Literature studies, choice of animal model and design of experiments

Systematic review is a structured, thorough and transparent way of doing a literature-search.

A systematic review is characterized by being

• Structured, Thorough, transparent
• address a specific research questions
• Transparent literature search and selection of papers
• Critical appraisal of papers

When you are planning an experiment it is common to do a literature search.

You must include more than 1 search component in your search. For biomedical research it is advised to include search for:

• Intervention: With focus on Welfare)-Impact of interventions and refinement of procedures
• Animal: With focus on basic biology, comparative aspects human vs animal and species-specific needs
• Disease or subject of interest: With focus on impact of the disease, prevalence of disease, pathogenesis and treatment of disease

Several relevant sources and/or databases are included in the search that is based on relevant keywords, including commonly used synonyms («MeSH terms»). The word “mouse” can have several synonyms that you have to include.  Similarly you find some for “rat” here.

Conclusions from several relevant studies are analyzed in a meta-analysis. This way evidence from several studies are used to confirm or reject theories for example between a drug and its effect on the body, disease-mechanisms et cetera that you don’t  easily get from a single study.  

Meta analyses are useful in planning animal experiments for example to evaluate if a model is really suitable to predict an outcome or relevant effects. If 50% of studies show an effect and the other 50% show an opposite effect, which studies shall we trust, or can we trust any of them? Based on this analysis you can make a more qualified evaluation whether or not it is reasonable to set up new animal studies or if it’s better to use other approaches to achieve more knowledge about a phenomenon.

Meta analyses are also useful and necessary in translational research from preclinical studies in animals to clinical studies in humans. Are conclusions from the animals studies so clear and reliable that they support continuous studies in patients, or do the result diverge in different directions? This can give useful information when you plan experiments both with regard to using the best model and to animal welfare.

Michael Festing has developed a teaching tool for planning of animal experiments including

• Proper design of animal experiments
• What factors influence the number of animals that are needed in a study
• Statistical methods
• How to save animal, time and resources by better planning of animal experiments

(recommended  web browsers explorer, safari)

There is also a lot of control questions and self tests included

Michael Festing is a toxicologist and has been working with problems related to design of animal experiments for several years and he is the author of the book “Design of Animal Experiments" (SAGE)

Read more about Michael Festing here.

Even when you base your group sizes in similar studies like:" We know from previous experiments that this study requires a minimum of n animals to obtain statistical power."

What are the relevant treatment effects, significance level and/or power in this study? Do you have an estimate of variation?

These numbers should be mentioned even if it's based on previous experience from similar studies.

Link to sample size calculator

Below you find some cartoon videos from YouTube illustrating some common misunderstandings and challenges with regards to design of animal experiments and evaluating the result.

• Biostatistics vs. Lab Research
• Power of the test, p-values, publication bias and statistical evidence
• What the p-value?