Preventive Veterinary Medicine 54 (2002) 265±278

A meta-analysis comparing the effect of vaccines

against Mycoplasma hyopneumoniae on
daily weight gain in pigs
C.S. Jensena,b,*, A.K. Ersbùllc, J.P. Nielsena
a
Department of Clinical Studies, The Royal Veterinary and Agricultural University,
Dyrlñgevej 88, DK-1870 Frederiksberg C, Denmark
b
Research Centre for the Management of Animal Production and Health (CEPROS),
Foulum, DK-8830 Tjele, Denmark
c
Department of Animal Science and Animal Health, The Royal Veterinary and Agricultural University,
GrùnnegaÊrdsvej 8, DK-1870 Frederiksberg C, Denmark
Received 23 February 2001; accepted 28 January 2002

Abstract

Our aims were to evaluate the published literature concerning the effect of swine vaccination
against Mycoplasma hyopneumoniae on the average daily weight gain (ADWG). This was done by
re-evaluating the in¯uence of selected factors on ADWG by a meta-analysis of published studies
from 1991 to 1999, ful®lling certain inclusion criteria.
With ADWG as the outcome, an analysis of variance was performed for such variables as
treatment, vaccination schedule, age during study, housing system and publication quality. Each
clinical trial was considered as a random effect and the numbers of pigs in each trial were weightings.
Of 63 published studies, 16 describing three commercial vaccines ful®lled the criteria for the meta-
analysis. Due to few studies with one of the vaccines …n ˆ 3†, only two vaccines were included.
Vaccinated pigs gained an average of 592 g …S:E: ˆ 15† with Stellamune1 and 590 g …S:E: ˆ 15†
with Suvaxyne1 compared to non-vaccinated pigs that gained an average of 569 g …S:E: ˆ 14†
…P < 0:01† when adjusted for age during the study. Vaccine type, vaccination schedule, housing
system and publication quality were not signi®cantly associated with ADWG. # 2002 Elsevier
Science B.V. All rights reserved.

Keywords: Meta-analysis; Mycoplasma hyopneumoniae; Vaccine; Swine; Respiratory disease

*
Corresponding author. Present address: Department of Clinical Studies, The Royal Veterinary and
Agricultural University, Dyrlñgevej 88, DK-1870 Frederiksberg C, Denmark. Tel.: ‡45-35-28-28-43;
fax: ‡45-35-28-28-38.
E-mail address: csj@kvl.dk (C.S. Jensen).

0167-5877/02/$ ± see front matter # 2002 Elsevier Science B.V. All rights reserved.
PII: S 0 1 6 7 - 5 8 7 7 ( 0 2 ) 0 0 0 0 5 - 3
266 C.S. Jensen et al. / Preventive Veterinary Medicine 54 (2002) 265±278

1. Introduction

Swine enzootic pneumonia causes major economic losses in the pig industry world-
wide (Ross, 1999). Mycoplasma hyopneumoniae is the primary agent responsible, but
secondary infections from Actinobacillus pleuropneumoniae, Pasteurella multocida and
Streptococcus suis also can increase the severity of the disease (Christensen et al.,
1999). In Denmark, 6.9 million doses of vaccines against M. hyopneumoniae were sold
in 1998, for an annual production of approximately 2 million pigs (Anon., 1998). Three
different commercial vaccines against M. hyopneumoniae were available in Denmark
in the study period: Stellamune11 (RespiSure1 outside Europe), Suvaxyn12 and
Hyoresp1.3 The vaccines were used to reduce the severity of pneumonia, and thereby
increase productivity by improving the average daily weight gain (ADWG) (Dayalu
et al., 1992; Dayalu, 1994; Nash, 1996). Both the Stellamune1 and Suvaxyn1 vaccines
contain inactivated whole cell cultures of M. hyopneumoniae combined with an oil
adjuvant.
Many studies have investigated the effect of vaccination on productivity (e.g. Vraa-
Andersen and Christensen, 1993; Charlier et al., 1994; Lium et al., 1994; Scheidt et al.,
1994; Trayer, 1994). However, several factors differed between studies, including the age
of pigs at vaccination, ef®cacy measure (e.g. ADWG and lung lesion), study periods,
housing system and group size. These variations between study designs make it dif®cult to
draw generalizable conclusions regarding the ef®cacy of any particular vaccine. Meta-
analysis is a method that takes this variability into account and increases the sample size by
combining results from several studies, thereby increasing the power to detect differences
in outcomes (Dickersin and Berlin, 1992).
Our main aim was to evaluate the effect of commercial vaccines against M. hyopneu-
moniae on the ADWG of growing pigs using a meta-analysis of published studies. Further,
the second aim was to examine the in¯uences of vaccine schedule, age of pigs, housing
system and publication quality.

