Two‐at‐a‐Time analysis in R [Design Issues]

posted by ElAlumno – 2019-03-22 22:59 (2125 d 03:16 ago) – Posting: # 20069
Views: 27,145

Thank you Helmut for your quick and detailed response!

I'm glad to hear that pseudo-periods have gone the way of the dodo. I agree with the EMA Biostatistics Working Party that it makes more sense (statistically) to maintain the original codings when leaving out data from irrelevant treatments (not to mention being less prone to human error than recoding). Also, I love their name; I bet the EMA Biostatistics Working Party could form a coalition government with the Slightly Silly Party, the Surprise Party, and the Rent Is Too Damn High Party. But I digress.

I don't have WinNonlin, but I tried analyzing Patterson & Jones' Example 4.5 with the Two‐at‐a‐Time Principle using R 1.1.456. My results were similar to the results you posted above, but not as close as I would have expected. I will post the code & table below. Do you see anything I am doing wrong or any obvious explanation of the differences? I don't know how WinNonlin handles missing values, so I tried a few variations (subjects with complete cases in the full dataset, and subjects with complete cases in each of the 3 incomplete-block subsets), but the differences persisted.

# 3-treatment design from Patterson & Jones (2017)
dta = read.table("exam45.dat", header=T)
dta[dta==99999] = NA
dta$Subj = factor(dta$subject)
dta$Per  = factor(dta$period)
dta$Seq  = factor(dta$sequence)
dta$Trt  = factor(dta$formula)
options(contrasts=c("contr.treatment","contr.poly"), digits=4)

# Data frame to store results
IBD = data.frame(Test = paste(rep(c("T vs. R", "S vs. R", "T vs. S"), each=2),
                              rep(c(" AUC", "Cmax"), times=3)),
                 PE = NA, LCI = NA, UCI = NA, CV = NA)

# AUC T vs R
muddle = lm(log(AUC)~Trt+Per+Seq+Subj, data=dta[dta$Trt!="S",])
IBD[1, 2:5] = c(100*exp(coef(muddle)["TrtT"]),
                100*exp(confint(muddle,c("TrtT"), level=.9)),
                100*sqrt(exp(summary(muddle)$sigma^2)-1))

# Cmax T vs R
muddle = lm(log(CMAX)~Trt+Per+Seq+Subj, data=dta[dta$Trt!="S",])
IBD[2, 2:5] = c(100*exp(coef(muddle)["TrtT"]),
                100*exp(confint(muddle,c("TrtT"), level=.9)),
                100*sqrt(exp(summary(muddle)$sigma^2)-1))

# AUC S vs R
muddle = lm(log(AUC)~Trt+Per+Seq+Subj, data=dta[dta$Trt!="T",])
IBD[3, 2:5] = c(100*exp(coef(muddle)["TrtS"]),
                100*exp(confint(muddle,c("TrtS"), level=.9)),
                100*sqrt(exp(summary(muddle)$sigma^2)-1))

# Cmax S vs R
muddle = lm(log(CMAX)~Trt+Per+Seq+Subj, data=dta[dta$Trt!="T",])
IBD[4, 2:5] = c(100*exp(coef(muddle)["TrtS"]),
                100*exp(confint(muddle,c("TrtS"), level=.9)),
                100*sqrt(exp(summary(muddle)$sigma^2)-1))

# AUC T vs S
muddle = lm(log(AUC)~Trt+Per+Seq+Subj, data=dta[dta$Trt!="R",])
IBD[5, 2:5] = c(100*exp(coef(muddle)["TrtT"]),
                100*exp(confint(muddle,c("TrtT"), level=.9)),
                100*sqrt(exp(summary(muddle)$sigma^2)-1))

# Cmax T vs S
muddle = lm(log(CMAX)~Trt+Per+Seq+Subj, data=dta[dta$Trt!="R",])
IBD[6, 2:5] = c(100*exp(coef(muddle)["TrtT"]),
                100*exp(confint(muddle,c("TrtT"), level=.9)),
                100*sqrt(exp(summary(muddle)$sigma^2)-1))

# Print table
print(IBD)


This produced the following table:

          Test     PE    LCI    UCI    CV
1 T vs. R  AUC 116.14 108.98 123.78 20.88
2 T vs. R Cmax 129.82 119.50 141.04 27.87
3 S vs. R  AUC 140.73 131.02 151.15 23.50
4 S vs. R Cmax 160.04 144.76 176.94 34.06
5 T vs. S  AUC  83.51  78.73  88.58 19.07
6 T vs. S Cmax  82.30  75.96  89.18 26.67

Complete thread:

UA Flag
Activity
 Admin contact
23,362 posts in 4,906 threads, 1,673 registered users;
97 visitors (0 registered, 97 guests [including 19 identified bots]).
Forum time: 02:16 CET (Europe/Vienna)

I have not failed 700 times. I have not failed once.
I have succeeded in proving
that those 700 ways will not work.    Thomas Alva Edison

The Bioequivalence and Bioavailability Forum is hosted by
BEBAC Ing. Helmut Schütz
HTML5