-
Notifications
You must be signed in to change notification settings - Fork 0
/
DCA_Natalizumab_Plot.R
190 lines (152 loc) · 10.4 KB
/
DCA_Natalizumab_Plot.R
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
########################################################################
################# Script for estimating the NB ####################
####### Under strategies: ######################
##### Treat none, treat all with N, treat all with DF, ############
##### Treat all with GA, and treat based on the model ##########
################ for one set of threshold values: #######################
################## T_DF=T_GA=10% ###################3
###### and a range between 19% - 40% for Natalizumab thresholds #################
######################### & Creation of relevant Plot ################################
##################################################################################3
#Entire ADdataset
RiskData$outcome<-RiskData$RELAPSE2year
n_studies<-c(nrow(RiskData[which(RiskData$STUDYID==1 & RiskData$TRT01A==1),]),nrow(RiskData[which(RiskData$STUDYID==1 & RiskData$TRT01A==4),]),
nrow(RiskData[which(RiskData$STUDYID==2 & RiskData$TRT01A==1),]),nrow(RiskData[which(RiskData$STUDYID==2 & RiskData$TRT01A==2),]),
nrow(RiskData[which(RiskData$STUDYID==2 & RiskData$TRT01A==4),]),nrow(RiskData[which(RiskData$STUDYID==3 & RiskData$TRT01A==3),]),
nrow(RiskData[which(RiskData$STUDYID==3 & RiskData$TRT01A==4),]))
n_events<-c(nrow(RiskData[which(RiskData$STUDYID==1 & RiskData$TRT01A==1 & RiskData$outcome==1),]), nrow(RiskData[which(RiskData$STUDYID==1 & RiskData$TRT01A==4 & RiskData$outcome==1),]),
nrow(RiskData[which(RiskData$STUDYID==2 & RiskData$TRT01A==1 & RiskData$outcome==1),]), nrow(RiskData[which(RiskData$STUDYID==2 & RiskData$TRT01A==2 & RiskData$outcome==1),]),
nrow(RiskData[which(RiskData$STUDYID==2 & RiskData$TRT01A==4 & RiskData$outcome==1),]), nrow(RiskData[which(RiskData$STUDYID==3 & RiskData$TRT01A==3 & RiskData$outcome==1),]),
nrow(RiskData[which(RiskData$STUDYID==3 & RiskData$TRT01A==4 & RiskData$outcome==1),]))
dataAD_Whole<-cbind(c(1,1,2,2,2,3,3),c(1,4,1,2,4,3,4),n_studies, n_events)
dataAD_Whole<-as.data.frame(dataAD_Whole)
colnames(dataAD_Whole)<-c("St","Dr","N_ran","N_rel")
dataAD_Whole_Pl<-dataAD_Whole[which(dataAD_Whole$Dr==4),]
## event rate in placebo arm in the whole dataset
m1<-metaprop(N_rel,N_ran,data=dataAD_Whole_Pl,sm="PLOGIT" )
#network meta-analysis in the whole dataset
TestPair <- pairwise(treat=Dr, event=N_rel, n=N_ran, data=dataAD_Whole, sm="RR", studlab=St, allstudies = TRUE)
net1 <- netmeta(TE, seTE, treat1, treat2, studlab, data = TestPair, sm = "RR", comb.random=F, comb.fixed=T, prediction=TRUE, ref=4)
# CREATING DATAFRAME THAT IS ONE LINE PER THRESHOLD PER all default STRATEGIES
nb=data.frame(seq(from=0.01, to=1, by=0.01))
names(nb)="threshold"
interv=nb
###Set threshold values for DF and Glatiramer
T_DF=0.10
T_GA=0.10
## MB for default strategies
##Estimation of the NB under "treat none" strategy
nb["Placebo"]=0
##Estimation of the NB under "treat all with Natalizumab" strategy using the corresponding threshold values
nb["Natalizumab"]<-(expit(m1$TE.fixed)-expit(m1$TE.fixed)*exp(net1$TE.fixed)[3,4])-nb$threshold
##Estimation of the NB under "treat all with DF" strategy using the corresponding threshold values
nb["Dimethyl FUmerate"]<-(expit(m1$TE.fixed)-expit(m1$TE.fixed)*exp(net1$TE.fixed)[1,4])-T_DF
##Estimation of the NB under "treat all with GA" strategy using the corresponding threshold values
