I believe good animal drug tests are already double-blind, at least if they rely at all on observation of the animals' behaviour to determine the outcome. Otherwise, the people monitoring outcomes will be affected in their interpretation of the animals' behavioural signals by their knowledge of which ones have received the drug.