User Documentation for the AOT425StatPgm Program

Prepared for the US Environmental Protection Agency

by Westat, May 2001; updated by USEPA June 2003

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

2. Installing the Program . . . . . . . . . . . . . . . . . . . . . . . . . 1

3. Overview of the OECD TC 425 Test Procedure . . . . . . . . . . . . . . . 3

4. Running the Program . . . . . . . . . . . . . . . . . . . . . . . . . . 4

5. Using the AOT425StatPgm for a Main Test . . . . . . . . . . . . . . . . 11

6. Using the Program for a Limit Test . . . . . . . . . . . . . . . . . . . 15

7. Options Menu Details . . . . . . . . . . . . . . . . . . . . . . . . . . 15

8. Recovering from Mistakes . . . . . . . . . . . . . . . . . . . . . . . . 17

9. Warning Messages . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

10. Smallest Usable Dose . . . . . . . . . . . . . . . . . . . . . . . . . . 18

11. Preparing the Reading Data Files and Importing Data from

Other Programs . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

12. Technical Details . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

1. - Introduction

The "Acute Oral Toxicity (Guideline 425) Statistical Program" (AOT425StatPgm)

is designed to be used with the acute oral toxicity testing procedure

presented in OECD GUIDELINE FOR THE TESTING OF CHEMICALS, Section 4: Health

Effects Test No. 425, Acute Oral Toxicity: Up-and-Down Procedure. For

convenience purposes, we will refer to that document as OECD TG 425 Procedure.

The AOT425StatPgm program performs the calculations required to complete the

test procedure.

The AOT425StatPgm program, a copy of this documentation, and a copy of the

testing guidelines can be found on the Internet at http://www.epa.gov/

oppfead1/harmonization. Information on how to download the program and

documentation from the Internet can be found at that site, or by searching the

Internet for "AOT425S" You will find a file named "AOT425Setup.exe" which is

downloaded before you follow the instructions in this manual. The program can

also be obtained on disk from the OECD or by contacting USEPA at the below

address. This document describes how to install the program, starting with

the "AOT425Setup.exe" or with two floppy disks, and how to use the program.

EPA and OECD are interested in improving the AOT425StatPgm program. If you

have any problems with the program or suggestions for improvement, please send

a message by mail to AOT425 Coordinator, USEPA(7506C), 1200 Pennsylvania Ave.

NW, Washington DC 20460, or send an email to AOT425@epa.gov. In addition,

user communications regarding problems may be submitted by commenting at the

Comments or Questions? link at http://www.epa.gov/pesticides/comments and

including the string AOT425 in your message. An entry will be added on the

page http://www.epa.gov/oppfead1/harmonization/ to highlight problems and

improvements that may have been identified.

1
2. - Installing the Program

In order to install and run AOT425StatPgm, you must have Windows 95 or higher

installed on your computer and a minimum of 5 megabytes of free hard disk

space. The Regional Settings of the computer (found on the Regional Settings

Control Panel from the Start/Settings button) must be set for English(United

States) before and during program use.

If you have installed a previous version of this program, uninstall the

previous version before installing this version. To uninstall the program,

open up the control panel and double click on the "Add/Remove Programs" icon.

Follow the instructions to remove the program. You may be asked several

questions during the process of uninstalling an old version of the program.

If asked, we recommend keeping shared files and ignoring any message similar

to "Can't delete AOT425StatPgm."

Installing the Program with Floppy Disks

Insert Disk #1 into the floppy drive on your computer. Using your mouse

pointer, click on the "Start" button at the lower left corner of your Windows

desktop, then click on the "Run" option. Type "A:\setup" in the text box, and

click the "Ok" button. Click the "OK" button to install the program files into

the C:\program files\AOT425StatPgm folder. Insert Disk #2 when requested by

the installer. A shortcut will automatically be placed in the Start Menu for

easy launch of AOT425StatPgm.

Installing the Program Starting with AOT425Setup.exe

The "AOT425Setup.exe" file from the Internet or a CD is a self-extracting

archive used to install the AOT425StatPgm on your computer. There are two

steps to install the program. First, run the "AOT425Setup.exe" program to put

the installation files into the "C:\Windows\Temp" directory. Second, run the

"Setup.exe" file in the "C:\Windows\Temp" directory to install the program on

your computer.

There are two ways to run the "AOT425Setup.exe" self-extraction program. You

can use Windows Explorer to open the folder containing the downloaded file,

and then double-click "AOT425 Setup.exe" to run the self-extraction program.

You may also click on the "Start" button in the lower left corner of your

Windows desktop, click on the "Run" option, and then type the full path of the

file location in the text box. Click "OK" to begin the extraction. You will

need to specify a folder for the installation files. The default folder is

"C:\Windows\Temp." Other locations can be used.

When you have extracted the files for the installation, there are two ways to

run the installation program, "Setup.exe". You can use Windows Explorer to

open the folder containing the installation files, and then double-click

"Setup.exe" to run the installation program (This file will have a

modification date of 1/16/1997). You may also click on the "Start" button in

the lower left corner of your Windows desktop, click on the "Run" option, and

then type the full path of the installation file location in the text box. The

files will be located in the C:\Windows\Temp directory unless another

directory was specified. The Windows Installer should appear. To begin the

installation process, click the large button in the installer window, as shown

below.

2
You will be prompted for a location in which to install the program. Select

"OK" to install the program to the default directory of "C:\Program

Files\AOT425StatPgm" or select another location. Select "OK" again to

complete program installation. A shortcut will automatically be placed in the

Start Menu for easy launch of AOT425StatPgm.

After running "Setup.exe" you can delete the installation files (in

C:\Windows\Temp or another selected directory).

3. - Overview of the OECD TG 425 Test Procedure

The purpose of the procedure is to test the short-term toxicity of compounds

to rodents. There are two forms of the test, the limit test and main test.

