www.elsevier.com/locate/vetmic

Veterinary Microbiology 110 (2005) 17–26

Spatial and temporal patterns of pig herds diagnosed with

Postweaning Multisystemic Wasting Syndrome (PMWS)

during the first two years of its occurrence in Denmark

Hakan Vigre a,*, Poul Bækbo b, Sven Erik Jorsal c, Vivi Bille-Hansen c,Anne-Grete Hassing b, Claes Enøe d, Anette Bøtner e

a The Danish Institute for Food and Veterinary Research, Department of Epidemiology and Risk Assessment,

Mørkhøj Bygade 19, DK-2860 Søborg, Denmarkb The National Committee for Pig Production, Danish Bacon & Meat Council, Vinkelvej 11,

DK-8620 Kjellerup, Denmarkc The Danish Institute for Food and Veterinary Research, Department of Veterinary Diagnostics and Research,

Bulowsvej 27, DK-1790 Copenhagen V, Denmarkd The national Committee for Pig Production, Danish Bacon and Meat Council, Axeltorv 3,

DK-1609 Copenhagen V, Denmarke The Danish Institute for Food and Veterinary Research, Department of Virology,

Lindholm, DK-4771 Kalvehave, Denmark

Received 31 January 2005; received in revised form 21 June 2005; accepted 11 July 2005

Abstract

The clinical syndrome Postweaning Multisystemic Wasting Syndrome (PMWS) in pigs has emerged globally during the last

decade. In October 2001, the first pig herd diagnosed with PMWS was reported in Denmark, and since then the number of herds

diagnosed with PMWS has increased markedly.

The etiology of PMWS is not well understood, but increased knowledge of the causal factors is prerequisite for applying

preventive interventions. In this study we described the temporal (time of diagnosis), spatial (location of herds) and spatio-

temporal pattern of Danish pig herds diagnosed with PMWS during the first two years after the first herd was diagnosed, and we

tested for spatial and spatio-temporal clustering using scan statistics.

The study population consisted of pig herds that during the study period (October 2001–September 2003) performed

diagnostic submissions to the two major veterinary diagnostic laboratories in Denmark (6724 herds). Of these, 277 herds were

diagnosed with PMWS. Two statistically significant spatial clusters of herds diagnosed with PMWS were identified. These

clusters included 11% and 8% of the study herds, respectively. Within these two clusters the relative risk for a herd to be

diagnosed with PMWS was twice as high as expected. One statistically significant spatio-temporal cluster was identified

between February and May 2002.

* Corresponding author. Tel.: +45 72347323.

E-mail address: hvi@dfvf.dk (H. Vigre).

0378-1135/$ – see front matter # 2005 Elsevier B.V. All rights reserved.

doi:10.1016/j.vetmic.2005.07.001

H. Vigre et al. / Veterinary Microbiology 110 (2005) 17–2618

We discuss different hypotheses that could explain why pig herds diagnosed with PMWS were clustered both spatially and

spatio-temporally, and conclude that the results support the hypothesis that PMWS is caused by introduction of a new,

unidentified, pathogen into the Danish pig production.

# 2005 Elsevier B.V. All rights reserved.

Keywords: PMWS; Denmark; Spatial; Temporal; Clustering; SaTScan

1. Introduction

Postweaning multisystemic wasting syndrome

(PMWS) is a recently described clinical condition

characterised by unthriftiness, wasting and increased

mortality among pigs in the post-weaned age group.

Within affected herds, also coughing, dyspnoea and

diarrhoea are frequent observed clinical signs. PMWS

was first identified in Canada in 1991 (Clark, 1997).

Since then, the distribution of PMWS has grown and

the syndrome is recognized worldwide. In Denmark,

both the veterinary practitioners and the diagnostic

laboratories have been aware of the clinical symptoms

and laboratory findings associated with PMWS since

the beginning of 2000 (Hassing et al., 2002), and in

October 2001, the first herd with PMWS was

diagnosed. Subsequently, all herds diagnosed with

PMWS have been registered at the Danish Institute of

Food and Veterinary Research, and between 2001 and

2003 the number of herds diagnosed with PMWS

increased markedly.

