartesian(xlim=c(-0.8,0.8))+
  geom_errorbar(aes(x=point, xmin=lb, 
                    xmax=ub),
                width=0, position=position_dodge(width = 0.5), colour="royalblue4")+
  scale_size_continuous(guide=FALSE)+
  geom_vline(xintercept=0, linetype='dashed')+
  facet_grid(~factor(population, levels=level_population))+
  ylab("")+
  xlab("SMD 95%CI (SMD>0: TAAR1 agonist better)")+
  theme_bw()


```

**Figure 9** Summary plot for efficacy outcomes at 4-6 weeks for the comparison of TAAR1 agonists vs. placebo in the two populations of patients with an acute episode of schizophrenia and Parkinson's disease psychosis. There were only available data for the timepoint of 3-13 weeks. The size of the points are proportional to the number of participants used in the analysis. SMD\>0 correspond to more symptom improvement by TAAR1 agonists compared to placebo. The findings of the random-effects model are presented. SMD: Standardized mean difference, 95%CI: 95% confidence interval. N=number of studies, n=number of participants.

### 2.5.11 Secondary outcome: Dropouts due to any reason

#### Timepoint 1: *1 day - 2 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}
forest(`meta_comp_taar1_vs_placebo_1 day-2 weeks_dropout_any`, 
       col.diamond = "green4",
       col.square =   "green3",
         leftcols = c("studlab",  "effect", "ci"),   
       rightcols = c("Overall"), rightlabs = c("Risk of bias"),   
                 xlim=c(0.2,5),
                 label.e = unique(`meta_comp_taar1_vs_placebo_1 day-2 weeks_dropout_any`$data$treat1_new), 
                 subgroup.hetstat = TRUE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = FALSE, prediction.subgroup = FALSE,
                 subgroup = FALSE, fixed=TRUE, random=TRUE,
                 label.left = "TAAR1 better", label.right="Placebo better")
```

**Figure 10** Forest plot for dropouts due to any reason for the comparison of TAAR1 agonist vs. placebo at 1 day - 2 weeks. Studies with no estimable effect size had 0 events in both arms. The overall risk of bias of the studies for this outcome is presented. OR=Odds ratio, 95%CI: 95% confidence intervals.

#### Timepoint 2: *3 weeks - 13 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=10, fig.height=5}
forest(`meta_comp_taar1_vs_placebo_3-13 weeks_dropout_any`, 
       col.diamond = "green4",
       col.square =   "green3",     
       label.c="placebo",
       rightcols = c("effect", "ci", "Overall"),rightlabs = c("OR", "95%-CI", "Risk of bias"),
                   xlim=c(0.2,5),
                 label.e = unique(`meta_comp_taar1_vs_placebo_3-13 weeks_dropout_any`$data$treat1_new), 
                 subgroup.hetstat = TRUE, overall.hetstat =TRUE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = TRUE, prediction.subgroup = FALSE, prediction = FALSE,
                 subgroup = TRUE, fixed=TRUE, random=TRUE,
                 label.left = "TAAR1 better", label.right="Placebo better")

```

**Figure 11** Forest plot for dropouts due to any reason for the comparison of TAAR1 agonist vs. placebo at 3-13 weeks (primary timepoint). The overall risk of bias of the studies for this outcome is presented. OR=Odds ratio, 95%CI: 95% confidence intervals.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.5.12 Secondary outcome: Dropouts due to adverse event

#### Timepoint 1: *1 day - 2 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}
forest(`meta_comp_taar1_vs_placebo_1 day-2 weeks_dropout_ae`, 
       col.diamond = "green4",
       col.square =   "green3",     
       label.c="placebo",
       leftcols = c("studlab",  "effect", "ci"),      
       rightcols = c("Overall"), rightlabs = c("Risk of bias"),   
                 label.e = unique(`meta_comp_taar1_vs_placebo_1 day-2 weeks_dropout_ae`$data$treat1_new), 
                 subgroup.hetstat = FALSE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = FALSE, prediction.subgroup = FALSE,
                 subgroup = FALSE, fixed=TRUE, random=TRUE,
                 label.left = "TAAR1 better", label.right="Placebo better")


```

**Figure 12** Forest plot for dropouts due to adverse event for the comparison of TAAR1 agonist vs. placebo at 1 day - 2 weeks. Studies with no estimable effect size had 0 events in both arms. The overall risk of bias of the studies for this outcome is presented. OR=Odds ratio, 95%CI: 95% confidence intervals.

#### Timepoint 2: *3 weeks - 13 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}
forest(`meta_comp_taar1_vs_placebo_3-13 weeks_dropout_ae`, 
       col.diamond = "green4",
       col.square =   "green3",     
       label.c="placebo",
       rightcols = c("effect", "ci", "Overall"),rightlabs = c("OR", "95%-CI", "Risk of bias"),
                   xlim=c(0.2,5),
                 label.e = unique(`meta_comp_taar1_vs_placebo_3-13 weeks_dropout_ae`$data$treat1_new), 
                 subgroup.hetstat = TRUE, overall.hetstat =TRUE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = TRUE, prediction.subgroup = FALSE, prediction = FALSE,
                 subgroup = TRUE, fixed=TRUE, random=TRUE,
                 label.left = "TAAR1 better", label.right="Placebo better")
```

**Figure 13** Forest plot for dropouts due to adverse event for the comparison of TAAR1 agonist vs. placebo at 3-13 weeks (primary timepoint). The overall risk of bias of the studies for this outcome is presented. OR=Odds ratio, 95%CI: 95% confidence intervals.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.5.13 Secondary outcome: Serious adverse event

#### Timepoint 1: *1 day - 2 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}
forest(`meta_comp_taar1_vs_placebo_1 day-2 weeks_serious`, 
       col.diamond = "firebrick4",
       col.square =   "firebrick3",     
       label.c="placebo",          
       leftcols = c("studlab",  "effect", "ci"),   
       rightcols = c("Overall"), rightlabs = c("Risk of bias"),   
                  xlim=c(0.2,5),
                 label.e = unique(`meta_comp_taar1_vs_placebo_1 day-2 weeks_serious`$data$treat1_new), 
                 subgroup.hetstat = TRUE, overall.hetstat =TRUE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = TRUE, prediction.subgroup = FALSE,
                 subgroup = TRUE, fixed=TRUE, random=TRUE,
                 label.left = "TAAR1 better", label.right="Placebo better")
```

**Figure 14** Forest plot for serious adverse events for the comparison of TAAR1 agonist vs. placebo at 1 day - 2 weeks. Studies with no estimable effect size had 0 events in both arms. The overall risk of bias of the studies for this outcome is presented. OR=Odds ratio, 95%CI: 95% confidence intervals.

#### Timepoint 2: *3 weeks - 13 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}
forest(`meta_comp_taar1_vs_placebo_3-13 weeks_serious`, 
       col.diamond = "firebrick4",
       col.square =   "firebrick3",     
       label.c="placebo",          
       rightcols = c("effect", "ci", "Overall"),rightlabs = c("OR", "95%-CI", "Risk of bias"),
                   xlim=c(0.2,5),
                 label.e = unique(`meta_comp_taar1_vs_placebo_3-13 weeks_serious`$data$treat1_new), 
                 subgroup.hetstat = TRUE, overall.hetstat =TRUE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = TRUE, prediction.subgroup = FALSE, prediction = FALSE,
                 subgroup = TRUE, fixed=TRUE, random=TRUE,
                 label.left = "TAAR1 better", label.right="Placebo better")

```

**Figure 15** Forest plot for serious adverse events for the comparison of TAAR1 agonist vs. placebo at 3-13 weeks (primary timepoint). The overall risk of bias of the studies for this outcome is presented. OR=Odds ratio, 95%CI: 95% confidence intervals

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.5.14 Secondary outcome: Death

#### Timepoint 1: *1 day - 2 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}
#forest(`meta_comp_taar1_vs_placebo_1 day-2 weeks_death`, 
 #      col.diamond = "firebrick4",
  #     col.square =   "firebrick3",     
   #    label.c="placebo",      
    #   leftcols = c("studlab",  "effect", "ci"),           
     #  rightcols = FALSE,
      #            xlim=c(0.2,5),
       #          label.e = unique(`meta_comp_taar1_vs_placebo_1 day-2 weeks_death`$data$treat1_new), 
        #         subgroup.hetstat = TRUE, overall.hetstat =TRUE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
         #        overall = TRUE, prediction.subgroup = FALSE,
          #       subgroup = TRUE, fixed=TRUE, random=TRUE,
           #      label.left = "TAAR1 better", label.right="Placebo better")

```

There were available usable data from four studies (i.e., [@RN116; @RN115; @RN123; @RN124]) but all of them reported 0 deaths so that could not provide estimable effect sizes with the current approach.

#### Timepoint 2: *3 weeks - 13 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}
forest(`meta_comp_taar1_vs_placebo_3-13 weeks_death`, 
       col.diamond = "firebrick4",
       col.square =   "firebrick3",     
       label.c="placebo",      
       rightcols = c("effect", "ci", "Overall"),rightlabs = c("OR", "95%-CI", "Risk of bias"),
                 label.e = unique(`meta_comp_taar1_vs_placebo_3-13 weeks_death`$data$treat1_new), 
                 subgroup.hetstat = FALSE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = FALSE, prediction.subgroup = FALSE, prediction = FALSE,
                 subgroup = FALSE, fixed=TRUE, random=TRUE,
                 label.left = "TAAR1 better", label.right="Placebo better")
```

**Figure 16** Forest plot for death for the comparison of TAAR1 agonist vs. placebo at 3-13 weeks (primary timepoint). Studies with 0 events in both arms could not provide estimable effect sizes with the current approach. The overall risk of bias of the studies for this outcome is presented. OR=Odds ratio, 95%CI: 95% confidence intervals.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.5.15 Secondary outcome: Any adverse event

#### Timepoint 1: *1 day - 2 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=5}
forest(`meta_comp_taar1_vs_placebo_1 day-2 weeks_adverse_event`, 
       col.diamond = "firebrick4",
       col.square =   "firebrick3",     
       label.c="placebo",      
        rightcols = c("Overall"), rightlabs = c("Risk of bias"),   
       leftcols = c("studlab",  "effect", "ci"),           
                 label.e = unique(`meta_comp_taar1_vs_placebo_1 day-2 weeks_adverse_event`$data$treat1_new), 
                 subgroup.hetstat = TRUE, overall.hetstat =TRUE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = TRUE, prediction.subgroup = FALSE, prediction=FALSE,
                 subgroup = TRUE, fixed=TRUE, random=TRUE,
                 label.left = "TAAR1 better", label.right="Placebo better")

```

**Figure 17** Forest plot for any adverse events for the comparison of TAAR1 agonist vs. placebo at 1 day - 2 weeks.The overall risk of bias of the studies for this outcome is presented. OR=Odds ratio, 95%CI: 95% confidence intervals

#### Timepoint 2: *3 weeks - 13 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}
forest(`meta_comp_taar1_vs_placebo_3-13 weeks_adverse_event`, 
       col.diamond = "firebrick4",
       col.square =   "firebrick3",     
       label.c="placebo",      
       rightcols = c("effect", "ci", "Overall"),rightlabs = c("OR", "95%-CI", "Risk of bias"),
                 label.e = unique(`meta_comp_taar1_vs_placebo_3-13 weeks_adverse_event`$data$treat1_new), 
                 subgroup.hetstat = TRUE, overall.hetstat =TRUE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = TRUE, prediction.subgroup = FALSE, prediction = FALSE,
                 subgroup = TRUE, fixed=TRUE, random=TRUE,
                 label.left = "TAAR1 better", label.right="Placebo better")

```

**Figure 18** Forest plot for any adverse event for the comparison of TAAR1 agonist vs. placebo at 3-13 weeks (primary timepoint). The overall risk of bias of the studies for this outcome is presented. OR=Odds ratio, 95%CI: 95% confidence intervals.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.5.16 Secondary outcome: Anticholinergic symptoms

#### Timepoint 1: *1 day - 2 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}
forest(`meta_comp_taar1_vs_placebo_1 day-2 weeks_anticholinergic_symptom`, 
       col.diamond = "firebrick4",
       col.square =   "firebrick3",     
       label.c="placebo",      
       leftcols = c("studlab",  "effect", "ci"),          
        rightcols = c("Overall"), rightlabs = c("Risk of bias"),   
                 label.e = unique(`meta_comp_taar1_vs_placebo_1 day-2 weeks_anticholinergic_symptom`$data$treat1_new), 
                 subgroup.hetstat = TRUE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = FALSE, prediction.subgroup = FALSE, prediction=FALSE,
                 subgroup = FALSE, fixed=TRUE, random=TRUE,
                 label.left = "TAAR1 better", label.right="Placebo better")

```

**Figure 19** Forest plot for anticholinergic side-effects (i.e., dry mouth in this study) for the comparison of TAAR1 agonist vs. placebo at 1 day - 2 weeks. The overall risk of bias of the studies for this outcome is presented. OR=Odds ratio, 95%CI: 95% confidence intervals.

#### Timepoint 2: *3 weeks - 13 weeks*

No usable data available from RCT.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.5.17 Secondary outcome: Hypotension

#### Timepoint 1: *1 day - 2 weeks*

No usable data available from RCT.

