Swine Respiratory Disease: Statistical Considerations - PowerPoint PPT Presentation

1 / 15
About This Presentation
Title:

Swine Respiratory Disease: Statistical Considerations

Description:

Multiple Pigs Allocated to Each Pen. Two Main Options for Allocating Pigs to Experimental Groups and Pens ... 320 pigs were enrolled in the study (80/site) ... – PowerPoint PPT presentation

Number of Views:107
Avg rating:3.0/5.0
Slides: 16
Provided by: lindato
Category:

less

Transcript and Presenter's Notes

Title: Swine Respiratory Disease: Statistical Considerations


1
Swine Respiratory DiseaseStatistical
Considerations
  • Janice Derr, Ph.D.
  • Center for Veterinary Medicine
  • U.S. Food and Drug Administration

2
Summary
  • Consider the statistical analysis at the study
    design stage
  • Make sure the study has adequate statistical
    power
  • Make sure the study design will result in a valid
    statistical analysis

3
Statistical Aspects of Designs for Field Studies
  • Multiple Sites
  • Multiple Pens at Each Site
  • Multiple Pigs Allocated to Each Pen
  • Two Main Options for Allocating Pigs to
    Experimental Groups and Pens (next slide)
  • Evaluation is in the Form Success/Failure

4
Options for Allocating Pigs to Experimental
Groups and Pens
  • Injectable article
  • Pigs from different experimental groups can be
    commingled in each pen
  • Article is in feed/water
  • Pigs from different experimental groups need to
    be in separate pens

5
The Experimental Unit
  • The smallest unit to which a treatment is
    randomized
  • Injectable article The pig can be the
    experimental unit
  • Article in feed or water The pen is the
    experimental unit
  • The pen is an important design factor due to
    dependence in response within the pen

6
Allocating Pigs to Pens Statistical
Considerations
  • Commingled experimental groups An equal number
    of pigs per experimental group per pen is the
    most efficient design.
  • Separate pens per experimental group An equal
    number of pigs per pen is the most efficient
    design.
  • Both designs Allow for losses during the study
    (pigs that will not be in the analysis).

7
Statisticians Can Provide Input to Questions
About Study Design
  • How Many Sites?
  • How Many Pens per Site?
  • How Many Pigs per Experimental Group per Site?
  • How Many Pigs per Pen?
  • (Other considerations also enter in to these
    decisions.)

8
Statisticians Need to Know
  • What is the variation in response among sites?
  • What is the variation in response among pens?
  • What is the level of dependence among responses
    from pigs in the same pen?
  • What is the minimum effect size of the test
    article compared with the negative control that
    is of interest?

9
A Hypothetical Study
  • Objective To evaluate the effectiveness of an
    injectable article in the treatment of Swine
    Respiratory Disease (SRD).
  • Study was conducted at four sites.
  • 320 pigs were enrolled in the study (80/site).
  • Each eligible pig was assigned to either test
    article or saline injection in a 11 ratio.
  • Pigs were commingled in pens, 4 pigs/experimental
    group/pen (8 pigs total/pen).
  • Several criteria were used to classify the animal
    as a Success or Failure after a set period of
    days.

10
Design Layout for Site 1
11
Hypothetical Results
12
Improvements in Statistical Methodology Impact
Field Studies in SRD
  • In the past, it was difficult to analyze
    success/failure responses from complex study
    designs.
  • At present, we have improved statistical
    methodology and software.
  • However, this methodology still has limitations.

13
Mixed Effects Analysis of Variance for
Success/Failure Responses
  • Correctly represents a binomial response (such as
    Success/Failure)
  • Allows for random effect of pen within site
  • Allows for random effect of site
  • Allows for dependence in response among animals
    in the same pen
  • Does not always run (with small numbers of
    animals)

14
Mixed Effects Analysis of Hypothetical Results
15
Summary
  • Consider the statistical analysis at the study
    design stage
  • Make sure the study has adequate statistical
    power
  • Make sure the study design will result in a valid
    statistical analysis
Write a Comment
User Comments (0)
About PowerShow.com