Effectiveness of Psychotherapies and Psychotropic Drugs against Mental Illness/Psychotherapies for suicide prevention

Effectiveness of Psychotherapies and Psychotropic Drugs against Mental Illness Psychotherapies for suicide prevention Psychotherapies for eating disorders Antidepressants Loading libraries library(dplyr) library(metafor) library(meta) Part 1: Data At https://docs.metapsy.org/databases/suicide-psyctr/ you can download documents in PDF format and data in ZIP format related to a meta-analysis conducted by researchers entitled "Suicide Prevention: Psychological Intervention vs. Control." This meta-analysis analyzed 199 clinical studies on the effectiveness of psychotherapy against suicidal ideation and suicide attempts, comparing groups receiving psychotherapy and control groups. The CSV dataset contains the Hedges' effect size "g" with its corresponding standard deviation for each study. The variable "condition_arm1" contains the type of psychotherapy used in the study: Cognitive: Cognitive (includes Cognitive Behavioral Therapy - CBT and Cognitive Therapy - CT). Psychodynamic: Psychodynamic. Mixed/other: Mixed or other. ASSIP: ASSIP (Attempted Suicide Short Intervention Program - Brief intervention program for those who have attempted suicide). Family: Family-related. Problem solving: Problem solving. Mindfulness: Mindfulness. Mentalization-based: Mentalization-based. EMDR and imagery-based: EMDR and imagery-based. Safety planning: Safety planning. Meaning-making: Searching/constructing meaning. CAMS: Collaborative Assessment and Management of Suicidality. Dialectic: Dialectical (referring to Dialectical Behavior Therapy - DBT). Interpersonal: Interpersonal , the variable condition_arm2 which contains the type of group the psychotherapy group is comparing to: Waiting list: Waiting list. CAU: Care As Usual (or Standard Treatment). Enhanced CAU: Enhanced Care As Usual (a form of standard treatment with additional elements). Supportive counseling: Supportive counseling. Other: Other. Placebo: Placebo To calculate the risk ratio (RR), I use the four variables "event_arm1, event_arm2, total_arm1, total_arm2," which are respectively: Number of events (responders, remission, and deterioration cases) in the first trial arm, Number of events (responders, remission, and deterioration cases) in the second trial arm, Total number of participants in arm 1 (event data), and Total number of participants in arm 2 (event data). The variable "outcome_type" contains ideation if the study concerns suicidal ideation, for which g is calculated, or attempts if the study concerns suicide attempts, for which RR is calculated. Loading and viewing the first 6 studies: df <- read.csv("data.csv", sep = ";", dec = ",", stringsAsFactors = FALSE) df %>% select(study,.g,.g_se,event_arm1,totaln_arm1, event_arm2,totaln_arm2, condition_arm1,condition_arm2,outcome_type) %>% head() study .g .g_se event_arm1 totaln_arm1 event_arm2 totaln_arm2 condition_arm1 condition_arm2 outcome_type Alavi 2013 -2.93050147860304 0.533409903311024 NA NA NA NA Cognitive Waiting list ideation Asbestos 2011 NA NA 6 18 7 17 Psychodynamic CAU attempts Andreoli 2016 NA NA 6 70 4 30 Psychodynamic CAU attempts Andreoli 2016 NA NA 3 70 4 30 Psychodynamic CAU attempts Ardashir 2018 -2.16004560814976 0.400887578748478 NA NA NA NA Cognitive CAU ideation Arnevik 2009 NA NA 1 58 3 46 Mixed/other Enhanced CAU attempts Part 2: Meta-analysis Calculating Hedges' g We calculate Hedges' g in meta-analyses of selected studies with outcome_type=ideation and perform the Egger test to assess whether there is publication bias due to the fact that studies with positive results and high g are more likely to be published than those with modest or negative results. Since the p-value is <0.05, the test does indeed