These tests are described below. The objective of the testing is to obtain

information on the median lethal dose (LD50) for the test substance. The

procedure is sequential; that is, whether to dose another animal depends on

the results of the previous test. Users may enter test results for one

animal, save the data, and then return several days later and enter test

results for the next animal. Once testing is complete, the program uses the

test results to calculate the estimated LD50. The user must use another

program to store and manage the test data. Only the information needed to

calculate the stopping criteria and the LD50 estimate are entered into the

AOT425StatPgm. The AOT425StatPgm calculates: 1) the recommended dose for the

next animal, 2) when to stop dosing animals, and 3) statistical estimates of

the LD50 and its confidence interval.

Overview of Limit Test

The limit test generally is used in cases where information indicates the test

substance is likely to be nontoxic below regulatory limit doses. Slightly

different test procedures are used for limit doses of 2000 mg/kg and 5000

mg/kg. In both cases, the general procedure is to dose one animal at the

limit dose (i.e., 2000 mg/kg or 5000 mg/kg). If the animal dies, the LD50 is

inferred to be less than the test dose, and the main test must be conducted to

determine the LD50. If the animal survives, additional animals are dosed.

The response of the additional animals will determine whether the LD50 is less

than or greater than the limit dose and if a main test must be conducted (see

paragraphs 22 - 30 of "OECD GUIDELINE FOR THE TESTING OF CHEMICALS, Section 4:

3
Health Effects Test No. 425, Acute Oral Toxicity: Up-and-Down Procedure" for a

complete description of the limit test).

Overview of Main Test

The main test is performed when the chemical is expected to be toxic or when

little or no information about its toxicity is available. Single animals are

dosed in sequence. The first animal receives a dose one step below the best

(or assumed) estimate of the LD50. If the animal survives, the second animal

receives a higher dose. If the first animal dies, the second animal receives

a lower dose. This sequence is repeated until one of three stopping criteria

is met. After recording the long-term outcomes for the animals, typically

after 2 weeks, the LD50 estimate and confidence interval (CI) may be

calculated (see paragraphs 22 and 31 - 35 of " OECD GUIDELINE FOR THE TESTING

OF CHEMICALS, Section 4: Health Effects Test No. 425, Acute Oral Toxicity: Up-

and-Down Procedure " for a complete description of the main test).

4. - Running the Program

When the installation has finished, you can start the program by clicking the

"Start" button in the lower left corner of the screen, scrolling up to

"Programs," and then clicking on the "AOT425StatPgm" icon in the menu.

The program has two windows. The first window, the Data Edit window, is used

primarily to enter and edit test data. The Data Entry window also shows the

recommended dose for the next test, and, if appropriate, the estimated LD50.

The second window, the Report window, shows the data, recommended dose,

estimated LD50, and confidence interval in a report form. The report can be

saved as a text file or printed.

Using the Data Edit Window

Figure 1 shows the screen after opening the program. In Figure 1 the

following parts of the Data Edit window are labeled in large bold/italic text:

the task bar, header area, data grid, and message box.

4
 Figure 1: Data Edit window

With the exception of the "Options" menu, the menus in the task bar act like

buttons. If you click on a menu, the program immediately performs that action

without first displaying a list of menu items. Unlike the other menus, the

Options menu has three menu items.

Throughout this document, menu commands and push-buttons on the task bar will

be referenced in [bracketed bold text]. This distinguishes actual commands

from regular text and minimizes confusion. The menu items and the functions

are as follows:

[New Test] Clears all entered information. Information for a new test can
then be entered.
[Load Data] loads previously saved test information from a file.
[Save Data] saves the current test information to a file.
[Get Report] displays a report of the current test in a Report window. The
report contains the file name, the header information, the recommended dose
progression, the actual dose and outcomes, a summary of the long-term
results, and the estimated LD50. If the recommended dosages are not
followed in the testing, the report will also contain a warning regarding
the methods used (see additional information in Section 9).
[Options] allows the user to set options for saving files, identifying the
smallest usable dose for testing, and qualifying the confidence intervals
for the LD50 estimate (see details in Section 7).

5
 [About AOT425StatPgm] provides information about the origin and date of
this program.
[Exit] quits the program.

Below the task bar, you will find the header information. The header

information describes the test being performed. The information describing

the test is entered into the white boxes. Each white box has text on a gray

background to the left of the box describing the information to be entered.

"Text/Substance" is a description of the test or substance being tested.

"Test Type" is the type of test being performed. The test type is defined
using a drop-down menu to select one of two options, either "Main" or
"Limit."

"Limit Dose" is the maximum dose that might be used. The limit dose is
defined using a drop-down menu to select one of two options, either "2000"
or "5000."

"LD50" is the best estimate of the LD50 before the test based on whatever
information is available. If you have no information from which to
estimate the LD50, enter the word Default.

"Sigma" is the best estimate of the standard deviation for the lethal dose,
in the log scale. Unless you have good information about sigma, use the
default value of 0.5.

When performing a limit test, the "LD50" and "sigma" are not used and cannot

be changed. When performing a main test, the "LD50" and "sigma" are used to

determine the recommended dose for each animal.

The "Data Grid" is located below the "Header Information." As the testing

progresses, you will enter information for each animal into the columns. The

data grid has the following six columns:

"Test Seq" is the test sequence. Note, the test sequence column cannot be
changed.

"Animal ID" is a character or numeric ID that can be used to link the data
in the AOT425StatPgm to data for the same animal in other data files.

The animal ID can be up to 40 characters long. The AOT425StatPgm does not

use the animal ID in the calculations. However, the ability to link the

data used in the analysis to other data files is important. Therefore, the

program requires that the Animal ID column be filled in.

"Dose (mg/kg)" is the actual dose of the tested substance that was
administered to the animal. The actual dose may be different than the
recommended dose for the animal.

The dose must be a positive number and can have up to 12 characters.

"Short-term outcome" is the animal response within 48 hours (or the

selected survival interval). Valid codes are O = lived, X = died. (Note,
this is the letter "O," not "zero.")

6
 "Long-term outcome" is the animal response at the end of a 14-day
observation period. Valid codes are O = lived, X = died.

"Program's data entry messages" has messages from the program to guide data
entry.

If the recommended dose for the next animal can be calculated, it will be

shown in a light blue box in the dose column below the information for the

last animal entered. The recommended dose is rounded to 2 or 3 significant

figures.