2. Materials and methods

2.1. Literature review

We restricted the review to studies investigating the effect of vaccines against

M. hyopneumoniae and published from 1991 to 1999. The studies were written in either
English or Danish, and identi®ed using computerized literature databases (BA/RRM,
AGRIS, CAB, CC, AGRICOLA, MEDLINE). All studies were reviewed to obtain
information on vaccine type used, sample size and assignment of pigs to vaccinated
and non-vaccinated groups, route of vaccination, vaccination schedule, housing systems,
and age during the study, ef®cacy measure and variance of ef®cacy measure.

1
Stellamune1 (RespiSure1 outside Europe) produced by Pfizer.
2
Suvaxyn1produced by Fort Dodge.
3
Hyoresp1 produced by Merial.
 C.S. Jensen et al. / Preventive Veterinary Medicine 54 (2002) 265±278 267

Table 1
Variables provided from the individual studies evaluating the effects of vaccination against M. hyopneumoniae,
and included in the meta-analysis

Variable Indicate Levels

ADWG ADWG Continuous

Trial A number given each trial
Treatment Which treatment the pigs were given
Trial sample size Number of pigs included in a
treatment group within a trial
Vaccination status Whether or not the pigs were
vaccinated
Age during study The period during which the daily
weight gain was measured
Housing system Housing system during the
study period
Vaccine schedule How old the pigs were at vaccination
Publication quality Whether the study was published in
an refereed publication or not
1±28
Non-vaccinated
Vaccinated with Stellamune1
Vaccinated with Suvaxyn1
Continuous
Yes
No
From birth to slaughter
From weaning to slaughter
From around 30 kg to slaughter
At the original farm
Moved to another farm
Pigs vaccinated twice, first time before weaning
Pigs vaccinated twice, first time at Weaning
Pigs vaccinated at other times
Yes
No

We used `study' to de®ne the published study (one publication), and a `trial' to de®ne the
®eld trials conducted within a study. Some studies included more than one trial and some
studies combined results from several herds into one estimate.

2.2. Inclusion criteria for studies

Only studies using one of three commercial intramuscular vaccines (e.g. Stellamune14
(RespiSure1 outside Europe), Suvaxyn15 and Hyoresp16) were included in the meta-
analysis. From this, only studies where vaccines were randomly assigned to pigs within
herds naturally infected with M. hyopneumoniae, and with vaccinated and non-vaccinated
pigs housed in the same facilities were included for further analysis. Further information
necessary for inclusion were the sample size of vaccinated and non-vaccinated pigs, and
both the age of pigs at vaccination and whether the pigs were moved or stayed on the
original farm. Finally, the studies also had to provide ADWG, and describe the period
during which the ADWG was measured.

