Databases save time and improve the quality of the design, management and processing of ecopathological surveys

Databases save time and improve the quality of the

design, management and processing of ecopathological

surveys

P Sulpice, F Bugnard, D Calavas

To cite this version:

P Sulpice, F Bugnard, D Calavas. Databases save time and improve the quality of the design,

management and processing of ecopathological surveys. Veterinary Research, BioMed Central,

1994, 25 (2-3), pp.120-126.

<

hal-00902181

>

HAL Id: hal-00902181

https://hal.archives-ouvertes.fr/hal-00902181

Submitted on 1 Jan 1994

HAL

is a multi-disciplinary open access

archive for the deposit and dissemination of

sci-entific research documents, whether they are

pub-lished or not.

The documents may come from

teaching and research institutions in France or

abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire

HAL

, est

destin´

ee au d´

epˆ

ot et `

a la diffusion de documents

scientifiques de niveau recherche, publi´

es ou non,

´

emanant des ´

etablissements d’enseignement et de

recherche fran¸

cais ou ´

etrangers, des laboratoires

publics ou priv´

es.

Databases

save

time

and

improve

the

quality

of

the

design,

management

and

processing

of

ecopathological

surveys

P

Sulpice

F

Bugnard,

D

Calavas

Centre

d’Écopathologie

Animale, 26, rue de la Baisse, 69100 Villeurbanne, France

Summary &horbar;

The

example

of an

ecopathological

_survey on

nursing

ewe mastitis shows that data bases have 4

complementary

functions: assistance

during

the

conception

of surveys;

follow-up

of surveys;

management

and

quality

control of data; and data

organization

for statistical

analysis.

This is made

possible by

the simultaneous

conception

of both the data base and the survey, and

by

the

inte-gration

of computer science into the work _{of the task group} that conducts the survey. This

methodology

helps

save time and

improve

the

quality

of data in

ecopathological

surveys.

ecopathology

/ survey / data

processing

/ data base /

sheep

Résumé &horbar; Les base de données :

gain

de _temps et de

qualité

dans la

conception,

la

gestion

et le traitement des

enquêtes

d’écopathologie.

En prenant

l’exemple

d’une

enquête d’écopathologie

sur les mammites des brebis aitaitantes, on montre _{que les} bases de données ont 4 fonctions

com-plémentaires :

aide à la _conception des

enquêtes,

suivi des

enquêtes, gestion

et contrôle de

qualité

des

données,

organisation

des données pour

l’analyse statistique.

Cela est rendu

possible

par une concep-tion simultanée de la base de données et de

l’enquête,

et

l’intégration

de la matière

informatique

au

groupe de travail

qui

élabore

l’enquête.

Cette

méthodologie

concourt _{à gagner du temps} et à

amélio-rer la

qualité

des données dans les

enquêtes d’écopathologie.

écopathologie

l enquête

l informatique

/ base de données / ovins

INTRODUCTION

Data bases become necessary in

eco-pathological

surveys because of the volume of the collected data and the structure of the information

(Lescourret et al, 1992).

At

first,

data bases were used at the Centre

d’Écopathologie

Animale after the

phase

of data collection. This late use

proved

to be

expensive

in

conception

time,

particularly

for the semantic

modelling

of data.

Further-*

Correspondence

and

reprints

more, the use of the data base after all the information has been collected allows

only

a late verification of the data without the

possibility

of

returning

to its

origin

to

improve

data

quality.

In order not to limit the definition of the function of the data base to the orga-nization of data

prior

to statistical

analysis,

its

conception

and construction must be

car-ried out at the same time as _{the survey.}

Thus,

the data bases

conceived,

gener-ated and used

by

the Centre

d’Ecopatholo-gie

Animale have 4

complementary

func-tions: assistance

during

the

conception

of surveys

by

a

computer-oriented approach;

follow-up

of the _survey in the

farm;

man-agement

and control of the data

quality;

and

organization

of data

prior

to statistical

analy-sis.

The

example

of the _survey on

nursing

ewe mastitis

(Calavas,

1992)

illustrates these 4 functions and

gives

a

general

idea of how this

approach

can

help

save time and

improve

the

quality

of the

ecopatho-logical

surveys.