You can move around the data grid using the tab key, enter key, and arrow

keys. Text can be deleted from a grid cell using the backspace key.

Data for the first animal is entered into the first row. Data for subsequent

animals are entered into successive rows until the testing stops. Each row in

the data grid contains the information for one animal. The information should

be entered in the order in which the animals are tested, as indicated by the

first column, "Test Sequence."

The message box is located at the bottom of the screen. It contains

instructions on how to use the program, instructions on when to stop testing,

and error messages. If there are errors in the header information, a message

will appear in the message box.

When the program starts, or when starting to enter information for a new test

(See [New Test]), the Test Type is "Main" and the Limit Dose is "2000." These

values must be changed when performing a limit test or if a limit dose of 5000

is to be used. The "LD50" and "Sigma" boxes show the default values. The

test description (Test/Substance) starts with the text "Enter test

description." All of the header information should be reviewed and corrected

before entering data into the data grid.

The program does not print the Data Edit window. To print the data, either

get a report and print the report, or hit the buttons "Alt" and "Print Screen"

simultaneously to transfer a picture of the data edit window to the clipboard.

Then paste the picture into another program (such as a word processor) and

print from the other program.

Using the Report Window

Clicking on the [Get Report] text in the task bar in the Data Edit window

opens up the Report window. Figure 2 below shows an example of a typical

Report window. The topmost section of the window contains a task bar (labeled

in large bold/italic text), with the following commands:

[New Test] - Exits the Report window, returns you to the Data Edit window
and clears current test data
[Print Report] - Prints Report window to the selected default printer
[Save Report] - Saves the Report window in a text file format
[Edit Data] - Exits Report window and returns you to the Data Edit window
[Exit] - Quits the program

7
If the report has not already been saved to a file, you will be prompted to

save the report when selecting [New Test], [Edit Data], and [Exit].

The following tips for printing may be helpful:

When clicking on [Print Report] the report text is sent immediately to the
default printer. The program does not use a print dialog box or allow you
to cancel printing.
To change the default printer or change printer options, select the START
menus Settings/Printers menu item. A window will appear. Select the
printer to use. Then use the File menu in the Printers window to set the
printer as the default printer or to modify the printer settings (by
selecting "Properties").
You may be able to facilitate reading the report into Microsoft by using
".doc" as the extension for the report file.
You may be able to facilitate reading the report into WordPerfect by using
".wpd" as the extension for the report file.
The mouse can be used to select text in the Report window. The selected
text can be copied to the clipboard by holding the Control key down while
pressing the C key on the keyboard. Text copied to the clipboard can be
pasted into a word processing program and printed from there.

Figure 2: Report window

8
Figure 3 shows the full text of a hypothetical Report window with different

sections of the report labeled. The upper section of the Report window shows

the date/time, the filename, and the last time the file was modified. The

header information follows the file information and includes the test

description, the test type, the limit dose, the assumed values for LD50 and

sigma, and the recommended dose progression. The information from the Data

Grid is shown in the middle part of the Report window. This includes the

animal ID, the dose, the short-term result, and the long-term result. The

animal data is followed by the dose recommendation, if any, and the stopping

criteria, if any.

If the sequence of doses and short-term outcomes are not consistent with the

OECD TG 425 procedure, one or more warnings may appear (see additional

information in Section 9). If all of the long-term outcomes have been

entered, the bottom of the report summarizes the long-term outcomes and

provides the estimated LD50 along with the confidence interval for the LD50.

9
AOT425statpgm (Version: 1.0) Test Results and Recommendations
Acute Oral Toxicity (OECD Test Guideline 425) Statistical Program DATA FILE
INFORMATION
Date/Time: Monday, May 21, 2001, 9:52:33 AM

Data file name: Case 5 33a 3 dead.dat

Last modified: 5/16/2001 12:05:42 PM

Test/Substance: Example Data.

Test type: Main Test HEADER

Limit dose (mg/kg): 2000 INFORMATION

Assumed LD50 (mg/kg): Default

Assumed sigma (mg/kg): 1

Recommended dose progression: 2000, 175, 17.5, 1.75

DATA:

Test Animal Dose Short-term Long-term

TEST DATA

Seq. ID (mg/kg) Result Result

____________________________________________ FROM DATA

1 1 175 O O GR ID
2 2 2000 X X
3 3 175 O O
4 4 2000 X X
5 5 175 O O
6 6 2000 O O
7 7 2000 X X
8 8 175 O O
9 9 2000 O O
10 10 2000 O O

11 11 2000 O O

____________________________________________

(X = Died, O = Survived)

Dose Recommendation: The main test is complete.

WARNING:

WARNING
The data may not be correct

A spacing between doses is not completely consistent with the

(IF APPLICABLE)
assumed sigma

Stopping criteria met: 3 at Limit Dose.

SUMMARY OF LONG-TERM RESULTS:

SUMMARY OF
Dose O X Total TEST DATA BY
______________________________________ DOSE
175 4 0 4

2000 4 3 7

______________________________________

All Doses 8 3 11

Statistical Estimate based on long term outcomes: STATISTICAL

ESTIMATE OF
Estimated LD50 = 2000 (The one dose with partial response).
LD50
95% PL Confidence interval is 797.7 to Greater than 20,000.

Figure 3: Example of Report window text with labels

10
5. - Using the AOT425StatPgm for a Main Test

The following steps provide a brief overview of how the program may be used

for a main test. Figure 7 at the end of this section illustrates the steps.

Enter the test description.

Select Main from the Test Type menu.
Select the Limit Dose from the Limit Dose menu.
For a main test, specify the assumed LD50 and sigma if values are known;
otherwise, use the default values.

Note that the first row of data grid contains the recommended dose for the

first animal.

Save the data (click on [Save Data], select an appropriate directory, enter
a file name (the default filename is work.dat or work_001.dat; we recommend
changing "work" to another more descriptive name).

The recommended dose for each animal in turn can be obtained from the data

grid or from the report. The report, if printed or saved, provides

documentation of the recommended dose.