Infection with porcine circovirus type-2 (PCV2) is

necessary for development of PMWS (Kennedy et al.,

2000; Krakowka et al., 2000; Ladekjær-Mikkelsen

et al., 2002). Infection with PCV2 is distributed

worldwide and subclinical infection probably occurs

in almost all pig herds (Pogranichniy et al., 2002;

Boisseson et al., 2004; Sibila et al., 2004). This

indicates that also other pathogens or factors other

than infection with PCV2 are necessary for develop-

ment of PMWS.

However, knowledge of the epidemiology of

PMWS and factors triggering the clinical appearance

of the syndrome is limited. Previous studies in animal

diseases have shown that spatio-temporal investiga-

tion of disease patterns can generate and support

hypotheses with respect to etiology and transmission

(outbreak of acute respiratory disease at cattle herds –

Nordstrom et al., 2000; leptospirosis among dogs –

Ward, 2002). In this paper we describe the temporal

(time of diagnosis), spatial (location of herds) and

spatio-temporal (interaction between space and time)

pattern of Danish pig herds diagnosed with PMWS

during the first two years after the first herd was

diagnosed. We also tested for spatial and spatio-

temporal clustering using scan statistics. Subse-

quently, we discuss three possible hypotheses that

can explain the pattern of the occurrence of the

disease.

2. Materials and methods

2.1. Study area (Denmark) and target population

(the Danish pig population)

The area of Denmark is 43,090 km2 and consists of

the peninsula Jutland, the major islands Funen,

Sealand, Lolland, Falster and Bornholm plus a number

of smaller islands. The Danish pig production is

characterised by relatively intensive production

systems and produces over 21 million slaughter pigs

annually (Danish Bacon and Meat Council, 2003). In

October 2001, 21,348 pig farms were registered in the

Danish Central Husbandry Register (CHR). (How-

ever, the CHR has a lag phase in recording closing

down pig herds, and according to Danmarks Statistik

(2003) in 2001 there were 12,936 pig herds in

Denmark.) The pig herd density in each of the 276

municipalities in Denmark (number of pig herds

registered in CHR located in the municipality divided

by the area (km2) of the municipality) is presented in

Fig. 1.

2.2. Study design

We used a retrospective longitudinal study to

describe the temporal, spatial and spatio-temporal

pattern of herds diagnosed with PMWS in Denmark

and to analyse for presence of spatial and spatio-

H. Vigre et al. / Veterinary Microbiology 110 (2005) 17–26 19

Fig. 1. Geographical map of Denmark and neighbouring countries, including information on the pig herd density (number of pig herds registered

in CHR per km2) in the 276 municipalities in Denmark.

temporal clusters of herds diagnosed with PMWS. The

study period started at the beginning of October 2001,

where the first herd with PMWS was diagnosed and

ended at the end of September 2003.

Theoretically, all pig herds in Denmark should have

been included in this study. However, not all pig herds

do have equal probability of submitting diagnostic

samples for laboratory examinations. Therefore, the

study population – study herds – consisted of the 6724

pig herds that between the beginning of October 2001

and the end of September 2003 performed one or

several diagnostic submissions either to the laboratory

run by the Danish Institute of Food and Veterinary

Research (DFVF) located at Sealand or to the

laboratory in Kjellerup, Jutland run by the Danish

Bacon and Meat Council. These two laboratories

process the majority (approximately 99%) of all

diagnostic samples submitted from Danish pig herds.

2.3. Definition of the diagnose PMWS

In this study the diagnosis of PMWS was made at

herd level. A herd was diagnosed with PMWS when

multinucleated cells and/or intracytoplasmatic inclu-

sions (histopathology) plus PCV2 (immunostaining of

cryostat sections) were detected in lymphoid organs of

pigs submitted for laboratory examination – either

with the purpose of specific examination for PMWS,

or with a disease description and post mortem findings

indicative of PMWS. The aggregation of diagnostic

information was done for all samples from pigs that

were submitted to the laboratories. Histopathology

and immunohistochemistry was performed at DFVF

according to previously described methods (Ladekjær-

Mikkelsen et al., 2002).

