PowerTOST

• Introduction
• Supported Designs
• Purpose
• Supported
  • Power and Sample Size
  • Methods
  • Helpers
• Defaults
  • Average Bioequivalence
  • Reference-Scaled Average Bioequivalence
  • Dose-Proportionality
  • Power Analysis
• Examples
  • Parallel Design
  • Crossover Design
  • Replicate Designs
  • Dose-Proportionality
  • Power Analysis
  • Speed Comparisons
• Installation
• Session Information

Version 1.5.6 built 2024-03-18 with R 4.3.3

Introduction

The package contains functions to calculate power and estimate sample size for various study designs used in (not only bio-) equivalence studies.

Supported Designs

#    design                        name    df
#  parallel           2 parallel groups   n-2
#       2x2               2x2 crossover   n-2
#     2x2x2             2x2x2 crossover   n-2
#       3x3               3x3 crossover 2*n-4
#     3x6x3             3x6x3 crossover 2*n-4
#       4x4               4x4 crossover 3*n-6
#     2x2x3   2x2x3 replicate crossover 2*n-3
#     2x2x4   2x2x4 replicate crossover 3*n-4
#     2x4x4   2x4x4 replicate crossover 3*n-4
#     2x3x3   partial replicate (2x3x3) 2*n-3
#     2x4x2            Balaam's (2x4x2)   n-2
#    2x2x2r Liu's 2x2x2 repeated x-over 3*n-2
#    paired                paired means   n-1

Codes of designs follow this pattern: treatments x sequences x periods.

Although some replicate designs are more ‘popular’ than others, sample size estimations are valid for all of the following designs:

Balaam’s design TR|RT|TT|RR should be avoided due to its poor power characteristics. The three period partial replicate design with two sequences TRR|RTR (a.k.a. extra-reference design) should be avoided because it is biased in the presence of period effects.

TOC ↩

Purpose

For various methods power can be calculated based on

• nominal α, coefficient of variation (CV), deviation of test from reference (θ₀), acceptance limits {θ₁, θ₂}, sample size (n), and design.

For all methods the sample size can be estimated based on

• nominal α, coefficient of variation (CV), deviation of test from reference (θ₀), acceptance limits {θ₁, θ₂}, target (i.e., desired) power, and design.

TOC ↩

Supported

Power and Sample Size

Power covers balanced as well as unbalanced sequences in crossover or replicate designs and equal/unequal group sizes in two-group parallel designs. Sample sizes are always rounded up to achieve balanced sequences or equal group sizes.

• Average Bioequivalence (with arbitrary fixed limits).
• ABE for Highly Variable Narrow Therapeutic Index Drugs by simulations: U.S. FDA, China CDE.
• Scaled Average Bioequivalence based on simulations.
  • Average Bioequivalence with Expanding Limits (ABEL) for Highly Variable Drugs / Drug Products: EMA, WHO and many others.
  • Average Bioequivalence with fixed widened limits of 75.00–133.33% if CV_wR >30%: Gulf Cooperation Council.
    
  • Reference-scaled Average Bioequivalence (RSABE) for HVDP(s): U.S. FDA, China CDE.
  • Iteratively adjust α to control the type I error in ABEL and RSABE for HVDP(s).
  • RSABE for NTIDs: U.S. FDA, China CDE.
• Two simultaneous TOST procedures.
• Non-inferiority t-test.
• Ratio of two means with normally distributed data on the original scale based on Fieller’s (‘fiducial’) confidence interval.
• ‘Expected’ power in case of uncertain (estimated) variability and/or uncertain θ₀.
• Dose-Proportionality using the power model.

TOC ↩

Methods

• Exact
  • Owen’s Q.
  • Direct integration of the bivariate non-central t-distribution.
• Approximations
  • Non-central t-distribution.
  • ‘Shifted’ central t-distribution.