Click [Get Report]

Figure 4: Report screen before entering data for the first animal

11
[Save Report] (the default file name is the name of the data file followed by

the word "Report" with the extension ."txt")

[Exit] the program.

After dosing the first animal and observing the short-term outcome:

Start the program.

Click [Load data] to load previously saved data (note the load file dialog
box starts from last used directory).
Enter data for the first animal (note there may be messages in the right
column). Enter the actual dose, which may not be the same as the
recommended dose.
Note recommended dose for the second animal and get a report, if desired.
[Save data].

Dose a second animal and observe

Start the program.

[Load data].
Enter the animal data.

Continue dosing and entering data until the recommended next dose says "Stop

Dosing" (see Figure 5).

Figure 5: Data Edit window with "Stop Testing" message

12
 [Exit] the program.

Observe animals for 14 days to obtain the long-term outcomes.

Start the program.

[Load data]
Enter the longer-term outcomes (should be entered as they become available.
Once several animals have been started, the entry of delayed deaths may
cause program running and report errors. See pages 18 and 23.)

The program says test is complete and calculates the LD50 estimate and its

confidence interval for the LD50.

[Get report].
[Print Report].
[Save Report].
[Exit] the program.

Figure 6: Final report at the end of the main test

13
The upper section of the report window (not shown here) contains the test

description, header information, and sequence of doses (similar to those shown

in Figures 2 and 3), discussed previously. The bottom section of the Report

window (shown here in Figure 6) contains the final dose recommendation

indicating that the test is complete, the stopping criteria, the summary of

long-term results, the statistical estimate of LD50, and the 95% confidence

interval for the LD50.

Lab Input Window Report window

Figure 7: Example program use flow for a main test

14
6. - Using the Program for a Limit Test

To use the program for a limit test, select the Test Type "Limit" from the

drop down menu in the header. If the limit test is to use a dose of 5000

mg/kg, select 5000 from the Limit Dose drop down menu. Figure 8 shows how the

screen looks when starting a limit test at 2000 mg/kg.

Figure 8: Data Edit window at start of limit test

Notice that the data grid is much shorter for a limit test, and the text boxes

for LD50 and sigma are disabled.

The sequence of steps when using the program for a limit test is very similar

to the sequence for a main test described in the last section, with the

exception that the limit test uses, at most, 5 animals.

When 3 animals die during a limit test, the limit test should stop. When the

limit test is being performed at 2000 mg/kg, version 1.0 of the program may

not identify this circumstance correctly. Testing should be stopped and the

conduct of a main test should be considered whenever a third death occurs.

7. - Options Menu Details

Clicking on [Options] in the task bar allows the user to access three separate

sets of options for running the program.

The first set of options, [File Saving] is shown in Figure 9.

15
 Figure 9: File Saving Options Menu

When the [Sequential Data File Numbering] option is checked, the program will

automatically assign a sequential number to each saved file. If data are

loaded from a file and the file name ends in "_###.dat" (where ### stand for a

number between 1 and 999), the default name when saving the file will have the

same name with the number increased by 1. If the data have not been saved

previously, the default file name will be "Work_001.dat."

To set the sequential numbering option, click on [Options] in the task bar,

then drag the pointer first down to [File saving], and then to [Sequential

Data File Numbering.] When you have selected this option, a checkmark will

appear next to [Sequential Data File Numbering] in the menu. Selecting

[Sequential Data File Numbering] again will de-select the option and the check

mark will disappear.

For example, if the file named "Data_001.dat" is loaded, the default (or

prompted) file name when saving the next time will be Data_002.dat. Each time

the file is saved the number will increase by one. For sequential numbering

to work, the file name must include an underscore character (_) followed by a

number and have the extension .dat. Thus, the number in the file name

"Data001.dat" would not be changed because there is no underscore character.

If you are about to save your data using a file name that already exists, the

following pop-up window will appear:

Clicking Yes causes the program to overwrite (or replace) the existing file.

If the [Overwrite File Backup] menu option is checked, the program will make a

copy of the existing file before replacing it. The name of the copy will be

"Backup of" followed by the name of the file.

For example, let's suppose you have loaded test data from a file called

"testdata.dat." You then record information from another animal and decide to

save the file under the same name, overwriting the previous version of the

file. If you have checked the [Overwrite File Backup] option, the program

will automatically save the previous version of the file under the name

"Backup of testdata.dat."

The second option in the [Options] menu is the [Smallest Usable Dose]. This

permits you to inform the program of practical limitations to your

16
experimental dosing. When you click on the [Smallest Usable Dose] menu

option, you see the box shown below in Figure 10:

Figure 10: Smallest Usable Dose dialog box

Enter the smallest usable dose (mg/kg) for the test substance and your

laboratory. Before using this option, see the discussion of the Smallest

Usable Dose in Section 10.

The third option in the [Options] menu allows you to select the confidence

intervals to be calculated. Figure 11 shows the menu box for selecting the

confidence intervals:

Figure 11: Confidence interval menu options

The 95% confidence interval will always be calculated, and the "95%" option

will always be checked in the menu. In addition to the "95%" confidence

interval, you may choose to calculate the "90%" and/or the "99%" confidence

intervals. To do this, you would click on [Options] in the task bar, then

drag the pointer down to [Confidence Intervals], and then click on the

additional confidence interval you wish to create. When you have selected an

additional confidence interval, a checkmark will appear next to the selected

confidence interval in the menu box, and the pull-down menu will close. To

select all three intervals, the menu must be opened twice. Clicking on a

selected confidence interval will de-select the confidence interval, and the

check mark will disappear.

17
8. - Recovering from Mistakes

The program creates two files that provide safeguards against data loss due to

mistakes and accidents. Both files are saved in the application directory,

that is, the directory in which the program is stored (usually "C:\Program

Files\AOT425StatPgm)."

If the power fails or the program otherwise quits abnormally, you will not be

able to save your data. However, the next time you start the program, the

program may be able to recover data that was unsaved at the time the program

quit. If so, the program will load the unsaved data. When the program is

running, any unsaved data are saved in the file "Recovered Data.dat" about

once every minute. This file is deleted when the program exits normally.