2.4. Geographical location of herds

The geographical locations of the pig herds were

obtained by merging the addresses of the herds to

geographical coordinates (National Survey and

Cadastre, 2004). The coordinates of the addresses

of 68 herds (of 6724 study herds) could not be

retrieved and these herds were excluded from the

statistical analysis. None of the 68 herds had a history

of PMWS.

H. Vigre et al. / Veterinary Microbiology 110 (2005) 17–2620

2.5. Collection of data concerning herd

characteristics

Data of herd type and herd size was extracted from

the Danish Central Husbandry Register (CHR). The

Danish SPF (specific pathogen free) Company

provided data concerning health status.

2.6. Descriptive spatial and temporal analysis

The laboratory data and the geographical informa-

tion were organised in one record per submission per

herd (the data contain one or several records per herd).

Each record was classified as a case (submission

fulfilling our criteria of PMWS) or a control

(submission not fulfilling our criteria of PMWS).

The case and control time, respectively, was defined as

the date of receiving the material for diagnostic

examination at the laboratories. Note that a herd could

be classified as control herd at one or multiple

submissions, and then in a later submission as a case

herd. After a herd was classified as a case herd, it was

censored from the study. Case and control time was

used in computation of month-specific proportion of

herds diagnosed with PMWS. Case time was also used

in the analysis of spatio-temporal clustering.

Data were organised, edited, manipulated and

summarised using the software SAS1, version 8

(Statistical Analysis System, 1999). From SAS, data

were exported as dBASE file for mapping purposes.

Using ArcView 8.3 GIS software (ESRI, 2002,

Redlands, CA) we produced maps displaying the

spatio-temporal distribution of herds diagnosed with

PMWS.

2.7. Analysing spatial and spatio-temporal

clustering

To investigate if the incidence of herds diagnosed

with PMWS were particularly high in some geogra-

phical areas during the study period (spatial cluster-

ing) and in some geographical areas during some

specific time-periods of the study period (spatio-

temporal clustering) we used scan statistics (Kulldorff,

1997). In the analysis we assumed that the herds

diagnosed with PMWS is a representative and an

unbiased spatial and temporal sample of all herds with

onset of PMWS in Denmark during the study period.

The data were organized into a case-file, a control-

file and a geographical coordinate-file and used as

input data in the spatio-temporal scanning software

SaTScan version 4.0 (Kulldorff, 2003). In this study

we used two different models to analyse the data – a

spatial Bernoulli model for case and control (non-

case) data and a space–time permutation model

utilising only case data.

In the spatial Bernoulli model, by scanning for

areas with significantly high proportions of cases we

investigated the data for spatial clusters, ignoring the

time of the diagnosis. The null-hypothesis in the

spatial Bernoulli model is that the number of cases in

any given area is binomially distributed with a

common prevalence parameter. By using scan statistic

we took the uneven density of pig farms in Denmark

into account when identifying spatial clusters of the

disease – a potential cause of confounding.

The spatial analysis in SaTScan imposes a circular

window on the map. In our analysis, the window was

in turn centered on the coordinates of each study herd

(both case and control). For each location, the radius

of the window varies in size from 0 to 50% of the total

population at risk.

To identify statistically significant spatial clusters

the likelihood ratio test was used. The likelihood

function was computed for each specific window, and

the one with the maximum likelihood constitute the

most likely cluster. The likelihood functions used in

this study and the likelihood ratio test are described

fully by Kulldorff (2003). Briefly, in the Bernoulli

model the value of the likelihood function for a

window increases by an increased difference between

proportion of cases within the window and outside the

window. The statistical significance of the identified

clusters was evaluated by comparing the values of the

likelihood function of identified clusters with the

maximum likelihoods in 999 random replicates of the

data set. The generation of random replicates was done

usingMonte Carlo sampling under the null-hypothesis

that the number of cases in any given area was

binomially distributed with a proportion equal to (total

number of cases/total number of study herds). The P-

value for each cluster identified in the real data was

obtained by comparing the likelihood value with the

maximum likelihoods from the random data sets. If the

likelihood value for a cluster identified in the real data

was in the most extreme 5% of all maximum

H. Vigre et al. / Veterinary Microbiology 110 (2005) 17–26 21

Table 1

Number and distribution of Danish pig herds diagnosed with PMWS (October 2001–September 2003) and distribution of all Danish pig herds,

respectively, according to herd type and health status

Pig herds diagnosed with PMWS All Danish pig herds

No. of herds Distribution (%) Distribution (%)