TOC ↩

Helpers

• Calculate CV from MSE or SE (and vice versa).
• Calculate CV from given confidence interval.
• Calculate CV_wR from the upper expanded limit of an ABEL study.
• Confidence interval of CV.
• Pool CV from several studies.
• Confidence interval for given α, CV, point estimate, sample size, and design.
• Calculate CV_wT and CV_wR from a (pooled) CV_w assuming a ratio of intra-subject variances.
• p-values of the TOST procedure.
• Analysis tool for exploration/visualization of the impact of expected values (CV, θ₀, reduced sample size due to dropouts) on power of BE decision.
• Construct design matrices of incomplete block designs.

TOC ↩

Defaults

• α 0.05, {θ₁, θ₂} (0.80, 1.25), target power 0.80. Details of the sample size search (and the regulatory settings in reference-scaled average bioequivalence) are shown in the console.
• Note: In all functions values have to be given as ratios, not in percent.

Average Bioequivalence

Conventional (unscaled)

Design "2x2" (TR|RT), exact method (Owen’s Q).

Highly Variable NTIDs (FDA, CDE)

Design "2x2x4" (TRTR|RTRT), upper limit of the confidence interval of σ_wT/σ_wR ≤2.5, approximation by the non-central t-distribution, 100,000 simulations.

Reference-Scaled Average Bioequivalence

Point estimate constraints (0.80, 1.25), homoscedasticity (CV_wT = CV_wR), scaling is based on CV_wR, design "2x3x3" (TRR|RTR|RRT), approximation by the non-central t-distribution, 100,000 simulations.

• EMA, WHO, Health Canada, and many other jurisdictions: Average Bioequivalence with Expanding Limits (ABEL).
• U.S. FDA, China CDE: RSABE.

Highly Variable Drugs / Drug Products

θ₀ 0.90.¹

EMA and many others

Regulatory constant 0.760, upper cap of scaling at CV_wR 50%, evaluation by ANOVA.

Health Canada

Regulatory constant 0.760, upper cap of scaling at CV_wR ~57.4%, evaluation by intra-subject contrasts.

Gulf Cooperation Council

Regulatory constant log(1/0.75)/sqrt(log(0.3^2+1)), widened limits 75.00–133.33% if CV_wR >30%, no upper cap of scaling, evaluation by ANOVA.

FDA, CDE

Regulatory constant log(1.25)/0.25, no upper cap of scaling, evaluation by linearized scaled ABE (Howe’s approximation).

Narrow Therapeutic Index Drugs (FDA, CDE)

θ₀ 0.975, regulatory constant log(1.11111)/0.1, implicit upper cap of scaling at CV_wR ~21.4%, design "2x2x4" (TRTR|RTRT), evaluation by linearized scaled ABE (Howe’s approximation), upper limit of the confidence interval of σ_wT/σ_wR ≤2.5.

Dose-Proportionality

β₀ (slope) 1+log(0.95)/log(rd) where rd is the ratio of the highest and lowest dose, target power 0.80, crossover design, details of the sample size search suppressed.

Power Analysis

Minimum acceptable power 0.70. θ₀; design, conditions, and sample size method depend on defaults of the respective approaches (ABE, ABEL, RSABE, NTID, HVNTID).

TOC ↩

Examples

Before running the examples attach the library.

library(PowerTOST)

If not noted otherwise, the functions’ defaults are employed.

Parallel Design

Power for total CV 0.35 (35%), group sizes 52 and 49.

power.TOST(CV = 0.35, n = c(52, 49), design = "parallel")
# [1] 0.8011186

Crossover Design

Sample size for assumed within- (intra-) subject CV 0.20 (20%).

sampleN.TOST(CV = 0.20)
# 
# +++++++++++ Equivalence test - TOST +++++++++++
#             Sample size estimation
# -----------------------------------------------
# Study design: 2x2 crossover 
# log-transformed data (multiplicative model)
# 
# alpha = 0.05, target power = 0.8
# BE margins = 0.8 ... 1.25 
# True ratio = 0.95,  CV = 0.2
# 
# Sample size (total)
#  n     power
# 20   0.834680

TOC ↩

Sample size for assumed within- (intra-) subject CV 0.40 (40%), θ₀ 0.90, four period full replicate study (any of TRTR|RTRT, TRRT|RTTR, TTRR|RRTT). Wider acceptance range for C_max (South Africa).

