Page 1

PAPERS & ARTICLES

tions (Figs 3a and 4b) are presumed to have been caused by

the microtome knife in cutting the calcified tissue. The cal-

cified inclusions within neurons (Figs 4a,b) were interpreted

as calcified aggregates of storage bodies. These novel inclu-

sions, the intense secondary calcification and the occurrence

of calcium oxalate uroliths and urinary tract pathology war-

rant further investigation.

References

AGMOULIS, D. P. (1985) The pathology of lysosomal storage diseases.

Pathology Anntua130, 247-285

BAUMKOTTER, J. F. & CANTZ, M. (1983) Inhibition of the activity in vitro

bv sulfated glycosaminoglyvcals anid other compounds. Biolhenlica

Bioplysica Acta 761, 163- 170

FISCHER, A., CARMICHAEL, K. P., MUNNELL, J. F., JHABVALA, P., THOMP-

SON, J. N., MATALON, R., JEZYK, P. F., WANG, P. & GIGER, U. (1998)

et

Sulfamidase deficiency in a family of dachshunids. A canine model of

mucopolysaccharidosisIIIA (Sanfilippo A). Pediatric Research 44, 74-82

JOLLY, R. I)., ALLAN, F. J., COLLETT, M. G., ROZAKLIS, T., MULLER, V. J. &

HOPWOOD, J. J. (2000) Mucopolysaccharidosis IIIA (Sanfilippo syndrome)

in a New Zealand huntawvay dog with ataxia. Neiv Zecclalatd Veteriniary Jouirnal

48, 144- 148

JOLLY, R. D. & WALKLEY, S. U. (1997) Lysosomal storage diseases of animals:

An essay in comparative pathology. VeteriMary Pathology 34, 527-548

LUNA, L. G. (1993) Histopathologic Methods anid Color Atlas of Special Stains

and Tissue Artifacts. American Histolabs Publicationi Departmenit. p 589

NEUFELD, E. F. & MUENZER, J. (1989) The mucopolysaccharidoses. In The

Metabolic Basis of Inherited Disease. Vol 2. Eds C. R. Scriver, A. L. Beaudet,

W. S. Sly, D. Valle. New York, NMcGraw Hill. pp 1565-1587

NISHIO, S., ABE, Y., WAKASUKI, A., IWATA, H., OCHI, K., TAKEUCHI, M.

& MATSUMOTO, A. (1985) Matrix glycosaminoglvcanis in uriniary stones.

JltornatilofUrology 134,503-505

WALKLEY, S. U. (11998) Cellular pathology of lysosomal storage diseases. Braint

Patholol0g8, 175- 183

WIN EPISCOPE 2.0: improved epidemiological

software for veterinary medicine

Veterinary Record (2001)

148, 567-572

M. Thrusfield, BVMS,

MSc, DTVM, CBiol, FIBiol,

MRCVS, Department of

Veterinary Clinical

Studies, University of

Edinburgh, Royal (Dick)

School ofVeterinary

Studies, Easter Bush

Veterinary Centre, Easter

Bush, Roslin, Midlothian

EH25 9RG

C. Ortega, DMV, PhD,

I. de Blas, DMV,

Department of Animal

Pathology, Faculty of

Veterinary Medicine,

University of Zaragoza,

C/Miguel Servet 177,

50013 Zaragoza, Spain

J. P. Noordhuizen, DVM,

PhD, Utrecht University,

Faculty ofVeterinary

Medicine, Department of

Ruminant Health Care,

PO Box 80151, 3508 TD

Utrecht, The Netherlands

K. Frankena, MSc, PhD,

Wageningen University,

Wageningen Institute of

Animal Sciences,

Quantitative Veterinary

Epidemiology Group,

PO Box 338, 6700 AH

Wageningen,

The Netherlands

M. THRUSFIELD, C. ORTEGA, 1. DE BLAS, J. P. NOORDHUIZEN, K. FRANKENA

Recent changes in veterinary medicine have required quantitative epidemiological techniques for designing

field surveys, identifying risk factors for multifactorial diseases, and assessing diagnostic tests. Several

relevant techniques are brought together in the package of veterinary epidemiological computer software,

WIN EPISCOPE 2.0, described in this paper. It is based on Microsoft Windows and includes modules for the

design and analysis of field surveys, control campaigns and observational studies, and a simple

mathematical model. It provides comprehensive 'Help' screens and should therefore be useful not only in

field investigations but also for teaching veterinary epidemiology.