nb["Glatiramer Acetate"]<-(expit(m1$TE.fixed)-expit(m1$TE.fixed)*exp(net1$TE.fixed)[2,4])-T_GA
###Start estimating net benefit for strategy "treat based on the model"
nb["Model"]<-NA
nb["OptimalNB"]<-NA
#The rule of deciding
for (t in 19:40){
RiskData$maxRDTh<-NA
RiskData$RecommendedTreatmentThreshold<-NA
for (i in 1:nrow(RiskData)){
RiskData$maxRDTh[i]<-max(RiskData$RDDF[i]-T_DF,RiskData$RDGA[i]-T_GA,RiskData$RDN[i]-nb$threshold[t],0)
RiskData$RecommendedTreatmentThreshold[i]<-which.max(c(RiskData$RDDF[i]-T_DF,RiskData$RDGA[i]-T_GA,RiskData$RDN[i]-nb$threshold[t],0))
}
## Creation of the congruent dataset, where the recommended treatment
#via the model is the actual given one
RiskDataCongruent<-RiskData[which(RiskData$Treatment==RiskData$RecommendedTreatmentThreshold),]
n_study1T1<-nrow(RiskDataCongruent[which(RiskDataCongruent$STUDYID==1 & RiskDataCongruent$TRT01A==1),])
n_study1T4<-nrow(RiskDataCongruent[which(RiskDataCongruent$STUDYID==1 & RiskDataCongruent$TRT01A==4),])
n_study2T1<-nrow(RiskDataCongruent[which(RiskDataCongruent$STUDYID==2 & RiskDataCongruent$TRT01A==1),])
n_study2T2<-nrow(RiskDataCongruent[which(RiskDataCongruent$STUDYID==2 & RiskDataCongruent$TRT01A==2),])
n_study2T4<-nrow(RiskDataCongruent[which(RiskDataCongruent$STUDYID==2 & RiskDataCongruent$TRT01A==4),])
n_study3T3<-nrow(RiskDataCongruent[which(RiskDataCongruent$STUDYID==3 & RiskDataCongruent$TRT01A==3),])
n_study3T4<-nrow(RiskDataCongruent[which(RiskDataCongruent$STUDYID==3 & RiskDataCongruent$TRT01A==4),])
r_study1T1<-nrow(RiskDataCongruent[which(RiskDataCongruent$STUDYID==1 & RiskDataCongruent$TRT01A==1 & RiskDataCongruent$outcome==1),])
r_study1T4<-nrow(RiskDataCongruent[which(RiskDataCongruent$STUDYID==1 & RiskDataCongruent$TRT01A==4 & RiskDataCongruent$outcome==1),])
r_study2T1<-nrow(RiskDataCongruent[which(RiskDataCongruent$STUDYID==2 & RiskDataCongruent$TRT01A==1 & RiskDataCongruent$outcome==1),])
r_study2T2<-nrow(RiskDataCongruent[which(RiskDataCongruent$STUDYID==2 & RiskDataCongruent$TRT01A==2 & RiskDataCongruent$outcome==1),])
r_study2T4<-nrow(RiskDataCongruent[which(RiskDataCongruent$STUDYID==2 & RiskDataCongruent$TRT01A==4 & RiskDataCongruent$outcome==1),])
r_study3T3<-nrow(RiskDataCongruent[which(RiskDataCongruent$STUDYID==3 & RiskDataCongruent$TRT01A==3 & RiskDataCongruent$outcome==1),])
r_study3T4<-nrow(RiskDataCongruent[which(RiskDataCongruent$STUDYID==3 & RiskDataCongruent$TRT01A==4 & RiskDataCongruent$outcome==1),])
dataAD_Congruent<-cbind(c(1,1,2,2,2,3,3),c(1,4,1,2,4,3,4),c(n_study1T1,n_study1T4,n_study2T1,n_study2T2,n_study2T4,n_study3T3,n_study3T4),
c(r_study1T1,r_study1T4,r_study2T1,r_study2T2,r_study2T4,r_study3T3,r_study3T4))
dataAD_Congruent<-as.data.frame(dataAD_Congruent)
colnames(dataAD_Congruent)<-c("St","Dr","N_ran","N_rel")
if (nrow(dataAD_Congruent[which(dataAD_Congruent$Dr==4 & dataAD_Congruent$N_ran!=0 ),])!=0) {
dataAD_Congruent_Pl<-dataAD_Congruent[which(dataAD_Congruent$Dr==4 & dataAD_Congruent$N_ran!=0 ),]
## event rate in placebo arm in the congruent dataset
m2<-metaprop(N_rel,N_ran,data=dataAD_Congruent_Pl,sm="PLOGIT" )
}
if (nrow(dataAD_Congruent[which(dataAD_Congruent$Dr==4 & dataAD_Congruent$N_ran!=0 ),])==0) {
dataAD_Congruent_Pl<-dataAD_Congruent[which(dataAD_Congruent$Dr==1 & dataAD_Congruent$N_ran!=0 ),]
## event rate in placebo arm in the congruent dataset
m2<-metaprop(N_rel,N_ran,data=dataAD_Congruent_Pl,sm="PLOGIT" )
}
#network meta-analysis in the congruent dataset