sampleN.TOST(CV = 0.40, theta0 = 0.90, theta1 = 0.75, design = "2x2x4")
# 
# +++++++++++ Equivalence test - TOST +++++++++++
#             Sample size estimation
# -----------------------------------------------
# Study design: 2x2x4 (4 period full replicate) 
# log-transformed data (multiplicative model)
# 
# alpha = 0.05, target power = 0.8
# BE margins = 0.75 ... 1.333333 
# True ratio = 0.9,  CV = 0.4
# 
# Sample size (total)
#  n     power
# 30   0.822929

TOC ↩

Sample size for assumed within- (intra-) subject CV 0.125 (12.5%), θ₀ 0.975. Narrower acceptance range for NTIDs (most jurisdictions).

sampleN.TOST(CV = 0.125, theta0 = 0.975, theta1 = 0.90)
# 
# +++++++++++ Equivalence test - TOST +++++++++++
#             Sample size estimation
# -----------------------------------------------
# Study design: 2x2 crossover 
# log-transformed data (multiplicative model)
# 
# alpha = 0.05, target power = 0.8
# BE margins = 0.9 ... 1.111111 
# True ratio = 0.975,  CV = 0.125
# 
# Sample size (total)
#  n     power
# 32   0.800218

TOC ↩

Sample size for equivalence of the ratio of two means with normality on the original scale based on Fieller’s (‘fiducial’) confidence interval.² Within- (intra-) subject CV_w 0.20 (20%), between- (inter-) subject CV_b 0.40 (40%).
Note the default α 0.025 (95% CI) of this function because it is intended for studies with clinical endpoints.

sampleN.RatioF(CV = 0.20, CVb = 0.40)
# 
# +++++++++++ Equivalence test - TOST +++++++++++
#     based on Fieller's confidence interval
#             Sample size estimation
# -----------------------------------------------
# Study design: 2x2 crossover
# Ratio of means with normality on original scale
# alpha = 0.025, target power = 0.8
# BE margins = 0.8 ... 1.25 
# True ratio = 0.95,  CVw = 0.2,  CVb = 0.4
# 
# Sample size
#  n     power
# 28   0.807774

TOC ↩

Replicate Designs

ABE

Conventional (unscaled)

Sample size for assumed within- (intra-) subject CV 0.45 (45%), θ₀ 0.90, three period full replicate study (TRT|RTR or TRR|RTT).

sampleN.TOST(CV = 0.45, theta0 = 0.90, design = "2x2x3")
# 
# +++++++++++ Equivalence test - TOST +++++++++++
#             Sample size estimation
# -----------------------------------------------
# Study design: 2x2x3 (3 period full replicate) 
# log-transformed data (multiplicative model)
# 
# alpha = 0.05, target power = 0.8
# BE margins = 0.8 ... 1.25 
# True ratio = 0.9,  CV = 0.45
# 
# Sample size (total)
#  n     power
# 124   0.800125

Note that the conventional model assumes homoscedasticity (equal variances of treatments). For heteroscedasticity we can ‘switch off’ all conditions of one of the methods for reference-scaled ABE. We assume a σ²-ratio of ⅔ (i.e., the test has a lower variability than the reference). Only relevant columns of the data frame shown.

reg <- reg_const("USER", r_const = NA, CVswitch = Inf,
                 CVcap = Inf, pe_constr = FALSE)
CV  <- CVp2CV(CV = 0.45, ratio = 2/3)
res <- sampleN.scABEL(CV=CV, design = "2x2x3", regulator = reg,
                      details = FALSE, print = FALSE)
print(res[c(3:4, 8:9)], digits = 5, row.names = FALSE)
#    CVwT    CVwR Sample size Achieved power
#  0.3987 0.49767         126         0.8052

Similar sample size because the pooled CV_w is still 0.45.