RECENT decades have witnessed changes in the practice of

veterinary medicine (Thrusfield 1998). In developed coun-

tries, the successful control of the major infectious diseases

and the resultant intensification of livestock enterprises has

produced a shift in interest towards complex, frequently non-

infectious, diseases, and increasing emphasis is being placed

on the health ofherds rather than individual animals (Brand

and others 1996). In contrast, in developing countries, the

control of infectious diseases is still the major problem and

progress needs to be made both in the measurement of dis-

ease frequencies, for example of trypanosomiasis in Africa

and Asia, and in the implementation of control and/or erad-

ication campaigns, for example for rinderpest (IAEA 1991).

These changes have required the application of quantita-

tive epidemiological procedures (Noordhuizen 1996), notably

the use of rigorous sampling theory when conducting field

surveys, and observational studies to identify the risk factors

associated with multifactorial diseases in both farm practice

and companion animal practice. Moreover, diagnostic tests

applied either to individual animals, or to animal populations

in eradication campaigns, can only be interpreted correctly

when their validity and reliability has been assessed. Despite

a long history in veterinary medicine, many diagnostic tests,

particularly serological tests, have not been so assessed.

The methods and statistical theory that underpin these

quantitative procedures have been described by Martin and

others (1987), Thrusfield (1997), and Noordhuizen and oth-

ers ( 1997), and several computer programs have been designed

to facilitate the necessary computations by veterinarians and

physicians. These include suites ofprograms, such as EPI INFO

(Dean and others 1995), PEPI (Gahlinger and Abramson 1995),

EPISCOPE (Frankena and others 1990) and EPIZOO (Kouba

1997), and many programs for specific, individual analyses

(The Epidemiology Monitor). However, these programs have

limitations, especially for users who lack a grounding in ana-

lytical techniques. The package WIN EPISCOPE 2.0 has therefore

been designed to combine the procedures that are commonly

used in the design and analysis ofepidemiological studies into

an easy-to-understand form based on Microsoft Windows.

The package provides the main computational procedures

used in the design and analysis ofsimple field surveys, in con-

trol campaigns and observational studies, and in the assess-

ment ofdiagnostic tests, and includes an introduction to basic

mathematical modelling ofinfectious diseases. For each com-

putation, comprehensive'Help' menus are provided which

describe the techniques and list, with references, the formu-

lae that are used. The package is an improved and expanded

version of EPISCOPE for MS DOS (Frankena and others 1990)

and of its first Microsoft Windows release, WIN EPISCOPE 1.0

(Ortega and others 1996).

MATERIALS AND METHODS

Basic structure of the program

WIN EPISCOPE 2.0 has been programmed with Borland Delphi

1.0, and can be run on IBM-compatible personal computers

with theMSWindows 3.1 operating system (or higher).

The Veterinary Record, May 5, 2001

567

Page 2

PAPERS & ARTICLES

ityl I -0 U RICS]a

w6'

Main module

Submodule

Files

Tests

Tests

Tests

Tests

Tests

Samples

Samples

Samples

Samples

Samples

Samples

Samples

Samples

Samples

Analysis

Analysis

Analysis

Analysis

Analysis

Analysis

Analysis

Analysis

Analysis

Models

Help

Agreement

Evaluation

Advanced evaluation

Cut-off value*

Multiple tests*

Estimate a mean

Estimate difference between means

Estimate a percentage

Estimate difference between percentages

Detection of disease

Threshold value*

Unmatched case-control

Matched case-control

Cohort*

Cross-sectional*

Stratified cross-sectional*

Case-control

Stratified case-control

Matched case-control

Cohort (cumulative incidence)

Cohort (incidence rate)

Stratified cohort (cumulative incidence)

Stratified cohort (incidence rate)

Reed-Frost

*Not available in earlier versions of EPISCOPE

Six modules can be accessed via the main menu bar, by

using either the computer's arrow keys or its 'mouse'. and

these give access to 'submodules' (Table 1). For example, the

module 'Tests' makes available the following submodules:

'Agreement, 'Evaluation', 'Advanced evaluation', 'Cut-off value'

and'Multiple tests' When a submodule is selected, a window

appears with two parts: 'Data entry' (either at the top or to

the left), and 'Results' (either at bottom or to the right).