2.3. Statistical analysis

The vaccination schedule was classi®ed into three groups: (1) pigs vaccinated twice,
with the ®rst vaccination before weaning; (2) pigs vaccinated twice, with the ®rst
4
Stellamune1 (RespiSure1 outside Europe) produced by Pfizer.
5
Suvaxyn1produced by Fort Dodge.
6
Hyoresp1produced by Merial.
268 C.S. Jensen et al. / Preventive Veterinary Medicine 54 (2002) 265±278

vaccination at weaning; (3) pigs vaccinated at other times. Variables used in the analysis
are listed in Table 1.
Heterogeneity of the effect of vaccination among the ®eld trials on the ADWG
was tested by performing an analysis of variance including trial, vaccine status and a
2-way interaction term between trial and vaccine status (Dickersin and Berlin, 1992).

Table 2
Vaccination studies …n ˆ 39† not meeting the inclusion criteria for the meta-analysis of the effect of vaccination
against M. hyopneumoniae on the ADWG

Reference authors Reason for exclusion based on first missing information

according to the list of inclusion criteria

Bili'c et al. (1992) Vaccination of sows

Bilic et al. (1996) No randomization at pig level
Blagovi'c et al. (1992) No randomization at pig level
Busse and BoÈhne (1996) No randomization at pig level
Charlier et al. (1994) No randomization at pig level
Chiu et al. (1992) No commercial vaccine used
Christensen and Vraa-Andersen (1993) Daily weight gain not reported
De Jong et al. (1996) No randomization at pig level
Diekman et al. (1994) No information concerning vaccine type used
Dohoo and Montgomery (1996) No randomization at pig level
Guadagnini et al. (1998) No randomization at pig level
Harrison and Bryant (1999) Combined vaccines
Jayappa et al. (1998) Combined vaccines
Karge et al. (1998) No randomization at pig level
Kobisch et al. (1994) Vaccination of sows
Maes et al. (1999) No randomization at pig level
Martinon et al. (1998a) Daily weight gain not reported
Martinon et al. (1998b) Daily weight gain not reported in figures
Martinon et al. (1998c) Daily weight gain not reported in figures
Miller et al. (1992) No commercial vaccine used
Moore et al. (1997) Daily weight gain not reported
Morrow et al. (1994) Daily weight gain not reported in figures
Murphy et al. (1993) Aerosol vaccine administration
MunÄoz et al. (1996a) Missing information of group size
MunÄoz et al. (1996b) Pigs not housed in same facilities
Okada et al. (1998) No commercial vaccine used
Petersen et al. (1991) Daily weight gain not reported
Petersen and Weiss (1992) Missing information on group size
Pihl and Bñkbo (1996) No randomization at pig level
Reynaud et al. (1992) No commercial vaccine used
Sheldrake et al. (1991) Intraperitoneal vaccine administration
Sheldrake et al. (1993) Intraperitoneal vaccine administration
Stripkovits and Miller (1993) Pigs not naturally infected
Thacker et al. (1998a) Vaccination of sows
Thacker et al. (1998b) Incomplete information on vaccine type used
Thacker et al. (1998c) Incomplete information on vaccine type used
Thacker et al. (1999) Combined vaccines
Weiss and Peterson (1992) Missing information on group size
Weng et al. (1992) Oral vaccine administration
 C.S. Jensen et al. / Preventive Veterinary Medicine 54 (2002) 265±278 269

A signi®cant interaction indicates heterogeneity. A funnel graph was made to check for
publication bias (Everitt, 1995).
With the ADWG as the outcome, an analysis of variance was performed with treatment,
age during study, vaccination schedule, housing system and publication quality as ®xed
effects. The variable describing `trial' was expressed as a random effect. The effect of
vaccination (both vaccines) on ADWG was evaluated using an orthogonal contrast
for comparison of vaccinated to non-vaccinated pigs. All variables were included in
the initial model. Backward elimination of the ®xed effects was performed to establish the
®nal model.