TOC ↩

Highly Variable Narrow Therapeutic Index Drug

Sample size assuming heteroscedasticity (CV_w 0.45, variance-ratio 2.5, i.e., the test treatment has a substantially higher variability than the reference). TRTR|RTRT according to the FDA’s guidances.^3,4,5 Assess additionally which one of the components (ABE, s_wT/s_wR-ratio) drives the sample size.

CV <- signif(CVp2CV(CV = 0.45, ratio = 2.5), 4)
n  <- sampleN.HVNTID(CV = CV, details = FALSE)[["Sample size"]]
# 
# +++++++++ FDA method for HV NTIDs ++++++++++++
#            Sample size estimation
# ----------------------------------------------
# Study design: 2x2x4 (TRTR|RTRT) 
# log-transformed data (multiplicative model)
# 1e+05 studies for each step simulated.
# 
# alpha  = 0.05, target power = 0.8
# CVw(T) = 0.549, CVw(R) = 0.3334
# True ratio = 0.95 
# ABE limits = 0.8 ... 1.25 
# 
# Sample size
#  n     power
# 50   0.812820
suppressMessages(power.HVNTID(CV = CV, n = n, details = TRUE))
#        p(BE)    p(BE-ABE) p(BE-sratio) 
#      0.81282      0.87052      0.93379

The ABE component shows a lower probability to demonstrate BE than the s_wT/s_wR component and hence, drives the sample size.

TOC ↩

ABEL

Sample size assuming homoscedasticity (CV_wT = CV_wR = 0.45).

sampleN.scABEL(CV = 0.45)
# 
# +++++++++++ scaled (widened) ABEL +++++++++++
#             Sample size estimation
#    (simulation based on ANOVA evaluation)
# ---------------------------------------------
# Study design: 2x3x3 (partial replicate)
# log-transformed data (multiplicative model)
# 1e+05 studies for each step simulated.
# 
# alpha  = 0.05, target power = 0.8
# CVw(T) = 0.45; CVw(R) = 0.45
# True ratio = 0.9
# ABE limits / PE constraint = 0.8 ... 1.25 
# EMA regulatory settings
# - CVswitch            = 0.3 
# - cap on scABEL if CVw(R) > 0.5
# - regulatory constant = 0.76 
# - pe constraint applied
# 
# 
# Sample size search
#  n     power
# 36   0.7755 
# 39   0.8059

TOC ↩

Iteratively adjust α to control the Type I Error.⁶ Heteroscedasticity (CV_wT 0.30, CV_wR 0.40, i.e., variance-ratio ~0.58), four period full replicate study (any of TRTR|RTRT, TRRT|RTTR, TTRR|RRTT), 24 subjects, balanced sequences.

scABEL.ad(CV = c(0.30, 0.40), design = "2x2x4", n = 24)
# +++++++++++ scaled (widened) ABEL ++++++++++++
#          iteratively adjusted alpha
#    (simulations based on ANOVA evaluation)
# ----------------------------------------------
# Study design: 2x2x4 (4 period full replicate)
# log-transformed data (multiplicative model)
# 1,000,000 studies in each iteration simulated.
# 
# CVwR 0.4, CVwT 0.3, n(i) 12|12 (N 24)
# Nominal alpha                 : 0.05 
# True ratio                    : 0.9000 
# Regulatory settings           : EMA (ABEL)
# Switching CVwR                : 0.3 
# Regulatory constant           : 0.76 
# Expanded limits               : 0.7462 ... 1.3402
# Upper scaling cap             : CVwR > 0.5 
# PE constraints                : 0.8000 ... 1.2500
# Empiric TIE for alpha 0.0500  : 0.05953
# Power for theta0 0.9000       : 0.805
# Iteratively adjusted alpha    : 0.03997
# Empiric TIE for adjusted alpha: 0.05000
# Power for theta0 0.9000       : 0.778

With the nominal α 0.05 the Type I Error will be inflated (0.05953). With the adjusted α 0.03997 (i.e., a ~92% CI) the TIE will be controlled, although with a slight loss in power (decreases from 0.805 to 0.778).
Consider sampleN.scABEL.ad(CV = c(0.30, 0.35), design = "2x2x4") to estimate the sample size preserving both the TIE and target power. In this example 26 subjects would be required.