Herd typea

Breeding and multiplying herd 3 1.1 1.6

Conventional production herd 263 94.9 92.3

Outdoor/organic herd 11 4.0 6.1

Health statusb

SPFc 30 10.8 8.7

MSd 85 30.7 21.2

Conventionale 162 58.5 70.1

a Herd type recorded in the Danish Central Husbandry Register.b Health status recorded in the Danish SPF Company.c Declared free from Enzootic pneumonia, Pleuropneumonia and Atrophic rhinitis.d Declared free from Pleuropneumonia and Atrophic rhinitis.e No disease declaration.

likelihoods, then the identified cluster was statistically

significant (Kulldorff, 2003).

In the spatial Bernoulli model, the relative risk

(RR) to be diagnosed with PMWS for a herd within an

identified cluster was calculated as the ratio of the

number of cases observed in the cluster and the

number of cases expected, assuming that the location

of herds diagnosed with PMWS was randomly

distributed.

Given the increase in incidence over the study

period, performing a space–time scan statistic to test

for spatio-temporal clusters would pick up the

increasing trend by assigning clusters during the

end of the study. Instead, to adjust the scan statistic for

the temporal trend we performed a space–time

permutation model (Kulldorff, 2003). In a space–time

permutation model, the number of observed cases in a

cluster is compared to what would have been expected

if there was no spatio-temporal interaction. Therefore,

we get a spatio-temporal cluster in a geographical area

if, during a specific time period, that area has a high

proportion of excess cases compared to surrounding

areas (the likelihood function of a spatio-temporal

window increases by an increased proportion of

excess cases in a specific area compared to surround-

ing area within a specific time-period). The temporal

dimension in our analysis was expressed with duration

of one month as the basic unit of analysis to reduce

computing time. The maximum spatial and the

maximum temporal cluster size were set to 50% of

the total population at risk and total study period,

respectively.

The P-value for each spatio-temporal cluster

identified in the real data by the space–time

permutation model was obtained following a similar

procedure as described for spatial clusters.

3. Results

Among the 6724 study herds, 277 herds were

diagnosed with PMWS (4% of the study herds)

according to the criteria outlined in this paper. The

most frequent number of pigs, from which diagnostic

samples were submitted for examination of PMWS

was five, but ranged between one and eight. The

distribution of the 277 pig herds diagnosed with

PMWS according to herd type and health status,

respectively, was comparable to the overall health

status of all Danish pig herds (Table 1). Among the

herds diagnosed with PMWS, 15.5% were specialised

in rearing growers and 12.6% reared only finishing

pigs. The number of weaners at farms diagnosed with

PMWS (Q1 = 600, median = 900, Q3 = 1500) was

slightly higher than the number of weaners in other

farms (Q1 = 500, median = 800, Q3 = 1300). The

month-specific proportion of herds diagnosed with

PMWS among the herds submitting samples to the

laboratories the month of concern is presented in

Fig. 2.

H. Vigre et al. / Veterinary Microbiology 110 (2005) 17–2622

Fig. 2. The month-specific percent of pig herds diagnosed with

PMWS among herds submitting samples for laboratory examination

the month of concern. The herds submitting samples include pig

herds submitting to the laboratories involved in this study.

The spatio-temporal pattern of the herds diagnosed

with PMWS is presented in Fig. 3. Around one year

passed from the first reported case until PMWS was

diagnosed in most parts of Denmark. However, no

cases were reported at Lolland, Falster or Bornholm

during the study period.

Two significant spatial clusters (P-value = 0.001,

0.031) of herds diagnosed with PMWS were identified

(Fig. 4). The spatial cluster located in south Jutland

has a radius of 49 km and included 732 study herds

(11% of the study population). Out of these herds 75

were diagnosed with PMWS during the study period.

The spatial cluster located in northwest Jutland has a

radius of 31 km and included 541 study herds (8% of

the study population). Out of these herds, 48 were

diagnosed with PMWS during the study period.