If you create a new file, or load and edit an existing file, and answer "No"

when asked if you want to save the data, the data are automatically saved to

the "Last Unsaved Data.dat" file. This file only saves data from the last

time the data were not saved. Older unsaved data will be lost. If you decide

you do want the data, find the "Last Unsaved Data.dat" file in the application

directory, copy it to your chosen data directory, and rename the file.

9. - Warning Messages

For a main test, the program's recommended dose and stopping rules are based

on the short-term outcome. However, the OECD TG 425 procedure has some

special cases where the dose sequence and stopping rules may be based on the

long-term outcome or other considerations. The special cases occur when an

animal unexpectedly dies late in the study (see portions of Guideline

paragraphs 23, 28, and 35). The program provides warnings that the

combination of the dose and short-term outcomes appears to be inconsistent

with the OECD TG 425 procedure. Possible situations when warnings occur

include the following:

In special cases where dosing is paused to observe the previously dosed

animals, the toxicologist may make decisions that are not consistent with
the program's automated recommendations based on the short-term outcomes
but that conform with good practice;
Special circumstances may arise requiring an informed decision by the
toxicologist and that decision may not be consistent with the OECD TG 425
procedure or may cause the final data to diverge from OECD TG 425
procedure; or
Errors may occur, such as errors in dosing or errors in recording the doses
and outcomes.

If the program presents a warning, we recommend that you review the data for

accuracy and review the program settings for suitability. If there are no

data entry problems, it is important to document the doses used and be sure

your laboratory records are clear.

There is also one situation with late deaths that may not present a warning,

but the user should recognize the procedure as erroneous and the estimate may

be incorrect. If dosing proceeds to include three animals at the limit dose,

but no deaths occur until after the first of these 3 is dosed, it may be that

no warning will appear. Unless the estimate is below the limit dose, it

should be disregarded.

18
10. - Smallest Usable Dose

The OECD 425 program allows the user to specify a smallest usable dose. The

smallest usable dose is the smallest dose of the test substance that the lab

can or will administer for the study. NOTE: The OECD TG 425 procedure does

not discuss the smallest usable dose. To strictly follow the OECD Guideline,

do not use this option.

The user can specify the smallest usable dose by selecting the [Smallest

Usable Dose] menu option on the [Options] menu in the Data Edit window.

If the smallest usable dose has not been specified and the user enters 1) a

dose below 2.0 mg/kg, or 2) a sequence of two doses below 5.0 mg/kg, a window

will pop up asking if there is a smallest usable dose for this test.

Answering "Yes" takes the user to the window to define the smallest usable

dose. The program will only ask once. If you answer "No" to the question by

mistake, go to the Options menu to define the smallest usable dose.

When the smallest usable dose is defined, then:

The recommended dose for the next test is never lower than the smallest
usable dose.
If there are three consecutive deaths at the smallest usable dose, the
program will ask if testing should stop.

If the answer is "No," a similar message will appear after four animals

have died at the smallest usable dose, then five, etc. until the user

answers "Yes" or another stopping criterion is met.

If at least one dose is below or close to the smallest usable dose, the
following warning will appear in the report:

At least one dose is close to the smallest usable dose. A smallest usable

dose is specified for this test. The OECD 425 procedure does not define or

discuss how to use the smallest usable dose.

11. - Preparing and Reading Data Files and Importing Data from Other Programs

The AOT425StatPgm program saves the data in a text file. The default file

extension for the data file is .dat (see Section 6). Any file extension can

be used as long as the file is a text file (i.e., in ASCII text formatting).

The format of the data in the data file is described below.

19
The Title/Substance, Test Type, Limit Dose, assumed LD50 and assumed sigma

values are each entered on separate lines. Each line begins with several

characters that describe what is on that line. The string of characters that

define what is on a line is called the "tag" for the line. The header

information can be in any order. When the file is loaded, header information

that is not defined within the data file is set to the default value in the

program when the file is loaded.

Following the header lines, the animal testing data (Animal ID, Dose, Short-

term Outcome, and Long-term Outcome) are stored as comma-delimited text, with

one line for each animal. That is, each animal's identification number, dose

level, short-term outcome, and long-term outcome are on a single line

separated by commas. Lines with animal testing data must contain at least one

comma and not start with one of the tags.

The tags that are defined are listed below. All tags end with the colon

character.

TITLE: for the Title/Substance description

TYPE: for the Test Type (either "Main" or "Limit")

LIMIT: for the Limit Dose (either "2000" or "5000")

ALD50: for the assumed LD50

SIGMA: for the assumed sigma

LOWDOSE: for the smallest usable dose (if it has been defined)

LOWSTOP: for the answer to whether to stop after N successive deaths at the

smallest usable dose, N or Y and number of successive doses, 3 or higher

integer. For example: LOWSTOP: N, 3

DEBUG: Sets a flag to provide the Log(LD50), Log of the 95% confidence bounds,

Slope, and maximum log likelihood in the report.

If the animal test data can be exported from another program as a comma

delimited ASCII text file, that file can be loaded into the AOT425StatPgm.

The user can then add the header information and save the complete data file

ready for AOT425StatPgm use. Alternatively, the header and animal test data

can be imported if the text file has the format used by the AOT425StatPgm

program.

The following is an example of a data file.

TITLE: Annex 2 Table 5 example

TYPE: Main

LIMIT: 5000

SIGMA: 0.5

ALD50: Default

1,175,O,O

2,550,O,O

3,1750,X,X

4,550,O,O

5,1750,X,X

6,550,O,O

7,1750,O,O

8,5000,X,X

9,1750,X,X

20
12. - Technical Details

This section provides details regarding the correspondence between the OECD TG

425 Guidelines and the calculations in the AOT415StatPgm program.

Technical Details on the Implementation of the OECD TG 425 Guidelines

The following paragraphs describe how the OECD TG 425 procedures are

implemented in the AOT425StatPgm program. The numbered paragraphs below

describe how the program implements the procedures in the corresponding

paragraph in the OECD TG 425 Guidelines. The unnumbered paragraphs following

the numbered paragraphs describe how the program handles situations not

explicitly described in the OECD TG 425 Guidelines.