TOC ↩

ABEL cannot be applied for AUC (except for the WHO). Hence, in many cases ABE drives the sample size. Four period full replicate study (any of TRTR|RTRT, TRRT|RTTR, TTRR|RRTT).

PK  <- c("Cmax", "AUC")
CV  <- c(0.45, 0.30)
# extract sample sizes and power
r1  <- sampleN.scABEL(CV = CV[1], design = "2x2x4",
                      print = FALSE, details = FALSE)[8:9]
r2  <- sampleN.TOST(CV = CV[2], theta0 = 0.90, design = "2x2x4",
                    print = FALSE, details = FALSE)[7:8]
n   <- as.numeric(c(r1[1], r2[1]))
pwr <- signif(as.numeric(c(r1[2], r2[2])), 5)
# compile results
res <- data.frame(PK = PK, method = c("ABEL", "ABE"),
                  n = n, power = pwr)
print(res, row.names = FALSE)
#    PK method  n   power
#  Cmax   ABEL 28 0.81116
#   AUC    ABE 40 0.80999

AUC drives the sample size.

For Health Canada it is the opposite (ABE for C_max and ABEL for AUC).

PK  <- c("Cmax", "AUC")
CV  <- c(0.45, 0.30)
# extract sample sizes and power
r1  <- sampleN.TOST(CV = CV[1], theta0 = 0.90, design = "2x2x4",
                    print = FALSE, details = FALSE)[7:8]
r2  <- sampleN.scABEL(CV = CV[2], design = "2x2x4",
                      print = FALSE, details = FALSE)[8:9]
n   <- as.numeric(c(r1[1], r2[1]))
pwr <- signif(as.numeric(c(r1[2], r2[2])), 5)
# compile results
res <- data.frame(PK = PK, method = c("ABE", "ABEL"),
                  n = n, power = pwr)
print(res, row.names = FALSE)
#    PK method  n   power
#  Cmax    ABE 84 0.80569
#   AUC   ABEL 34 0.80281

Here C_max drives the sample size.

TOC ↩

Sample size assuming homoscedasticity (CV_wT = CV_wR = 0.45) for the widened limits of the Gulf Cooperation Council.

sampleN.scABEL(CV = 0.45, regulator = "GCC", details = FALSE)
# 
# +++++++++++ scaled (widened) ABEL +++++++++++
#             Sample size estimation
#    (simulation based on ANOVA evaluation)
# ---------------------------------------------
# Study design: 2x3x3 (partial replicate)
# log-transformed data (multiplicative model)
# 1e+05 studies for each step simulated.
# 
# alpha  = 0.05, target power = 0.8
# CVw(T) = 0.45; CVw(R) = 0.45
# True ratio = 0.9
# ABE limits / PE constraint = 0.8 ... 1.25 
# Widened limits = 0.75 ... 1.333333 
# Regulatory settings: GCC 
# 
# Sample size
#  n     power
# 54   0.8123

TOC ↩

RSABE

HVD(P)s

Sample size for a four period full replicate study (any of TRTR|RTRT, TRRT|RTTR, TTRR|RRTT) assuming heteroscedasticity (CV_wT 0.40, CV_wR 0.50, i.e., variance-ratio ~0.67). Details of the sample size search suppressed.

sampleN.RSABE(CV = c(0.40, 0.50), design = "2x2x4", details = FALSE)
# 
# ++++++++ Reference scaled ABE crit. +++++++++
#            Sample size estimation
# ---------------------------------------------
# Study design: 2x2x4 (4 period full replicate)
# log-transformed data (multiplicative model)
# 1e+05 studies for each step simulated.
# 
# alpha  = 0.05, target power = 0.8
# CVw(T) = 0.4; CVw(R) = 0.5
# True ratio = 0.9
# ABE limits / PE constraints = 0.8 ... 1.25 
# Regulatory settings: FDA 
# 
# Sample size
#  n    power
# 20   0.81509

TOC ↩

NTIDs (FDA, CDE)

Sample size assuming heteroscedasticity (CV_w 0.10, variance-ratio 2.5, i.e., the test treatment has a substantially higher variability than the reference). TRTR|RTRT according to the FDA’s guidance.⁷ Assess additionally which one of the three components (scaled ABE, conventional ABE, s_wT/s_wR-ratio) drives the sample size.