Within these two clusters the relative risk for a herd to

be diagnosed with PMWS was twice as high as

expected (RR: 2.5 and 2.1, respectively).

In the space–time permutation model one sig-

nificant spatio-temporal cluster was identified (P-

value = 0.001) in northwest Jutland between February

and May 2002 (Fig. 4). Within this geographical area

during this period there were 11 herds diagnosed with

PMWS, whereas under the hypothesis of no space–

time interaction only one case was expected. The

cluster was located within one of the significant spatial

clusters identified in the spatial analysis. The

characteristics concerning herd type, herd size and

health status of these 11 herds were not different

compared to other herds diagnosed with PWMS or to

the overall characteristics of Danish pig herds.

4. Discussion

This study describes and analyses the spatial and

temporal patterns of herds diagnosed with PMWS

during the first two years (October 2001–September

2003) of its occurrence in the Danish pig production.

However, PMWS has probably occurred in Denmark

before 2001. One year before, two herds were

clinically suspected for PMWS, but the histopatho-

logical criteria as previously outlined in this paper

were not completely fulfilled in pigs examined from

those herds (Hassing et al., 2002). Recently, at DFVF

we detected histopathological changes consistent with

PMWS as well as PCV2 in formalin fixed tissue

samples from a pig from 1989. The samples came

from a pig herd affected with wasting, pneumonia and

diarrhoea. So, PMWS probably may have occurred

long before 2001, but during 2001–2003 there seemed

to be a significant increase in the occurrence of PMWS

in some part of the country.

We used scan statistic to identify spatial and spatio-

temporal clusters of herds diagnosed with PMWS.

Generally, with scan statistics it is possible to pinpoint

the general location and duration of a cluster. However,

the exact boundaries of the clusters remain unclear.

The most important results in this study were:

(i) D

uring the first two years the incidence of herds

diagnosed with PMWS increased markedly and

unpredictably.

(ii) I

n the spatial Bernoulli model, two geographical

clusters were identified. This indicated that

during the first two years with PMWS in the

Danish pig production the risk for herds to be

diagnosed with PMWS was significantly higher

in these two geographical areas compared to the

remaining part of the study population.

(iii) I

n the space–time permutation model, one spatio-

temporal cluster was identified. In herds within

this area, the relative risk of being diagnosed with

PMWS during the first half of 2002 compared to

other areas was significantly higher than during

other time-periods.

H. Vigre et al. / Veterinary Microbiology 110 (2005) 17–26 23

Fig. 3. Map of Denmark showing the location of herds that was diagnosed with PMWS within four different periods of the study period.

Below, we discuss three hypotheses that might

explain these results.

4.1. Hypothesis 1: introduction of a new pathogen

The identification of a significant spatio-temporal

cluster early in the apparent disease outbreak

supports the hypothesis that PMWS is caused by a

new pathogen that has been introduced to one or

several pig herds in Denmark. After a local outbreak,

the pathogen has been spread to most parts of

Denmark during the first two years after introduc-

tion. The hypothesis is further supported by the

result obtained in the spatial model – the two

identified geographical clusters are located in the

geographical areas with the highest pig herd

densities in Denmark. If the transmission of a

potential new pathogen is partly airborne, then the

H. Vigre et al. / Veterinary Microbiology 110 (2005) 17–2624

Fig. 4. The map of Denmark with the results from the SaTScan

analysis showing (i) the geographical location of the statistically

significant spatial clusters (during the whole study period) as black

circles, and (ii) the geographical location of the statistically sig-

nificant spatio-temporal cluster as a grey circle. Herds diagnosed

with PMWS during the study period are shown on the map as black

dots.

infectious pressure – and subsequently the risk of

onset with PMWS – will be higher in geographical

areas with high density of pig herds compared to

other geographical areas. Because most veterinary

practitioners work in specific geographical areas

transmission of a new pathogen by the veterinarian

might also result in spatial and spatio-temporal

clustering of the syndrome. However, since most

pig herds in Denmark perform sanitary routines at the

level of SPF herds, the risk of transmitting infectious

agents passively from herd to herd by personnel is

expected to be very small. The geographically

independent increased incidence of onset of PMWS

throughout Denmark can be ascribed to transmission

of a pathogen by movement of living pigs between

herds.

