Abk�rzung zur Hauptnavigation
Abk�rzung zu den Newsmeldungen
Abk�rzung zu den Topstories
Inhaltsbereich
### Legend for the Output-File

## Schnellinfo

## Featured

ADCCT is a collection of SAS-macros to simulate the statistical properties of adaptive seamless two-stage designs with treatment selection.
The Programs allows to compare k treatments to a control group (with k=2 and 3). The structure of the programs is modular to allow for
flexible extensions and modifications.

A more detailed description of the methods implemented can be found in the following papers:

- Adaptive Designs for Confirmatory Clinical Trials

Bretz F*, Koenig F*, Brannath W, Glimm E, Posch M

Statistics In Medicine 2009 Apr 15;28(8):1181-217.

* The first two authors (in alphabetic order) have made equal contributions to this paper. - Adaptive Dunnett Tests for Treatment Selection

Franz Koenig, Werner Brannath, Frank Bretz, and Martin Posch

Statistics in Medicine. 2008, May 10;27(10):1612-25.

The main module is the program “main_module.sas“, which starts the SAS-IML environment and calls all other sub-macros. All relevant settings are made in the beginning of this module.

All sub-macros are stored in the file “macros_two_stage_ADCCT.sas”. Before running the simulations with “main_module.sas” you have to run “macros_two_stage_ADCCT.sas”.

All simulations presented in section 5 "A simulation study" of the the paper by Bretz, Koenig et al (2009) were performed with the SAS - macros ADCCT. Modified versions of the file "main_module.sas" with the specifications to run the simulations in Section 5 will are available on this webpage for download.

- Comparison of 2 active treatments versus control - Figure 3: Script file main_module_V1.00_examle_figure3.sas and SAS-Code for EPS-figures
- Comparison of 3 active treatments versus control - Figure 6: Script file main_module_V1.00_examle_figure6.sas and SAS-Code for EPS-figures

Legal disclaimer: The SAS macros ADCCT are freeware and are offered by the authors in the spirit of enhancing research. The macros are presented with absolutely no warranty whatsoever.

Manual ADCCT (not available)

*COMB_FUNC: *

comb_func= **1 **Inverse Normal Combination Function

comb_func= **2 **Fishers Combination Function

comb_func= **. **no adaptive combination test

*Methode *

/* multiplicity adjustment strategies for the combination tests */

/* methods 1-6 and 101-106 are applied for comb_func=1 or 2. */

methode= **1 **Dunnett

methode= **2 **p_adjusted by Bonferroni

methode= **4 **Sidak

methode= **5 **p_hier

methode= **6 **p_simes

methode= **101 **1st stage Dunnett, 2nd estimated hierarchical

methode= **102 **1st stage Bonferroni, 2nd estimated hierarchical

methode= **104 **1st stage Sidak, 2nd estimated hierarchical

methode= **105 **1st stage fix hierarchical, 2nd estimated hierarchical

methode= **106 **1st stage Simes, 2nd estimated hierarchical

/* standard seperate phII/phII studies */

/* for testing only the observations of the second stage are used */

methode= **21 **seperate Bonferroni with nr of selected treatments

methode= **22 **seperate Dunnett with nr of selected treatments

/* one stage test (form ally the p-value of a dropped treatment is set to p=1*/

/* all observations are used for testing */

methode= **31 **Bonferroni

methode= **32 **Dunnett

/* the conditional error aprroach to adaptive treatment selection */

/* (new working paper - not published yet */

methode= **11 **p-values of the second stage are calculated via conditional error

concept

(equal to orginal Dunnett test if no treatments are selected and no sample size reallocation is performed)

methode= **12 **p-value corrected by Dunnett using the data only \r\nof the second

stage

*ALPHA *

significance level

*N *

perplanned total no of observations per treatment group

*n1 *

preplanned first stage sample size

*n2 *

preplanned second stage sample size (N=n1+n2)

*SIGMA *

Standard deviation

*MU0 *

mean value mu for treatment group **0 **(control group)

*MU1 *

mean value mu for treatment group **1 **(active treatment)

*MU2 *

mean value mu for treatment group **2 **(active treatment)

*MU3 *

mean value mu for treatment group 3 (active treatment)

*REJ@LEAST_ONE *

reject at least one null hypotheses

*REJ_ALO_H0 *

reject at least one true null hypotheses (simulated level)

*REJ_ALO_H1 *

reject at least one false null hypotheses (simulated power)

*REJ_ALL *

*rej_tr1 *

probability of a rejection for treatment **1 **

*P_SEL_TR0 *

*p_sel_tr1 *

probability of a selection for treatment **1 **

*rej_trX *

probability of a rejection for treatment **X **

*p_sel_trX *

probability of a selection for treatment **X**