CV <- signif(CVp2CV(CV = 0.10, ratio = 2.5), 4)
n  <- sampleN.NTID(CV = CV)[["Sample size"]]
# 
# +++++++++++ FDA method for NTIDs ++++++++++++
#            Sample size estimation
# ---------------------------------------------
# Study design:  2x2x4 (TRTR|RTRT) 
# log-transformed data (multiplicative model)
# 1e+05 studies for each step simulated.
# 
# alpha  = 0.05, target power = 0.8
# CVw(T) = 0.1197, CVw(R) = 0.07551
# True ratio     = 0.975 
# ABE limits     = 0.8 ... 1.25 
# Implied scABEL = 0.9236 ... 1.0827 
# Regulatory settings: FDA 
# - Regulatory const. = 1.053605 
# - 'CVcap'           = 0.2142 
# 
# Sample size search
#  n     power
# 32   0.699120 
# 34   0.730910 
# 36   0.761440 
# 38   0.785910 
# 40   0.809580
suppressMessages(power.NTID(CV = CV, n = n, details = TRUE))
#        p(BE)  p(BE-sABEc)    p(BE-ABE) p(BE-sratio) 
#      0.80958      0.90966      1.00000      0.87447

The s_wT/s_wR component shows the lowest probability to demonstrate BE and hence, drives the sample size.

TOC ↩

Compare that with homoscedasticity (CV_wT = CV_wR = 0.10):

CV <- 0.10
n  <- sampleN.NTID(CV = CV, details = FALSE)[["Sample size"]]
# 
# +++++++++++ FDA method for NTIDs ++++++++++++
#            Sample size estimation
# ---------------------------------------------
# Study design:  2x2x4 (TRTR|RTRT) 
# log-transformed data (multiplicative model)
# 1e+05 studies for each step simulated.
# 
# alpha  = 0.05, target power = 0.8
# CVw(T) = 0.1, CVw(R) = 0.1
# True ratio     = 0.975 
# ABE limits     = 0.8 ... 1.25 
# Regulatory settings: FDA 
# 
# Sample size
#  n     power
# 18   0.841790
suppressMessages(power.NTID(CV = CV, n = n, details = TRUE))
#        p(BE)  p(BE-sABEc)    p(BE-ABE) p(BE-sratio) 
#      0.84179      0.85628      1.00000      0.97210

Here the scaled ABE component shows the lowest probability to demonstrate BE and drives the sample size – which is much lower than in the previous example.

TOC ↩

Comparison with fixed narrower limits applicable in other jurisdictions. Note that a replicate design is not mandatory – reducing the chance of dropouts and requiring less administrations

CV  <- 0.10
# extract sample sizes and power
r1  <- sampleN.NTID(CV = CV, print = FALSE, details = FALSE)[8:9]
r2  <- sampleN.TOST(CV = CV, theta0 = 0.975, theta1 = 0.90,
                    design = "2x2x4", print = FALSE, details = FALSE)[7:8]
r3  <- sampleN.TOST(CV = CV, theta0 = 0.975, theta1 = 0.90,
                    design = "2x2x3", print = FALSE, details = FALSE)[7:8]
r4  <- sampleN.TOST(CV = CV, theta0 = 0.975, theta1 = 0.90,
                    print = FALSE, details = FALSE)[7:8]
n   <- as.numeric(c(r1[1], r2[1], r3[1], r4[1]))
pwr <- signif(as.numeric(c(r1[2], r2[2], r3[2], r4[2])), 5)
# compile results
res <- data.frame(method = c("FDA/CDE", rep ("fixed narrow", 3)),
                  design = c(rep("2x2x4", 2), "2x2x3", "2x2x2"),
                  n = n, power = pwr, a = n * c(4, 4, 3, 2))
names(res)[5] <- "adm. #" # number of administrations
print(res, row.names = FALSE)
#        method design  n   power adm. #
#       FDA/CDE  2x2x4 18 0.84179     72
#  fixed narrow  2x2x4 12 0.85628     48
#  fixed narrow  2x2x3 16 0.81393     48
#  fixed narrow  2x2x2 22 0.81702     44