#### Timepoint 2: *3 weeks - 13 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}
forest(`meta_comp_taar1_vs_placebo_3-13 weeks_hypotension`, 
       col.diamond = "firebrick4",
       col.square =   "firebrick3",     
       label.c="placebo",      
       rightcols = c("effect", "ci", "Overall"),rightlabs = c("OR", "95%-CI", "Risk of bias"),
                 label.e = unique(`meta_comp_taar1_vs_placebo_3-13 weeks_hypotension`$data$treat1_new), 
                 subgroup.hetstat = FALSE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = FALSE, prediction.subgroup = FALSE, prediction = FALSE,
                 subgroup = TRUE, fixed=TRUE, random=TRUE,
                 label.left = "TAAR1 better", label.right="Placebo better")

```

**Figure 20** Forest plot for hypotension for the comparison of TAAR1 agonist vs. placebo at 3-13 weeks (primary timepoint). The overall risk of bias of the studies for this outcome is presented. OR=Odds ratio, 95%CI: 95% confidence intervals.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.5.18 Secondary outcome: Dizziness

#### Timepoint 1: *1 day - 2 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}
forest(`meta_comp_taar1_vs_placebo_1 day-2 weeks_dizziness`, 
       col.diamond = "firebrick4",
       col.square =   "firebrick3",     
       label.c="placebo",      
       leftcols = c("studlab",  "effect", "ci"),          
        rightcols = c("Overall"), rightlabs = c("Risk of bias"),   
                 label.e = unique(`meta_comp_taar1_vs_placebo_1 day-2 weeks_dizziness`$data$treat1_new), 
                 subgroup.hetstat = TRUE, overall.hetstat =TRUE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = TRUE, prediction.subgroup = FALSE, prediction=FALSE,
                 subgroup = TRUE, fixed=TRUE, random=TRUE,
                 label.left = "TAAR1 better", label.right="Placebo better")

```

**Figure 21** Forest plot for dizziness for the comparison of TAAR1 agonist vs. placebo at 1 day - 2 weeks. The overall risk of bias of the studies for this outcome is presented. OR=Odds ratio, 95%CI: 95% confidence intervals.

#### Timepoint 2: *3 weeks - 13 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}
forest(`meta_comp_taar1_vs_placebo_3-13 weeks_dizziness`, 
       col.diamond = "firebrick4",
       col.square =   "firebrick3",     
       label.c="placebo",      
       rightcols = c("effect", "ci", "Overall"),rightlabs = c("OR", "95%-CI", "Risk of bias"),
                 label.e = unique(`meta_comp_taar1_vs_placebo_3-13 weeks_dizziness`$data$treat1_new), 
                 subgroup.hetstat = TRUE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = FALSE, prediction.subgroup = FALSE, prediction = FALSE,
                 subgroup = TRUE, fixed=TRUE, random=TRUE,
                 label.left = "TAAR1 better", label.right="Placebo better")

```

**Figure 22** Forest plot for dizziness for the comparison of TAAR1 agonist vs. placebo at 3-13 weeks (primary timepoint). The overall risk of bias of the studies for this outcome is presented. OR=Odds ratio, 95%CI: 95% confidence intervals.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.5.19 Secondary outcome: Nausea or vomiting

#### Timepoint 1: *1 day - 2 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}
forest(`meta_comp_taar1_vs_placebo_1 day-2 weeks_nausea_vomitting`, 
       col.diamond = "firebrick4",
       col.square =   "firebrick3",     
       label.c="placebo",      
       leftcols = c("studlab",  "effect", "ci"),          
        rightcols = c("Overall"), rightlabs = c("Risk of bias"),  
                 label.e = unique(`meta_comp_taar1_vs_placebo_1 day-2 weeks_nausea_vomitting`$data$treat1_new), 
                 subgroup.hetstat = TRUE, overall.hetstat =TRUE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = TRUE, prediction.subgroup = FALSE, prediction=FALSE,
                 subgroup = TRUE, fixed=TRUE, random=TRUE,
                 label.left = "TAAR1 better", label.right="Placebo better")

```

**Figure 23** Forest plot for nausea or vomiting for the comparison of TAAR1 agonist vs. placebo at 1 day - 2 weeks. The overall risk of bias of the studies for this outcome is presented. OR=Odds ratio, 95%CI: 95% confidence intervals.

#### Timepoint 2: *3 weeks - 13 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}
forest(`meta_comp_taar1_vs_placebo_3-13 weeks_nausea_vomitting`, 
       col.diamond = "firebrick4",
       col.square =   "firebrick3",     
       label.c="placebo",      
       rightcols = c("effect", "ci", "Overall"),rightlabs = c("OR", "95%-CI", "Risk of bias"),
                 label.e = unique(`meta_comp_taar1_vs_placebo_3-13 weeks_nausea_vomitting`$data$treat1_new), 
                 subgroup.hetstat = TRUE, overall.hetstat =TRUE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = TRUE, prediction.subgroup = FALSE, prediction = FALSE,
                 subgroup = TRUE, fixed=TRUE, random=TRUE,
                 label.left = "TAAR1 better", label.right="Placebo better")


```

**Figure 24** Forest plot for nausea or vomiting for the comparison of TAAR1 agonist vs. placebo at 3-13 weeks (primary timepoint). The overall risk of bias of the studies for this outcome is presented. OR=Odds ratio, 95%CI: 95% confidence intervals.

It should be noted that there appears to be some heterogeneity in the findings between ulotaront and ralmitaront. This heterogeneity may potentially be driven by the fact that ulotaront is also a 5-HT1A receptor partial agonist, which is associated with gastrointestinal symptoms.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.5.20 Secondary outcome: QTc prolongation

#### Timepoint 1: *1 day - 2 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}
forest(`meta_comp_taar1_vs_placebo_1 day-2 weeks_qtc_prolongation`, 
       col.diamond = "firebrick4",
       col.square =   "firebrick3",     
       label.c="placebo",      
       leftcols = c("studlab",  "effect", "ci"),          
        rightcols = c("Overall"), rightlabs = c("Risk of bias"),  
                 label.e = unique(`meta_comp_taar1_vs_placebo_1 day-2 weeks_qtc_prolongation`$data$treat1_new), 
                 subgroup.hetstat = TRUE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = FALSE, prediction.subgroup = FALSE, prediction=FALSE,
                 subgroup = TRUE, fixed=TRUE, random=TRUE,
                 label.left = "TAAR1 better", label.right="Placebo better")

```

**Figure 25** Forest plot for QTc prolongation for the comparison of TAAR1 agonist vs. placebo at 1 day - 2 weeks. Studies with not estimable effect sizes had 0 events in both groups. The overall risk of bias of the studies for this outcome is presented. OR=Odds ratio, 95%CI: 95% confidence intervals.

#### Timepoint 2: *3 weeks - 13 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}

#forest(`meta_comp_taar1_vs_placebo_3-13 weeks_qtc_prolongation`, 
 #      col.diamond = "firebrick4",
  #     col.square =   "firebrick3",     
   #    label.c="placebo",      
    #   rightcols = c("effect", "ci", "Overall"),rightlabs = c("OR", "95%-CI", "Risk of bias"),
     #            label.e = unique(`meta_comp_taar1_vs_placebo_3-13 weeks_qtc_prolongation`$data$treat1_new), 
      #           subgroup.hetstat = TRUE, overall.hetstat =TRUE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
       #          overall = TRUE, prediction.subgroup = FALSE, prediction = FALSE,
        #         subgroup = TRUE, fixed=TRUE, random=TRUE,
         #        label.left = "TAAR1 better", label.right="Placebo better")

```

There were available data in @RN110 , which reported 0 events in both groups, so that an effect size cannot be estimated with the current approach.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.5.21 Secondary outcome: QTc interval in msec

#### Timepoint 1: *1 day - 2 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}
forest(`meta_comp_taar1_vs_antipsychotic_1 day-2 weeks_qtc_interval`, 
       col.diamond = "firebrick4",
       col.square =   "firebrick3",     
       label.c="placebo",      
       leftcols = c("studlab",  "effect", "ci"),   leftlabs = c("Study", "MD msec", "95%-CI"),       
        rightcols = c("Overall"), rightlabs = c("Risk of bias"),  
                 label.e = unique(`meta_comp_taar1_vs_antipsychotic_1 day-2 weeks_qtc_interval`$data$treat1_new), 
                 subgroup.hetstat = FALSE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = FALSE, prediction.subgroup = FALSE, prediction=FALSE,
                 subgroup = FALSE, fixed=TRUE, random=TRUE,
                 label.left = "TAAR1 better", label.right="Placebo better")
```

**Figure 26** Forest plot for QTc interval in msec for the comparison of TAAR1 agonist vs. placebo at 1 day - 2 weeks. The overall risk of bias of the studies for this outcome is presented. MD:Mean difference, 95%CI: 95% confidence intervals

#### Timepoint 2: *3 weeks - 13 weeks*

No usable data available from RCT.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.5.22 Secondary outcome: Weight increased

#### Timepoint 1: *1 day - 2 weeks*

No usable data available from RCT.

#### Timepoint 2: *3 weeks - 13 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}
forest(`meta_comp_taar1_vs_placebo_3-13 weeks_weight_increased`, 
       col.diamond = "firebrick4",
       col.square =   "firebrick3",     
       label.c="placebo",      
       rightcols = c("effect", "ci", "Overall"),rightlabs = c("OR", "95%-CI", "Risk of bias"),
                 label.e = unique(`meta_comp_taar1_vs_placebo_3-13 weeks_weight_increased`$data$treat1_new), 
                 subgroup.hetstat = TRUE, overall.hetstat =TRUE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = TRUE, prediction.subgroup = FALSE, prediction = FALSE,
                 subgroup = FALSE, fixed=TRUE, random=TRUE,
                 label.left = "TAAR1 better", label.right="Placebo better")

```

**Figure 27** Forest plot for weight gain for the comparison of TAAR1 agonist vs. placebo at 3-13 weeks (primary timepoint). The overall risk of bias of the studies for this outcome is presented. OR=Odds ratio, 95%CI: 95% confidence intervals.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.5.23 Secondary outcome: Weight in kg

#### Timepoint 1: *1 day - 2 weeks*

No usable data available from RCT.

#### Timepoint 2: *3 weeks - 13 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}
forest(`meta_comp_taar1_vs_placebo_3-13 weeks_weight`, 
       col.diamond = "firebrick4",
       col.square =   "firebrick3",     
       label.c="placebo",      
       rightcols = c("effect", "ci", "Overall"),rightlabs = c("MD kg", "95%-CI", "Risk of bias"),
                 label.e = unique(`meta_comp_taar1_vs_placebo_3-13 weeks_weight`$data$treat1_new), 
                 subgroup.hetstat = TRUE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = FALSE, prediction.subgroup = FALSE, prediction = FALSE,
                 subgroup = FALSE, fixed=TRUE, random=TRUE,
                 label.left = "TAAR1 better", label.right="Placebo better")



```

**Figure 28** Forest plot for weight change in kg for the comparison of TAAR1 agonist vs. placebo at 3-13 weeks (primary timepoint). The overall risk of bias of the studies for this outcome is presented. MD:Mean difference, SD: standard deviation, 95%CI: 95% confidence intervals.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.5.24 Secondary outcome: Hyperprolactinemia

#### Timepoint 1: *1 day - 2 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}
#forest(`meta_comp_taar1_vs_placebo_1 day-2 weeks_hyperprolactinemia`, 
 #      col.diamond = "firebrick4",
  #     col.square =   "firebrick3",     
   #    label.c="placebo",      
    #   rightcols = c("effect", "ci", "Overall"),
     #            label.e = unique(`meta_comp_taar1_vs_placebo_1 day-2 weeks_hyperprolactinemia`$data$treat1_new), 
      #           subgroup.hetstat = TRUE, overall.hetstat =TRUE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
       #          overall = FALSE, prediction.subgroup = FALSE, prediction=FALSE,
        #         subgroup = TRUE, fixed=TRUE, random=TRUE,
         #        label.left = "TAAR1 better", label.right="Placebo better")

```

There were available data from one study [@RN116] reporting 0 events in both arms so that an effect size cannot be estimated with the current approach.

#### Timepoint 2: *3 weeks - 13 weeks*

No usable data available from RCT.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.5.25 Secondary outcome: Prolactin levels

#### Timepoint 1: *1 day - 2 weeks*

No usable data available from RCT.

#### Timepoint 2: *3 weeks - 13 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}

forest(`meta_comp_taar1_vs_placebo_3-13 weeks_prolactin`, 
       col.diamond = "firebrick4",
       col.square =   "firebrick3",     
       label.c="placebo",      
       rightcols = c("effect", "ci", "Overall"),rightlabs = c("MD ng/ml", "95%-CI", "Risk of bias"),
                 label.e = unique(`meta_comp_taar1_vs_placebo_3-13 weeks_prolactin`$data$treat1_new), 
                 subgroup.hetstat = FALSE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = FALSE, prediction.subgroup = FALSE, prediction = FALSE,
                 subgroup = FALSE, fixed=TRUE, random=TRUE,
                 label.left = "TAAR1 better", label.right="Placebo better")

```

**Figure 29** Forest plot for prolactin levels in ng/ml for the comparison of TAAR1 agonist vs. placebo at 3-13 weeks (primary timepoint). SD was imputed from the dataset of antipsychotic trials in Huhn et al 2021. The overall risk of bias of the studies for this outcome is presented. MD:Mean difference, SD=Standard deviation, 95%CI: 95% confidence intervals.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.5.26 Secondary outcome: Akathisia

Terms considered included: restlessness (when was used to describe akathisia)

#### Timepoint 1: *1 day - 2 weeks*

No usable data available from RCT.