The OECD TG 425 Guidelines are not always explicit about 1) when and if to

drop doses from the dose progression, 2) how to handle actual doses that are

not equal from the recommended dose, and 3) whether the stopping criteria are

based on the short- or long-term outcomes. In addition, the program 1)

provides warnings when the data are not consistent with the OECD TG 425

Guidelines and 2) allows specification of a Smallest Usable Dose. The

implementation of these items is described below.

7. For the main test, the recommended starting dose is 175 mg/kg if the

LD50 is set to "Default." The default sigma is 0.5, corresponding to

half-log units between recommended doses. Sigma is used to calculate

the recommended dose progression.

13. For the main test, the "48-hour survival" pattern used to determine

whether and how to dose the next animal is the "Short-term Outcome" in

the data grid. The "Status of the animals at termination" is the "Long-

term Outcome" in the data grid. Once a stopping criterion has been met

and the long-term outcomes for all animals are entered into the data

grid, the program will calculate the LD50. The LD50 results are based

on the long-term outcome. The LD50 is calculated using maximum

likelihood except as specified in paragraph 42(b).

The program calculates the LD50 if all long-term outcomes are entered,

even if no stopping criterion is met. If no stopping criterion is met,

the LD50 is labeled as preliminary. The LD50 estimate is displayed in

the message box on the Data Edit Window if data (including long-term

outcomes) for four or more animals are entered in the data grid.

The program assumes that the animals are observed for a total of 14 days

to obtain the long-term outcome (see paragraph 27).

23. For a limit test at 2000 mg/kg, the program recommends conducting the

main test if the short- or long-term outcome for the first animal is "X"

(a death). The LD50 assessment is based on the long-term outcome.

If the first animal dies in the short- or long-term and data are entered

for additional animals, the program recommends conducting the main test

regardless of the outcomes for the additional animals.

The program recommends starting the main test using a dose of 175 mg/kg

(i.e., setting LD50 = "Default"). This dose recommendation is based on

paragraph 7.

21
24. If the pattern in the long-term outcome is one of those in paragraph 24,

and all animals are tested at 2000 mg/kg, the program says "The LD50 is

less than 2000 mg/kg" and recommends conducting the main test.

If the recorded doses are not exactly equal to the 2000 mg/kg, the

program says the LD50 is less than the maximum of the recorded doses.

The program recommends starting the main test using a dose of 175 mg/kg

(i.e., setting LD50 = "Default"). This dose recommendation is based on

paragraph 7.

If the first animal survives and the next three animals die, version 1.0

of the program may erroneously recommend continued testing. If this

occurs, stop dosing. The LD50 is less than 2000 mg/mg (but an under-

estimated value) in this instance and it is recommended that the

starting dose of the main test be 175 mg/kg.

25. If the pattern in the long-term outcome is one of those in paragraph 25,

and all animals are tested at 2000 mg/kg, the program says "The LD50 is

greater than 2000 mg/kg."

If the recorded doses are not exactly equal to 2000 mg/kg, the program

says the LD50 is greater than the minimum of the recorded doses.

In some instances where deaths are delayed, a main test may achieve a

sequence of 3 consecutive animals at the limit dose. In these cases, if

all animals show short-term survival, even after long-term outcomes are

entered the program may erroneously say The LD50 is greater than 2000

mg/kg. This result is incorrect.

27. The program recommendations are based on the long-term outcomes if they

are entered; otherwise, they are based on the short-term outcomes. For

a limit test at 5000 mg/kg, the program recommends conducting the main

test if the long-term outcome for the first animal is "X" (a death). If

the long-term (or short-term) outcome for the first animal is "O"

(survival) and two more animals are tested and both have long-term (or

short-term) outcomes of "O," the program recommends "Stop dosing

animals." If all three animals survive in the long-term, the program

says "The LD50 is greater than 5000 mg/kg."

If the first animal dies in the short- or long-term observation and data

are entered for additional animals, the program recommends conducting

the main test regardless of the outcomes for the additional animals.

The program recommends starting the main test using a dose of 175 mg/kg

(i.e., setting LD50 = "Default"). This dose recommendation is based on

paragraph 7.

If the second or third animal dies after the short-term observation and

before the long-term observation, the recommendation "Stop dosing

animals" changes to "Continue testing" until either three animals

survive or three animals die in the long-term. See paragraph 28.

28. Decisions about whether to dose another animal are based on the long-

term outcomes, if they are entered, otherwise on the short-term outcomes

22
 (evaluated separately for each animal). Thus, if either of the second

or third animal dies in either the short- or long-term observation, the

program will recommend testing an additional animal until either three

animals have survived or three have died in the long-term observation.

As the short- and long-term data are obtained and entered, a

recommendation to "Stop dosing animals" may change to a recommendation

to dose another animal. Dosing continues until either three animals

have survived or three have died in the long term.

29. If the pattern in the long-term outcome is one of those in paragraph 29,

and all animals are tested at 5000 mg/kg, the program says "The LD50 is

less than 5000 mg/kg" and "A main test may be needed."

If the recorded doses are not exactly equal to the 5000 mg/kg, the

program says the LD50 is less than the maximum of the recorded doses.

30. If the pattern in the long-term outcome is one of those in paragraph 30,

and all animals are tested at 5000 mg/kg, the program says "The LD50 is

greater than 5000 mg/kg."

If the recorded doses are not exactly equal to 5000 mg/kg, the program

says the LD50 is greater than the minimum of the recorded doses.

In some instances where deaths are delayed, a main test may achieve a

sequence of 3 consecutive animals at the limit dose. In these cases, if

all animals show short-term survival, even after long-term outcomes are

entered the program may erroneously say The LD50 is greater than 5000

mg/kg. This result is incorrect.

31. For a main test, the program's calculations are not affected if the time

interval for assessing the short-term outcome is different for different

animals. Assessment (and entry into the program) of the short-term

outcome should be delayed until one is confident of the survival of the

previously dosed animal.