Hypothesis 1 is also supported by the result in a

previous study investigating the impact of potential

risk factors upon the occurrence of PMWS (Cook

et al., 2001). In that study the risk of PMWS was

significantly higher in pig herds in close proximity

(within 3 km) to other affected herds. Also larger

herds and those that purchased greater numbers of

replacement breeding stock were more likely to report

PMWS.

Even though the results strongly support the

hypothesis that the spread of PMWS is related to

introduction of a new pathogen, there is so far no

reported microbiological evidence confirming this.

Instead, the molecular characterisation of PCV2

strains from PMWS-affected and non-affected pig

herds has not identified any molecular marker of

virulence in PCV2 that can explain the sudden

outbreak and spread throughout the world (Pogra-

nichniy et al., 2002; Boisseson et al., 2004). These

studies converge on the conclusion that the PMWS

epidemics do not originate in the appearance of a new

PCV2 variant, but is caused by strains with endemic

occurrence in the pig industry. Also, in an experi-

mental study, pigs inoculated with PCV2 strains from

non PMWS-affected herds developed clinical signs of

PMWS (Allan et al., 2002). Therefore Allan et al.

(2004) speculated whether it is possible that changes

in the global pig industry have resulted in up-

regulating a common sub-clinical PCV2 infection to

PMWS. This leads us to the second hypothesis.

4.2. Hypothesis 2: change in management

Even if events of diseases may be found to cluster in

space and/or time, the clustering is not necessarily a

result of a contagious process involving local transmis-

sion (Carpenter, 2001). Another hypothesis that might

explain the apparent outbreak of PMWS is that a

progressive change of management factors in pig herds

throughout Denmark have resulted in occurrence of a

new clinical syndrome – PMWS – whereas the causal

agent – PCV2 – has been present in the pig population

for several years. This hypothesis is supported by

observations from the Swedish pig production (Wallg-

ren et al., 2004) that revealed no evident links (trade or

close distance) between affected herds, but when

PMWS was diagnosed there were always significant

management problems on the farms.

However, it is difficult to explain from this hy-

pothesis the presence of significant spatial and spatio-

temporal clusters of herds diagnosedwith PMWS in the

western part of Denmark, and that PMWS was almost

absent in the south-eastern part of Denmark during the

first two years. We found that the distribution of pig

herds diagnosed with PMWS according to herd type

and health status was similar to the overall distribution

of Danish pig herds. Further, wewould have expected a

H. Vigre et al. / Veterinary Microbiology 110 (2005) 17–26 25

more even geographical distribution of the disease

syndrome if it was related solely to management

factors.

Nevertheless, significant effect of different manage-

ment factors on PMWShas been reported byRose et al.

(2003). They found that large pens in weaning facilities

increased the odds for PMWS, whereas long empty

periods in farrowing and weaning facilities decreased

the odds for PMWS. These results are in accordance

with Danish experience from the field concerning

treatment of clinical PMWS. This indicates that the

degree of problems with PMWS depends on manage-

ment factors,which againmight be in accordancewith a

disease syndrome with multifactorial causes.

4.3. Hypothesis 3: spatial and temporal variation

in sensitivity of the diagnostic procedures

In the data analysis and the discussion whether

PMWS is caused by a new pathogen or change in

management, we assume that the herds diagnosed with

PMWS at the laboratories represented a random

spatial and temporal sample of all pig herds that

utilised the laboratories during the study period.

However, the spatial and temporal pattern of herds

diagnosed with PMWS might be a result of variation

in sensitivity in the diagnostic procedure.

In Denmark, most veterinary practitioners work in

specific geographical areas, and variation in how

intensively veterinary practitioners utilize the

laboratories, might bias the result of the spatial

analysis. By using only herds from which diagnostic

samples were submitted to the laboratories during

the study period we partly adjusted for that bias.

