French health recording systems: possible new valorizations of events
recorded by farmers
X. Bourrigan1, S. Mattalia1, C. Bouissel2, R. Dremaux3, M. Dupres4 J.J. Evard5, X. Gouraud6,
M. Legay7, L. Maurin8, B. Schmitt9, J.M. Gautier1, P. Roussel1 & G. Blériot1
1
Institut de l’Elevage, 149 rue de Bercy, 75012 Paris, France
[email protected] (Corresponding Author)
2
GDS de Haute-Saône, 17 Quai Yves Barbier, 70000 Vesoul, France
3
SYNELIA, 3595 Route de Tournai, 59501 Douai, France
4
CA de Saône et Loire, 59, rue du 19 mars 1962, CS 70610, 71010 Mâcon, France
5
GDS du Lot, 430 avenue Jean Jaurès, 46004 Cahors, France
6
SNGTV, 5 rue Moufle, 75011 Paris, France
7
CRA de Normandie, 6 rue des Roquemonts, 14000 Caen, France
8
GDS de Bretagne, 13 rue du Sabot, BP 28, 22440 Ploufragan, France
9
GDS de Moselle, 64 avenue André Malraux, 57000 Metz, France
Abstract
A study of 7 health information systems aimed to analyze the consistency of events recorded
by farmers, in order to use them for genetic evaluations or for new herd management tools.
From 2007 to 2012, these software were used by 18 303 cattle farmers. The diseases recorded
are more or less detailed according to the tools. A classification is proposed using ICAR
recommendations and separating curative and preventive treatments. The herds recording
events regularly and with information on diversified diseases were selected, assuming that they
record information exhaustively. 15.6 % of herds met these requirements; they represented
56.3 % of the events. The results of annual and monthly prevalence, calculated on 16 types of
diseases and on 9 categories of animals are consistent with those previously reported in the
literature. For AI bulls with daughters present in at least 2 tools, the proportion of each disease
among those recorded per progeny was compared between tools. Ratios were quite
homogeneous, which is a good sign regarding the consistency of the records among tools. These
results show that harmonization of health records is possible. The health events could be used
to develop new genetic evaluations on health traits or new references used for benchmarking.
Keywords: health recording systems, diseases, classification, prevalence
Introduction
The development of genomics brings new perspectives for selection and herd
management. A reference population of few tens of thousands of animals genotyped and
phenotyped enables the development a genomic evaluations. Genetic trend is expected to
increase, thanks to the reduction in generation interval, particularly for low heritability traits
such as those related to animal health. In this context, phenotyping new traits becomes strategic
(Boichard et al., 2015). Many countries have taken this opportunity to implement the recording
of health events, either by veterinaries (Frandsen, 2013), or by farmers (Egger-Danner et al.,
2012, Koeck et al., 2015). Such data are used both for genetic evaluations and for herd
management, in order to encourage the collect information.
In France, farmers must record all the veterinary treatments (regulation of 5 June 2000).
They are also encouraged to record health events. Recording can be done by many forms: paper,
software, web application… IT solutions are proposed by collective organizations (milk
recording or animal health support organizations) located in several French areas, by
veterinarians groups or by independent suppliers.
In order to look for potential uses of such data in genomic evaluations, France Génétique
Elevage conducted a study of health events recorded in the information systems of collective
organizations. This paper presents an overview of the study and the prospects of this work.
Material and methods
The study of each tool was performed in 2 steps: i) a presentation of the general
characteristics of the information systems, ii) a description of the data recorded.
Questionnaire on recording of health data
8 collective organizations described the use of their software and the data recorded in
their information system (number of farms equipped, compulsory data, reference list of
diseases, valorizations proposed and actions to encourage farmers to record health data…).
Analysis of volume and relevant of diseases
These information systems aim first at recording treatments of cattle animals for
regulation reasons. As it was not possible to ask each farmer the authorization for using these
sensitive data, and even if only events (not the treatments) were used for the study, the animal
and herd identities had to be re-numbered within tool, in order to ensure the farmer’s anonymity.
Only the animal sire’s identity with health data was transmitted without re-numbering.
Health events not corresponding to diseases were deleted (eg. dehorning). The data of 7
information systems among 8 could be analyzed, on a reference period from 2007 to 2012.
Definitions of health events were more or less detailed according to the tools. In order to
facilitate the comparison between tools, events were grouped according to 20 classes defined
by national experts in animal health. These classes were based on those defined by ICAR
(ICAR, 2012). 16 classes among 20 corresponded to curative events; data corresponding to the
4 others (preventive treatments) were excluded from the further analyses. Several diseases were
included within the same class. For instance, the class “metabolic and digestive disorders”
includes acidosis, ketosis and milk fever.