TOC ↩

Dose-Proportionality

CV 0.20 (20%), doses 1, 2, and 8 units, assumed slope β₀ 1, target power 0.90.

sampleN.dp(CV = 0.20, doses = c(1, 2, 8), beta0 = 1, targetpower = 0.90)
# 
# ++++ Dose proportionality study, power model ++++
#             Sample size estimation
# -------------------------------------------------
# Study design: crossover (3x3 Latin square) 
# alpha = 0.05, target power = 0.9
# Equivalence margins of R(dnm) = 0.8 ... 1.25 
# Doses = 1 2 8 
# True slope = 1, CV = 0.2
# Slope acceptance range = 0.89269 ... 1.1073 
# 
# Sample size (total)
#  n     power
# 18   0.915574

Note that the acceptance range of the slope depends on the ratio of the highest and lowest doses (i.e., it gets tighter for wider dose ranges and therefore, higher sample sizes will be required).
In an exploratory setting wider equivalence margins {θ₁, θ₂} (0.50, 2.00) were proposed,⁸ translating in this example to an acceptance range of 0.66667 ... 1.3333 and a sample size of only six subjects.

TOC ↩

Power Analysis

Explore impact of deviations from assumptions (higher CV, higher deviation of θ₀ from 1, dropouts) on power. Assumed within-subject CV 0.20 (20%), target power 0.90. Plot suppressed.

res <- pa.ABE(CV = 0.20, targetpower = 0.90)
print(res, plotit = FALSE)
# Sample size plan ABE
#  Design alpha  CV theta0 theta1 theta2 Sample size Achieved power
#     2x2  0.05 0.2   0.95    0.8   1.25          26      0.9176333
# 
# Power analysis
# CV, theta0 and number of subjects leading to min. acceptable power of ~0.7:
#  CV= 0.2729, theta0= 0.9044
#  n = 16 (power= 0.7354)

If the study starts with 26 subjects (power ~0.92), the CV can increase to ~0.27 or θ₀ decrease to ~0.90 or the sample size decrease to 10 whilst power will still be ≥0.70.
However, this is not a substitute for the ‘Sensitivity Analysis’ recommended in ICH-E9,⁹ since in a real study a combination of all effects occurs simultaneously. It is up to you to decide on reasonable combinations and analyze their respective power.

TOC ↩

Speed Comparisons

Performed on a Xeon E3-1245v3 3.4 GHz, 8 MB cache, 16 GB RAM, R 4.3.3 64 bit on Windows 7.

ABE

2×2 crossover design, CV 0.17. Sample sizes and achieved power for the supported methods (the 1^st one is the default).

    method  n   power time (s)
     owenq 14 0.80568  0.00128
       mvt 14 0.80569  0.11778
noncentral 14 0.80568  0.00100
   shifted 16 0.85230  0.00096

The 2^nd exact method is substantially slower than the 1^st. The approximation based on the noncentral t-distribution is slightly faster but matches the 1^st exact method closely. Though the approximation based on the shifted central t-distribution is the fastest, it might estimate a larger than necessary sample size. Hence, it should be used only for comparative purposes.

ABEL

Four period full replicate study (any of TRTR|RTRT, TRRT|RTTR, TTRR|RRTT), homogenicity (CV_wT = CV_wR 0.45). Sample sizes and achieved power for the supported methods.

              function              method  n   power time (s)
        sampleN.scABEL    ‘key’ statistics 28 0.81116   0.1348
 sampleN.scABEL.sdsims subject simulations 28 0.81196   2.5377

Simulating via the ‘key’ statistics is the method of choice for speed reasons.
However, subject simulations are recommended if

• the partial replicate design (TRR|RTR|RRT) is planned and
• the special case of heterogenicity CV_wT > CV_wR is expected.