Table 3
Vaccination studies …n ˆ 13† included in the meta-analysis on the effect of vaccines against M. hyopneumoniae
on the ADWG
Reference Vaccine No. of ADWG ADWG Age Housing Vaccine schedule
authors type used pigsa vaccinatedb non-vaccinatedb during during (weeks after
studyc triald farrowing)

Dalloli et al. (1998) Suvaxyn1 634 712 690 3 1 4 ‡ 7

Diekman et al. (1999) Stellamune1 108 760 780 3 1 1 ‡ 4
Stellamune1 108 700 700 3 1 1 ‡ 4
Hogg (1991) Suvaxyn1 100 707 675 3 2 1 ‡ 3
Le Grand and Stellamune1 911 815 799 3 1 1 ‡ 4
Kobisch (1996)
Lium et al. (1994) Suvaxyn1 120 709 719 3 1 1 ‡ 4
Stellamune1 211 716 719 3 1 1 ‡ 4
Maes et al. (1998c)e Stellamune1 468 608 583 3 1 1 ‡ 4
Martinod (1996) Stellamune1 240 662 599 2 1 1 ‡ 3
Pejsak et al. (1992) Stellamune1 306 418 390 1 1 1 ‡ 3
Stellamune1 327 422 404 1 1 4 ‡ 7
Stellamune1 210 412 387 1 1 1 ‡ 3
Scheidt et al. (1994)f Stellamune1 100 826 766 3 1 6 ‡ 8
Stellamune1 100 826 766 3 1 1 ‡ 3
Trayer (1994) Stellamune1 988 811 790 3 2 4 ‡ 6
Uhlenhopp (1991) Suvaxyn1 132 680 653 3 1 1 ‡ 3
Vraa-Andersen and Suvaxyn1 200 554 544 2 1 1 ‡ 3
Christensen (1993)g
Suvaxyn1 200 568 544 2 1 4 ‡ 6
Suvaxyn1 200 628 621 2 1 1 ‡ 3
Suvaxyn1 200 638 621 2 1 4 ‡ 6
Suvaxyn1 200 623 611 2 1 1 ‡ 3
Suvaxyn1 200 630 611 2 1 4 ‡ 6
Suvaxyn1 200 588 585 2 1 1 ‡ 3
Suvaxyn1 200 605 585 2 1 4 ‡ 6
Suvaxyn1 200 618 594 2 1 1 ‡ 3
Suvaxyn1 200 635 594 2 1 4 ‡ 6
Wallgren et al. (1998) Suvaxyn1 252 638 607 3 1 10 ‡ 12
Suvaxyn1 302 660 607 3 1 10 ‡ 12
a
Total number of vaccinated and non-vaccinated pigs used in the trials.
b
ADWG for pigs given by the reference authors.
c
From birth to slaughter (1), from weaning to slaughter (2), during finishing (3).
d
On the original farm (1), moved to another farm (2).
e
Identical to Maes et al. (1998b).
f
Identical to Scheidt et al. (1993).
g
Identical to Vraa-Andersen (1994a) and Vraa-Andersen et al. (1994b).
270 C.S. Jensen et al. / Preventive Veterinary Medicine 54 (2002) 265±278

The unit of concern was the group of pigs assigned to a treatment within a trial. Trials
conducted at the same herd were treated as independent. All analyses were performed as
weighted analyses of variance using trial sample size as weights (Armitage and Berry,
1987). The assumptions for performing the analyses of ADWG were evaluated using
residual plots and test for normality. All analyses were performed using the procedure
PROC MIXED from the statistical package SAS1 (Statistical Analysis System, SAS
Institute, 1996). A signi®cance level of 5% was used.