FUNCTIONS OF DATA BASES AT THE CENTRE

D’ECOPATHOLOGIE

ANIMALE

Assistance

during conception

_{of surveys}

A survey and its data base are

developed

simultaneously

and

interactively.

From the very

beginning,

the

computer

_scientist

par-ticipates

in the work of the task group in

charge

of the survey

(Calavas, 1992).

This allows the

development

of a _synergy and

an _{interaction between the survey and the}

database,

as well as the distribution of

con-ceptual

elements,

such as a

hypothesis

scheme

(path analysis),

a

project

of the

pro-tocol,

models of

questionnaires,

and a

con-ceptual

data model

(fig

1

).

The

conceptual approach

previous

to the

generation

of data bases

(Tardieu

et

al,

1983)

has a

large impact

on the

conception

of the _survey. In some cases, the

conceptual

data model built from the

hypothesis

scheme,

the

projects

of

questionnaires

and the data collection

protocol

all shed

light

upon a non-functional

organization

of the survey

(data coding,

identification of

indi-viduals,

and coherence of

questionnaires).

Furthermore, peculiarities

that make the progress of the survey easier in the

field,

must be taken into account in the data base. For

example,

in this survey, the ewes were

identified

by

a collar number

(to

facilitate

pinpointing

a

sample

of ewes monitored

dur-ing

the entire

survey)

and/or

by

a

perennial

identification

number,

an ear

tag,

or

tattoo-ing

(for

animals affected

by

mastitis and added to the

sample

in the course of the

sur-vey),

which necessitated

taking

into account

2 different

identifying

attributes

(fig

2).

Follow-up

of surveys

The

study

of the information flow allows

computerization

of certain

operations

of the survey

management,

for

example,

letters

sent to different

participants

_{of the survey}

(reminders

to return

questionnaires

and survey

agreements)

and tools of the pro-tocol

follow-up

(identification

labels for the

biological

samples

and lists of the ewes

examined

during

the first visit in order to

facilitate the

follow-up

of these animals

dur-ing

subsequent visits).

This also allows the automation of the return of information to

the breeders and the interviewers in the

course of the _{survey, eg} results of

sero-logical analysis

and

preliminary

statistical

analyses.

Management

and control of data

quality

The fact that the data base is

operational

when the _survey starts in the field allows a

continuous

operation

of

rereading/correction

and data

entry

as soon as the _{first survey}

documents return to the Centre

d’fco-pathologie

Animale.

Information control at different levels

(identification,

intrinsic

coherence,

and coherence between

variables)

allows the detection of errors and inconsistencies

unno-ticed

during

the

rereading

of

questionnaires

(Sulpice, 1992).

This also makes it

possi-ble to _go

immediately

back to

the

source of

information in the case of error, or

Organization

of data

The data base constitutes the

protected

and

autonomous

primary layer

of the _survey

data,

ie includes all collected

data,

as well as an exhaustive data

dictionary.

The data base

management system

becomes a server for the data to be

exported

to the

sta-tistical

analysis

software

(eg,

SAS and

SPADN).

On their way

back,

the

synthetic

data

generated by

the statistical

analysis

(eg,

calculated data and risk

factors)

are

integrated

in a

secondary

layer.

In the

end,

these 2 layers

of information constitute the referential data base built from the _survey.

SAVING TIME AND IMPROVING THE QUALITY OF SURVEY DATA

Saving

time

The _synergy

implemented

in the concep-tion of the survey is aimed at

sharing

the

feasibility

constraints of the _survey

(eg,

the way the visits progress, the

special

organi-zation of a notation document in the form of a

table,

or the

necessity

of a double iden-tification of the

ewes)

and the data base

(eg,

division into entities and associations

to be

implemented).

There are no ’head losses’ insofar as

data

processing

is

integrated

in the task group that conducts the survey,

computer-ized data

processing

being

one of the

dis-ciplines

of the task group and the

computer

scientist

representing

one of the

professions

(Rosner, 1984).