32. The assumed LD50 is the toxicologist's best estimate of the LD50. The

recommended dose for the first animal is one step below the assumed

LD50, where a step is a factor of the antilog of the assumed sigma, and

sigma is the reciprocal of the estimated slope of the dose-response

curve. The default sigma is 0.5, giving a default dose progression

factor of about 3.2. If no estimate of the substance's lethality is

available, set the assumed LD50 to "Default" to get a recommended dose

for the first animal of 175 mg/kg. Using the default values, the

program uses the dose progression shown in paragraph 32.

For the corresponding value of sigma, the program uses the dose

progressions shown in paragraph 32 and Annex 2. To get the dose

progressions in paragraph 32 and Annex 2, doses close to the limit dose

may be dropped from the progression. The program drops from the dose

progression the dose below the limit dose if log10(limit dose/dose) <

Min(0.15, sigma/2).

33. Dosing continues until one of the three criteria in paragraph 33 is met

or until the fifteenth animal has been dosed. The calculations for the

stopping criteria are based on the short-term outcomes. For criteria

23
 (b) and (c), a reversal occurs if any of the following are true: 1) a

death is followed by a survival at a lower dose, 2) a survival is

followed by a death at a higher dose, and 3) a death is followed by a

survival at a dose greater than or equal to the limit dose.

For evaluating the stopping criterion in paragraph 33(a), a dose is

considered to be equal to the limit dose if the dose is closer to the

limit dose than to other doses in the dose sequence.

Criterion 33(c) is implemented following the discussion and examples in

Annex 3.

34. When any one of the stopping criteria have been meet and the long-term

outcomes (outcomes at test termination) have been entered, the program

calculates the LD50.

If the long-term outcomes are entered for all animals, the program will

calculate the LD50. If the stopping criteria have not been met based on

the short-term outcomes, the LD50 estimate will be labeled as

preliminary. The LD50 estimate is displayed in the message box on the

Data Edit window if data (including long-term outcomes) for four or more

animals are entered in the data grid.

40. For each dose, the AOT425StatPgm report shows the number of animals used

and the number showing signs of toxicity.

41. The LD50 is calculated using maximum likelihood, when possible, except

in the exceptional cases described in paragraph 42. The statistical

calculations find the combination of LD50 and slope that maximizes the

likelihood.

If the data do not have a positive dose-response relationship and

paragraph 42 does not apply, the program says, "The data do not show a

positive dose response relationship." The OECD TG 425 procedure can

result in data with zero or negative dose response in rare circumstances

(such as 1) starting at a very low dose with a small sigma; if all

animals survive, dosing may not stop before reaching the limit dose, or

2) testing of doses near the limit dose when stopping criteria 33(c) is

met and the data show zero dose response).

If deaths occur late in the observation period, the program may not

recognize that the dose-response relationship is not positive. In such

cases, which can include progression to a limit dose, the toxicologist

is cautioned to review for plausibility any program statement regarding

the LD50 estimate.

42. The procedures used in special cases are listed below.

When more than one subparagraph within paragraph 42 applies, it is not

clear which to use. The program assumes that paragraph 42(c) applies to

tests where the limit dose is the only dose with partial response,

regardless of the stopping criteria. Paragraph 42(a) applies in other

cases where the stopping criterion is paragraph 33(a).

42(a) If 1) testing stopped based on the criterion in paragraph 33(a) or after

dosing 15 animals, 2) the data do not show a positive dose-response

24
 relationship, and 3) the upper bound dose (limit dose) ended testing,

then the LD50 is reported to be above the limit dose.

Final dose is considered to be equal to the limit dose if the final dose

is closer to the limit dose than to the next lower dose in the dose

progression.

If the data show a positive dose-response relationship, the LD50 is

estimated using paragraph 42(c) or using maximum likelihood.

42(b) If all the dead animals have a lower dose than all surviving animals,

the reported confidence interval for the LD50 is the range from the

highest dose for all surviving animals to the lowest dose for all dead

animals. The LD50 estimate is the value of the LD50 that maximizes the

likelihood when the slope is fixed at 1/assumed sigma.

The confidence interval in this case is not a profile likelihood

confidence interval and is labeled as "Approximate." The same

confidence interval is reported regardless of the confidence level.

The LD50 estimate obtained by maximizing the likelihood with fixed slope

may not be within the reported confidence interval. In this case, the

program provides the message: "Warning: The 95% confidence interval does

not include the estimated LD50. The results should be used with caution.

The assumptions used to estimate the LD50 and/or its confidence interval

may not be appropriate for the data."

42(c) If the surviving and dead animals have only one dose in common and all

other dead animals (if any) have a higher dose and all other live

animals (if any) have lower doses, the reported LD50 is the common

dose.

43. When maximum likelihood can be used to calculate the LD50, the

AOT425StatPgm program calculates the LD50 using maximum likelihood. Use

of other programs is not needed.

44. Stopping rule 33(c) is calculated by the program. Separate calculation,

possibly using a spreadsheet program, is not required. If this stopping

criterion is met, the LD50 is calculated as described in paragraphs 41

and 42.

If this criterion is met, the program may not be able to calculate the

LD50 by maximum likelihood (see paragraphs 41 and 42).

45. The AOT425StatPgm program calculates confidence intervals for point

estimates of the LD50. Except for data covered by paragraph 42(b),

confidence intervals are calculated using profile likelihood.

47. The upper bound on the profile likelihood confidence interval may be

infinite, meaning that an upper bound does not exist. Upper confidence

bounds greater than 20,000 mg/kg or infinite bounds are reported as

"Greater than 20,000." The lower bound can be very close to zero.

Lower confidence bounds that are below 0.0001 are reported as zero.

48. The AOT425StatPgm program provides information, particularly the LD50

estimate, that can be incorporated into the test report.

25
 The program does not routinely provide the maximum likelihood slope

estimate. Any slope estimate obtained from the OECD TG 425 procedure is

not reliable.

Implementation of Warnings

The program provides warnings if the reported doses and outcomes are not

consistent with the recommended doses. For calculating the warnings, for a

main test a dose is considered equal to the recommended dose if the dose is

closer to the recommended dose than to other doses in the dose progression.

For the limit test, a dose is considered equal to the limit dose if it is

within a factor of 3.2 of the limit dose.