However, because our study was carried out by

exclusively using herds that had submitted diag-

nostic samples to the two laboratories there is a

potential selection bias because herds with other

health problems than PMWS or herds that utilize the

laboratories might not be evenly distributed in

Denmark. In areas with a general low frequency of

submitting diagnostic samples there might be a

relatively higher proportion (spatial clustering) of

herds diagnosed with PMWS compared to areas

were the submission frequency is higher. To evaluate

the importance of this potential bias, we reanalysed

the Bernoulli model using all herds registered in

CHR as control. Two spatial clusters were identified in

this analysis, and these clusters were geographi-

cally distributed almost equal to the two clusters

identified in the Bernoulli model using the herds

from which diagnostic samples were submitted as

controls. This similarity indicates that the size of the

potential bias is limited.

However, due to the inconvenience related to the

PMWS diagnosis for some herds, there tend to be

some reluctance to submit material for laboratory

examination. A geographical related difference in

attitude to submitting material for laboratory exam-

inations of PMWS could not be completely excluded.

The temporal pattern of an increased incidence of

herds diagnosed with PMWS may be a direct result of

lack of sensitivity in the diagnostic procedures of

PMWS early in the course of the outbreak – especially

concerning the awareness of the disease and lack of

clinical experiences among the veterinary practi-

tioners. In general, identification of disease outbreaks

are often characterised by a lag phase, ranging from

few days to several months. The lag phase is a

consequence of initial low levels of suspicion of the

disease. Later in the course of an outbreak the intensity

of reporting often increased because of increased

awareness of the disease resulting in more submis-

sions for laboratory diagnostic examination. However,

at the same time confirmation rates may have

decreased (and subsequently also the registered

number of herds diagnosed with PMWS) because of

increased confidence in clinical diagnosis, which does

not request additional laboratory diagnostic examina-

tion.

5. Conclusion

The spatial and spatio-temporal analysis in this

study were not able to differentiate definitively

between presented hypotheses, because, spatial and

temporal variation in the sensitivity of the diagnostic

procedures of the disease might have contributed to

some extent in the pattern of spread of PMWS through

space and time. However, we conclude that the study

results support the hypothesis that PMWS is caused by

introduction a new, yet unidentified, pathogen into the

Danish pig production more than they support the

hypothesis that PMWS is caused by changes in

management.

H. Vigre et al. / Veterinary Microbiology 110 (2005) 17–2626

Acknowledgement

The authors thank Birgitta Svensmark and Kristian

Barfod at the Danish Bacon and Meat Council for

supporting the project by making data from the

laboratory, run by Danish Bacon and Meat Council at

Kjellerup, available.

References

Allan, G.M., McNeilly, F., Meehan, B., Kennedy, S., Johnston, D.,

Ellis, J., Krakowka, S., Fossum, C., Wattrang, E., Wallgren, P.,

2002. Reproduction of PMWS with a 1993 Swedish isolate of

PCV-2. Vet. Rec. 150, 255–256.

Allan, G.M., McNeilly, F., Ellis, J., Krakowka, S., Bøtner, A.,

McCullough, K., Nauwynck, H., Kennedy, S., Meehan, B.,

Charreyre, C., 2004. PMWS: experimental model and co-infec-

tions. Vet. Microbiol. 98, 165–168.

Boisseson, C., Beven, V., Bigarre, L., Thiery, R., Rose, N., Eveno,

E., Madec, F., Jestin, A., 2004. Molecular characterization of

Porcine circovirus type 2 isolates from post-weaning multi-

systemic wasting syndrome-affected and non-affected pigs. J.

Gen. Virol. 85, 293–304.

Carpenter, T.E., 2001. Methods to investigate spatial and temporal

clustering in veterinary epidemiology. Prev. Vet. Med. 48, 303–

320.

Clark, E.G., 1997. Post-weaning multisystemic wasting syndrome.

In: Proceedings of the American Association of Swine Practi-

tioners, pp. 499–501.

Cook, A.J.C., Pascoe, S.R., Wilsmith, J.W., Gresham, A.C.J., 2001.

A case:control study of Post-weaning Multi-systemic Wasting

Syndrome (PMWS) and Porcine Dermatitis Nepropathy Syn-

drome (PDNS). Pig J. 48, 53–60.

Danish Bacon and Meat Council, statistics 2003. November

21, 2004. http://www.danskeslagterier.dk/smmedia/UK_stat_

pdf?mb_GUID=254A7C6F-508C-4B9A-8009-

63FF0AD0893C.pdf.