9 animal categories were defined: 4 in dairy breeds (cows, heifers, calves less than 2
months old, bulls) and 5 in beef breeds (same 4 categories than for dairy cattle + young bulls
less than 9 months old).
The herds assumed to record health events exhaustively were identified using criteria
based on the diversity of disorders and the regularity of recording data. They should have at
least events in to 3 different classes of disorders recorded each year between 2009 and 2012.
The annual and monthly prevalence was calculated within class of disorders, category of
animals and period. Prevalence Pijk is calculated using the following formula:
Pijk = Nocijk / Ntotij
With Nocijk: number of animals of the category i, with at least one occurrence of disease
k during the period j
Ntotij: total number of animals of the category i present during period j
Moreover, a study of progeny of AI Holstein bulls was performed, in order to assess the
harmonization of recorded data between information systems. Only diseases related to lactating
cows could be analyzed. Selected bulls should have at least 500 daughters with disorders
recorded in at least 2 different information systems.
Unfortunately, the pedigree of animals without any disorder was not available in the
transmitted files, and this information could not be recovered using the national genetic
information system, since the animal IDs had been re-numbered. Therefore the prevalence of
disorders within progeny of bulls was not possible to calculate, and it was replaced by the
proportion (prop) of each disorder among those recorded for the daughters of each bull:
prop
= Nd
Nd
With Ndijk: number of daughters of the bull i, with disorders j, recorded in tool k.
In this study, the category “dairy cows” was divided into 3 classes, according to parity
(1st, 2nd, 3rd and later).
Results
General characteristics of the information systems
Health events are recorded by farmers using web applications developed for herd
management purposes. The number of farms subscribing to applications for animal health
recording increased by 20% between 2009 and 2011, with a total of 37 300 farmers in 2011.
The way to record health events shows similarities between tools. Recording a disease
without treatment is always possible. Recording a treatment without the associated disease is
impossible for 3 tools. The date and name of the disease are always required and all tools have
a reference list of diseases. This list contains from 50 to 516 choices depending on the tool,
with a total of 898 different diseases among the 8 tools described in the survey.
Enhance of data is made within herd, through historic and annual health reports. Only one
tool proposes a comparison with a reference group of herds. In order to strengthen recording,
these tools are connected with other applications (milk recording organizations, veterinarians).
Analysis of the recorded health events
7 tools transmitted 10 384 788 health events from 25 736 cattle farmers. 18 303 cattle
farmers recorded 6 938 477 health events for 1 613 353 bovine animals (76.4% dairy breeds)
over the period 2007-2012. In 2012, the herds and the animals of the study represented 5.8 %
of the French cattle herds and 2.8 % of the bovines in France (Institut de l’Elevage, 2012). The
curative health events represented 59.2% of the data.
2 528 herds with “exhaustive” records were selected. They represented 15.6% of the total
number of herds and 56.3% of the recorded diseases. 80.5% of animals were dairy cattle, with
65 cows per herd. The farmers recorded each year 282 health events from 9 different categories
on average (against 129 health events for all the herds without selection). In 2012, the main
diseases were mastitis (22.2%), digestive disorders (15.9%), respiratory diseases (13.5%),
general infectious diseases (7.0%), post-partum diseases (5.5%) locomotory disorders (4.6%).
The annual prevalence are stable over 4 studied years. They are presented in table 1 for 3
categories of dairy animals in 2012. The profiles of monthly prevalence are also rather stable
over years and consistent with our expectations (e.g.: figure 1 for dairy cows in 2012).
Table 1: Results of annual prevalence (%) in 2012
Categories of animals and number
Types of diseases
Mastitis
Digestive disorders
Post-partum diseases
Locomotory disorders
General infectious diseases
Respiratory diseases
Reproductive disorders
Metabolic diseases
Skin diseases
Others neonatal diseases
Dairy cows
143 278
29.1
9.1
9.1
7.1
4.3
5.4
4.5
3.6
2.7
Dairy heifers
119 228
Dairy calves
21 428
8.0
39.4
0.6
4.5
8.6
1.2
0.3
1.7
1.0
2.2
23.7
0.4
1.1
4.8
6,0%
5,0%
4,0%
3,0%
2,0%
1,0%
0,0%
jan
feb
mar
apr
may
jun
jul
aug
sept
oct
nov
Mastitis
Respiratory diseases
Digestive disorders
Locomatory disorders
dec
Figure 1: Evolution of monthly prevalence (%) in 2012 for dairy cows
45 AI Hosltein bulls had more than 500 daughters with records in at least 2 information
systems. For each tool, the proportion of mastitis among all recorded diseases was calculated,
and it was compared to the same proportion calculated with the 6 other tools together. We
observed a good consistency between the proportions obtained with each tool when compared
to the proportion calculated with the 6 other ones together. An illustration is shown on figure 2
with one given tool (R² = 75%). The other diseases presented a poorer consistency (e.g. figure
2 for locomotory disorders, R² = 41%). However, it remains acceptable, regarding to the low
heritability of these traits, to the low frequency of these disorders and to the fact that these tools
are located in different regions with heterogeneous management systems.