#### Timepoint 2: *3 weeks - 13 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}

forest(`meta_comp_taar1_vs_placebo_3-13 weeks_akathisia`, 
       col.diamond = "firebrick4",
       col.square =   "firebrick3",     
       label.c="placebo",      
       rightcols = c("effect", "ci", "Overall"),rightlabs = c("OR", "95%-CI", "Risk of bias"),
                 label.e = unique(`meta_comp_taar1_vs_placebo_3-13 weeks_akathisia`$data$treat1_new), 
                 subgroup.hetstat = TRUE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = FALSE, prediction.subgroup = FALSE, prediction = FALSE,
                 subgroup = FALSE, fixed=TRUE, random=TRUE,
                 label.left = "TAAR1 better", label.right="Placebo better")


```

**Figure 30** Forest plot for akathisia for the comparison of TAAR1 agonist vs. placebo at 3-13 weeks (primary timepoint). The overall risk of bias of the studies for this outcome is presented. OR:Odds ratio, 95%CI: 95% confidence intervals.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.5.27 Secondary outcome: Extrapyramidal symptoms

#### Timepoint 1: *1 day - 2 weeks*

No usable data available from RCT.

#### Timepoint 2: *3 weeks - 13 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}
forest(`meta_comp_taar1_vs_placebo_3-13 weeks_eps_symptoms`, 
       col.diamond = "firebrick4",
       col.square =   "firebrick3",     
       label.c="placebo",      
       rightcols = c("effect", "ci", "Overall"),rightlabs = c("OR", "95%-CI", "Risk of bias"),
                 label.e = unique(`meta_comp_taar1_vs_placebo_3-13 weeks_eps_symptoms`$data$treat1_new), 
                 subgroup.hetstat = FALSE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = FALSE, prediction.subgroup = FALSE, prediction = FALSE,
                 subgroup = FALSE, fixed=TRUE, random=TRUE,
                 label.left = "TAAR1 better", label.right="Placebo better")



```

**Figure 31** Forest plot for extrapyramidal side effects (use of antiparkinsonian medication as a proxy) for the comparison of TAAR1 agonist vs. placebo at 3-13 weeks (primary timepoint). The overall risk of bias of the studies for this outcome is presented. MD:Mean difference, SD: standard deviation, 95%CI: 95% confidence intervals. Furthermore, there were narrative descriptions from a conference abstract [@RN109] that ulotaront did not differ from placebo in extrapyramidal symptoms with low absolute frequencies in two unpublished phase III trials [@RN111; @RN112]. Moreover, ulotaront was reported not to deteriorate movement symptoms in participants with Parkinson's disease psychosis as measured with the Unified Parkinson Disease Rating Scale Part III [@RN13].

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.5.28 Secondary outcome: Anxiety

#### Timepoint 1: *1 day - 2 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}

#forest(`meta_comp_taar1_vs_placebo_1 day-2 weeks_anxiety`, 
 #      col.diamond = "firebrick4",
  #     col.square =   "firebrick3",     
   #    label.c="placebo",      
    #   leftcols = c("studlab",  "effect", "ci"),   
     #            label.e = unique(`meta_comp_taar1_vs_placebo_1 day-2 weeks_anxiety`$data$treat1_new), 
      #           subgroup.hetstat = TRUE, overall.hetstat =TRUE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
       #          overall = FALSE, prediction.subgroup = FALSE, prediction=FALSE,
        #         subgroup = TRUE, fixed=TRUE, random=TRUE,
         #        label.left = "TAAR1 better", label.right="Placebo better")


```

There were usable data for one study [@RN115], which reported 0 events in both groups so that an effect size could not be estimated with the current approach.

#### Timepoint 2: *3 weeks - 13 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}
forest(`meta_comp_taar1_vs_placebo_3-13 weeks_anxiety`, 
       col.diamond = "firebrick4",
       col.square =   "firebrick3",     
       label.c="placebo",      
       rightcols = c("effect", "ci", "Overall"),rightlabs = c("OR", "95%-CI", "Risk of bias"),
                 label.e = unique(`meta_comp_taar1_vs_placebo_3-13 weeks_anxiety`$data$treat1_new), 
                 subgroup.hetstat = TRUE, overall.hetstat =TRUE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = TRUE, prediction.subgroup = FALSE, prediction = FALSE,
                 subgroup = FALSE, fixed=TRUE, random=TRUE,
                 label.left = "TAAR1 better", label.right="Placebo better")

```

**Figure 32** Forest plot for anxiety as adverse event for the comparison of TAAR1 agonist vs. placebo at 3-13 weeks (primary timepoint). The overall risk of bias of the studies for this outcome is presented. OR:Odds ratio, 95%CI: 95% confidence intervals.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.5.29 Secondary outcome: Agitation

#### Timepoint 1: *1 day - 2 weeks*

No usable data available from RCT.

#### Timepoint 2: *3 weeks - 13 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}
forest(`meta_comp_taar1_vs_placebo_3-13 weeks_agitation`, 
       col.diamond = "firebrick4",
       col.square =   "firebrick3",     
       label.c="placebo",      
       rightcols = c("effect", "ci", "Overall"),rightlabs = c("OR", "95%-CI", "Risk of bias"),
                 label.e = unique(`meta_comp_taar1_vs_placebo_3-13 weeks_agitation`$data$treat1_new), 
                 subgroup.hetstat = TRUE, overall.hetstat =TRUE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = TRUE, prediction.subgroup = FALSE, prediction = FALSE,
                 subgroup = FALSE, fixed=TRUE, random=TRUE,
                 label.left = "TAAR1 better", label.right="Placebo better")


```

**Figure 33** Forest plot for agitation for the comparison of TAAR1 agonist vs. placebo at 3-13 weeks (primary timepoint). The overall risk of bias of the studies for this outcome is presented. OR:Odds ratio, 95%CI: 95% confidence intervals.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.5.30 Secondary outcome: Headache

#### Timepoint 1: *1 day - 2 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}

forest(`meta_comp_taar1_vs_placebo_1 day-2 weeks_headache`, 
       col.diamond = "firebrick4",
       col.square =   "firebrick3",     
       label.c="placebo",      
       leftcols = c("studlab",  "effect", "ci"),     
        rightcols = c("Overall"), rightlabs = c("Risk of bias"),  
                 label.e = unique(`meta_comp_taar1_vs_placebo_1 day-2 weeks_headache`$data$treat1_new), 
                 subgroup.hetstat = TRUE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = FALSE, prediction.subgroup = FALSE, prediction=FALSE,
                 subgroup = TRUE, fixed=TRUE, random=TRUE,
                 label.left = "TAAR1 better", label.right="Placebo better")
```

**Figure 34** Forest plot for headache for the comparison of TAAR1 agonist vs. placebo at 1 day to 2 weeks. The overall risk of bias of the studies for this outcome is presented. OR:Odds ratio, 95%CI: 95% confidence intervals.

#### Timepoint 2: *3 weeks - 13 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}

forest(`meta_comp_taar1_vs_placebo_3-13 weeks_headache`, 
       col.diamond = "firebrick4",
       col.square =   "firebrick3",     
       label.c="placebo",      
       rightcols = c("effect", "ci", "Overall"),rightlabs = c("OR", "95%-CI", "Risk of bias"),
                 label.e = unique(`meta_comp_taar1_vs_placebo_3-13 weeks_headache`$data$treat1_new), 
                 subgroup.hetstat = TRUE, overall.hetstat =TRUE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = TRUE, prediction.subgroup = FALSE, prediction = FALSE,
                 subgroup = FALSE, fixed=TRUE, random=TRUE,
                 label.left = "TAAR1 better", label.right="Placebo better")

```

**Figure 35** Forest plot for headache for the comparison of TAAR1 agonist vs. placebo at 3-13 weeks (primary timepoint). The overall risk of bias of the studies for this outcome is presented. OR:Odds ratio, 95%CI: 95% confidence intervals.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.5.31 Secondary outcome: Sedation

Terms that were considered as reported in the trials: fatigue, sedation, somnolence

#### Timepoint 1: *1 day - 2 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}
forest(`meta_comp_taar1_vs_placebo_1 day-2 weeks_sedation`, 
       col.diamond = "firebrick4",
       col.square =   "firebrick3",     
       label.c="placebo",         
        rightcols = c("Overall"), rightlabs = c("Risk of bias"),  
       leftcols = c("studlab",  "effect", "ci"),   
                 label.e = unique(`meta_comp_taar1_vs_placebo_1 day-2 weeks_sedation`$data$treat1_new), 
                 subgroup.hetstat = TRUE, overall.hetstat =TRUE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = TRUE, prediction.subgroup = FALSE, prediction=FALSE,
                 subgroup = TRUE, fixed=TRUE, random=TRUE,
                 label.left = "TAAR1 better", label.right="Placebo better")

```

**Figure 36** Forest plot for sedation for the comparison of TAAR1 agonist vs. placebo at 1 day to 2 weeks. The overall risk of bias of the studies for this outcome is presented. OR:Odds ratio, 95%CI: 95% confidence intervals.

#### Timepoint 2: *3 weeks - 13 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}

forest(`meta_comp_taar1_vs_placebo_3-13 weeks_sedation`, 
       col.diamond = "firebrick4",
       col.square =   "firebrick3",     
       label.c="placebo",      
       rightcols = c("effect", "ci", "Overall"),rightlabs = c("OR", "95%-CI", "Risk of bias"),
                 label.e = unique(`meta_comp_taar1_vs_placebo_3-13 weeks_sedation`$data$treat1_new), 
                 subgroup.hetstat = TRUE, overall.hetstat =TRUE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = TRUE, prediction.subgroup = FALSE, prediction = FALSE,
                 subgroup = FALSE, fixed=TRUE, random=TRUE,
                 label.left = "TAAR1 better", label.right="Placebo better")

```

**Figure 37** Forest plot for sedation for the comparison of TAAR1 agonist vs. placebo at 3-13 weeks (primary timepoint). The overall risk of bias of the studies for this outcome is presented. OR:Odds ratio, 95%CI: 95% confidence intervals.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.5.32 Secondary outcome: Insomnia

#### Timepoint 1: *1 day - 2 weeks*

No usable data available from RCT.

#### Timepoint 2: *3 weeks - 13 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}

forest(`meta_comp_taar1_vs_placebo_3-13 weeks_insomnia`, 
       col.diamond = "firebrick4",
       col.square =   "firebrick3",     
       label.c="placebo",      
       rightcols = c("effect", "ci", "Overall"),rightlabs = c("OR", "95%-CI", "Risk of bias"),
                 label.e = unique(`meta_comp_taar1_vs_placebo_3-13 weeks_insomnia`$data$treat1_new), 
                 subgroup.hetstat = TRUE, overall.hetstat =TRUE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = TRUE, prediction.subgroup = FALSE, prediction = FALSE,
                 subgroup = FALSE, fixed=TRUE, random=TRUE,
                 label.left = "TAAR1 better", label.right="Placebo better")

```

**Figure 38** Forest plot for insomnia for the comparison of TAAR1 agonist vs. placebo at 3-13 weeks (primary timepoint). The overall risk of bias of the studies for this outcome is presented. The overall risk of bias of the studies for this outcome is presented. OR:Odds ratio, 95%CI: 95% confidence intervals.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.5.33 Summary outcome plot: Dropouts and side-effects

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=8, warning=FALSE}

ae_categories<-ae_categories %>% 
  select(event_text, MedDRA_an, MedDRA_group) %>%  
  rename(outcome=MedDRA_an) 

safety_primary<-meta_outcome %>% left_join(ae_categories) %>%  #I need to change a bit 
  mutate(event_text=ifelse(outcome=="death", "Mortality", 
                           ifelse(outcome=="serious", "Serious event", 
                                  ifelse(outcome=="dropout_any", "Dropouts any reason",
                                         ifelse(outcome=="dropout_ae", "Dropouts adverse event", event_text)))),
         MedDRA_group=ifelse(outcome=="death" | outcome=="serious", "serious",
                             ifelse(outcome=="dropout_any" | outcome=="dropout_ae", "dropouts", MedDRA_group))) %>% 
  filter(!is.na(TE.random)) %>% filter(!is.na(event_text)) %>%
  mutate(point=TE.random,
         lb=TE.random-1.96*seTE.random,
         ub=TE.random+1.96*seTE.random) %>% unique() %>%
  filter(comparison=="taar1_vs_placebo") %>% 
  mutate(n_text=paste0("N=", k,", n=",n))

safety_primary_plot <-safety_primary

safety_primary_plot$event_text <- factor(safety_primary_plot$event_tex, levels=unique(safety_primary_plot$event_tex))
level_order<- sort(unique(safety_primary_plot$event_text))
lever_order_meddra<-c("dropouts", "any", "serious", "autonomic", "endocrinic", "neuromuscular", "neuropsychiatric")

safety_primary_plot$event_text <- factor(safety_primary_plot$event_tex, levels=unique(safety_primary_plot$event_tex))
level_order<- sort(unique(safety_primary_plot$event_text))
lever_order_meddra<-c("dropouts", "any", "serious", "autonomic", "endocrinic", "neuromuscular", "neuropsychiatric")

safety_primary_plot<-safety_primary_plot %>% mutate(type_outcome=ifelse(outcome=="dropout_any" | outcome=="dropout_ae", "dropout", "side"))

ggplot(data=safety_primary_plot, 
       aes(y=factor(event_text, level=rev(level_order)), x=exp(point),colour=type_outcome, shape=timepoint))+
  geom_text(aes(x=0.03, label=n_text), colour="black" , position=position_dodge(width = 0.9))+
  geom_point(aes(size=n), position=position_dodge(width = 0.5))+
  geom_errorbar(aes(x=exp(point), xmin=exp(TE.random-1.96*seTE.random), 
                    xmax=exp(TE.random+1.96*seTE.random), colour=type_outcome,shape=timepoint),
                width=0, position=position_dodge(width = 0.5))+
  geom_vline(xintercept=1, linetype='dashed')+
  scale_size_continuous(guide=FALSE)+
  facet_grid(factor(MedDRA_group, levels=lever_order_meddra)~., scales = "free", space = "free")+
  scale_x_log10() +
  ylab("")+
  scale_shape_manual(values=c(1,19))+
  scale_colour_manual(guide=FALSE, values=c("green4", "firebrick4"))+
  xlab("OR 95%CI (OR>1: TAAR1 agonists worse)")+
  theme_bw()


```

**Figure 39** Summary forest plot for dropouts and side-effects for the comparison of TAAR1 agonists compared to placebo and for the two timepoints with usable data. It should be noted that different populations have been pooled in the analysis of dropouts and adverse events including schizophrenia, Parkinson's disease psychosis, narcolepsy-cateplexy and healthy volunteers. The size is proportional to the number of participants in the analysis. The findings of the random-effects model are presented. OR\>1 corresponds to more events with TAAR1 agonists compared to placebo. OR: Odds ratio, 95%CI: 95% confidence intervals. N=number of studies, n=number of participants.

## 2.6 Comparison 2: *TAAR1 agonist* vs *antipsychotic*

### 2.6.1 Primary outcome: Overall symptoms

#### Timepoint 1: *1 day - 2 weeks*

No usable data available from RCT.

#### Timepoint 2: *3 weeks - 13 weeks* (primary timepoint)