Two graphic buttons enable the user to 'Calculate' or

'Close', and tabular results can be converted to graphs by

using a simple graphic button.

Source texts

The definitions of the statistical terms, parameters and for-

mulae used by the submodules have been obtained from

several standard epidemiological and statistical texts

(Snedecor and Cochran 1980, Fleiss 1981, Schlesselman

1982, Rothman 1986, Martin and others 1987, Levy and

Lemeshow 1991, Lwanga and Lemeshow 1991, Armitage

and Berry 1994, Noordhuizen and others 1997, Thrusfield

1997).

Data entry

Each submodule includes a field for inputting data, either

directly or by'pasting' from other packages. Other parame-

ters may also be changed. For example, the precision of sam-

ple estimates may be altered by selecting different levels of

confidence, and the statistical power of a study may be var-

ied. The program has built-in checks to ensure that imper-

missible data are not entered, for example, values outside the

permitted ranges or non-numerical characters.

File module

The first module,'Files', allows the working language (cur-

rently either English or Spanish) to be selected and the printer

settings to be defined. It also makes it possible to 'paste' data

from other applications, and to print the screen's current

display when a submodule is in use.

Tests module

This module has five submodules for the assessment ofdiag-

nostic tests:

Agreement Compares two diagnostic tests, and computes

the 'kappa' statistic as a measure of the agreement beyond

chance between them, and its specified confidence interval.

When applied to two or more runs of the same test, it is a

measure ofthe test's repeatability (reliability).

Evaluation Computes diagnostic sensitivity and specificity

(the primary estimators of the validity of a test), true and

apparent prevalence, and positive and negative predictive

values with their confidence intervals.

Youden's J statistic (a summary measure of validity) and its

confidence interval.

It also computes

Advanced evaluation Explores the relationships between

sensitivity, specificity, and true or apparent prevalence by

holding two of the three parameters constant while the third

increases in increments of 10 per cent. The results can be

displayed graphically.

Cut-off value Applies to tests that are based on continuous

variables, for example, antibody titres. The numbers of

animals of known disease status that are scored positive and

negative for different values of the variable are entered in a

data table. The sensitivity, specificity and predictive values

are displayed, with their confidence intervals, for different

cut-off values

of the

variable. A

Characteristic

(ROC)

curve and

interval can also be displayed.

Receiver Operating

associated

confidence

Multiple tests Computes the sensitivity, specificity and

predictive values of a test strategy when two independent

tests are applied under either a parallel test interpretation

(when an animal is classified as affected if positive to either

test) or a serial interpretation (when an animal is classified

as affected only if positive to both tests). Independent tests

assess different indicators of disease. For example, a test for

an antibody would be independent of a test for a pathogen

but two tests that each identify an antibody would not be

independent. Consideration of dependency

(Smith 1995) and is beyond the scope of this package.

is complex

Samples module

This module computes sample sizes for animal health surveys

and observational studies and has nine submodules:

Estimate a mean Determines the size of a single sample

required to estimate a population mean with a given level of

confidence and absolute precision (absolute error). The size

of the population from which the sample will be drawn may

be specified. An estimate of the standard deviation

required.

is

Estimate difference between means Calculates the

sample size required to detect a difference between the

means of two populations with

confidence and power. The expected means of the two

populations and an estimate of the common standard

deviation are required. The sample sizes are computed for

independent and related samples, and for one-tailed and

two-tailed hypotheses.

a

specified

level of

Estimate a percentage Determines the sample size required

to estimate a percentage, for example a prevalence, with a

specified absolute precision (absolute error) at a given level of

confidence. Alternatively, the precision can be calculated for a

specified sample size. The results of both calculations can be

presented graphically for various percentages.

Estimate difference between percentages Displays the

sample size required to estimate a specified difference

The Veterinary Record, May 5, 2001

568

Page 3

PAPERS & ARTICLES

between two percentages with a specified level of confidence

and power for one-tailed and two-tailed hypotheses.

Detection of disease Determines the sample size required, at

a specified level of confidence, to identify at least one affected

animal when a minimum percentage of animals can be

assumed to be affected, for example in eradication campaigns.

