Today, this series of articles closes with a description of the Guidance on the assessment of the efficacy adopted in April 2018 following a public hearing.
A – still too short – summary of the almost 30-page manuscript would exceed the space available here. That is why we are concentrating below on the structure of the Guideline and the fundamental new points by comparison with the Guideline which applied previously.
The Guideline is sub-divided into three major chapters detailing the requirements made of (1.) the nature of the evidence of efficacy for the individual additive categories and function groups, (2.) the number of in-vivo trials, resulting from the animal species/animal category and their number, in which the additive is to be used, and (3.) the performance of in-vivo efficacy studies with statistical considerations, followed by three relatively short chapters (4.) on in-vitro studies, (5.) a template for structuring the reports on the studies, and finally (6.) on the performance of literature studies.
The evidence of efficacy can basically also be provided through references from literature, provided that the additive described in the publication is (largely) identical with the additive for which the application is being made.
By comparison with the Guideline that has been valid so far, the list of additives for which no documentary evidence of efficacy in the target animal is necessary is now longer. It covers most of the nutritional additives (all amino acids, the generally recognised vitamins and trace elements and urea). It also includes authorised food additives (flavouring substances, colouring substances) if it can be assumed that their effect is the same in foods and feeds.
The efficacy of additives which influence the characteristic of feedstuffs (for example anti-caking agents, bonding agents) can be demonstrated by in-vitro studies. This concerns most technological additives and those colouring substances that add colour to the feed. For additives that are to take effect on or in the animal, the evidence of efficacy must generally be provided by in-vivo studies. Exceptions to this rule are the mycotoxin binders, for which a combination of in-vitro and in-vivo studies can be accepted.
Short-term studies are recognised as evidence of efficacy for additives that lend colour to foods originating from the animal and most enzymes (in the functional group of digestibility promoters). Digestibility studies are only considered to be evidence of efficacy if certain experimental parameters (such as for example P-retention for phytases, convertible energy of the feed for polysaccharidases, ileal digestibility of amino acids for proteases) are investigated. Such precise definitions are not to be understood as a schoolmasterly approach, but instead as an aid for the applicant.
The transfer of the evidence of efficacy from one animal category to another is regulated more clearly. The list of animal categories to which it is transferred automatically has been broadened, but is still relatively small by comparison with the wishes of the applicants. This can be illustrated by an example – when an application is made to use a zootechnical additive for all terrestrial animal species (poultry, pigs, cattle, rabbits, horses, minks), up to now at least 24 efficacy studies have been necessary. Now the number is 18 (and perhaps only eight according to the ideas of the applicant).
A central question remaining was how many studies (showing a significant improvement of a parameter due to the additive) are necessary if the application relates to all animal species or to a certain selection of animal species. The principle that it is only possible to conclude that an additive has the capacity (potential) to be effective if at least three studies (with statistically significant effects) on an animal species/category are presented has remained unchanged. Here it was necessary to find a compromise between scientific scepticism about extrapolations on the one hand and the justified interests of the industry (and to a certain extent of animal welfare). It is fundamentally possible to extrapolate within a type of production (growth or (re)production) and animal species. For example, if the efficacy has been demonstrated by three studies with broiler chicks, the efficacy is automatically assumed for growth in laying chicks, turkeys, geese, ducks, quails, pheasants, guinea fowl, partridges and ostriches, as well as for ornamental birds. However, if on the other hand the application is made for use of an additive in poultry altogether (in other words also for layers), three studies are required for growing (broiler chicks) and three studies for layers (layer hens). This principle is also applied for cattle and swine. For additives to be used in all fish, three studies on salmonids must be presented and one each on three other kinds of fish.
The authors
The authors – Gerhard Flachowsky and Jürgen Gropp – have for a long time been members of the EFSA Panel on Additives, Products and Substances used in Animal Feed (FEEDAP). They report on the work of the FEEDAP Panel. The articles contain personal representations and views, not those of EFSA or the EU, and therefore are not necessarily the same as the opinions of EFSA or the EU. As experts working for EFSA, the authors are subject to certain secrecy obligations.

The number and scope of necessary studies with coccidiostats has also been reduced (from at least seven with three field trials to six without field trials), at the same time with an increase in the demands made of precision and informative value of the individual studies.
The recommendations on the experimental design are based on the tried and tested principle of comparison of a control group with a group receiving the additive (in the lowest recommended dose). Fundamentally no distinction is made here between administering with feed or administering with water. It is evident from the sub-chapter on Statistics that a statistician should be involved already during the planning of the trials and thus before commencement of the trials. As additives (with the exception of coccidiostats) are to be effective in healthy animals (REG 1831/2003), the status “healthy” required a supplementary definition. The mortality rate should not exceed the standards usual in today’s animal husbandry in Europe. Trials in which the trial animals were given preventive treatment with antibiotics or other antimicrobial substances will hardly be recognised. If treatment of animals during the trial becomes necessary for reasons of animal welfare or for economic reasons, care must be taken to avoid interaction between the therapy (medication) and the additive.
Meta-analysis, a statistical method of joint evaluation for so far at least four studies, of which in the worst case no individual one could significantly document efficacy statistically, but which displayed minimal numerical differences as significant thanks to a high number of cases, is now only evaluated as one study and accordingly can no longer stand for all three studies.
The FEEDAP Panel expects that thanks to better studies, the newly worded Guidance, resulting from the experience of more than ten years of assessing applications, will help to simplify applications and shorten the time required by FEEDAP for processing them, which will render queries and reworking by the applicant as well as additional trials unnecessary.