```{r echo=F, fig.fullwidth=T, fig.width=12, fig.height=4}
forest(`meta_comp_taar1_vs_antipsychotic_3-13 weeks_overall`, 
       col.diamond = "royalblue4",
       col.square =   "royalblue1",
                 rightcols = c("effect", "ci","Overall"), 
                 rightlabs = c("SMD", "95%-CI", "Risk of bias"),  
       xlim=c(-0.8,0.8),
       label.c="risperidone",
                 label.e = unique(`meta_comp_taar1_vs_antipsychotic_3-13 weeks_overall`$data$treat1_new), 
                 subgroup.hetstat = TRUE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = FALSE, prediction.subgroup = FALSE,
                 subgroup = TRUE, fixed=TRUE,
                 label.left = "Antipsychotic better", label.right="TAAR1 better")

```

**Figure 40** Forest plot for overall symptoms (primary outcome, measureed with PANSS total) for the comparison of TAAR1 agonist vs. antipsychotics at 3-13 weeks (primary timepoint). The overall risk of bias of the studies for this outcome is presented. SMD: Standardized mean difference, SD: Standard deviation, 95%CI: 95% confidence interval.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.6.2 Secondary outcome: Positive symptoms

#### Timepoint 1: *1 day - 2 weeks*

No usable data available from RCT.

#### Timepoint 2: *3 weeks - 13 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}
forest(`meta_comp_taar1_vs_antipsychotic_3-13 weeks_positive`, 
       col.diamond = "royalblue4",
       col.square =   "royalblue1",
                 rightcols = c("effect", "ci","Overall"), 
                 rightlabs = c("SMD", "95%-CI", "Risk of bias"),  
       xlim=c(-0.8,0.8),
       label.c="risperidone",
                 label.e = unique(`meta_comp_taar1_vs_antipsychotic_3-13 weeks_positive`$data$treat1_new), 
                 subgroup.hetstat = TRUE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = FALSE, prediction.subgroup = FALSE,
                 subgroup = TRUE, fixed=TRUE,
                 label.left = "Antipsychotic better", label.right="TAAR1 better")
```

**Figure 41** Forest plot for positive symptoms for the comparison of TAAR1 agonist vs. antipsychotics (measured with PANSS positive) at 3-13 weeks (primary timepoint). The overall risk of bias of the studies for this outcome is presented. SMD: Standardized mean difference, SD: Standard deviation, 95%CI: 95% confidence interval.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.6.3 Secondary outcome: Negative symptoms

#### Timepoint 1: *1 day - 2 weeks*

No usable data available from RCT.

#### Timepoint 2: *3 weeks - 13 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}
forest(`meta_comp_taar1_vs_antipsychotic_3-13 weeks_negative`, 
       col.diamond = "royalblue4",
       col.square =   "royalblue1",
                 rightcols = c("effect", "ci","Overall"), 
                 rightlabs = c("SMD", "95%-CI", "Risk of bias"),  
       xlim=c(-0.8,0.8),
       label.c="risperidone",
                 label.e = unique(`meta_comp_taar1_vs_antipsychotic_3-13 weeks_negative`$data$treat1_new), 
                 subgroup.hetstat = TRUE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = FALSE, prediction.subgroup = FALSE,
                 subgroup = TRUE, fixed=TRUE,
                 label.left = "Antipsychotic better", label.right="TAAR1 better")
```

**Figure 42** Forest plot for negative symptoms for the comparison of TAAR1 agonist vs. antipsychotics (measured with PANSS negative) at 3-13 weeks (primary timepoint). The overall risk of bias of the studies for this outcome is presented. SMD: Standardized mean difference, SD: Standard deviation, 95%CI: 95% confidence interval.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.6.4 Secondary outcome: Depressive symptoms

#### Timepoint 1: *1 day - 2 weeks*

No usable data available from RCT.

#### Timepoint 2: *3 weeks - 13 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}
forest(`meta_comp_taar1_vs_antipsychotic_3-13 weeks_depression`, 
       col.diamond = "royalblue4",
       col.square =   "royalblue1",
                 rightcols = c("effect", "ci","Overall"), 
                 rightlabs = c("SMD", "95%-CI", "Risk of bias"),  
       label.c = "risperidone",
       xlim=c(-0.8,0.8),
                 label.e = unique(`meta_comp_taar1_vs_antipsychotic_3-13 weeks_depression`$data$treat1_new), 
                 subgroup.hetstat = TRUE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = FALSE, prediction.subgroup = FALSE,
                 subgroup = TRUE, fixed=TRUE,
                 label.left = "Antipsychotic better", label.right="TAAR1 better")

```

**Figure 43** Forest plot for depression symptoms for the comparison of TAAR1 agonist vs. antipsychotics (measured with PANSS anxiety/depression factor) at 3-13 weeks (primary timepoint). The overall risk of bias of the studies for this outcome is presented. SMD: Standardized mean difference, SD: Standard deviation, 95%CI: 95% confidence interval.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.6.5 Secondary outcome: Cognitive symptoms

#### Timepoint 1: *1 day - 2 weeks*

No usable data available from RCT.

#### Timepoint 2: *3 weeks - 13 weeks*

No usable data available from RCT.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.6.6 Secondary outcome: Quality of life

#### Timepoint 1: *1 day - 2 weeks*

No usable data available from RCT.

#### Timepoint 2: *3 weeks - 13 weeks*

No usable data available from RCT.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.6.7 Secondary outcome: Functioning

#### Timepoint 1: *1 day - 2 weeks*

No usable data available from RCT.

#### Timepoint 2: *3 weeks - 13 weeks*

No usable data available from RCT.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.6.8 Secondary outcome: Response to treatment

#### Timepoint 1: *1 day - 2 weeks*

No usable data available from RCT.

#### Timepoint 2: *3 weeks - 13 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}

forest(`meta_comp_taar1_vs_antipsychotic_3-13 weeks_response`, 
       col.diamond = "royalblue4",
       col.square =   "royalblue1",
                 rightcols = c("effect", "ci","Overall"), 
                 rightlabs = c("OR", "95%-CI", "Risk of bias"),  
       xlim=c(0.2,5),
       label.c="risperidone",
                 label.e = unique(`meta_comp_taar1_vs_antipsychotic_3-13 weeks_response`$data$treat1_new), 
                 subgroup.hetstat = TRUE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = FALSE, prediction.subgroup = FALSE,
                 subgroup = TRUE, fixed=TRUE,
                 label.left = "Antipsychotic better", label.right="TAAR1 better")


```

**Figure 44** Forest plot for response to treatment (\>50% symptom reduction in PANSS total from baseline) for the comparison of TAAR1 agonist vs. antipsychotics at 3-13 weeks (primary timepoint). The overall risk of bias of the studies for this outcome is presented. OR: Odds ratio, 95%CI: 95% confidence interval.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.6.9 Secondary outcome: Relapse

#### Timepoint 1: *1 day - 2 weeks*

No usable data available from RCT.

#### Timepoint 2: *3 weeks - 13 weeks*

No usable data available from RCT.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.6.10 Summary outcome plot: Efficacy outcomes

```{r echo=F, fig.fullwidth=T, fig.width=10, fig.height=5, warning=FALSE}

outcome_categories<-outcome_names %>% 
  select(outcome, event_text) 

efficacy_primary_2<-meta_outcome %>% left_join(outcome_categories) %>% 
  filter(outcome=="cognition" | outcome=="depression" | outcome=="negative" | outcome=="overall" | outcome=="positive") %>% #efficacy outcome
  filter(!is.na(TE.random)) %>% filter(!is.na(event_text)) %>%
  mutate(point=TE.random,
         lb=TE.random-1.96*seTE.random,
         ub=TE.random+1.96*seTE.random) %>% 
  unique() %>%
  filter(comparison=="taar1_vs_antipsychotic") %>% 
  mutate(n_text=paste0("N=", k,", n=",n))

efficacy_primary_2$population <- factor(efficacy_primary_2$population, levels=unique(efficacy_primary_2$population))
efficacy_primary_2$event_text <- factor(efficacy_primary_2$event_tex, levels=unique(efficacy_primary_2$event_tex))
level_outcomes<- c("Overall symptoms", "Positive symptoms", "Negative symptoms", "Depressive symptoms", "Cognitive impairement")
level_population<-c("Schizophrenia spectrum", "Parkinson Disease Psychosis")

ggplot(data=efficacy_primary_2, 
       aes(y=factor(event_text, level=rev(level_outcomes)), x=point))+
  geom_text(aes(x=0.65, label=n_text), colour="black" , position=position_dodge(width = 0.9))+
  geom_point(size=3.4,position=position_dodge(width = 0.5), colour="royalblue4")+ #Size proportional to the number of participants
  coord_cartesian(xlim=c(-0.8,0.8))+
  geom_errorbar(aes(x=point, xmin=lb, 
                    xmax=ub),
                width=0, position=position_dodge(width = 0.5), colour="royalblue4")+
  scale_size_continuous(guide=FALSE)+
  geom_vline(xintercept=0, linetype='dashed')+
  facet_grid(~factor(population, levels=level_population))+
  ylab("")+
  xlab("SMD 95%CI (SMD>0: TAAR1 agonists better)")+
  theme_bw()


```

**Figure 45** Summary plot for efficacy outcomes at 4 weeks for the comparison of TAAR1 agonists versus antipscyhotcs in patients with an acute episode of schizophrenia. There were only data for the timepoint of 3-13 weeks. There were only data for ralmitaront as TAAR1 agonist and risperidone as control group. SMD\>0 correspond to more symptom improvement by TAAR1 agonists compared to antipsychotic. SMD: Standardized mean difference, 95%CI: 95% confidence interval. N=number of studies, n=number of participants.

### 2.6.11 Secondary outcome: Dropouts due to any reason

#### Timepoint 1: *1 day - 2 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}
forest(`meta_comp_taar1_vs_antipsychotic_1 day-2 weeks_dropout_any`, 
       col.diamond = "green4",
       col.square =   "green3",    
       xlim=c(0.2,5),
       label.c="amisulpride",      
       rightcols = c("effect", "ci", "Overall"),rightlabs = c("OR", "95%-CI", "Risk of bias"),
                 label.e = unique(`meta_comp_taar1_vs_antipsychotic_1 day-2 weeks_dropout_any`$data$treat1_new), 
                 subgroup.hetstat = TRUE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = FALSE, prediction.subgroup = FALSE,
                 subgroup = TRUE, fixed=TRUE,
                 label.left = "TAAR1 better", label.right="Antipsychotic better")



```

**Figure 46** Forest plot for dropouts due to any reason for the comparison of TAAR1 agonist vs. antipsychotics at 1 day to 2 weeks. The overall risk of bias of the studies for this outcome is presented. OR: Odds ratio, 95%CI: 95% confidence interval.

#### Timepoint 2: *3 weeks - 13 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}
forest(`meta_comp_taar1_vs_antipsychotic_3-13 weeks_dropout_any`, 
       col.diamond = "green4",
       col.square =   "green3",
                 rightcols = c("effect", "ci","Overall"), 
                 rightlabs = c("OR", "95%-CI", "Risk of bias"),  
       label.c = "risperidone",
       xlim=c(0.2,5),
                 label.e = unique(`meta_comp_taar1_vs_antipsychotic_3-13 weeks_dropout_any`$data$treat1_new), 
       
                 subgroup.hetstat = TRUE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = FALSE, prediction.subgroup = FALSE,
                 subgroup = TRUE, fixed=TRUE,
                 label.left = "TAAR1 better", label.right="Antipsychotic better")

```

**Figure 47** Forest plot for dropouts due to any reason for the comparison of TAAR1 agonist vs. antipsychotics at 3-13 weeks (primary timepoint). The overall risk of bias of the studies for this outcome is presented. OR: Odds ratio, 95%CI: 95% confidence interval.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.6.12 Secondary outcome: Dropouts due to adverse event

#### Timepoint 1: *1 day - 2 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}
#forest(`meta_comp_taar1_vs_antipsychotic_1 day-2 weeks_dropout_ae`, 
 #      col.diamond = "green4",
  #     col.square =   "green3",     
   #    label.c="amisulpride",      
    #   rightcols = c("effect", "ci", "Overall"),rightlabs = c("OR", "95%-CI", "Risk of bias"),
     #            label.e = unique(`meta_comp_taar1_vs_antipsychotic_1 day-2 weeks_dropout_ae`$data$treat1_new), 
      #           subgroup.hetstat = TRUE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
       #          overall = FALSE, prediction.subgroup = FALSE,
        #         subgroup = TRUE, fixed=TRUE,
         #        label.left = "TAAR1 better", label.right="Antipsychotic better")


```

There were usable data in [@RN115], whcih reported 0 events in both groups and an effect size cannot be estimated with the current approach.