Sharing

conceptual

elements

during

the

conception stage

of the _survey

helps

val-orize the work of

preparing

the

analysis

doc-uments. For

example,

a

diagram

of the

physical

data model is used to

give

a

syn-thetic view of the collected data and can be used as a work document

during

the

con-ception phase

and as an element of the

study report

(fig

2).

The automation of the data

management

tasks allows a reduction in the time

spent

on the administrative

management

of the survey

(eg,

movements of the

documents,

survey progress, and

mail)

and

thus, greater

importance

can be

given

to the

follow-up

of the data collection

(eg, rereading

the

docu-ments,

respecting

the deadlines of the

pro-tocol,

and

making

inquiries).

The

opportunity

to

continuously

_carry out

operations

of

rereading/correction

and data

entry

allows the statistical

analysis

to be started at the return of the final document and sometimes even in the course of the survey when the documents are

split

_{up. It}

is therefore

possible

to

considerably

reduce the time between the arrival of the infor-mation at the Centre

d’!copathologie

Ani-male and the

beginning

of the statistical

analysis,

thus

helping

to obtain results much

quicker

(fig

1

).

Improving

the

quality

of the survey data

First,

due to the

computer-oriented

approach

preceding

its

generation,

the data base

helps improve

the

quality

of the data col-lection

during

the

conception stage

of the survey:

feasibility

of

protocols;

coherence of the

questionnaires;

study

of the infor-mation flow.

The data base then

helps

collections of data

by using

the

follow-up

tools which it

generates (owing

to the

generated

lists,

it is

possible

to reduce the number of ewes

lost

during

the _survey because of identi-fication

errors).

Its main mission at the level of the data collection is to

provide

control of the data

reliability

at the time of their

entry

and allow a

quick

return to the source of the information in the case of an error. For

example,

4

questionnaires

about the

weaning,

flock

contacts)

and 4 598 individ-ual forms filled out

during

the first visit of the _survey were verified and entered into the data base

immediately

after

being

received at the Centre

d’Ecopathologie

Ani-male.

Concerning

the

questionnaires,

it was

found out that 2.82% of the data was

miss-ing

(176

out of 6

235).

Reminders were sent

immediately

and 169

missing

data items

were recovered

(to

reach the level _{of 1 per} 1 000

missing data).

In what concerns the

ewes, each animal was described

by

20

variables,

totalling

91 960 data items

(4 598

x

20).

The rate of the

missing

data in all filled-out forms was less than 1 _per 1 000

(89

out of 91

960).

Finally, during

statistical

analysis,

the data base allows a transfer to the statisti-cal

software,

both files

containing

the _survey data and a data

dictionary (with

the name,

label and

description

of the classes of

vari-ables).

The data

dictionary

provides,

in an

univocal way, a reference to each variable based on the

precise

wording

_{of the}

ques-tions;

this excludes all risk of

ambiguity

in the

analyzed

variables

(no

errors

possible

in the

wording

or in the value of each

class).

Moreover,

all

editing

necessary for the

anal-ysis

is documented. This

provides security

and

comfort,

especially taking

into account

the

difficulty

of

giving only

8-character

names to variables

(in

the survey there

were, for

example,

119 individual variables relative to the ewes, of which 59 character-ized the _mammary

glands,

with notation of lesions on each

quarter

of the _mammary

gland

twice

during

the survey; thus it was

difficult to have 59

explicit

names based on a 8-character

label).

Illustrative

figures

on the _survey of

nursing

ewe mastitis

Some indicators will

help appreciate

the time saved and the

quality improved

in this survey. The

conception

and the

generation

of the data base with the data base

man-agement system

Dataflex

(Data

Access

Cor-poration,

1991)

required

15

working days

for a

computer

scientist. The database was

operational

at the time of the data collec-tion

(fig

1

the

statistical

analysis

started in

July

1993 on the data collected

during

the first

visit,

that is to _say 34 d after the return

of the last

questionnaire,

and

immediately

after the return of the last reminder. The

quality

of the collected data can be evalu-ated

by

the final rate of the

missing

data after the reminders

(less

than 1 _per 1

000).