Alternatively, either the maximum number ofanimals affected

in a population can be calculated when a sample of specified

size identifies no cases, or the probability ofidentifying at least

one positive animal in a sample of a given size can be

calculated when a specified percentage of the population is

assumed to be affected. The results of these three calculations

can be presented graphically for ranges ofthe specified values.

Threshold value Calculates the number of animals that

need to be sampled from a group (for example, a farm)

selected from a population (for example, the farms in a

region), to determine whether the group has a prevalence

above or below a specified value in the population, and to

identify a level above or below this value, at which specified

action should be taken, with a user-defined level of

confidence and statistical power.

Unmatched case-control and Matched case-control

Calculates the number ofcases and controls to be selected in

a case-control study to detect a disease/factor relationship,

expressed in terms of a minimum odds ratio, with a defined

level of confidence and power. The matched option is

selected when the cases and controls are matched with

respect to potential confounders, for example, age, breed or

sex; the sample size can be determined for both one-tailed

and two-tailed hypotheses.

Cohort Calculates the numbers of animals exposed and not

exposed to a factor in a cohort study that are needed to

detect a disease/factor relationship, expressed in terms of a

minimum relative risk, with a defined level of confidence

and power.

Analysis module

The fourth module is concerned with the analysis ofthe three

main types of observational study; cross-sectional, case-

control and cohort. It estimates the main measures of asso-

ciation between disease and risk factors: the prevalence ratio,

odds ratio and relative risk, and some secondary measures.

The module has nine submodules:

Cross-sectional Calculates the prevalence ratio, odds ratio,

attributable risk, attributable proportion and attributable

proportion in the exposed group.

It also calculates approximate logarithmic and test-based

95 per cent confidence intervals for the prevalence ratio and

odds ratio, with an indication of their validity.

Stratified cross-sectional Allows up to 10 strata to be

defined for a potential confounder, and computes the crude

and adjusted prevalence ratio and odds ratio, and their

interval estimates. A Breslow-Day test for homogeneity of

the prevalence ratio and odds ratio is applied to assist the

investigator in distinguishing between confounding factors

and interactions.

Case-control Estimates the odds ratio and its selected

approximate

logarithmic

and

intervals, and the attributable proportion and attributable

proportion in the exposed group.

test-based

confidence

Stratified case-control Makes the same calculations as

'Stratified cross-sectional' but for the odds ratio only.

Matched case-control Allows up to five controls per case

to be matched, and calculates the same parameters as'Case-

control'.

Cohort (cumulative incidence) Estimates the relative risk

and odds ratio, and their selected approximate logarithmic

and test-based confidence

incidence-derived data. The attributable risk, attributable

proportion and attributable proportion in the exposed

group are also calculated.

intervals, using cumulative-

Cohort (incidence rate) Estimates the relative risk and its

selected approximate logarithmic and test-based confidence

intervals by using true incidence rates, that is with 'animal-

time' denominators. The attributable

proportion and attributable proportion in the exposed

group are also calculated.

risk, attributable

Stratified cohort (cumulative incidence) Makes the same

calculations

cumulative incidence-derived relative risk replacing the

prevalence ratio.

as 'Stratified cross-sectional' but with the

Stratified cohort (incidence rate) Uses incidence rate-

derived data, rather than cumulative incidence values.

Models module

The final analytical module simulates the progress of a sim-

ple epidemic of infectious disease by using the Reed-Frost

model. The model requires the number of cases, the number

ofimmune animals and the number ofsusceptible animals at

the beginning of the epidemic, and the probability of infec-

tion (contact rate). The results can be displayed either as a

table listing the number of animals in each of the three cate-

gories over time, or graphically.

Help module

The last module,'Help', provides an index of the different

parts of the program. Each submodule also contains its own

'Help'. detailing the purpose of the submodule, the data

required for the calculations, the formulae used in the com-

putations, and guidance on the interpretation of the results.

Further reading is also sometimes listed.

The 'Help' of some submodules includes examples of the

use of the program, selected either from the users' manual

available with the earlier version of EPISCOPE (Frankena and

others 1990) or from a clinical veterinary epidemiology man-

ual (Ortega and others 1995).