#### Timepoint 2: *3 weeks - 13 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}

forest(`meta_comp_taar1_vs_antipsychotic_3-13 weeks_dropout_any`, 
       col.diamond = "green4",
       col.square =   "green3",
                 rightcols = c("effect", "ci","Overall"), 
                 rightlabs = c("OR", "95%-CI", "Risk of bias"),  
       label.c = "risperidone",
       xlim=c(0.2,5),
                 label.e = unique(`meta_comp_taar1_vs_antipsychotic_3-13 weeks_dropout_any`$data$treat1_new), 
                 subgroup.hetstat = TRUE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = FALSE, prediction.subgroup = FALSE,
                 subgroup = TRUE, fixed=TRUE,
                 label.left = "TAAR1 better", label.right="Antipsychotic better")


```

**Figure 48** Forest plot for dropouts due to adverse events for the comparison of TAAR1 agonist vs. antipsychotics at 3-13 weeks (primary timepoint). The overall risk of bias of the studies for this outcome is presented. OR: Odds ratio, 95%CI: 95% confidence interval.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.6.13 Secondary outcome: Serious adverse event

#### Timepoint 1: *1 day - 2 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}
forest(`meta_comp_taar1_vs_antipsychotic_1 day-2 weeks_serious`, 
       col.diamond = "firebrick4",
       col.square =   "firebrick3",     
       label.c="amisulpride",      
       rightcols = c("effect", "ci", "Overall"),rightlabs = c("OR", "95%-CI", "Risk of bias"),
      xlim=c(0.2,5),
                 label.e = unique(`meta_comp_taar1_vs_antipsychotic_1 day-2 weeks_serious`$data$treat1_new), 
                 subgroup.hetstat = TRUE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = FALSE, prediction.subgroup = FALSE,
                 subgroup = TRUE, fixed=TRUE,
                 label.left = "TAAR1 better", label.right="Antipsychotic better")


```

**Figure 49** Forest plot for serious adverse events for the comparison of TAAR1 agonist vs. antipsychotics at 1 day to 2 weeks. The overall risk of bias of the studies for this outcome is presented. OR: Odds ratio, 95%CI: 95% confidence interval.

#### Timepoint 2: *3 weeks - 13 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}

forest(`meta_comp_taar1_vs_antipsychotic_3-13 weeks_serious`, 
       col.diamond = "firebrick4",
       col.square =   "firebrick3",
                 rightcols = c("effect", "ci","Overall"), 
                 rightlabs = c("OR", "95%-CI", "Risk of bias"),  
       label.c = "risperidone",
       xlim=c(0.2,5),
                 label.e = unique(`meta_comp_taar1_vs_antipsychotic_1 day-2 weeks_serious`$data$treat1_new), 
                 subgroup.hetstat = TRUE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = FALSE, prediction.subgroup = FALSE,
                 subgroup = TRUE, fixed=TRUE,
                 label.left = "TAAR1 better", label.right="Antipsychotic better")


```

**Figure 50** Forest plot for serious adverse events for the comparison of TAAR1 agonist vs. antipsychotics at 3-13 weeks (primary timepoint). The overall risk of bias of the studies for this outcome is presented. OR: Odds ratio, 95%CI: 95% confidence interval.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.6.14 Secondary outcome: Death

#### Timepoint 1: *1 day - 2 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}
#forest(`meta_comp_taar1_vs_antipsychotic_1 day-2 weeks_death`, 
 #      col.diamond = "firebrick4",
  #     col.square =   "firebrick3",     
   #    label.c="amisulpride",      
    #   rightcols = c("effect", "ci", "Overall"),rightlabs = c("OR", "95%-CI", "Risk of bias"),
     # xlim=c(0.2,5),
      #           label.e = unique(`meta_comp_taar1_vs_antipsychotic_1 day-2 weeks_death`$data$treat1_new), 
       #          subgroup.hetstat = TRUE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
        #         overall = FALSE, prediction.subgroup = FALSE,
         #        subgroup = TRUE, fixed=TRUE,
          #       label.left = "TAAR1 better", label.right="Antipsychotic better")

```

There were usable data in [@RN115], whcih reported 0 events in both groups and an effect size cannot be estimated with the current approach.

#### Timepoint 2: *3 weeks - 13 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}

#forest(`meta_comp_taar1_vs_antipsychotic_3-13 weeks_death`, 
 #      col.diamond = "firebrick4",
  #     col.square =   "firebrick3",
   #              rightcols = c("effect", "ci","Overall"), 
    #             rightlabs = c("OR", "95%-CI", "Risk of bias"),  
     #  label.c = "risperidone",
      # xlim=c(0.2,5),
       #          label.e = unique(`meta_comp_taar1_vs_antipsychotic_1 day-2 weeks_death`$data$treat1_new), 
        #         subgroup.hetstat = TRUE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
         #        overall = FALSE, prediction.subgroup = FALSE,
          #       subgroup = TRUE, fixed=TRUE,
           #      label.left = "TAAR1 better", label.right="Antipsychotic better")
```

There were usable data in [@RN113], whcih reported 0 events in both groups and an effect size cannot be estimated with the current approach.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.6.15 Secondary outcome: Any adverse event

#### Timepoint 1: *1 day - 2 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}
forest(`meta_comp_taar1_vs_antipsychotic_1 day-2 weeks_adverse_event`, 
       col.diamond = "firebrick4",
       col.square =   "firebrick3",     
       label.c="amisulpride",      
       rightcols = c("effect", "ci", "Overall"),rightlabs = c("OR", "95%-CI", "Risk of bias"),
      #xlim=c(0.2,5),
                 label.e = unique(`meta_comp_taar1_vs_antipsychotic_1 day-2 weeks_adverse_event`$data$treat1_new), 
                 subgroup.hetstat = TRUE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = FALSE, prediction.subgroup = FALSE,
                 subgroup = TRUE, fixed=TRUE,
                 label.left = "TAAR1 better", label.right="Antipsychotic better")

```

**Figure 51** Forest plot for any adverse event for the comparison of TAAR1 agonist vs. antipsychotics at 1 day to 2 weeks. The overall risk of bias of the studies for this outcome is presented. OR: Odds ratio, 95%CI: 95% confidence interval.

#### Timepoint 2: *3 weeks - 13 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}

forest(`meta_comp_taar1_vs_antipsychotic_3-13 weeks_adverse_event`, 
       col.diamond = "firebrick4",
       col.square =   "firebrick3",
                 rightcols = c("effect", "ci","Overall"), 
                 rightlabs = c("OR", "95%-CI", "Risk of bias"),  
       label.c = "risperidone",
       xlim=c(0.2,5),
                 label.e = unique(`meta_comp_taar1_vs_antipsychotic_3-13 weeks_adverse_event`$data$treat1_new), 
                 subgroup.hetstat = TRUE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = FALSE, prediction.subgroup = FALSE,
                 subgroup = TRUE, fixed=TRUE,
                 label.left = "TAAR1 better", label.right="Antipsychotic better")

```

**Figure 52** Forest plot for any adverse event for the comparison of TAAR1 agonist vs. antipsychotics at 3-13 weeks (primary timepoint). The overall risk of bias of the studies for this outcome is presented. OR: Odds ratio, 95%CI: 95% confidence interval.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.6.16 Secondary outcome: Anticholinergic symptoms

#### Timepoint 1: *1 day - 2 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}
forest(`meta_comp_taar1_vs_antipsychotic_1 day-2 weeks_anticholinergic_symptom`, 
       col.diamond = "firebrick4",
       col.square =   "firebrick3",     
       label.c="amisulpride",      
       rightcols = c("effect", "ci", "Overall"),rightlabs = c("OR", "95%-CI", "Risk of bias"),
      #xlim=c(0.2,5),
                 label.e = unique(`meta_comp_taar1_vs_antipsychotic_1 day-2 weeks_anticholinergic_symptom`$data$treat1_new), 
                 subgroup.hetstat = TRUE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = FALSE, prediction.subgroup = FALSE,
                 subgroup = TRUE, fixed=TRUE,
                 label.left = "TAAR1 better", label.right="Antipsychotic better")

```

**Figure 53** Forest plot for anticholinergic side effects (e.g., dry mouth in this study) for the comparison of TAAR1 agonist vs. antipsychotics at 1 day to 2 weeks. The overall risk of bias of the studies for this outcome is presented. OR: Odds ratio, 95%CI: 95% confidence interval.

#### Timepoint 2: *3 weeks - 13 weeks*

No usable data available from RCT.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.6.17 Secondary outcome: Hypotension

#### Timepoint 1: *1 day - 2 weeks*

No usable data available from RCT.

#### Timepoint 2: *3 weeks - 13 weeks*

No usable data available from RCT.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.6.18 Secondary outcome: Dizziness

#### Timepoint 1: *1 day - 2 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}
forest(`meta_comp_taar1_vs_antipsychotic_1 day-2 weeks_dizziness`, 
       col.diamond = "firebrick4",
       col.square =   "firebrick3",     
       label.c="amisulpride",      
       rightcols = c("effect", "ci", "Overall"),rightlabs = c("OR", "95%-CI", "Risk of bias"),
     # xlim=c(0.2,5),
                 label.e = unique(`meta_comp_taar1_vs_antipsychotic_1 day-2 weeks_dizziness`$data$treat1_new), 
                 subgroup.hetstat = TRUE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = FALSE, prediction.subgroup = FALSE,
                 subgroup = TRUE, fixed=TRUE,
                 label.left = "TAAR1 better", label.right="Antipsychotic better")


```

**Figure 54** Forest plot for dizziness for the comparison of TAAR1 agonist vs. antipsychotics at 1 day to 2 weeks. The overall risk of bias of the studies for this outcome is presented. OR: Odds ratio, 95%CI: 95% confidence interval.

#### Timepoint 2: *3 weeks - 13 weeks*

No usable data available from RCT.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.6.19 Secondary outcome: Nausea or vomiting

Terms that were considered as reported in the trials: nausea, vomiting, emesis

#### Timepoint 1: *1 day - 2 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}
forest(`meta_comp_taar1_vs_antipsychotic_1 day-2 weeks_nausea_vomitting`, 
       col.diamond = "firebrick4",
       col.square =   "firebrick3",     
       label.c="amisulpride",      
       rightcols = c("effect", "ci", "Overall"),rightlabs = c("OR", "95%-CI", "Risk of bias"),
     # xlim=c(0.2,5),
                 label.e = unique(`meta_comp_taar1_vs_antipsychotic_1 day-2 weeks_nausea_vomitting`$data$treat1_new), 
                 subgroup.hetstat = TRUE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = FALSE, prediction.subgroup = FALSE,
                 subgroup = TRUE, fixed=TRUE,
                 label.left = "TAAR1 better", label.right="Antipsychotic better")

```

**Figure 55** Forest plot for nausea or vomitting for the comparison of TAAR1 agonist vs. antipsychotics at 1 day to 2 weeks.It should be noted that the study [@RN115] examined ulotaront that is also a partial agonsit of 5-HT1A causing potentially gastrointestinal side-effects. In comparison, ralmitaront is not a partial agonist of the 5-HT1A receptor. The overall risk of bias of the studies for this outcome is presented. OR: Odds ratio, 95%CI: 95% confidence interval.

#### Timepoint 2: *3 weeks - 13 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}

forest(`meta_comp_taar1_vs_antipsychotic_3-13 weeks_nausea_vomitting`, 
       col.diamond = "firebrick4",
       col.square =   "firebrick3",
                 rightcols = c("effect", "ci","Overall"), 
                 rightlabs = c("OR", "95%-CI", "Risk of bias"),  
       label.c = "risperidone",
      # xlim=c(0.2,5),
                 label.e = unique(`meta_comp_taar1_vs_antipsychotic_3-13 weeks_nausea_vomitting`$data$treat1_new), 
                 subgroup.hetstat = TRUE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = FALSE, prediction.subgroup = FALSE,
                 subgroup = TRUE, fixed=TRUE,
                 label.left = "TAAR1 better", label.right="Antipsychotic better")
```

**Figure 56** Forest plot for nausea or vomitting for the comparison of TAAR1 agonist vs. antipsychotics at 3-13 weeks (primary timepoint). It should be noted that the study [@RN113] examined ralmitaront, which compared to ulotaront is not a partial agonist of the 5-HT1A receptor that can be associated with gastrointestinal side effects. The overall risk of bias of the studies for this outcome is presented. OR: Odds ratio, 95%CI: 95% confidence interval.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.6.20 Secondary outcome: QTc prolongation

#### Timepoint 1: *1 day - 2 weeks*

No usable data available from RCT.

#### Timepoint 2: *3 weeks - 13 weeks*

No usable data available from RCT.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.6.21 Secondary outcome: QTc interval in msec

#### Timepoint 1: *1 day - 2 weeks*

No usable data available from RCT.

#### Timepoint 2: *3 weeks - 13 weeks*

No usable data available from RCT.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.6.22 Secondary outcome: Weight increased

#### Timepoint 1: *1 day - 2 weeks*

No usable data available from RCT.

#### Timepoint 2: *3 weeks - 13 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}
forest(`meta_comp_taar1_vs_antipsychotic_3-13 weeks_weight_increased`, 
       col.diamond = "firebrick4",
       col.square =   "firebrick3",
                 rightcols = c("effect", "ci","Overall"), 
                 rightlabs = c("OR", "95%-CI", "Risk of bias"),  
       label.c = "risperidone",
      # xlim=c(0.2,5),
                 label.e = unique(`meta_comp_taar1_vs_antipsychotic_3-13 weeks_weight_increased`$data$treat1_new), 
                 subgroup.hetstat = TRUE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = FALSE, prediction.subgroup = FALSE,
                 subgroup = TRUE, fixed=TRUE,
                 label.left = "TAAR1 better", label.right="Antipsychotic better")
```

**Figure 57** Forest plot for weight increase for the comparison of TAAR1 agonist vs. antipsychotics at 3-13 weeks (primary timepoint). The overall risk of bias of the studies for this outcome is presented. OR: Odds ratio, 95%CI: 95% confidence interval.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.6.23 Secondary outcome: Weight in kg

#### Timepoint 1: *1 day - 2 weeks*

No usable data available from RCT.

