The number and diversity of additives available for animal feed has dramatically increased during the last decades, particularly since the ban of antibiotic and hormonal growth promoters. Evaluating the efficacy of feed additives is an art in itself.
Benzoni Gaëlle at InVivo Nutrition et Santé Animales, France, Baéza Elisabeth at INRA, France
The European livestock industry, and more particularly the poultry production sector are continuously looking at improving their techno-economic performance. In the meanwhile they are trying to take into account regulatory affairs and society requests such as animal welfare or the environment. The use of appropriate additives can generate added value for different players of the sector.
The feed additives market has evolved a lot since the 50s, with first the use of antibiotics as growth promoters, then enzymes were proposed in the 80-90s, followed by prebiotics, probiotics, acidifiers and then vegetal extracts in the following decades. This alternative feed additives range has boomed and diversified due to the rational use of antibiotics in the 70s (positive list), followed by the reduction of their use in the 90s and finally their ban in Europe in 2006. Nowadays, feed additives from various origins are on the market, with or without registration file, scientific evaluation, or guaranteed composition and with a wide price range.
Despite this, potential users, whether it be farmers, integrators or feed formulators, have to decide whether they want to use an additive or not and to choose it properly.Providing reliable information therefore enabling evaluation of the efficacy of the marketed feed additives is the goal.
Use of additives
Assessing feed additive efficacy requires precisely identifying factors that limit technical and economical performances of the farms and therefore being able to choose the most adequate product which is able to target these points. If the request is not precise enough, the choice of the additive can lead to a non-adapted product. For example, improving FCR is a general aim. It is important to know why the FCR is poor. If it is due to bad digestibility of raw materials, the choice will be orientated to the use of enzymes. If linked to bad absorption of nutrients, products acting on gut mucosa protection or gut flora regulation would be more suitable.
Moreover, according to users, the main problem regarding the use of additives is variability of the results. However this variability is not specific to additives currently on the market. Antibiotic additives are facing the same problem.
Could part of this variability be linked to the fact that additives are tested with too general expectations or in conditions that do not correspond to their mode of action? To correctly evaluate an additive, it is essential to know for which objective it is being tested for. This is obvious for additives whose nature and function are the same: thus we expect from a coccidiostat a control of coccidiosis. Users expect from an enzyme a better digestibility of some raw materials or nutrients. However, when talking about vegetal extracts, it refers only to the nature of the additive and does not give any indication on its function. Consequently, it is necessary to know the exact composition in vegetal active molecules to conclude on the function of the product and on the adapted context to test it to get optimum effects.
How to conduct a trail
Once the additive is chosen in relation to its function and its precise target, it is necessary to assess its efficacy in a suitable and objective trial. First and foremost a protocol has to be established in order to avoid false positive and false negative results. False negative results, or an under evaluation of the potential of the additive, can be obtained when the additive is used under conditions where the product cannot express its full potential, when the targeted problem is not present or on contrary when the problem's intensity is too high. For example, it is possible to get false negative results when an anticoccidian product is tested on animals that are not infected with coccidiosis or when an enzyme is used with a raw material that is not its substrate. Similarly, evaluation of a gut flora regulator, whether it is an essential oil, a live bacteria or another product from different origin, has to be used on animals with unbalanced gut flora. False negatives are often obtained when gut flora regulators are tested in optimal sanitary conditions. To illustrate, Benzoni and et al. (2010) have tested, in the same trial, a gut flora regulator based on activated copper additive (B-Safe) both on broilers in good sanitary conditions and on broilers with non-optimal conditions aiming at increasing sanitary conditions. Presented in Figure 1 are the results of different conditions in the study on efficacy of the products.
In this type of trial, the challenge had to fit the desired objective. For example, generate an unbalanced gut flora through a bacterial inoculation or through breeding stress will not allow evaluating the same thing. Moreover the intensity of the challenge on reduction of zootechnical performances had to remain realistic when compared to technical results obtained on the field. Evaluating efficacy of an additive in extreme conditions is likely to lead to false results. Good criteria to assess admissibility of a trial is the proximity of performance of control batch without any additive with the average performance obtained by farmers. If performances are too far from the average field, trial results have to be considered cautiously. The use of a positive control group, receiving a recognised efficient product for the concerned criterion, can also attest for the admissibility of the trial. If the reference product does not lead to expected results, trial conditions and especially intensity of the challenge can be questioned. A contemporary negative control group, raised in same conditions, except the addition of the product provide also a guarantee against false positive results or against over evaluation of the efficacy of product. Indeed, authors sometimes conclude on the efficacy of a product by comparing its effect with a reference product. However if there is no negative control group, it is not possible to affirm that the positive control group (reference product) has improved performance. Thus it is risky to assume that the product is efficient. Negative control group is thus an essential admissibility condition for a trial. Ideal situation is to have a negative control group raised at the same time and in the same house than trial group. This is quite easy in experimental farms where animals receiving the various treatments are raised in different pens in the same house.
On field, it is possible to isolate animals through the set-up of pens among the rest of the flock or though selection of cages for animals raised in cages. Those isolated animals will be raised in same conditions than the rest of the batch but can be fed with different feed. Thus it will be possible to conclude on the effect of the additive. When it is possible to have a contemporary negative control but only in another house, the trial will be conclusive only if it is repeated on two successive batches with an inversion of houses between the two batches.