Implementation of the Smallest Usable Dose

The program allows the use of a smallest usable dose. The implementation of

the smallest usable dose is discussed in Section 9. If the smallest usable

dose has been defined, a dose is considered to be equal to the smallest usable

dose if that dose is closer to the smallest usable dose than other doses in

the dose sequence.

In paragraph 33, a reversal occurs if a survival is followed by a death at a

dose less than or equal to the smallest usable dose. In paragraph 42(a), if

1) testing stops based on three or more consecutive deaths at the smallest

usable dose and the user says to stop or at the fifteenth animal, 2) the data

do not show a positive dose response relationship, and 3) the smallest usable

dose ended testing, then the LD50 is reported to be less than the smallest

usable dose.

Program Constants and Limits

In general, the program will read in any ASCII test file, regardless of

whether the contents of the file are valid for performing the calculations.

Limits on data entry include:

Any printable characters can be entered into the Test/substance, LD50,

and sigma boxes.

Any printable characters except the following characters can be entered

from the keyboard into the data grid: , ~ % & ( | { and }.

Any characters except a comma can be read from a data file into the data

grid.

When reading from a data file, any text for the test type that, when

converted to upper case is not 'MAIN' or 'LIMIT' will be ignored and the

default 'Main' will be used.

When reading from a data file, any text for the limit dose other than

'2000' or '5000' will be ignored and the default '2000' will be used.

Although there is no formal limit on the number of characters in the

Test/substance description, we recommend using fewer than 400 characters.

Very long descriptions have produced problems on some computer-printer

combinations.

26
If the data read from a file is not consistent with the file format for a data

file, extra information is ignored. The program warns the user if the data

files have data for more than 15 animals for a main test or 5 animals for a

limit test.

The data read from the data file or entered from the keyboard are displayed in

the data edit window. If the data are not valid for the calculations, the

program will present messages in the right column or the lower message box.

The user can edit the data in the Data Edit window.

The following defines valid data for the calculations:

The number of characters in the test/substance description should be

less than 32,000.

An animal ID can have up to 40 characters.

A dose can be any number from 1e-308 to 1e308.

A short and long-term outcome must be lower or upper case "O" ("Oh" not

zero) and "X." These values will be converted to upper case by the

program.

The LD50 must be a number between 0.0000000001 and 100000 or the word

"Default."

Sigma must be between .01 and 4.0.

Unless otherwise specified by the user, the calculations use 1e-300 as the

smallest usable dose.

When using sequential file numbering, the number must be between 000 and 999

inclusive.

The program will always provide the 95% confidence interval, if it can be

calculated. The user can optionally select a 90%, 99%, or both a 90% and 99%

confidence interval. Other confidence levels cannot be specified.

The likelihood and profile likelihood calculations use a Newton-Raphson

algorithm. For calculating the LD50, the model uses slope and intercept

parameters. For calculating the profile likelihood, the model uses the LD50

and slope parameters. The starting values for the parameters cannot be

specified by the user. When calculating the LD50 with a fixed slope, the

starting value for the LD50 is Log10(dose for the last animal)/sigma. For

all other cases, the starting value is zero.

When fitting a model, the likelihood is assumed to be converged if the

relative change in the log likelihood is less than 1e-8 from one iteration to

the next. After the likelihood has converged, iterations continue until the

change in slope on the last iteration is less than half the change in the

previous iteration or less than 0.1% of the maximum slope estimate. In no

case are more than 40 iterations performed to obtain the parameter estimates.

If the likelihood does not converge in 40 iterations, a warning message is

presented. The algorithm uses step halving to minimize the likelihood in rare

cases where the Newton-Raphson step does not reduce the likelihood. In no

27
case are more than 40 step halvings performed. In practice, the algorithm

generally converges quickly without requiring step halving. As part of the

algorithm, the program converts Z values from a standard normal distribution

to probability (P) values. The function used accepts Z values from 37 to 37.

P for values below 37 is zero. If a Z value is less than 37, it is replaced

by 37. Similarly, if the Z value is greater than 37, it is replaced by 37.

In effect this minimizes the influence of outliers, avoids taking the log of

zero, and allows convergence in rare cases.

The program calculates some simple statistics to determine if the profile

likelihood confidence bounds are infinite or finite. For finite confidence

bounds, the program steps away from the LD50 estimate, calculating the profile

likelihood, until a bounded range is obtained that encloses the confidence

bound. In the log dose scale, when stepping away from the LD50, the first

step has a length equal to the range of the data. Each successive step is

bigger than the last step by a factor of 2. If the bounded region around the

confidence bound is not found after 40 such steps, the bound is reported as

infinite. The confidence bound is located by successive bisection (on the log

dose scale) of the bounded region. Up to 40 bisections are performed. The

confidence bound converges if the relative difference between the critical

likelihood for defining the confidence interval and the estimate is less than

1e-8 or if the length of the bounded interval is less than 0.0001 on the log10

scale.

The constants determining the convergence of the likelihood and the confidence

bounds cannot be changed by the user.

The recommended dose is rounded to somewhat more than two significant figures

unless the doses are closely spaced, in which case more significant figures

are used.

Internally the LD50 and the confidence bounds are stored as double precision

floating point numbers. When these are presented in the report, values below

0.0001 are presented as zero, values above 20,000 are presented as "Greater

than 20,000." Values are presented to at least three significant figures;

however, trailing zeroes beyond the decimal are dropped.

For a selected data file, more exact estimates of the LD50 and confidence

intervals can be obtained by editing from the data file and adding a line

starting with "DEBUG:. Using this file, the log of the LD50 and confidence

bounds will be shown at the bottom of the report. Based on comparison of the

AOT425StatPgm output to other programs, the LD50 estimates from the debug

information should be good to at least six significant figures and the profile

likelihood confidence interval bounds should be good to at least four

significant figures.

The AOT425StatPgm program fits the probit model and gets profile likelihood

confidence intervals for almost any data, regardless of whether the data

follow the OECD TG 425 procedure. It may be necessary to edit the data file

by adding a line with "DEBUG:" to get the parameter estimates.

28