#### Timepoint 2: *3 weeks - 13 weeks*

No usable data available from RCT.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.6.24 Secondary outcome: Hyperprolactinemia

#### Timepoint 1: *1 day - 2 weeks*

No usable data available from RCT.

#### Timepoint 2: *3 weeks - 13 weeks*

No usable data available from RCT.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.6.25 Secondary outcome: Prolactin levels

#### Timepoint 1: *1 day - 2 weeks*

No usable data available from RCT.

#### Timepoint 2: *3 weeks - 13 weeks*

No usable data available from RCT.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.6.26 Secondary outcome: Akathisia

#### Timepoint 1: *1 day - 2 weeks*

No usable data available from RCT.

#### Timepoint 2: *3 weeks - 13 weeks*

No usable data available from RCT.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.6.27 Secondary outcome: Exrapyramidal symptoms

#### Timepoint 1: *1 day - 2 weeks*

No usable data available from RCT.

#### Timepoint 2: *3 weeks - 13 weeks*

No usable data available from RCT.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.6.28 Secondary outcome: Anxiety

#### Timepoint 1: *1 day - 2 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}
forest(`meta_comp_taar1_vs_antipsychotic_1 day-2 weeks_anxiety`, 
       col.diamond = "firebrick4",
       col.square =   "firebrick3",     
       label.c="amisulpride",      
       rightcols = c("effect", "ci", "Overall"),rightlabs = c("OR", "95%-CI", "Risk of bias"),
      #xlim=c(0.2,5),
                 label.e = unique(`meta_comp_taar1_vs_antipsychotic_1 day-2 weeks_anxiety`$data$treat1_new), 
                 subgroup.hetstat = TRUE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = FALSE, prediction.subgroup = FALSE,
                 subgroup = TRUE, fixed=TRUE,
                 label.left = "TAAR1 better", label.right="Antipsychotic better")

```

**Figure 58** Forest plot for anxiety as adverse event for the comparison of TAAR1 agonist vs. antipsychotics at 1 day to 2 weeks. The overall risk of bias of the studies for this outcome is presented. OR: Odds ratio, 95%CI: 95% confidence interval.

#### Timepoint 2: *3 weeks - 13 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}

forest(`meta_comp_taar1_vs_antipsychotic_3-13 weeks_anxiety`, 
       col.diamond = "firebrick4",
       col.square =   "firebrick3",
                 rightcols = c("effect", "ci","Overall"), 
                 rightlabs = c("OR", "95%-CI", "Risk of bias"),  
       label.c = "risperidone",
       xlim=c(0.2,5),
                 label.e = unique(`meta_comp_taar1_vs_antipsychotic_3-13 weeks_anxiety`$data$treat1_new), 
                 subgroup.hetstat = TRUE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = FALSE, prediction.subgroup = FALSE,
                 subgroup = TRUE, fixed=TRUE,
                 label.left = "TAAR1 better", label.right="Antipsychotic better")

```

**Figure 59** Forest plot for anxiety as adverse event for the comparison of TAAR1 agonist vs. antipsychotics at 3-13 weeks (primary timepoint). The overall risk of bias of the studies for this outcome is presented. OR: Odds ratio, 95%CI: 95% confidence interval.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.6.29 Secondary outcome: Agitation

#### Timepoint 1: *1 day - 2 weeks*

No usable data available from RCT.

#### Timepoint 2: *3 weeks - 13 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}
forest(`meta_comp_taar1_vs_antipsychotic_3-13 weeks_agitation`, 
       col.diamond = "firebrick4",
       col.square =   "firebrick3",
                 rightcols = c("effect", "ci","Overall"), 
                 rightlabs = c("OR", "95%-CI", "Risk of bias"),  
       label.c = "risperidone",
       #xlim=c(0.2,5),
                 label.e = unique(`meta_comp_taar1_vs_antipsychotic_3-13 weeks_agitation`$data$treat1_new), 
                 subgroup.hetstat = TRUE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = FALSE, prediction.subgroup = FALSE,
                 subgroup = TRUE, fixed=TRUE,
                 label.left = "TAAR1 better", label.right="Antipsychotic better")


```

**Figure 60** Forest plot for agitation for the comparison of TAAR1 agonist vs. antipsychotics at 3-13 weeks (primary timepoint). The overall risk of bias of the studies for this outcome is presented. OR: Odds ratio, 95%CI: 95% confidence interval.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.6.30 Secondary outcome: Headache

#### Timepoint 1: *1 day - 2 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}
forest(`meta_comp_taar1_vs_antipsychotic_1 day-2 weeks_headache`, 
       col.diamond = "firebrick4",
       col.square =   "firebrick3",     
       label.c="amisulpride",      
       rightcols = c("effect", "ci", "Overall"),rightlabs = c("OR", "95%-CI", "Risk of bias"),
     # xlim=c(0.2,5),
                 label.e = unique(`meta_comp_taar1_vs_antipsychotic_1 day-2 weeks_headache`$data$treat1_new), 
                 subgroup.hetstat = TRUE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = FALSE, prediction.subgroup = FALSE,
                 subgroup = TRUE, fixed=TRUE,
                 label.left = "TAAR1 better", label.right="Antipsychotic better")

```

**Figure 61** Forest plot for headache as adverse event for the comparison of TAAR1 agonist vs. antipsychotics at 1 day to 2 weeks. The overall risk of bias of the studies for this outcome is presented. OR: Odds ratio, 95%CI: 95% confidence interval.

#### Timepoint 2: *3 weeks - 13 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}
forest(`meta_comp_taar1_vs_antipsychotic_3-13 weeks_headache`, 
       col.diamond = "firebrick4",
       col.square =   "firebrick3",
                 rightcols = c("effect", "ci","Overall"), 
                 rightlabs = c("OR", "95%-CI", "Risk of bias"),  
       label.c = "risperidone",
       xlim=c(0.2,5),
                 label.e = unique(`meta_comp_taar1_vs_antipsychotic_3-13 weeks_headache`$data$treat1_new), 
                 subgroup.hetstat = TRUE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = FALSE, prediction.subgroup = FALSE,
                 subgroup = TRUE, fixed=TRUE,
                 label.left = "TAAR1 better", label.right="Antipsychotic better")

```

**Figure 62** Forest plot for headache for the comparison of TAAR1 agonist vs. antipsychotics at 3-13 weeks (primary timepoint). The overall risk of bias of the studies for this outcome is presented. OR: Odds ratio, 95%CI: 95% confidence interval.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.6.31 Secondary outcome: Sedation

#### Timepoint 1: *1 day - 2 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}
forest(`meta_comp_taar1_vs_antipsychotic_1 day-2 weeks_sedation`, 
       col.diamond = "firebrick4",
       col.square =   "firebrick3",     
       label.c="amisulpride",      
       rightcols = c("effect", "ci", "Overall"),rightlabs = c("OR", "95%-CI", "Risk of bias"),
     # xlim=c(0.2,5),
                 label.e = unique(`meta_comp_taar1_vs_antipsychotic_1 day-2 weeks_sedation`$data$treat1_new), 
                 subgroup.hetstat = TRUE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = FALSE, prediction.subgroup = FALSE,
                 subgroup = TRUE, fixed=TRUE,
                 label.left = "TAAR1 better", label.right="Antipsychotic better")
```

**Figure 63** Forest plot for sedation for the comparison of TAAR1 agonist vs. antipsychotics at 1 day to 2 weeks. The overall risk of bias of the studies for this outcome is presented. OR: Odds ratio, 95%CI: 95% confidence interval.

#### Timepoint 2: *3 weeks - 13 weeks*

No usable data available from RCT.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.6.32 Secondary outcome: Insomnia

#### Timepoint 1: *1 day - 2 weeks*

No usable data available from RCT.

#### Timepoint 2: *3 weeks - 13 weeks*

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4}
forest(`meta_comp_taar1_vs_antipsychotic_3-13 weeks_insomnia`, 
       col.diamond = "firebrick4",
       col.square =   "firebrick3",
                 rightcols = c("effect", "ci","Overall"), 
                 rightlabs = c("OR", "95%-CI", "Risk of bias"),  
       label.c = "risperidone",
       xlim=c(0.2,5),
                 label.e = unique(`meta_comp_taar1_vs_antipsychotic_3-13 weeks_insomnia`$data$treat1_new), 
                 subgroup.hetstat = TRUE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = FALSE, prediction.subgroup = FALSE,
                 subgroup = TRUE, fixed=TRUE,
                 label.left = "TAAR1 better", label.right="Antipsychotic better")
```

**Figure 64** Forest plot for insomnia for the comparison of TAAR1 agonist vs. antipsychotics at 3-13 weeks (primary timepoint). The overall risk of bias of the studies for this outcome is presented. OR: Odds ratio, 95%CI: 95% confidence interval.

#### Timepoint 3: *\> 13 weeks*

No usable data available from RCT.

### 2.6.33 Summary outcome plot: Dropouts and side-effects

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=8, warning=FALSE}
safety_primary_2<-meta_outcome %>% left_join(ae_categories) %>%  #I need to change a bit 
  mutate(event_text=ifelse(outcome=="death", "Mortality", 
                           ifelse(outcome=="serious", "Serious event", 
                                  ifelse(outcome=="dropout_any", "Dropouts any reason",
                                         ifelse(outcome=="dropout_ae", "Dropouts adverse event", event_text)))),
         MedDRA_group=ifelse(outcome=="death" | outcome=="serious", "serious",
                             ifelse(outcome=="dropout_any" | outcome=="dropout_ae", "dropouts", MedDRA_group))) %>% 
  filter(!is.na(TE.random)) %>% filter(!is.na(event_text)) %>%
  mutate(point=TE.random,
         lb=TE.random-1.96*seTE.random,
         ub=TE.random+1.96*seTE.random) %>% unique() %>%
  filter(comparison=="taar1_vs_antipsychotic") %>% 
  mutate(n_text=paste0("N=", k,", n=",n))

safety_primary_plot_2 <-safety_primary_2

safety_primary_plot_2$event_text <- factor(safety_primary_plot_2$event_tex, levels=unique(safety_primary_plot_2$event_tex))
level_order_2<- sort(unique(safety_primary_plot_2$event_text))
lever_order_meddra<-c("dropouts", "any", "serious", "autonomic", "endocrinic", "neuromuscular", "neuropsychiatric")

safety_primary_plot_2<-safety_primary_plot_2 %>% mutate(type_outcome=ifelse(outcome=="dropout_any" | outcome=="dropout_ae", "dropout", "side"))

ggplot(data=safety_primary_plot_2, 
       aes(y=factor(event_text, level=rev(level_order_2)), x=exp(point),colour=type_outcome, shape=timepoint))+
  geom_text(aes(x=0.01, label=n_text), colour="black" , position=position_dodge(width = 0.9))+
  geom_point(aes(size=n), position=position_dodge(width = 0.5))+
  geom_errorbar(aes(x=exp(point), xmin=exp(TE.random-1.96*seTE.random), 
                    xmax=exp(TE.random+1.96*seTE.random), colour=type_outcome,shape=timepoint),
                width=0, position=position_dodge(width = 0.5))+
  geom_vline(xintercept=1, linetype='dashed')+
  scale_size_continuous(guide=FALSE, range=c(1,3))+
  facet_grid(factor(MedDRA_group, levels=lever_order_meddra)~., scales = "free", space = "free")+
  scale_x_log10() +
  ylab("")+
  scale_shape_manual(values=c(1,19))+
  scale_colour_manual(guide=FALSE, values=c("green4", "firebrick4"))+
  xlab("OR 95%CI (OR>1: TAAR1 agonists worse)")+
  theme_bw()



```

**Figure 65** Summary forest plot for dropouts and side-effects for the comparison of TAAR1 agonists compared to antipsychotics and for the two timepoints with usable data. There were data for single doses of ralmitaront versus amisulpride in healthy volunteers at 1 day, and ulotaront versus risperidone in acute schizophrenia or schizoaffective disorder at 4 weeks. The size of the points is proportional to the number of participants. OR\>1 corresponds to more events with TAAR1 agonists compared to antipsychotics. OR: Odds ratio, 95%CI: 95% confidence interval. N=number of studies, n=number of participants.

## 2.7 Mechanistic insights into the effects of TAAR1 agonism for psychosis

There were no usable data available from RCTs on neurobiological measures relevant to psychosis and neurotransmitter signalling that could be analyzed as secondary outcomes of this review. Therefore, to be able to provide additional insights into the mechanism of TAAR1 agonism for psychosis, we sought to describe any potentially available data from both controlled and uncontrolled studies, irrespective of whether they have been included or not in the review (see above).

Additional evidence from two neuroimaging studies [@RN115; @RN132] generally supports the notion that TAAR1 agonists can ameliorate the elevated striatal dopamine synthesis and release capacity that is considered central in the development of symptoms of psychosis [@howes2022].

@RN115 was double-blind, three-arm parallel RCT examining the effects of a single dose of ulotaront 50mg compared to placebo and amisulpride 400mg on fMRI blood oxygen level-dependent (BOLD) striatal responses to reward processing, as measured during the anticipation phase of the Monetary Incentive Delay (MID) task (primary outcome). BOLD striatal responses to reward during the anticipatory phase of MID were selected as the primary outcome due to their potential relation to increased dopamine release in the striatum. The study enrolled 96 healthy volunteers, 51 with low schizotypy traits and 45 with high, as measured with the Schizotypal Personality Questionnaire (scores \<10 indicate low schizotypy traits, \>=40 high schizotypy traits) to examine interactions due to the potential shared increased striatal activity between positive symptoms and schizotypal traits. The interventions were administered 2 hours before the fMRI. Moreover, the study measured responses to other brain regions, the outcome phase of MID, BOLD responses to the N-back task for working memory, responses to an emotional test battery, resting state connectivity, and arterial spin labelling.