A period effect is possible in this type of trial but can be reduced by repeating the trial on numerous batches. When it is impossible to get a contemporary negative control, it is still possible to compare, in the same period of the year, farms performances receiving the control feed and on farms receiving the feed including the additive. In this case, farms have to be as similar as possible (same genetic, type of house, average performance). The more the breeding conditions differ, the more farms are needed in the trial in order to compensate for the variability of the biological response and to be able to see significant differences among performances.
Recording performance of a control and a test group provides average performance that can seem to be different. However this means comparison cannot lead to a reliable conclusion on effects of an additive. As illustrated in Figure 2, variability of data has to be taken into account to conclude. Indeed, in figures 2a and 2b, the difference of average weight between diet A and B are the same but variability's are different. As a result, in figure 2a, diet B achieves significantly higher live weight than diet A, while in figure 2b the diet has no significant effect on broiler's live weight.
Variability is estimated through standard deviation, or through coefficient of variation (CV expressed as %, CV = 100 x standard deviation / mean). By definition, several data are necessary to evaluate population variability and a single data per group is not enough. For example, to evaluate population growth, the most reliable process is to individually weigh all animals because no bias is created by animal selection. When setting up a trial, the number of animal to include (N) can be calculated. It depends on the expected trial accuracy (F), on the population variability (CV) and on the least significant difference that we want to be able to detect (d, expressed as %).This number is calculated thanks to following formula: N = F x CV² / d². F is a constant that depends on α risk (risk to conclude that means are different when they are not) and on β risk (risk to conclude that means are not different when they are). Parameters currently used are: F=8, corresponding to α=0,05 and β=0,5. For more reliability, we can use F=21, corresponding to α=0,05 and β=0,1.
When there is a high number of animals, weighing in small groups represents a suitable alternative to individual weighing. Size of the group is calculated according to the total number of animals and to the number of data necessary originating from the above formula. When weighing of the whole flock is not possible, even in groups, individual weighing of a percentage of the animals can be performed (survey weighing). Number of animals to be weighed can also be calculated with the above formula. This method can be applied to all types of datas: weight, feed consumption, egg numbers, litter scores, pododermatitis lesions.
Using reverse of the previous formula enables to calculate the least significant difference that can be detected : d = CV x (F / N)1/2. Measurable difference is all the more smaller as the variability is lower and the number of measurements is higher. To limit the number of measurements while still being able to detect a small difference, the experimental design has to avoid effects that are not due to the tested additive. Moreover, the formula, including CV and number of observations, allows us to get an idea of the significance of the results, without processing a complex statistical analysis: if difference between average weight of control group and average weight of test group is 2% and that the calculated value for "d" is 5%, the probability of a significant effect of the additive on weight is low. Thus, it will not be possible to assess the efficacy of the product in this experimental design as the 2% difference is due to global variability and not to the additive itself.
Global criteria for assessment of additive efficacy
Information above enables us to establish conclusive feed additives tests or to assess the admissibility of a trial presented by a feed additive supplier. However the results of a trial gives information about the effect of an additive in a specific situation (1 genetic, 1 sanitary condition, 1 feed programme) and does not allow evaluating total capacity of the additive to alleviate a limitating breeding point in different contexts. It is then necessary to repeat trials under different conditions to get a more accurate evaluation of the product. Compilation of different trials data's can then be processed in a meta-analysis in order to evaluate the global effect of the additive in many different situations. This meta-analysis can be completed by a holo-analysis allowing modelling of additive's response. It proposes an equation about the potential improvement of performance allowed by the product (for example the % of growth improvement compared to control) according to the characteristics of the farm. On a more practical point of view, seven simple criteria points allow you to assess the efficacy of an additive (Rosen, 2004). First of all, the number of trials run on the product under different condition has to be at a minimum of 30. The number of trials that do not include a negative control has to be at maximum of five. Scientific comity revised publications on the additive is a positive point. Within all the trials presented, the additive has to improve significantly in at least seven out of 10 trials. Variation coefficient of the response has to be in the 100 to 200% range. Optimal dose to maximise additive effects on a chosen criterion (final weight, or FCR) has to be determined. Providing a model for additive response according to breeding conditions is also an advantage. Finally, calculation of return on investment is compulsory. This is generally expressed by n:1 ratio (n€ gain for 1€ invested).
References on request
Points for thorough evaluation
- An optimum use of an additive requires a precise diagnostic of the farm in order to identify the different points that limit performance and to choose the right mode of action, and the right additive to be remove them.
- Additives have to be tested in conditions where they can show their full effect. It can be in presence of coccidia for a coccidiostat or in poor sanitary conditions for a gut flora regulator. Specific poor conditions have to remain realistic and close to those found on commercial farms.
- A negative control group, preferentially contemporaneous is compulsory. A positive control group can support interpretation of results.
- Both for own trials or trials presented by a feed additive supplier, results have to be screened with statistical analysis or at minimum, results have to present average figures, standard deviations, and number of observations. This is compulsory to get a reliable conclusion on the efficacy of the product.
- Several trials have to be set up under different conditions. Indeed, intensity and homogeneity of the global answer will allow to conclude on the product efficacy.
- Techno-economic interest of the additive will be evaluated not only according to its efficacy but also according to its potential return on investment.
- Awareness that the combined use of several additives can lead to interactions (positive or negative). Finalise additive evaluation with some association tests.