Downloading the package from the Internet

WIN EPISCOPE 2.0 is available free from the websites of the

Veterinary Faculty of the University of Zaragoza (Spain) at

http://infecepi.unizar.es/pages/ratio/soft_uk.htm, Wageningen

Agricultural University (The Netherlands) at http://www.zod.

wau.nl/genr/epi.html, and the Royal Dick School of

Veterinary Studies, University of Edinburgh at http://www.

clive.ed.ac.uk/winepiscope. Each site provides instructions for

downloading the program.

RESULTS

Examples of some results generated by WIN EPISCOPE from

data derived from published studies are given below.

Example 1: Diagnostic test cut-off values-

diagnosis of Neospora caninum infection in dairy

cattle (Davison and others 1999)

Neospora caninum is now recognised as a major infectious

cause ofbovine abortion in many countries. In the UK a com-

The Veterinary Record, May 5, 2001

569

Page 4

PAPERS & ARTICLES

(a

FIG 2: Sample output

from Tests:Multiple

tests

FIG 1: Sample output from Tests: Cut-off value. (a) ROC curve,

(b) Results at a given cut-off

mercial antibody-detection ELISA (MAST Diagnostics) is used

to diagnose the infection. The results of this test are a con-

tinuous variable, per cent positivity (pp), and so various cut-

offvalues can be used to define an animal's status. The results

can be assessed by drawing a ROC curve describing the rela-

tionship between pairs of true-positive rates (sensitivity) on

the vertical axis, and false-positive rates (100-specificity) per

cent on the horizontal axis, for a range ofcut-offvalues. Tests

with ROC curves furthest into the top left-hand corner of the

graph are better tests, and, for a given test, the'best' cut-off

in terms of minimising the proportion of animals that are

misclassified, is the point on the ROC closest to the top left-

hand corner.

Fig Ia shows the ROC curve for the ELISA forN caninum for

cut-off values up to a maximum recorded value of 107 pp,

based on an investigation of 114 infected cattle and 199 unin-

fected cattle. This curve suggests that the test is good. However,

the cut-off values of 20, 25 and 30 PP are all close to the top

left-hand corner, and the most appropriate value can be

selected by determining the sensitivity and specificity for each

ofthem by altering the cut-offvalue; in this way the 20 PP cut-

off value is shown to maximise the sensitivity and specificity

(Fig lb). In practice, the selection ofthe most appropriate cut-

off value also depends on several other factors, including the

prevalence and the impact, for example, the cost of false-pos-

itive and false-negative results (Smith 1995).

Example 2: Multiple testing strategies-diagnosis

of hepatobiliary disorders in cats (based on Smith

1995, after Center and others 1986)

Feline hepatobiliary disorders are difficult to diagnose clini-

cally in their early stages, but high serum activities of alka-

line phosphatase (ALP) and y-glutamyl transferase (GGT), can

provide evidence of hepatic disease. However, an assessment

of 69 affected cats and 20 healthy animals showed that their

individual diagnostic sensitivities (50 per cent and 86 per

cent, respectively) are not high. Thus, the application ofeither

test alone would fail to diagnose the condition in 50 per cent

and 14 per cent of cases, respectively. Entry of their specifici-

ties (ALP 93 per cent, GGT 67 per cent) and sensitivities reveals

that the proportion of affected cats that would be diagnosed

with the condition, that is, the sensitivity of the test, would

increase to 93 per cent if both enzymes were measured (Fig

2), and that at the same time the negative predictive value (the

proportion of test-negative animals that are unaffected)

would increase. Both enzymes should therefore be measured

to rule out the disease.

Example 3: Detection of disease - Pan African

rinderpest campaign (IAEA 1991)

During the stage of serosurveillance in the Pan African rinder-

pest campaign, it was necessary to sample herds to determine

whether unvaccinated animals had seroconverted, that is, had

been exposed to natural infection. On the basis of previous

knowledge, an endemic seroprevalence of at least 5 per cent

was considered to be plausible in herds exposed to natural

infection. In one case, the campaigners needed to know how

many ofa herd of80 cattle had to be sampled to be 95 per cent

confident ofdetecting at least one seropositive animal, that is,

one ofthe four animals assumed to have been infected ifnat-

ural exposure had occurred. Forty-two animals were required

(Fig 3). The graph to the right of the figure shows the num-

bers ofanimals that would have been required ifdifferent lev-

els of endemic seroprevalence had been assumed.

