# Information ----------------------------------------------------------------- ## Project title -------------------------------------------------------------- ### Charting the pitfalls of disproportionality analysis ## Data ----------------------------------------------------------------------- ### VigiBase, accessed 31st December 2024, deduplicated, non-domestic ## Authors -------------------------------------------------------------------- ### Michele Fusaroli, Daniele Sartori, Eugene P. van Puijenbroek, G. Niklas Norén ## Version -------------------------------------------------------------------- ### Set up: 2024-11-13 ### Last update: 2025-03-31 # Set up ---------------------------------------------------------------------- ## upload DiAna --------------------------------------------------------------- library(DiAna) FAERS_version="VigiBase" ## Project_path --------------------------------------------------------------- DiAna_path <- here::here() project_path <- paste0(DiAna_path,"/projects/Pitfalls") if(!file.exists(project_path)){dir.create(project_path)} project_path <- paste0(project_path,"/") ## Input ---------------------------------------------------------------------- import("DRUG") import("REAC") import("DEMO") Demo <- Demo[,init_fda_dt:=as.character(init_vb_date)] import("INDI") import("OUTC") import_MedDRA() import_ATC() smq_narrow <- extractSMQ() ## polio vaccine, growth retardation -------------------------------------------------- drug <- "polio vaccine" event <- "growth retardation" restrictions <- list("crude" = unique(Demo$primaryid), "neonates" = unique(Demo[age_in_days < 28]$primaryid), "infants" = unique(Demo[age_in_days >= 28 & age_in_days < 30 * 24]$primaryid), "children" = unique(Demo[age_in_days >= 30 * 24 & age_in_days < 11*365]$primaryid), "other" = unique(Demo[age_in_days >= 11*365]$primaryid), "unknown" = unique(Demo[is.na(age_in_days)]$primaryid)) disproportionality_df <- data.table() for (n in 1:length(restrictions)) { t <- restrictions[n] t_name <- names(t) t_pids <- unlist(t) df <- disproportionality_analysis( drug_selected = unlist(drug), reac_selected = unlist(event), temp_drug = Drug[role_cod %in% c("PS", "SS")], restriction = t_pids )[, nested := t_name] disproportionality_df <- rbindlist(list(disproportionality_df, df), fill = TRUE) } disproportionality_df <- disproportionality_df[,expected:=(D*E/(D_E+D_nE+nD_E+nD_nE))] render_forest(disproportionality_df,facet_v="nested",point_size = 5,xcoord_lims = c(-6,6)) create_dag("polio_vaccine","growth_retardation",label_inquiry = "", confounder_path = list(nodes=list("young_age"),signs=list("+","+"), label="Confounding by Age")) ## insulin, thrombotic stroke | diabetes -------------------------------------- drug <- "insulin nos" event <- list("ischaemic stroke"=list(smq_narrow[["Ischaemic central nervous system vascular conditions (SMQ)"]])) diabetes <- smq_narrow[["Hyperglycaemia/new onset diabetes mellitus (SMQ)"]] restrictions <- list("crude" = unique(Demo$primaryid), "diabetes" = unique(Indi[indi_pt%in%diabetes]$primaryid)) disproportionality_df <- data.table() for (n in 1:length(restrictions)) { t <- restrictions[n] t_name <- names(t) t_pids <- unlist(t) df <- disproportionality_analysis( drug_selected = unlist(drug), reac_selected = unlist(event), temp_drug = Drug[role_cod %in% c("PS", "SS")], restriction = t_pids )[, nested := t_name] disproportionality_df <- rbindlist(list(disproportionality_df, df), fill = TRUE) } no no disproportionality_df <- disproportionality_df[,expected:=(D*E/(D_E+D_nE+nD_E+nD_nE))] render_forest(disproportionality_df,facet_v="nested",point_size = 5,xcoord_lims = c(-6,6)) ## hydrochlorothiazide cough | ACEI --------------------------------------- drug <- "hydrochlorothiazide" event <- "cough" restrictions <- list("crude" = unique(Demo$primaryid), "no_ACEI" = setdiff(Demo$primaryid,Drug[substance%in%ATC[Class4 == "ACE inhibitors, plain"]$substance]$primaryid)) disproportionality_df <- data.table() for (n in 1:length(restrictions)) { t <- restrictions[n] t_name <- names(t) t_pids <- unlist(t) df <- disproportionality_analysis( drug_selected = unlist(drug), reac_selected = unlist(event), temp_drug = Drug[role_cod %in% c("PS", "SS")], restriction = t_pids )[, nested := t_name] disproportionality_df <- rbindlist(list(disproportionality_df, df), fill = TRUE) } disproportionality_df <- disproportionality_df[,expected:=(D*E/(D_E+D_nE+nD_E+nD_nE))] render_forest(disproportionality_df[nested%in%c("crude","no_ACEI")],facet_v="nested",point_size = 5,xcoord_lims = c(-6,6)) ## finasteride, breast cancer -------------------------------------------------- drug <- "finasteride" event <- "breast cancer" restrictions <- list("naive" = unique(Demo$primaryid), "male" = unique(Demo[sex=="M"]$primaryid), "female" = unique(Demo[sex=="F"]$primaryid), "unknown" = unique(Demo[is.na(sex)]$primaryid)) disproportionality_df <- data.table() for (n in 1:length(restrictions)) { t <- restrictions[n] t_name <- names(t) t_pids <- unlist(t) df <- disproportionality_analysis( drug_selected = unlist(drug), reac_selected = unlist(event), temp_drug = Drug[role_cod %in% c("PS", "SS")], restriction = t_pids )[, nested := t_name] disproportionality_df <- rbindlist(list(disproportionality_df, df), fill = TRUE) } disproportionality_df <- disproportionality_df[,expected:=(D*E/(D_E+D_nE+nD_E+nD_nE))] render_forest(disproportionality_df,facet_v="nested",point_size = 5,xcoord_lims = c(-6,6)) ## ceftriaxone, hepatitis -------------------------------------------------- drug <- "ceftriaxone" event <- "hepatitis" restrictions <- list("naive" = unique(Demo$primaryid), "neonates" = unique(Demo[age_in_days < 28]$primaryid), "infants" = unique(Demo[age_in_days >= 28 & age_in_days < 30 * 24]$primaryid), "children" = unique(Demo[age_in_days >= 30 * 24 & age_in_days < 11*365]$primaryid), "teenagers" = unique(Demo[age_in_days >= 11*365 & age_in_days < 18*365]$primaryid), "adults" = unique(Demo[age_in_days >= 18*365 & age_in_days < 65*365]$primaryid), "young old" = unique(Demo[age_in_days >= 65*365 & age_in_days < 75*365]$primaryid), "middle old" = unique(Demo[age_in_days >= 75*365 & age_in_days < 85*365]$primaryid), "oldest old" = unique(Demo[age_in_days >= 85*365]$primaryid), "unknown" = unique(Demo[is.na(age_in_days)]$primaryid)) disproportionality_df <- data.table() for (n in 1:length(restrictions)) { t <- restrictions[n] t_name <- names(t) t_pids <- unlist(t) df <- disproportionality_analysis( drug_selected = unlist(drug), reac_selected = unlist(event), temp_drug = Drug[role_cod %in% c("PS", "SS")], restriction = t_pids )[, nested := t_name] disproportionality_df <- rbindlist(list(disproportionality_df, df), fill = TRUE) } disproportionality_df <- disproportionality_df[,expected:=(D*E/(D_E+D_nE+nD_E+nD_nE))] render_forest(disproportionality_df,facet_v="nested",point_size = 5,xcoord_lims = c(-6,6)) ## mmr vaccine, autism ----------------------------------------------------- drug <- "measles vaccine;mumps vaccine;rubella vaccine" event <- "autism spectrum disorder" df <- disproportionality_trend(drug, event, temp_drug=Drug[role_cod%in%c("PS","SS")]) plot_disproportionality_trend(df) df <- df[,expected:=(D*E/TOT)] render_forest(df[,substance:="MMR vaccine"][period<2002&period>1996][ , IC_signal:= factor(ifelse(IC_lower<0,"no SDR","SDR"), levels=c("not enough cases","no SDR","SDR"),ordered=TRUE) ][,period:=as.factor(period)],facet_v = "period", point_size = 5,xcoord_lims = c(-6,6)) ## venlafaxine with rhabdomyolisis-------------------------------------------- drug <- "venlafaxine" event <- "rhabdomyolysis" restrictions <- list("crude" = unique(Demo$primaryid), "no_statins" = setdiff(Demo$primaryid,intersect(Drug[substance%in%ATC[Class4 == "HMG CoA reductase inhibitors"]$substance]$primaryid, Reac[pt=="rhabdomyolysis"]$primaryid))) temp <- disproportionality_trend("venlafaxine","rhabdomyolysis", temp_drug=Drug[role_cod%in%c("PS","SS")],min_2004 = FALSE) plot_disproportionality_trend(temp) temp <- temp[,expected:=(D*E/TOT)] temp1 <- disproportionality_trend("venlafaxine","rhabdomyolysis", temp_drug=Drug[role_cod%in%c("PS","SS")],restriction = setdiff(Demo$primaryid,intersect(Drug[substance%in%ATC[Class4 == "HMG CoA reductase inhibitors"]$substance]$primaryid, Reac[pt=="rhabdomyolysis"]$primaryid)),min_2004 = FALSE) plot_disproportionality_trend(temp1) temp1 <- temp1[,expected:=(D*E/TOT)] temp3 <-rbindlist(list(temp[,substance:="venlafaxine"][,event:="rhabdomyolysis"][,IC_signal:= factor(ifelse(IC_lower<0,"no SDR","SDR"),levels=c("not enough cases","no SDR","SDR"),ordered=TRUE) ][ as.numeric(as.character(period))>2001&as.numeric(as.character(period))<2011 ][,period:=as.factor(period)][,period:=paste0(period,"crude")],temp1[,substance:="venlafaxine"][,event:="rhabdomyolysis"][, IC_signal:= factor(ifelse(IC_lower<0,"no SDR","SDR"),levels=c("not enough cases","no SDR","SDR"),ordered=TRUE) ][ as.numeric(as.character(period))>2001&as.numeric(as.character(period))<2011 ][,period:=as.factor(period)][,period:=paste0(period,"no_statins")])) temp3 <-rbindlist(list(temp[,substance:="venlafaxine"][,event:="rhabdomyolysis"][,nested:="crude"][,IC_signal:= factor(ifelse(IC_lower<0,"no SDR","SDR"),levels=c("not enough cases","no SDR","SDR"),ordered=TRUE) ][ as.numeric(as.character(period))>2001 ][,period:=as.factor(period)],temp1[,substance:="venlafaxine"][,event:="rhabdomyolysis"][,nested:="no_statin"][, IC_signal:= factor(ifelse(IC_lower<0,"no SDR","SDR"),levels=c("not enough cases","no SDR","SDR"),ordered=TRUE) ][ as.numeric(as.character(period))>2001 ][,period:=as.factor(period)])) plot_disproportionality_trend(temp3) #check proportion of statins for rhabdomyolysis length(unique(Drug[primaryid%in%Reac[pt==event]$primaryid][role_cod%in%c("PS","SS")][substance %in% ATC[Class4 == "HMG CoA reductase inhibitors"]$substance]$primaryid))/length(unique(Reac[pt==event]$primaryid)) length(intersect(Drug[primaryid%in%Reac[pt==event]$primaryid][role_cod%in%c("PS","SS")][substance %in% ATC[Class4 == "HMG CoA reductase inhibitors"]$substance]$primaryid,Demo[init_fda_dt<2009]$primaryid))/length(intersect(Reac[pt==event]$primaryid,Demo[init_fda_dt<2009]$primaryid)) ## lanadelumab with hereditary angioedema----------------------------------- drug <- "lanadelumab" event <- "hereditary angioedema" df <- disproportionality_analysis( drug_selected = drug, reac_selected = event, temp_drug = Drug[role_cod %in% c("PS", "SS")] )[, nested := event] disproportionality_df <- df[,expected:=(D*E/(D_E+D_nE+nD_E+nD_nE))] render_forest(disproportionality_df,facet_v="event", point_size = 5,xcoord_lims = c(-11,11)) ## tisagenlecleucel, respiratory tract infection ------------------------------ drug <- "tisagenlecleucel" event <- c("pt"=list("respiratory tract infection"), "HLTs"=list(c(MedDRA[hlt%in%c("upper respiratory tract infections", "lower respiratory tract and lung infections")]$pt, "respiratory tract infection")), "SOC" = list(c(MedDRA[soc=="infections and infestations"]$pt)), "pneumonia"="pneumonia", "sinusitis"="sinusitis") df <- disproportionality_analysis( drug_selected = drug, reac_selected = event, temp_drug = Drug[role_cod %in% c("PS", "SS")] )[, nested := event] no disproportionality_df <- df[,expected:=(D*E/(D_E+D_nE+nD_E+nD_nE))] render_forest(disproportionality_df,facet_v="event", point_size = 5,xcoord_lims = c(-6,6)) resp_terms_used <- Reac[pt%in%event$HLTs][primaryid%in%Drug[ role_cod%in%c("PS","SS")][substance=="tisagenlecleucel"]$primaryid][ ,.N,by="pt"][order(-N)] ## efalizumab with neuropathy----------------------------------- drug <- "efalizumab" event <- c("neuropathy peripheral"=list("neuropathy peripheral"), "HLGT"=list(MedDRA[hlgt=="peripheral neuropathies"]$pt), "guillain-barre syndrome"="guillain-barre syndrome", "demyelinating polyneuropathy"="demyelinating polyneuropathy") df <- disproportionality_analysis( drug_selected = drug, reac_selected = event, temp_drug = Drug[role_cod %in% c("PS", "SS")] )[, nested := event] no disproportionality_df <- df[,expected:=(D*E/(D_E+D_nE+nD_E+nD_nE))] render_forest_new(disproportionality_df,facet_v="event", point_size = 5,xcoord_lims = c(-6,6)) ## montelukast, photophobia ------------------------------ drug <- "montelukast" event <- "photophobia" df <- disproportionality_analysis( drug_selected = drug, reac_selected = event, temp_drug = Drug[role_cod %in% c("PS", "SS")] )[, nested := event] disproportionality_df <- df[,expected:=(D*E/(D_E+D_nE+nD_E+nD_nE))] render_forest(disproportionality_df,facet_v="event", point_size = 5,xcoord_lims = c(-6,6)) pids_m_p <- intersect(Drug[substance==drug&role_cod%in%c("PS","SS")]$primaryid, Reac[pt==event]$primaryid) Demo[primaryid%in%pids_m_p][,.N,by="occr_country"] Drug[primaryid%in%pids_m_p][role_cod%in%c("PS","SS")][,.(substance=paste0(substance,collapse="; ")),by="primaryid"][,.N,by=substance] ## docetaxel with anhedonia --------------------------------------------------- drug <- "docetaxel" event <- "anhedonia" pids_d_a <- intersect(Drug[role_cod%in%c("PS","SS")][substance==drug]$primaryid, Reac[pt==event]$primaryid) Reac[primaryid%in%pids_d_a][,.N,by="pt"][order(-N)] Demo[primaryid%in%pids_d_a][,init_fda_dt:=as.numeric(substr(init_fda_dt,0,4))][,.N,by="init_fda_dt"][order(-N)] Demo[primaryid%in%pids_d_a][,.N,by="occp_cod"][order(-N)] Demo[primaryid%in%pids_d_a][,.N,by="occr_country"][order(-N)] df <- disproportionality_analysis( drug_selected = drug, reac_selected = event, temp_drug = Drug[role_cod %in% c("PS", "SS")] )[, nested := event] disproportionality_df <- df[,expected:=(D*E/(D_E+D_nE+nD_E+nD_nE))] render_forest(disproportionality_df,facet_v="event",point_size = 3,xcoord_lims = c(-6,6)) doc_pids <- unique(Drug[substance=="docetaxel"][role_cod%in%c("PS","SS")]$primaryid) alo_pids <- unique(Reac[pt=="alopecia"]$primaryid) alo_doc <- intersect(alo_pids,doc_pids) pids_7_events <- intersect(intersect(intersect(intersect(intersect(intersect(alo_pids,Reac[pt=="anhedonia"]$primaryid), Reac[pt=="anxiety"]$primaryid), Reac[pt=="discomfort"]$primaryid), Reac[pt=="emotional distress"]$primaryid), Reac[pt=="injury"]$primaryid), Reac[pt=="pain"]$primaryid) length(pids_7_events) length(intersect(pids_7_events,doc_pids)) df <- disproportionality_analysis( drug_selected = drug, reac_selected = event, temp_drug = Drug[role_cod %in% c("PS", "SS")], restriction = setdiff(Demo$primaryid,pids_7_events) )[, nested := event] disproportionality_df <- df[,expected:=(D*E/(D_E+D_nE+nD_E+nD_nE))] render_forest(disproportionality_df,facet_v="event",point_size = 5,xcoord_lims = c(-6,6)) ## tisagenlecleucel, anemia -------------------------------------------------- drug <- "tisagenlecleucel" event <- "anaemia" restrictions <- list("naive" = unique(Demo$primaryid), "leukemias" = unique(Indi[indi_pt%in%MedDRA[hlgt=="leukaemias"]$pt]$primaryid)) disproportionality_df <- data.table() for (n in 1:length(restrictions)) { t <- restrictions[n] t_name <- names(t) t_pids <- unlist(t) df <- disproportionality_analysis( drug_selected = unlist(drug), reac_selected = unlist(event), temp_drug = Drug[role_cod %in% c("PS", "SS")], restriction = t_pids )[, nested := t_name] disproportionality_df <- rbindlist(list(disproportionality_df, df), fill = TRUE) } disproportionality_df <- disproportionality_df[,expected:=(D*E/(D_E+D_nE+nD_E+nD_nE))] render_forest(disproportionality_df,facet_v="nested",point_size = 5,xcoord_lims = c(-6,6)) Drug[primaryid%in%Indi[indi_pt%in%MedDRA[hlgt=="leukaemias"]$pt]$primaryid][,.N,by="substance"][order(-N)] tot <- length(unique(Drug[primaryid%in%Indi[indi_pt%in%MedDRA[hlgt=="leukaemias"]$pt]$primaryid]$primaryid)) length(unique(Drug[primaryid%in%Indi[indi_pt%in%MedDRA[hlgt=="leukaemias"]$pt]$primaryid][substance=="ibrutinib"]$primaryid))/tot length(unique(Drug[primaryid%in%Indi[indi_pt%in%MedDRA[hlgt=="leukaemias"]$pt]$primaryid][substance=="imatinib"]$primaryid))/tot length(unique(Drug[primaryid%in%Indi[indi_pt%in%MedDRA[hlgt=="leukaemias"]$pt]$primaryid][substance=="venetoclax"]$primaryid))/tot ## pembrolizumab and cardiac failure----------------------------------- drug <- "pembrolizumab" event <- "cardiac failure" df <- disproportionality_analysis( drug_selected = drug, reac_selected = event, temp_drug = Drug[role_cod %in% c("PS", "SS")] )[, nested := "crude"] df1 <- disproportionality_analysis( drug_selected = drug, reac_selected = event, temp_drug = Drug[role_cod %in% c("PS", "SS")], restriction = unique(Outc[outc_cod!="OT"]$primaryid) )[, nested := "only_serious"] disproportionality_df <- rbindlist(list(df,df1)) disproportionality_df[,expected:=(D*E/(D_E+D_nE+nD_E+nD_nE))] render_forest(disproportionality_df,facet_v="nested",point_size = 5,xcoord_lims = c(-6,6))