The findings of the primary outcome of this study are reported here as they were deemed most relevant to dopaminergic signalling. Ulotaront appeared to be associated with smaller BOLD striatal responses compared to placebo (left striatum: -0.13%, p=0.03, SMD=-0.55, 95%CI: -1.05, -0.05; right striatum: -0.12%, p=0.06, SMD=-0.44, 95%CI: -0.94, 0.06) but not to amisulpride (left: -0.07%, p=0.38, SMD=-0.22, 95%CI: -0.71, 0.27; right: -0.03%, p=0.66, SMD: -0.11, 95%CI: -0.60, 0.38) in the anticipation of loss, and did not differ from both placebo (left: -0.07%, p=0.30, SMD=-0.25, 95%CI: -0.75, 0.25; right: -0.07%, p=0.27, SMD=-0.25, 95%CI: -0.75, 0.25) and amisulpride (left: -0.03%, p=0.65, SMD=-0.11, 95%CI: -0.60, 0.38; right: -0.02%, p=0.74, SMD=-0.08, 95%CI: -0.57, 0.41) in the anticipation of win in the MID task. There were also no differences in BOLD insular responses (anticipation of loss - ulotaront vs. placebo left 0.00%, p=0.99, SMD=0.00, 95%CI: -0.50, 0.50 and right -0.10%, p=0.16 SMD=-0.36, 95%CI:-0.86, 0.14. ulotaront vs. amisulpride left 0.01%, p=0.94, SMD=0.02, 95%CI: -0.48, 0.52 and right -0.08%, p=0.42, SMD=-0.20, 95%CI: -0.70, 0.30; anticipation of win - ulotaront vs. placebo left: 0.08%, p=0.18, SMD=0.34, 95%CI: -0.16, 0.84, and right -0.01%, p=0.90, SMD=-0.04, 95%CI: -0.54, 0.46, ulotaront vs. amisulpride left 0.02%, p=0.84, SMD=0.05, 95%CI: -0.45, 0.55, and right 0.03%, p=0.72, SMD=0.09, 95%CI: -0.40, 0.58), but ulotaront seemed to increase BOLD responses in the medial orbital frontal cortex in anticipation of loss compared to placebo (left: 0.19%, p=0.07, SMD=0.44, 95%CI: -0.06, 0.94; right: 0.17%, p=0.10, SMD=0.43, 95%CI: -0.07, 0.93) and amisulpride (left: 0.25%, p=0.03, SMD=0.56, 95%CI: 0.06, 1.06; right: 0.22%, p=0.05, SMD=0.50, 95%CI: 0.00, 1.00) and in anticipation of win compared to placebo (left: 0.25%, p=0.01, SMD=0.64, 95%CI: 0.14, 1.14; right: 0.24%, p=0.01, SMD=0.68, 95%CI: 0.18, 1.18), but not clearly to amisulpride (left: 0.10%, p=0.30, SMD=0.26, 95%CI: -0.24, 0.76; right: 0.11%, p=0.19, SMD=0.33, 95%CI: -0.16, 0.82). Interactions with schizotypy traits were not reported for the primary outcome, but ulotaront was reported to reverse in comparison to placebo the aberrant increased BOLD response in the right insula observed in participants with high schizotypy traits but not in those with low schizotypy traits.

However, it is unclear if the primary outcome was selected *a priori* or *a posteriori*, as no primary outcome was reported in the registration of the study on clinicaltrials.gov, thus biases due to seletected reported outcomes cannot be excluded. There were also differences between the interventions in some of the other measures, with ulotaront being reported as associated with the modification of signals in brain regions associated with cognitive symptoms in schizophrenia. Moreover, ulotaront was associated with sedation in 60% of the participants, more frequently than placebo (see **Figure 36**) and amisulpride (see **Figure 63**) that impaired reaction times in the tasks but was not correlated with BOLD responses. Nevertheless, the impact of this sedative effect on these findings is unclear.

In conclusion, this study suggests that ulotaront might be able to reduce dopamine release in situations that are increased, like the anticipatory phase of MID, and potentially in psychosis.

@RN132 was an open-label single-arm trial examining the effects of a 2-week treatment with ulotaront (50-75mg/d) as an add-on to antipsychotic treatment on dopamine synthesis capacity in clinically stable patients with schizophrenia (an uncontrolled study that was not included in this meta-analysis, with available data from a conference abstract in ACNP 2024 [@RN109]]). This study included 14 men and 5 women (aged 18-45 years, mean age 32.9 years) with clinically stable schizophrenia (DSM-5) on a stable dose of antipsychotics for at least 3 weeks before screening. Three additional patients were enrolled but did not complete the baseline scan, so they were excluded from the study. The patients underwent an F-DOPA PET scan both at baseline and after 2 weeks of treatment with ulotaront (50-75mg/d) as an add-on to antipsychotic treatment. At baseline, the patients had a mean PANSS total score of 79.3 (SD=7.3), PANSS positive subscale of 19.2 (SD=4.6), PANSS Marder positive factor of 24.3 (SD=4.6), and CGI-S of 4.2 (SD=0.6), indicating that the patients had on average symptoms of moderate severity.

After 2 weeks of treatment with add-on ulotaront, there was a decrease in dopamine synthesis capacity in the striatum as indicated by the F-DOPA PET scan by -3.98% from baseline (95% CI: -8.68%, 0.72%), with the 95% CI not excluding the null effect, and the striatal subregions ranging from -3.38% in the associative striatum, -3.89% in the sensorimotor striatum, to -5.79% in the limbic striatum. This reduction was considered to be clinically meaningful, and although it was not correlated with the relative reduction in PANSS total scores (r=0.19, p\>0.05), it was correlated with improvements in the PANSS Marder positive factor (r=0.49, p\<0.05).

The study can be considered to have a high risk of bias due to the single-arm design, lack of control conditions, and the absence of blinding, yet the measurement of the outcome can be considered objective.

In conclusion, this study suggests that ulotaront may be able to reduce the dopamine synthesis capacity in clinically stable, but still symptomatic, people with schizophrenia, and the decrease in dopamine synthesis could be associated with the improvement of symptoms.

No other controlled or uncontrolled study (see *included and excluded studies* in the supplementary files of the review) reported any neurobiological measure. It should be noted that another study, i.e., NCT02699372, consisting of double-blind trials examining single and multiple ascending doses of ralmitaront compared to placebo in healthy volunteers (see *Table 1*), utilized neuropsychological tasks but did not couple them with neuroimaging or other neurobiological measures to provide further insights. There were no usable data for this study for any of the outcomes; only brief and narrative descriptions of the findings of these tasks were available in the study protocol of @RN11 . These findings describe that ralmitaront improved reward-related behavioral tasks compared to placebo but lacked additional information.

## 2.8 Subgroup analyses and meta-regressions

There were less than 5 studies available for the analyses of the primary outcome, precluding the exploration of potential reasons of heterogeneity with subgroup analyses and meta-regressions.

Nevertheless, we explored potential differences between ulotaront and ralmitaront in improving overall symptoms of acute schizophrenia, and potential dose-effects by plotting the effect sizes for each fixed-dose level of TAAR1 agonist compared to placebo. However, we did not conduct any formal subgroup test or dose-response meta-analysis due to the limited number of studies.

Moreover, it should be noted that we presented the effect sizes in each specific subgroup of participants for all outcomes, but we did not perform a formal subgroup test

### 2.8.1 Exploration of differences between ulotaront and ralmitaront in their effects on overall symptoms of acute schizophrenia

```{r echo=F, fig.fullwidth=T, fig.width=12, fig.height=4}
forest(meta_comp_drug, 
       col.diamond = "royalblue4",
       col.square =   "royalblue1",
                 rightcols = c("effect", "ci","Overall"), 
                 rightlabs = c("SMD", "95%-CI", "Risk of bias"),  
       xlim=c(-0.8,0.8),
                 label.e = unique(meta_comp_drug$data$treat1_new), 
       label.c=unique(meta_comp_drug$data$treat2_new),
                 subgroup.hetstat = TRUE, overall.hetstat =FALSE, test.subgroup.random = FALSE,test.subgroup.fixed = FALSE,
                 overall = FALSE, prediction.subgroup = FALSE,
                 subgroup = TRUE, fixed=TRUE,
                 label.left = "Placebo better", label.right="TAAR1 better")
```

**Figure 66** Forest plot presenting the effects of ulotaront and ralmitaront compared to placebo on overall symptoms in acute schizophrenia (measured with PANSS total) at 4-6 weeks separately. SMD: Standardized mean difference, 95%CI: 95% confidence intervals, SD: standard deviation.

### 2.8.2 Exploration of dose-effects of ulotaront and ralmitaront on overall symptoms of acute schizophrenia

```{r echo=F, fig.fullwidth=T, fig.width=11, fig.height=4, warning=FALSE}
ggplot(data=data_doses, aes(x=dose,
                                 y=-yi,
                                 ymin=-yi+1.96*sqrt(vi), ymax=-yi-1.96*sqrt(vi), 
                                 group=study_name, colour=study_name))+
  geom_pointrange(position=position_dodge(width = 1.2), aes(shape=schedule), size=1)+
  xlim(c(0,150))+
  coord_cartesian(ylim=c(-0.5,0.5), expand=TRUE)+
  xlab("dose (mg//d)")+
  ylab("SMD 95%CI (SMD>0: TAAR1 agonists better")+
  scale_shape_manual(values = c(15,0))+
  geom_hline(yintercept = 0, linetype="dashed")+
  facet_wrap(~drug_name,  scales = "free", nrow=2)+
  theme_bw()
```

**Figure 67** Effect sizes for each dose level of ulotaront and ralmitaront compared to placebo in improving overall symptoms of acute schizophrenia at 4-6 weeks are presented separately. The findings of each study are presented separately, and the effect sizes were corrected for using the same control condition in each study. It should be noted that @RN110 was a two-arm flexible-dosing study, and its effect size is placed in the middle of the range of the administered dose (50mg/d to 75mg/d). SMD: Standardized mean difference, 95%CI: 95% confidence intervals.

In the fixed-dosing studies [@RN111; @RN112; @RN113], higher doses (i.e., ulotaront 100mg/d and ralmitaront 150mg/d) may be associated with more improvement compared to smaller doses. However, the effect sizes at these high doses still appear to be small, with a standardized mean difference (SMD) of approximately 0.2. In the flexible dosing study by @RN110 , which utilized doses of ulotaront up to 75mg/d, small-to-medium effect sizes were observed. Importantly, the available data are limited. Therefore, the findings are inconclusive.

## 2.9 Sensitivity analyses

We did not conduct the following sensitivity analyses for the primary outcome due to the limited data: 1) Restricting the analysis to studies with low risk of bias, since there were 4 studies with low risk of bias in acute schizophrenia [@RN110; @RN111; @RN112; @RN113], and 1 study with high risk of bias in Parkinson's disease psychosis (1 study with high risk of bias) [@RN13]. 2) Excluding imputed data, since no imputed data were used in the analysis of the primary outcome. There were no imputed data in the analysis of the primary outcome.

Due to the limited number of studies, we also presented a fixed-effects model for all outcomes in addition to the random-effects model to assess the robustness of the findings. The findings were robust in the majority of cases. For the primary outcome in the meta-analysis of 4 studies in acute schizophrenia [@RN110; @RN111; @RN112; @RN113], the point estimate of the effect sizes did not differ between random- and fixed-effects (SMD:0.15 and 0.13, respectively), but the 95% CI in the fixed-effects model was narrower, excluding the null effect (95% CI: -0.05, 0.34, and 0.02, 0.05, respectively). Nonetheless, the interpretation of the findings that TAAR1 agonists have little to no effect compared to placebo in improving overall symptoms does not change

## 2.10 Reporting bias

Reporting bias was evaluated using the ROB-ME tool [@page2023], taking into account the usable data from the eligible studies identified in the search, the comprehensiveness of the search, and potential patterns of missingness and small-study effects. We followed the algorithm to assign a low or high risk of reporting bias, or to express some concerns, by answering signalling questions related to the domains mentioned above.

### 2.10.1 Results matrix

We identified the total number of eligible and completed studies, with or without usable data, that could contribute to dropout and safety data for each comparison (as shown in **Table 1** and **Table 2**): 1) There were 12 studies with 626 participants for the comparison of TAAR1 agonists versus placebo over a period of 1 day to 2 weeks. 2) There were 2 studies with 100 participants for the comparison of TAAR1 agonists versus placebo over the same duration. 3) For the comparison of TAAR1 agonists versus placebo over 3-13 weeks, there were 6 studies involving 1550 participants, including also 1 study (NCT03669640) focusing predominantly on negative symptoms. 4) Finally, it should be noted that one study (NCT04115319) with 475 participants could contribute to the comparison of TAAR1 agonists versus other antipsychotics for periods exceeding 13 weeks. However, no usable data for this analysis were available from any other study.

The results matrix for the eligible studies, which were identified in the search, for all outcomes with usable data from at least one study, can be found below in **Table 2**.

```{r echo=FALSE}
knitr::kable(results_matrix)
```

**Table 2** Results matrix for the primary outcome. ✓: A study result is available for inclusion in the meta-analysis.; \~:No study result is available for inclusion in the meta-analysis, for a reason unrelated to the P value, magnitude or direction of the result.; ?: Unclear whether an eligible study result was generated or unclear reason of missing evidence. Narrative descriptions of the effects of drugs as reported in the studies are mentioned here to facilitate the judgement of regarding the reason of missingness. \* @RN111 included two parts, i.e., adults (sample size 435) and adolescents (sample size unclear but it was probably 28; planned 90). It should be noted that all of the other studies included adult populations. na: not applicable for this outcome (e.g., efficacy outcomes in healthy volunteers or after a single dose in stable patients).