Example 4: Sample size in unmatched case-

control studies - breed disposition to canine

pyometra (Niskanen and Thrusfield 1998)

The investigators wished to identify the breed disposition to

canine pyometra in the breeds of dog that constituted 1 per

cent or more ofthe dog population ofFinland. The minimum

breed-specific risk to be detected was an approximate two-

FIG 3: Sample output

from Samples:

Detection of disease

The Veterinary Record, May 5, 2001

570

Page 5

PAPERS & ARTICLES

FIG 4: Sample output

from Samples:

Unmatched case-

control studies

fold increase, relative to the risk in the total dog population,

with the level of confidence set at 95 per cent and a statisti-

cal power of 80 per cent. Clinical records of approximately

1100 cases were available.

Exploration of various values of the input parameters

indicated that approximately 10 controls would be required

for each case, that is, approximately 11,000 controls with a

two-tailed hypothesis (Fig 4).

Example 5: Sample size in cohort studies - the

relationship between sexual status and acquired

urinary incontinence in bitches (Thrusfield and

Holt 1998)

The possibility that the neutering ofbitches increases the risk

of acquired urinary incontinence was investigated in a five-

year cohort study. The investigators wished to detect a rela-

tive risk of at least 2, with the level ofconfidence and statistical

power set at 95 per cent and 80 per cent, respectively. The

examination of a clinic's records suggested that the cumula-

tive incidence of the condition in entire bitches during the

period ofthe study was likely to be about 1 per cent. The entry

ofthese parameters in the relevant submodule indicated that

approximately 2300 animals were required in each cohort

(neutered and entire) (Fig 5).

Example 6: Reed-Frost model-control of Johne's

disease (Collins and Morgan 1991 a)

Johne's disease (paratuberculosis) can be controlled at herd

level by a test-and-cull programme. The economic justifica-

FIG 6: Sample output from Models: Reed-Frost model

FIG 5: Sample output from Samples: Cohort studies

tion of a control programme depends on several factors,

including the prevalence of the infection, which can be mod-

elled over time by using the Reed-Frost model. This model

can investigate the effect ofchanging some ofthe input para-

meters, for example, the number of cattle initially infected

and the probability of calves becoming infected, to determine

to which ofthem the progress of infection in the herd is most

'sensitive'.

Fig 6 shows the progression of the infection in successive

years in a herd consisting of 30 infectious cows and 100 sus-

ceptible calves, with a contact rate (the probability that an

infected cow will infect a calf) of 0-025 (2-5 per cent). Cattle

do not become immune to paratuberculosis and so the

'immune' class in the model represents affected animals. The

results indicate that there is a peak of cases within the first

year, but that four ofthe calves remain uninfected after three

years. Further simulations indicate that the spread of paratu-

berculosis in infected herds is most sensitive to the contact

rate, supporting the recommendations that Johne's disease

should be controlled by minimising the contacts between

cows and calves.

The modellers refined the simulations to include, for

example, the risk ofpurchasing Mycobacterium paratubercu-

losis-infected cattle as replacements. The combination of the

results of the Reed-Frost model with the analysis of all other

relevant factors indicated that a test-and-cull control pro-

gramme should be profitable when the herd prevalence before

testing was more than 5 per cent (Collins and Morgan 199lb).

DISCUSSION

This new version of WIN EPISCOPE includes several submod-

ules which were not available in the earlier versions (Table 1)

and as a result it provides a more comprehensive package for

the design and analysis of epidemiological investigations.

Furthermore, data from other epidemiological packages such

as EPI INFO (Dean and others 1995), and databases and spread-

sheets such as MS ACCESS and MS EXCEL can be input into the

program.

The design of the program as a MSWindows application,

and the structure of the modules' screens, with menu-driven

data entry and results on the same screen, makes it easy to use.

The graphical presentation of some of the results, the avail-

ability of comprehensive'Help' pages detailing the formulae

that are used, the data that are required, and the interpreta-

The Veterinary Record, May 5, 2001

am=.

... MM-

571

VMMMIM

M-

Page 6

PAPERS & ARTICLES

tion that should be applied to the results, makes the program

valuable for teaching veterinary epidemiology, as more quan-

titative epidemiology is introduced into veterinary curricula

(WHO 1991).

