-
Notifications
You must be signed in to change notification settings - Fork 6
/
README.Rmd
130 lines (96 loc) · 5.88 KB
/
README.Rmd
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
---
output: github_document
---
<!-- README.md is generated from README.Rmd. Please edit that file -->
```{r setup, include = FALSE}
knitr::opts_chunk$set(
collapse = TRUE,
comment = "#>",
fig.path = "man/figures/README-",
out.width = "100%"
)
```
```{r, echo = F}
suppressPackageStartupMessages(library(multiarm))
```
# multiarm <img src='man/figures/multiarm.png' align="right" height="139" />
`r badger::badge_devel("mjg211/multiarm", "blue")`
`r badger::badge_code_size("mjg211/multiarm")`
`r badger::badge_lifecycle("maturing", "blue")`
`r badger::badge_custom("contributions", "welcome", "blue")`
## Description
`multiarm` provides functions to assist with the design of single- and multi-stage multi-arm clinical trials.
In either case the available functions allow for sample size determination, trial simulation, analytical operating characteristic calculation, and the production of several informative plots.
Note that an R Shiny graphical user interface is also available for designing single-stage trials; it can be accessed within R using `multiarm::gui()`, or online at [https://mjgrayling.shinyapps.io/multiarm](https://mjgrayling.shinyapps.io/multiarm).
Additional information on this app can be found in [Grayling and Wason (2020)](https://doi.org/10.1186/s12885-020-6525-0).
## Getting started
You can install the latest development version of `multiarm` from [Github](https://github.com/) with:
```{r, eval = F}
devtools::install_github("mjg211/multiarm")
```
An introductory example of how to make use of some of the package's core functionality can be found below.
More detailed support, for single-stage designs, is available in the package vignette, which can be accessed with `vignette("single_stage", package = "multiarm")`.
See also [Grayling and Wason (2020)](https://doi.org/10.1186/s12885-020-6525-0).
For further help, please email [email protected].
## Details
In total, 46 functions are currently available.
Their naming conventions are such that several character strings are joined together, separated by underscores.
The first string indicates the purpose of the function (i.e., what type of calculation it performs):
- `build_###_###()`: Build multi-arm clinical trial design objects, like those returned by the `des_###_###()` functions.
For use when a specific design is of interest.
- `des_###_###()`: Determine the sample size required by a particular type of multi-arm clinical trial design.
- `gui()`: Provides a graphical user interface to design determination.
- `opchar_###_###()`: Determine the operating characteristics (power, family-wise error-rates, etc.) of a supplied multi-arm clinical trial design, via multivariate normal integration.
- `plot.multiarm_des_###_###()`: Produce informative plots (power, false discovery rate curves, etc.) relating to a supplied multi-arm clinical trial design.
- `sim_###_###()`: Empirically estimate the operating characteristics (power, family-wise error-rates, etc.) of a supplied multi-arm clinical trial design, via simulation.
The second indicates the design:
- `###_dtl_###()`: Relate to multi-stage drop-the-losers designs. See, e.g., [Wason *et al* (2017)](https://doi.org/10.1177/0962280214550759).
- `###_gs_###()`: Relate to group-sequential multi-arm multi-stage designs. See, e.g., [Magirr *et al* (2012)](https://doi.org/10.1093/biomet/ass002).
- `###_ss_###()`: Relate to single-stage designs. See, e.g., [Grayling and Wason (2020)](https://doi.org/10.1186/s12885-020-6525-0).
The third indicates what type of outcome the function is for:
- `###_###_bern()`: Assume a Bernoulli distributed primary outcome.
- `###_###_norm()`: Assume a normally distributed primary outcome.
- `###_###_pois()`: Assume a Poisson distributed primary outcome.
## Example: Single-stage design for a normally distributed outcome
Typically, a `des_###_###()` function would be used first to identify a design for the trial parameters of interest.
For example, consider designing a single-stage trial for:
- Three experimental treatment arms (see `K`).
- Desiring a family-wise error-rate of at most 5\%, controlling using Dunnett's correction (see `alpha` and `correction`).
- Desiring marginal power to reject each null hypothesis of 80\% for a clinically relevant difference of 1 (see `beta`, `delta1`, and `power`).
- Assuming the standard deviation of all responses is 1 (see `sigma`).
- Allocating patients equally to each arm (see `ratio`).
To compute the design, we would run:
```{r des}
des <- des_ss_norm(K = 3,
alpha = 0.05,
beta = 0.2,
delta1 = 1,
sigma = rep(1, 4),
ratio = rep(1, 3),
correction = "dunnett",
power = "marginal")
```
Then, the total required sample size is:
```{r N}
des$N
```
In addition, the operating characteristics under the *global null*, *global alternative*, and each of the *least favourable configurations*, can be accessed with:
```{r opchar}
des$opchar
```
Useful plots can also be produced with `plot.multiarm_des_ss_norm()` as follows:
```{r plot, dpi = 600}
plot(des)
```
## References
Grayling MJ, Wason JMS (2020) A web application for the design of multi-arm clinical trials.
*BMC Cancer* 20:80.
DOI: [10.1186/s12885-020-6525-0](https://doi.org/10.1186/s12885-020-6525-0).
PMID: [32005187](https://www.ncbi.nlm.nih.gov/pubmed/32005187).
Magirr D, Jaki T, Whitehead J (2012) A generalized Dunnett test for multi-arm multi-stage clinical studies with treatment selection.
*Biometrika* 99(2):494--501.
DOI: [10.1093/biomet/ass002](https://doi.org/10.1093/biomet/ass002).
Wason J, Stallard N, Bowden J, Jennison C (2017) A multi-stage drop-the-losers design for multi-arm clinical trials.
*Stat Meth Med Res* 26(1):508--524.
DOI: [10.1177/0962280214550759](https://doi.org/10.1177/0962280214550759).
PMID: [25228636](https://www.ncbi.nlm.nih.gov/pubmed/25228636).