have publication bias, so we use Duval and Tweedie's "trim and fill" procedure, which estimates the number of missing studies and recalculates g, obtaining g = -0.21 with a 95% confidence interval (CI): [-0.33, -0.09]. Therefore, in this case, the effect of psychotherapy on suicidal ideation is small but statistically significant because the CI does not include 0: # --- ANALYSIS FOR SUICIDAL IDEATION (Hedges' g) --- # The protocol calculates Hedges' g to measure the difference between groups. # The data are already present in the .g and .g_se columns of the CSV file. data_ideation <- df %>% filter(outcome_type=="ideation") ma_ideation <- metagen(TE = .g, seTE = .g_se, studlab = study, data = data_ideation, sm = "SMD", method.tau = "REML", # Metodo per la stima dell'eterogeneità random = TRUE) # --- PUBLICATION BIAS ASSESSMENT --- # Egger test metabias(ma_ideation, method.stat = "egger") Linear regression test of funnel plot asymmetry Test result: t = -2.89, df = 104, p-value = 0.0046 Bias estimate: -1.2020 (SE = 0.4152) Details: - multiplicative residual heterogeneity variance (tau^2 = 2.8600) - predictor: standard error - weight: inverse variance - reference: Egger et al. (1997), BMJ m_g<- trimfill(ma_ideation) print(paste("g =",round(m_g$TE.random,2),"; 95%CI: [",round(m_g$lower.random,2),",",round(m_g$upper.random,2),"]")) "g = -0.21; 95%CI: [ -0.33, -0.09]" Measuring g for the various psychotherapies on the data_ideation data frame, we find that EMDR and imagery-based therapy has the highest g (g=-1.3), ASSIP: ASSIP (Attempted Suicide Short Intervention Program) (g=-0.73), CAMS: CAMS (Collaborative Assessment and Management of Suicidality) (g=-0.46) etc. , so the effect of psychotherapy on suicidal ideation, barring publication bias, ranges from high to low: for (p in unique(data_ideation$condition_arm1)) { df_post <- data_ideation %>% filter(condition_arm1==p) m_pst<-metagen(TE = .g, seTE = .g_se, data = df_post, studlab = study, sm = "SMD", random = TRUE) print(paste(p,": g=",round(m_pst$TE.random,2),"; 95%CI: [",round(m_pst$lower.random,2),",",round(m_pst$upper.random,2),"]")) } [1] "Cognitive : g= -0.39 ; 95%CI: [ -0.57 , -0.22 ]" [1] "Mixed/other : g= -0.22 ; 95%CI: [ -0.34 , -0.09 ]" [1] "Family : g= -0.25 ; 95%CI: [ -0.44 , -0.06 ]" [1] "Problem solving : g= -0.42 ; 95%CI: [ -0.56 , -0.29 ]" [1] "Mindfulness : g= -0.11 ; 95%CI: [ -0.37 , 0.16 ]" [1] "Psychodynamic : g= 0.01 ; 95%CI: [ -0.21 , 0.24 ]" [1] "Safety planning : g= -0.41 ; 95%CI: [ -0.76 , -0.07 ]" [1] "ASSIP : g= -0.73 ; 95%CI: [ -1.53 , 0.06 ]" [1] "EMDR and imagery based : g= -1.3 ; 95%CI: [ -2.78 , 0.19 ]" [1] "CAMS : g= -0.46 ; 95%CI: [ -0.72 , -0.21 ]" [1] "Dialectic : g= -0.42 ; 95%CI: [ -0.94 , 0.1 ]" Calculation of the Risk Ratio (RR) Calculating the RR in meta-analysis of the selected studies yields RR=0.74, so in this case, the group treated with psychotherapy has a reduction in risk of 1-0.74=26% higher risk of attempting suicide compared to the control group: # SUICIDE ATTEMPTS (RR) --- # First filter the data for the outcome 'attempts' data_attempts <- df %>% filter(outcome_type == "attempts") ma_attempts_sub <- metabin( event.e = event_arm1, n.e = totaln_arm1, event.c = event_arm2, n.c = totaln_arm2, studlab = study, data = data_attempts, sm = "RR", subgroup = condition_arm1, method.tau = "DL", random = TRUE ) summary(ma_attempts_sub) Number of studies: k = 93 Number of observations: o = 39735 (o.e = 20000, o.c = 19735) Number of events: e = 2618 RR 95%-CI z p-value Common effect model 0.8440 [0.7862; 0.9061] -4.68 < 0.0001 Random effects model 0.7462 [0.6607; 0.8429] -4.71 < 0.0001 Quantifying heterogeneity (with 95%-CIs): tau^2 = 0.0791; tau = 0.2813; I^2 = 37.3% [19.2%; 51.4%]; H = 1.26 [1.11; 1.43] Test of heterogeneity: Q d.f. p-value 146.73 92 0.0003