WIN EPISCOPE 2.0 is not suitable for advanced epidemio-

logical studies, for example those requiring multivariate

analyses, which require not only more sophisticated statisti-

cal packages but also considerable statistical knowledge.

However, it should satisfy the main requirements for many

epidemiological studies and for the formal teaching of epi-

demiology at basic and intermediate levels.

The authors are preparing a manual of exercises to com-

plement the package, which will be included on the websites

from which it can be downloaded.

ACKNOWLEDGEMENTS

The authors thank the Government of Arag6n, Spain, for

supporting the development of WIN EPISCOPE 2.0 (project

CONSI+D). The programme is a part of the activities of the EPI-

DECON and RATIO networks. Dr Helen Davison, formerly of

the Centre for Tropical Veterinary Medicine, University of

Edinburgh, provided the raw data for the ROC curve.

I.................................................I

References

ARMITAGE, P. & BERRY, G. (1994) Statistical Methods in Medical Research.

3rd edn. Oxford, Blackwell Science

BRAND, A., NOORDHUIZEN, J. P. T. M. & SCHUKKEN, Y. H. (1996) Herd

Health and Production Management in Dairy Practice. Wageningen,

Wageningen Pers

CENTER, S. A., BALDWIN, B. H., DILLINGHAM, S., ERB, H. N. & TENNANT,

B. C. (1986) Diagnostic value of serum y-glutamyl transferase and alkaline

phosphatase activities in hepatobiliary disease in the cat. Journal of the

American Veterinary Medical Association 188, 507-510

COLLINS, M. T. &MORGAN, I. R. (1991a) Epidemiological model of paratu-

berculosis in cattle. Preventive Veterinary Medicine 11, 131-146

COLLINS, M. T. & MORGAN, I. R. (1991b) Economic decision analysis model

of a paratuberculosis test and cull program. Journal oftheAmerican Veterinary

Medical Association 199, 1724-1729

DAVISON, H. C., GREINER, M. & TREES, A. J. (1999) Quantitative analyses

of Neospora caninum serological data obtained from dairy cattle. Eds E. A.

Goodall, M. V. Thrusfield. In Proceedings of the Society for Veterinary

Epidemiology and Preventive Medicine, March 24 to 27, 1999. pp 172-181

DEAN, A. D., DEAN, J. A., COULOMBIER, D., BRENDEL, K. A., SMITH, D. C.,

BURTON, A. H., DICKER, R. C., SULLIVAN, K., FAGAN, R. F. & ARNER,

T. G. (1995) Epi Info Version 6: A Word-Processing, Database and Statistics

Program for Public Health on IBM-Compatible Micro-Computers. Atlanta,

Centers for Disease Control and Prevention

THE EPIDEMIOLOGY MONITOR. Software Inventory/Library. 2560 Whisper

Wind Court, Roswell, GA 30076, USA

FLEISS, J. L. (1981) Statistical Methods for Rates and Proportions. New York,

Wiley and Sons

FRANKENA, K., NOORDHUIZEN, J. P., WILLEBERG, P., VAN VOORTHUY-

SEN, P. F. & GOELEMA, J. 0. (1990) EPISCOPE: Computer programs in vet-

erinary epidemiology. Veterinary Record 126, 573-576

GAHLINGER, P. M. & ABRAMSON, J. H. (1995) Computer Programs for

Epidemiologic Analysis: PEPI. Stone Mountain, USD

IAEA (1991) The sero-monitoring of rinderpest throughout Africa. Phase one.