TOC ↩

Installation

You can install the released version of PowerTOST from CRAN with

package <- "PowerTOST"
inst    <- package %in% installed.packages()
if (length(package[!inst]) > 0) install.packages(package[!inst])

… and the development version from GitHub with

# install.packages("remotes")
remotes::install_github("Detlew/PowerTOST")

Skips installation from a github remote if the SHA-1 has not changed since last install. Use force = TRUE to force installation.

TOC ↩

Session Information

Inspect this information for reproducibility. Of particular importance are the versions of R and the packages used to create this workflow. It is considered good practice to record this information with every analysis.
Version 1.5.6 built 2024-03-18 with R 4.3.3.

options(width = 66)
sessionInfo()
# R version 4.3.3 (2024-02-29 ucrt)
# Platform: x86_64-w64-mingw32/x64 (64-bit)
# Running under: Windows 10 x64 (build 19045)
# 
# Matrix products: default
# 
# 
# locale:
# [1] LC_COLLATE=German_Germany.utf8 
# [2] LC_CTYPE=German_Germany.utf8   
# [3] LC_MONETARY=German_Germany.utf8
# [4] LC_NUMERIC=C                   
# [5] LC_TIME=German_Germany.utf8    
# 
# time zone: Europe/Berlin
# tzcode source: internal
# 
# attached base packages:
# [1] stats     graphics  grDevices utils     datasets  methods  
# [7] base     
# 
# other attached packages:
# [1] PowerTOST_1.5-6
# 
# loaded via a namespace (and not attached):
#  [1] cubature_2.1.0    compiler_4.3.3    fastmap_1.1.1    
#  [4] cli_3.6.2         tools_4.3.3       htmltools_0.5.7  
#  [7] rstudioapi_0.15.0 yaml_2.3.8        Rcpp_1.0.12      
# [10] mvtnorm_1.2-4     rmarkdown_2.26    knitr_1.45       
# [13] xfun_0.42         digest_0.6.35     rlang_1.1.3      
# [16] evaluate_0.23

TOC ↩

1. Tóthfalusi L, Endrényi L. Sample Sizes for Designing Bioequivalence Studies for Highly Variable Drugs. J Pharm Pharmacol Sci. 2012; 15(1): 73–84. doi:10.18433/j3z88f. Open access. ↩
2. Fieller EC. Some Problems In Interval Estimation. J Royal Stat Soc B. 1954; 16(2): 175–85. JSTOR:2984043. ↩
3. U.S. Food and Drug Administration, Office of Generic Drugs. Draft Guidance on Dabigatran Etexilate Mesylate. Recommended Jun 2012; Revised Sep 2015, Jul 2017. Online. ↩
4. U.S. Food and Drug Administration, Office of Generic Drugs. Draft Guidance on Rivaroxaban. Recommended Sep 2015. Online. ↩
5. U.S. Food and Drug Administration, Office of Generic Drugs. Draft Guidance on Edoxaban Tosylate. Recommended May 2017; Revised Mar 2020. Online. ↩
6. Labes D, Schütz H. Inflation of Type I Error in the Evaluation of Scaled Average Bioequivalence, and a Method for its Control. Pharm Res. 2016; 33(11): 2805–14. doi:10.1007/s11095-016-2006-1. ↩
7. U.S. Food and Drug Administration, Center for Drug Evaluation and Research. Draft Guidance for Industry. Bioequivalence Studies with Pharmacokinetic Endpoints for Drugs Submitted Under an ANDA. August 2021. Online. ↩
8. Hummel J, McKendrick S, Brindley C, French R. Exploratory assessment of dose proportionality: review of current approaches and proposal for a practical criterion. Pharm. Stat. 2009; 8(1): 38–49. doi:10.1002/pst.326. ↩
9. International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use. ICH Harmonised Tripartite Guideline. E9. Statistical Principles for Clinical Trials. 5 February 1998. Online. ↩