CONCLUSION

The

synergetic

and simultaneous

organi-zation between the

ecopathological

survey and the data base reinforces the role of the data bases. This determines the 4

comple-mentary

functions of the databases:

as-sistance in the

conception

of _surveys; tech-nical

management

of surveys;

management

and control of the data

quality;

and

organi-zation of data for statistical

analysis.

Thus,

the data bases set _up

by

the

Cen-tre

d’tcopathologie

Animale for each of its surveys conform with:

(i)

its

guidelines,

as

they

reduce the time of

processing

the

eco-pathological

surveys and

generate

a ref-erential data base built from the survey; and

(ii)

the chosen

methology

_{of data}

pro-cessing

integrated

into the work of a

pluridisciplinary

and

pluriprofessional

task group that conducts the surveys and their

follow-up.

REFERENCES

Calavas D

(1992) Conception

de

1’enqu6te.

In: Mammites des brebis allaitantes. Centre

d’Écopathologie

Animale, Villeurbanne, 11,

l,pp143

Data Access

Corporation

₍₁₉₉₁₎ Dataflex User’s Guide. Miami, USA, pp 732

Lescourret F, Perochon L, Coulon JB,

Faye

B,

Landais E

(1992) Modelling

and information

system using

the MERISE method for

agri-cultural research: the

example

of a database for a

study

on

performances

in

dairy

cows.

Agric Syst 38,

149-173

Rosner G

(1984) Rapport

relatif a I’etude de fais-abiiite du

projet

de creation d’un centre

regional

d’6copathologie

muiti-especes.

GIE Lait Viande Rhone

Alpes, Lyon,

pp 74

Vet Res (1994) 25, 126-129

&copy; Elsevier/INRA

Sulpice

P

(1992) Appropriation

de l’outil

informa-tique

par des

enqu6teurs

au cours d’une

experi-mentation de saisie sur site. Actes 2! Coll Eur

Agrimatica Informatique

et

t6l6matique

agri-coles. Methodes et conduites de

projets

Char-bonnibres-les-Bains, France, 2-3 juillet, Tele Promotion Rurale Rh6ne-Alpes, 263-271 Tardieu H, Rochfeld A, Colleti R

(1983)

La

metho-de MERISE. Tome I. Principes et outils. Les

Editions d’Organisation,

Paris, France, pp 320

Deer-herd

health and

production

profiling

in New Zealand. I.

Study design

L

Audigé

PR

Wilson,

RS

Morris

Department

of

Veterinary

Clinical _Sciences,

Massey

University,

Palmerston North, New Zealand

Summary &horbar;

A

2-year

observational

study

was conducted on 15 commercial reddeer farms in the North Island of New Zealand, to

provide

health and

production

data, and

identify

risk factors for outcomes

in-cluding

health,

reproduction,

venison and velvet antler

production.

About 2 700 hinds, 2 400 wea-ner deer and 1 500 _stags were monitored.

Daily

_management

practices,

deer

performance

and health

problems

were recorded. At

3-monthly

visits,

samples

were collected from deer, pastures and soils. Data

were

processed by

multivariable statistical

techniques

to

identify

the most important factors

affecting

production

and health.

epidemiology

/ observational

study

/ farmed deer / New Zealand / multivariable statistical

ana-lysis

Résumé &horbar; Profil de santé et de

production

dans les

élevages

de cerfs

élaphe

en

Nouvelle-Zélande. 1. Protocole

d’enquête.

Une

enquête

d’observation de 2 années était en cours dans 15

élevages

de cerfs

élaphe

dans l’ile du Nord en Nouvelle-Zélande afin d’en connaître les

profils

de santé et de

production,

et les facteurs de _risques associés aux

paramètres

de santé, de

reproduction,

et de

production

de venaison et de bois en velours. Environ 2 700 biches,

2 400 jeunes

cerfs sevrés et i 500 cerfs étaient individuellement suivis. Les

pratiques

d’élevage,

les

performances

zootech-niques

et les

problèmes

de santé étaient

enregistrés quotidiennement.

Cerfs, patures et sols étaient

prélevés

au cours de visites trimestrielles. Les données étaient

analysées

_{par méthodes statistiques} multivariées.

épidémiologie lenquête

d’observationlCervidéslNouvelle-Zélandelanalyse statistique

mut-tivariée

*