### 2.10.2 Funnel plots and examination of small-study effects

There were insufficient data for the primary outcome but also other outcomes (i.e., fewer than 10 studies) to examine small-study effects with funnel plots. Thus, we cannot explore the pattern of missing evidence or conduct sensitivity analysis to explore the potential impact of the missing evidence. Thus, we replied the following signalling questions as ROB-ME-4.7: "probably no" and ROB-ME-4.8: "probably no".

### 2.10.3 Coverage of the search strategy

We conducted a comprehensive search for both published and unpublished studies in multiple electronic databases, supplemented by manual searches of the reference lists of included studies and previous reviews. Screening was carried out by at least two independent reviewers. All studies identified in the search were conducted by the pharmaceutical industry. Although we contacted them, we have not received additional data for the first iteration. Considering that all eligible studies are experimental and were conducted recently, it is expected that most, if not all, have been registered, thus facilitating their detection during searches in clinical trial registries. Nonetheless, it's possible that small phase I studies focusing on pharmacokinetics may not have been registered and could therefore have been overlooked. Consequently, we do not anticipate that our search has missed any major or significant studies (ROB-ME-3.1: "no", ROB-ME-3.2: "probably yes", ROB-ME-3.3:"probably yes", ROB-ME-4.5: "no").

### 2.10.4 ROB-ME evaluation

Data were available for all eligible studies concerning the primary outcome, i.e., answers to the ROB-ME signalling question 4.1: "no", ROB-ME-4.3: "yes" for adolescents, ROB-ME-4.4: "no", thus leading to low risk of reporting bias - considering also the comprehensiveness of the search.

The results matrix in **Table 2** could also apply to the secondary efficacy outcomes in acute psychosis, although some studies did not have usable data for these outcomes. Specifically, the two failed phase III trials, [@RN111; @RN112], did not report effects on symptom domains. Consequently, these analyses would be prone to bias due to missing evidence (ROB-ME-4.1: "yes", ROB-ME-4.2: "yes", thus leading to *high risk* of reporting bias).

The reasons for the missingness of data for these outcomes can be unclear, but in most cases, it is due to either not being measured or not reported for reasons other than their P-values, magnitude, and direction of effects, yet this cannot be excluded. Therefore, we employed, as a general rule, the following approach in response to ROB-ME-4.1: "no" (as none was generally considered to have reported usable data for these outcomes due to their P-value, magnitude, or direction), ROB-ME-4.3: 'yes', ROB-ME-4.4: ''NI" if less than 50% of the participants had usable data, suggesting the possibility that the findings could notably change if the missing data had been added to the analysis. The cut-off of 50% of the participants was selected to balance out that in most cases, the findings were not missing due to unfavorable findings (see **Table 2**) but also to consider the possibility that the findings may be likely to change with the addition of missing evidence, given the small number of studies and participants in their estimates. In this situation, we assigned some concerns or *moderate risk* of missing evidence given that ROB-ME-4.5: "no" (see above, 2.10.3), and patterns of missing evidence or sensitivity analyses could not be conducted. However, the impact of this on the magnitude and direction of the findings can be unclear. If there were data from more than 50% of the participants, we responded to ROB-ME-4.4: 'probably no', leading to *low risk* given that ROB-ME-4.5: "no" , and patterns of missing evidence or sensitivity analyses could not be conducted.

Finally, *low risk* of reporting bias was judged for the outcomes used to provide mechanistic insights for the same reasons described above. The search was comprehensive and no other study was found to measure neurobiological outcomes.

## 2.11 Summary of evidence

### 2.11.1 Approach to evaluate the confidence in the evidence

For each comparison, outcome and timepoint, we presented the summary of the evidence by presenting the summary of the association, biases due to study limitations, biases due to reporting bias, biases due to indirectness and other biases.

**Summary of the association**: We presented the number of studies (N), participants (n) contributing to the analysis, the point estimates and 95% confidence intervals for both random- and fixed-effects models. We also made a statement about dose-effects for the primary outcome (**Figure 67**).

**Within-study biases**: This refers to the study limitations measured with the risk of bias assessments using RoB2 [@RN64]. The majority of the studies had a low risk of bias, or at most, some concerns across domains (mainly in missing outcome data, and also in the randomization process, period, and carry-over effects in @RN124) and outcomes, except for one small study [@RN13] with a high risk of bias due to missing outcome data for the efficacy outcomes. Thus, we assigned "high risk" for within-study biases when the efficacy outcomes were solely based on [@RN13]. On other occasions where studies had a low risk or at most some concerns, mainly in the domain of missing outcome data, we assigned "low risk" of within-study biases when no more than 50% of the studies were rated with some concerns, and "moderate risk" otherwise. In most cases, some concerns were raised due to missing outcome data. A recent meta-epidemiological study suggested that biases due to missing outcome data might underestimate the effects [@wang2024]. In schizophrenia trials, there is evidence that participants who dropped out may have a worse prognosis compared to those who stayed in the trial [@rabinowitz2008]. However, there were generally no clear differences in dropouts between TAAR1 agonists and control conditions, although TAAR1 may be associated with higher dropout rates compared to placebo (**Figure 11**). Therefore, we judged that a high risk of bias due to missing outcome data in efficacy outcomes in @RN13, could have possibly overestimated the effects of TAAR1 agonists compared to placebo in favor of the former. However, in other cases, when there were some concerns for safety, it is unclear whether missing outcome data could have led to biased findings.

**Across-study bias**: This is referring to reporting bias, and we used the approach mentioned above to assign "low risk", "moderate risk" or "high risk".

**Indirectness**: We assigned "no concerns" regarding indirectness for the efficacy outcomes, as there were no issues related to population, interventions, control conditions, outcomes, or settings in the clinical trials involving adults with an acute episode of schizophrenia or Parkinson's disease psychosis. However, potential indirectness in terms of population characteristics may exist in the analysis of other outcomes. This is because we pooled data from diverse populations, including those with schizophrenia spectrum disorders, Parkinson's disease psychosis, narcolepsy-cataplexy, and healthy volunteers. Although the majority of adverse events might not vary significantly across these groups, we considered "moderate risk" regarding indirectness when the analyses included fewer than 50% of participants with schizophrenia. This is because schizophrenia can be considered the primary population to which these findings would be applied. Nevertheless, the impact on the magnitude and direction of the effects remains unclear.

**Other biases**: We considered additional biases, such as potential factors that could impact the magnitude and direction of the findings. Specifically, we regarded "moderate risk" for other biases for the primary outcome and response to treatment in adults with an acute episode of schizophrenia due to potential problems in the execution of two phase III trials on ulotaront during COVID-19 [@RN111; @RN112]. These trials had high placebo effects of about 14 points and 19 points reduction in PANSS total scores (compared to a mean of approximately 6 points in antipsychotic trials, [@leucht2018]), suggesting that potential factors associated with placebo effects could have diluted effect sizes and underestimated the effects. A conference abstract at ACNP 2024 [@RN109] reported that pooled analyses of these two trials [@RN111; @RN112] using participants enrolled before COVID-19 found effect sizes similar to those in the phase II trial of @RN110 . Potential explanations, among others, could include the pressure for recruitment during the pandemic that might have led to baseline inflation and less careful recruitment, difficulties in the execution and monitoring of the trial such as protocol deviations and co-treatments, or other factors [@leucht2013; @leucht2019]. It should be noted that another trial was conducted during COVID-19 [@RN113], but the potential impact of the pandemic on its execution has not been reported. Nonetheless, the possibility that the findings have been underestimated cannot be excluded.

### 2.11.2 Summary of evidence tables

#### Summary of evidence table for efficacy outcomes

##### TAAR1 agonists versus placebo in adults with an acute episode of schizophrenia

```{r echo=FALSE, fig.width=12, fig.height=8}
knitr::kable(soe_sch_ta1_pl_eff)
```

**Table 3** Summary of evidence table for the comparison of TAAR1 agonists versus placebo for efficacy outcomes in adults with an acute episode of schizophrenia. There were data for ulotaront and ralmitaront. There is an uncertain/unclear dose-response indication (see **Figure 67**). Across-study bias was evaluated using the ROB-ME approach [@page2023]. N=number of studies, n=number of participants. SMD: Standardized mean difference, OR: Odds ratio, 95%CI: 95% confidence interval.

##### TAAR1 agonists versus antipsychotics in adults with an acute episode of schizophrenia

```{r echo=FALSE, fig.width=12, fig.height=8}
knitr::kable(soe_sch_ta1_apd_eff)
```

**Table 4** Summary of evidence table for the comparison of TAAR1 agonists versus antipsychotics for efficacy outcomes in adults with an acute episode of schizophrenia. There were data for ralmitaront versus risperidone in a single study [@RN113]. Across-study bias was evaluated using the ROB-ME approach [@page2023]. N=number of studies, n=number of participants. SMD: Standardized mean difference, OR: Odds ratio, 95%CI: 95% confidence interval.

##### TAAR1 agonists versus placebo for efficacy outcomes in Parkinson's disease psychosis

```{r echo=FALSE, fig.width=12, fig.height=8}
knitr::kable(soe_pd_ta1_pl_eff)
```

**Table 5** Summary of evidence table for the comparison of TAAR1 agonists versus placebo for efficacy outcomes in Parkinson's disease psychosis. There were data from a single study for ulotaront [@RN13]. Across-study bias was evaluated using the ROB-ME approach [@page2023]. N=number of studies, n=number of participants. SMD: Standardized mean difference, OR: Odds ratio, 95%CI: 95% confidence interval.

#### Summary of evidence table for dropouts and side-effects

##### TAAR1 agonists versus placebo for dropouts and side-effects in adults with psychosis, other mental health conditions and healthy volunteers

```{r echo=FALSE, fig.width=12, fig.height=8}
knitr::kable(soe_all_ta1_pl_saf)
```

**Table 6** Summary of evidence table for the comparison of TAAR1 agonists versus placebo for dropouts and side-effects in adults with psychosis, other mental health conditions and healthy volunteers. There were data for ulotaront and ralmitaront. Across-study bias was evaluated using the ROB-ME approach [@page2023]. N=number of studies, n=number of participants. MD: Mean difference, OR: Odds ratio, 95%CI: 95% confidence interval.

##### TAAR1 agonists versus antipsychotics for dropouts and side-effects in adults with psychosis, other mental health conditions and healthy volunteers

```{r echo=FALSE, fig.width=12, fig.height=8}
knitr::kable(soe_all_ta1_apd_saf)
```

**Table 7** Summary of evidence table for the comparison of TAAR1 agonists versus antipsychotics for dropouts and side-effects in adults with psychosis, other mental health conditions and healthy volunteers. There were data from two studies, one for ulotaront vs. amisulpride (single dose) [@RN115], and one for ralmitaront vs. risperidone (4 weeks) [@RN113]. Across-study bias was evaluated using the ROB-ME approach [@page2023]. N=number of studies, n=number of participants. OR: Odds ratio, 95%CI: 95% confidence interval.

#### Summary of evidence table for the mechanistic insights of TAAR1 agonism for psychosis

Evidence from two neuroimaging studies [@RN115; @RN132] generally supports the notion that TAAR1 agonists can ameliorate the elevated striatal dopamine synthesis and release capacity that is considered central in the development of symptoms of psychosis [@howes2022]. The summary of the evidence of these studies is presented below.

```{r echo=FALSE, fig.width=12, fig.height=8}
knitr::kable(soe_mechanistic)
```

**Table 8** Summary of evidence table for the mechanistic insights of TAAR1 agonism for psychosis. The findings of two studies [@RN115; @RN132] with available neurobiological measures relevant to psychosis are narratively reported. These include the double-blind randomized controlled trial of @RN115 with the fMRI BOLD striatal responses in the anticipatory phase of the Monetary Incentive Delay (MID) task (proxy for increased dopaminergic signalling in the striatum), and the open label single-arm study of @RN132 with F-DOPA PET in the striatum. Across-study bias was evaluated using the ROB-ME approach [@page2023]. N=number of studies, n=number of participants. SMD=Standardized mean difference, 95%CI: 95% confidence interval. p-value: p-value for the correlation coefficient.

# 3. Abbreviations

BOLD: Blood Oxygen Level-Dependent

DB: Double-Blind

DSM-IV-TR/DSM-5: Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision / Fifth Edition

ECG: Electrocardiogram

F-DOPA: Fluorodopa

GAD: Generalized Anxiety Disorder

GRADE approach: Grading of Recommendations Assessment, Development, and Evaluation approach

kg: Kilograms

MDD: Major Depressive Disorder

MD: Mean Difference

MID: Monetary Incentive Delay task

mg: Milligrams

MMSE: Mini-Mental State Examination

msec: Milliseconds

na: Not applicable/available

ng/ml: Nanograms per Milliliter

N: Number of studies

n: Number of participants

NNH: Number Needed to Harm

NPI: Neuropsychiatric Inventory

OR: Odds Ratio

PANSS: Positive and Negative Syndrome Scale

PET: Positron Emission Tomography

REML: Restricted Maximum Likelihood

RCT: Randomized Controlled Trial

RoB2: Risk of Bias 2

ROB-ME: Risk Of Bias due to Missing Evidence

SAPS-PD: Scale for the Assessment of Positive Symptoms in Parkinson's Disease

SD: Standard Deviation

SMD: Standardized Mean Difference

TAAR1: Trace amine-associated receptor 1

# 4. References

```{r echo=FALSE}
grateful::cite_packages(output = "paragraph", pkgs = "Session", out.dir = ".")
```