Proceedings

of

final

research

FAO/IAEA/SIDA/OAU/BAR/PARC

co-ordinated

Bingerville, Cote dIvoire, November 19-23, 1990. IAEA-TECDOC-623. Vienna,

International Atomic Energy Agency

KOUBA, V. (1997) Computerised methods for animal health risk assessment

using the EPIZOO 2.6 program. Revue Scientifique et Technique- Office

International des Epizooties 16, 793-799

LEVY, P. S. & LEMESHOW, S. (1991) Sampling of Populations: Methods and

Applications. New York, John Wiley & Sons

LWANGA, S. K. & LEMESHOW, S. (1991) Sample Size Determination in

Health Studies: a Practical Manual. Geneva, World Health Organization

MARTIN,

Epidemiology: Principles and Methods. Ames, Iowa State University Press

NISKANEN, M. & THRUSFIELD, M. V. (1998) Associations between age, par-

ity, hormonal therapy and breed, and pyometra in Finnish dogs. Veterinary

a

co-ordination

meeting

of

the

research

programme,

S. W., MEEK, A. H. & WILLEBERG,

P. (1987) Veterinary

Record143, 493-498

NOORDHUIZEN, J. P. T. M. (1996) Introduction to the EPIDECON course

'Models and Quantitative Methods

Application ofQuantitative Methods in Veterinary Epidemiology, Advanced

Seminar, October

1996. Zaragoza, Instituto Agronomico

Mediterraneo de Zaragoza. I- 1-I-5

NOORDHUIZEN, J. P. T. M., FRANKENA, K., VAN DER HOOFD, C. M. &

GRAAT,E. A. M.(1997) Application ofQuantitative Methods inVeterinary

Epidemiology. Wageningen, Wageningen Pers

ORTEGA, C., DE BLAS, I., FRANKENA, K. & NOORDHUIZEN, J. P. T. M.

(1996) WIN EPISCOPE 1.0: su aplicacion en investigaciones epidemiologicas

de tipo cuantitativo. XIV Reunion Cientifica de la Sociedad Espanola de

Epidemiologia (SEE), Zaragoza. pp 1-3

ORTEGA, C., NOORDHUIZEN, J. P. T.M., STARK, K. D. C. &THRUSFIELD,

M. (1995) Manual de Epidemiologia Aplicada Para el Veterinario Clinico.

Barcelona, Esteve-Temas de Veterinaria

ROTHMAN, K. J. (1986) Modern Epidemiology. Boston and Toronto, Little

Brown

SCHLESSELMAN, J. J. (1982) Case-Control Studies: Design, Conduct,Analysis.

NewYork, Oxford University Press

SMITH, R. D. (1995) Veterinary Clinical Epidemiology. A Problem-Oriented

Approach. 2nd edn. Boca Raton, CRC Press. pp 42-43, 59-60,62-64

SNEDECOR, G. W. & COCHRAN, W. G. (1980) Statistical Methods. 7th edn.

Ames, Iowa State University Press

THRUSFIELD, M. (1997) Veterinary Epidemiology. Revised 2nd edn. Oxford,

Blackwell Science

THRUSFIELD, M. V. (1998) Epidemiology-prospects in the 21st century. In

Towards Livestock Disease Diagnosis and Control in the 21st Century.

Proceedings of an international symposium on the diagnosis and control of

livestock diseases using nuclear and related techniques. International Atomic

Energy Agency/Food and Agriculture Organization. Vienna, April 7 to 11,

1997.Vienna, International Atomic Energy Agency. pp 269-289

THRUSFIELD, M. V. & HOLT, P. E. (1998) Acquired urinary incontinence in

bitches: its incidence and relationship to neutering practices. Journal ofSmall

Animal Practice 39, 559-566

WHO (1991) Consultation on development and training in veterinary epi-

demiology, October 9 to 11, 1990, Hanover, FRG. WHO/CDS/VPH/91.76.

Geneva, World Health Organization. pp 1-28

in Veterinary Epidemiology'

In

14 to 25,

_ ABSTRACT

Lufenuron for treating fungal

infections of dogs and cats

THE results of giving 138 dogs and 159 cats a single oral dose

of lufenuron for the treatment of dermatophytosis or super-

ficial dermatomycoses were assessed by comparing them with

18 dogs and 42 cats with the same conditions which were not

treated. In the treated dogs the mean times to obtaining neg-

ative fungal culture results and to the resolution of the gross

lesions were 14-5 and 20-8 days, respectively, and in the cats

the comparable periods were 8-3 and 12 days. There were no

adverse reactions to the drug. In the control animals the

lesions took approximately 90 days to resolve.

BEN-ZIONY, Y. & ARZI, B. (2000) Use of lufenuron for treating fungal infec-

tions of dogs and cats: 297 cases (1997-1999). Journal of the American

Veterinary Medical Association 217, 1510-1513

The Veterinary Record, May 5, 2001

572