Figure 2: Proportion of mastitis (left figure) and of locomotory disorders (right figure) among
all recorded diseases within bull progeny: comparison of results of the tool X with the 6 other
tools (dairy cows in 3rd lactation and more)
Discussion
The classes of diseases used in the 7 tools are very heterogeneous, which does not
facilitate a common valorization. However, a grouping of the diseases is possible.
The relevance of the prevalences calculated in this study is difficult to assess, since very
few references are available in France. A study of Fourichon (2001) on 200 Holstein and
Normande herds indicates a prevalence of 44.1% for clinical mastitis, 17.3% for post-partum
diseases, 10.9% for locomotory disorders, 9.7% for metabolic diseases, 5.1% for digestive
disorders and 2.6% for respiratory disorders. The regions and breeds covered by our study are
more diversified, herd management has probably changed within the last decade as well as the
genetic level of animals. However the results of annual prevalence are globally consistant with
our results, at least for the most important post-partum diseases.
The criteria used to select herds with “exhaustive” records may be improved. However
the consistency of monthly prevalences between tools, the big diversity of health events
recorded in the selected herds show their relevance.
Finaly, the study of consistency of records between tools using data from progeny of AI
common bulls would have been more relevant, if a prevalence could have been calculated
instead of a proportion among diseases. However, first results are encouraging; such records
could probably be used for genetic evaluations, and interesting references (e.g. monthy
prevalences within breed and parity) could be developped by pooling these data.
Conclusion
The study shows that health data recorded by farmers could be used for genetic
evaluations or advices. The current valorizations are made within each tool and they are limited
by the volume of relevant data. A better valorization would be possible by grouping events of
different tools, according to the classification proposed by ICAR. It would motivate the farmers
to improve the recording of health events.
This study raised two crucial points:
- a clear and common communication involving all the actors is necessary, as it was
mentioned in other experiences (Blériot & Thomas , 2013, Egger-Danner et al., 2012),
- health data are very sensitive. Farmers should be able to indicate via their IT application
whether they accept (or not) that their data are used for a given collective valorization.
List of References
Blériot G. & G. Thomas, 2013. Projet PARABOV (2011-2012) – Bilan de la collecte. Institut
de l’Elevage, CR 001378002, 47p.
Boichard D., V. Ducrocq & S. Fritz, 2015. Sustainable dairy cattle selection in the genomic era.
J. Anim. Breed. Genet 132 (2): 135-143
Egger-Danner C., B. Fuerst-Waltl, W. Obritzhauser., C. Fuerst, H. Schwarzenbacher, B.
Grassauer, M. Mayerhofer & A. Koeck, 2012. Recording of direct health traits in
Austria-experience report with emphasis on aspects of availability for breeding
purposes. J. Dairy Sci. 95(5) 2765-2777.
Fourichon C., H. Seegers, N. Bareille & F. Beaudeau, 2001. Estimation des pertes et de l’impact
économique consécutifs aux principaux troubles de la santé en élevage bovin laitier.
Rencontres Recherches Ruminants, Paris, France 137-143.
Frandsen J., 2013. Presentation of Danish system of registration and use of health data. ICAR
Technical Meeting Aarhus (Denmark), 29-31 May 2013.
ICAR., 2012. Guidelines for recording, evaluation and genetic improvement of health traits,
418-424.
Institut de l’Elevage., 2012. Bovine key numbers - National Database for Identification. 4p.
Koeck A., J. Jamrozik, G.J. Kistemaker, F.S. Schenkel, R.K. Moore, D.M. Lefebvre, D.F.
Kelton & F. Miglior, 2015. Development of genetic evaluations for metabolic
disease traits for Canadian dairy cattle. Interbull Bulletin 49, 76-79.