Danmarks Statistik, 2003. November 21, 2004. http://www.stat-

bank.dk/statbank5a/default.asp?w=1024.

Hassing, A.G., Bøtner, A., Ladekjær-Mikkelsen, A.-S., Bækbo, P.,

Jorsal, S.E., Bille-Hansen, V., 2002. Postweaning multisystemic

wasting syndrome in Denmark. In: Proceedings of the 17th

Congress of the International Pig Veterinary Society, June 2–5,

2002. Ames, Iowa, USA.

Kennedy, S., Moffett, D., McNeilly, F., Meehan, B., Ellis, J.,

Krakowka, S., Allan, G.M., 2000. Reproduction of lesions of

postweaning multisystemic wasting syndrome by infection of

conventional pigs with porcine circovirus type 2 alone or in

combination with porcine parvovirus. J. Comp. Pathol. 122, 9–

24.

Krakowka, S., Ellis, J.A., Meehan, B., Kennedy, S., McNeilly, F.,

Allan, G., 2000. Viral wasting syndrome of swine: experimental

reproduction of postweaning multisystemic wasting syndrome

in gnotobiotic swine by coinfection with porcine circovirus 2

and porcine parvovirus. Vet. Pathol. 37, 254–263.

Kulldorff, M., 1997. A spatial scan statistic. Commun. Stat.: Theory

Methods 26, 1481–1496.

Kulldorff, M., 2003. SaTScan v4.0 Software for the spatial and

space-time scan statistics. Information Management Services,

Inc. http://www.satscan.org/.

Ladekjær-Mikkelsen, A.S., Nielssen, J., Stadejek, T., Storgaard, T.,

Krakowka, S., Ellis, J., McNeilly, F., Allan, G., Bøtner, A., 2002.

Reproduction of postweaning multisystemic wasting syndrome

(PMWS) in immunostimulated and non-immunostimulated 3-

week-old piglets experimentally infected with porcine circovirus

type 2 (PCV2). Vet. Microbiol. 89, 97–114.

National Survey and Cadastre, 2004. Copenhagen, Denmark.

Nordstrom, M., Pfeiffer, D.U., Jarp, J., 2000. A space-time cluster

investigation of an outbreak of acute respiratory disease in

Norwegian cattle herds. Prev. Vet. Med. 47, 107–119.

Pogranichniy, R.M., Yoon, K.J., Harms, P.A., Sorden, S.D., Daniels,

M., 2002. Case-control study on the association of porcine

circovirus type 2 and other swine viral pathogens with post-

weaning multisystemic wasting syndrome. J. Vet. Diagn. Invest.

14, 449–456.

Rose, N., Larour, G., Le Diguerher, G., Eveno, E., Jolly, J.P.,

Blanchard, P., Oger, A., Le Dimna, M., Jestin, A., Madec, F.,

2003. Risk factors for porcine post-weaning multisystemic

wasting syndrome (PMWS) in 149 French farrow-to-finish

herds. Prev. Vet. Med. 61, 209–225.

Sibila, M., Calsamiglia, M., Segales, J., Blanchard, P., Badiella, L.,

le Dimna, M., Jestin, A., Domingo, M., le Dimna, M., 2004. Use

of a polymerase chain reaction assay and an ELISA to monitor

porcine circovirus type 2 infection in pigs from farms with and

without postweaning multisystemic wasting syndrome. Am. J.

Vet. Res. 65, 88–92.

Wallgren, P., Hasslung, F., Bergstrom, G., Linder, A., Belak, K.,

Hard af Segerstad, C., Stampe, M., Molander, B., Bjornberg

Kallay, T., Norregard, E., Ehlorsson, C.J., Tornquist, M., Fos-

sum, C., Allan, G.M., Robertsson, J.A., 2004. Postweaning

multisystemic wasting syndrome—PMWS. The first year with

the disease in Sweden. Vet. Q. 26, 170–187.

Ward, M.P., 2002. Clustering of reported cases of leptospirosis

among dogs in the United States and Canada. Prev. Vet. Med.

56, 215–226.