if (nrow(dataAD_Congruent[which(dataAD_Congruent$Dr==4 & dataAD_Congruent$N_ran!=0 ),])!=0) {
TestPair <- pairwise(treat=Dr, event=N_rel, n=N_ran, data=dataAD_Congruent, sm="RR", studlab=St, allstudies = TRUE)
net2 <- netmeta(TE, seTE, treat1, treat2, studlab, data = TestPair, sm = "RR", comb.random=F, comb.fixed=T, prediction=TRUE, ref=4)
}
if (nrow(dataAD_Congruent[which(dataAD_Congruent$Dr==4 & dataAD_Congruent$N_ran!=0 ),])==0) {
TestPair <- pairwise(treat=Dr, event=N_rel, n=N_ran, data=dataAD_Congruent, sm="RR", studlab=St, allstudies = TRUE)
net2 <- netmeta(TE, seTE, treat1, treat2, studlab, data = TestPair, sm = "RR", comb.random=F, comb.fixed=T, prediction=TRUE, ref=1)
}
n_total<-nrow(RiskDataCongruent)
n_pl<-nrow(RiskDataCongruent[which(RiskDataCongruent$Treatment==4),])
n_N<-nrow(RiskDataCongruent[which(RiskDataCongruent$Treatment==3),])
n_DF<-nrow(RiskDataCongruent[which(RiskDataCongruent$Treatment==1),])
n_GA<-nrow(RiskDataCongruent[which(RiskDataCongruent$Treatment==2),])
if (n_N!=0) {
e_N<-n_N*expit(m2$TE.fixed)* exp(net2$TE.nma.fixed[which(net2$treat1==3)])
}
if (n_N==0) {
e_N<-0
}
if (n_DF!=0 && nrow(dataAD_Congruent[which(dataAD_Congruent$Dr==4 & dataAD_Congruent$N_ran!=0 ),])!=0 ) {
e_DF<-n_DF*expit(m2$TE.fixed)* exp(net2$TE.nma.fixed[which(net2$treat1==1)])
e_DF<-unique(e_DF)
}
if (n_DF!=0 && nrow(dataAD_Congruent[which(dataAD_Congruent$Dr==4 & dataAD_Congruent$N_ran==0 ),])!=0 ) {
e_DF<-n_DF*expit(m2$TE.fixed)
}
if (n_DF==0) {
e_DF<-0
}
if (n_GA!=0) {
e_GA<-n_GA*expit(m2$TE.fixed)* exp(net2$TE.nma.fixed[which(net2$treat1==2)])
}
if (n_GA==0) {
e_GA<-0
}
if (n_pl!=0) {
e_pl<-n_pl*expit(m2$TE.fixed)
}
if (n_pl==0) {
e_pl<-0
}
e_total<- ( e_pl + e_N + e_DF + e_GA)/n_total
nb[t,6]<-(expit(m1$TE.fixed)-e_total)-((n_N*nb$threshold[t]+n_DF*T_DF+n_GA*T_GA)/n_total)
nb[t,7]<-max(nb[t,2],nb[t,3],nb[t,4], nb[t,5], nb[t,6])
}
#getting maximum number of avoided interventions
ymax=max(nb[11:30,3:6],na.rm = TRUE)
ymin=min(nb[11:30,3:6],na.rm = TRUE)
# inializing new benfit plot with treat all with Natalizumab option
plot(x=nb$threshold[19:40], y=nb$Natalizumab[19:40], type="l",lwd=4,col="seagreen3",xlim=c(0.19, 0.40), ylim=c(ymin, ymax+0.2), xlab="Threshold probability", ylab=paste("Net Benefit"))
# adding treat none option
lines(x=nb$threshold[19:40], y=nb$Placebo[19:40],col="steelblue3",lwd=4)
lines(nb$threshold[19:40],y=nb$`Dimethyl FUmerate`[19:40],col="lightgoldenrod2",lwd=4)
lines(nb$threshold[19:40],y=nb$`Glatiramer Acetate`[19:40],col="orange",lwd=4)
lines(nb$threshold[19:40],y=nb$Model[19:40],lwd=4, col="red")
#lines(nb$threshold[20:40], y=c(nb$Natalizumab[20],nb$Model[21:28],nb$`Dimethyl FUmerate`[29:40]) , type="l",col=c("light blue", "red", "orange"),lwd=2)
lines(nb$threshold[19:20], y=nb$OptimalNB[19:20]+0.007, type="l",lty = "dashed",col="black",lwd=4)
lines(nb$threshold[20:21], y=nb$OptimalNB[20:21]+0.007, type="l",lty = "dashed",col="black",lwd=4)
lines(nb$threshold[21:22], y=nb$OptimalNB[21:22]+0.007, type="l",lty = "dashed",col="black",lwd=4)
lines(nb$threshold[22:24], y=nb$OptimalNB[22:24]+0.007, type="l",lty = "dashed",col="black",lwd=4)
lines(nb$threshold[24:25], y=nb$OptimalNB[24:25]+0.007, type="l",lty = "dashed",col="black",lwd=4)
lines(nb$threshold[25:40], y=nb$OptimalNB[25:40]+0.007, type="l",lty = "dashed",col="black",lwd=4)
#then add the legends
#legend("topright", c("Treat none","Treat based on the model","Treat all with Natalizumab","Treat all with Dimethyl Fumerate","Highest Net Benefit"), cex=0.8, col=c(),text.font=4, lwd=2)