3. Results

Sixty-three relevant publications evaluating the effect of swine respiratory vaccines were
identi®ed. Of these, three were review papers (Dayalu et al., 1992; Dayalu, 1994; Nash,
1996) and six were of trials for which the results also were published elsewhere (Vraa-
Andersen, 1994a; Vraa-Andersen et al., 1994b; Scheidt et al., 1993; Bili'c et al., 1994;
Maes et al., 1998a,b). Of the remaining 54 publications, 39 studies were not used in the

Table 4
Herd specific information …n ˆ 13† from studies included in the meta-analysis on the effect of vaccines against
M. hyopneumoniae (MH) on the ADWG

Reference No. of herds Production system Other respiratory Herd size

infections
Pigs Sows
a b c
Dalloli et al. (1998) 1 ± AP , PM ± 900
Diekman et al. (1999) 1 All-in all-out PRRS, APa, PMb ± ±
1 Continuous PRRS, APa, PMb ± ±
Hogg (1991) 1 ± ± ± 560
Le Grand and 1 ± PMb ± ±
Kobisch (1996)
Lium et al. (1994) 4 Conventional herds, ± ± 30±75
continuous
Maes et al. (1998c) 5 Continuous Influenza, PRRS 850±3120 ±
Martinod (1996) 3 ± APa ± ±
Pejsak et al. (1992) 2 ± ± 650±920
Scheidt et al. (1994) 1 ± PMb ± ±
Trayer (1994) 1 All-in all-out in finishing PMb 2000 ±
Uhlenhopp (1991) 1 ± APa ± 200
Vraa-Andersen and 1 SPFd herd infected with MH APa 3000 ±
Christensen (1993)
1 SPF herd infected with MH APa 2500 ±
1 SPF herd infected with MH APa, PMb, ARe 6000 ±
1 Conventional herd APa, ARe 5000 ±
1 Conventional herd APa, ARe 120000 ±
Wallgren et al. (1998) 2 ± ± 85±95
a
A. pleuropneumoniae.
b
P. multocidae.
c
The symbol `±' indicates that the information is not available.
d
Specific pathogen free.
e
Atrophic rhinitis.
 C.S. Jensen et al. / Preventive Veterinary Medicine 54 (2002) 265±278 271

Table 5
Description of the methods, and findings on lung lesions reported for field trials included in the meta-analysis

Reference Detection methods of Findings regarding lung lesions

lung lesions
Non-vaccinateda Vaccinateda

Dalloli et al. (1998) Each lobe separately See reference See reference
Lung disease 0.70% 0.70%
Diekman et al. (1999) Mean lung score 1.30% 1.70%
Mean lung score 10.90% 5.10%
Hogg (1991) Lesion score 1.79 0.78
Le Grand and Kobisch (1996) Prevalence of pneumonia 50.00% 33.00%
Lium et al. (1994) Mean area affected 3.27% 0.85%
Mean area affected 3.27% 2.12%
Maes et al. (1998a,b) Mean area affected 17.08% 5.97%
Martinod (1996) Lung score 15.90% 4.70%
Pejsak et al. (1992) Lung score 7.90 4.40
Lung score 9.70 5.80
Lung score 12.70 6.30
Scheidt et al. (1994) Lesion score 12.00% 6.00%
Lesion score 12.00% 4.00%
Trayer (1994) Lung damage 10.41% 5.71%
Uhlenhopp (1991) Mean lesion score 23.68 16.06
Vraa-Andersen and Reported in another paperb ± ±
Christensen (1993)
Wallgren et al. (1998) Percentage of lung affected 24.60% 4.60%
Percentage of lung affected 24.60% 8.30%
a
Findings of lung lesions in non-vaccinated and vaccinated pigs.
b
Christensen and Vraa-Andersen (1993).

analysis for various reasons (Table 2). Thus, 14 studies (equaling 28 trials) met the
inclusion criteria (Table 3); 12 trials tested Stellamune1, 16 Suvaxyn1 and ®ve Hyoresp1.
Hyoresp1 was excluded from the analysis due to an insuf®cient number of trials (Kyriakis
et al., 1999 and Reynaud et al., 1998). Variance of ADWG was reported only in three
studies and therefore could not be used in our analysis (Lium et al., 1994; Wallgren et al.,
1998; Maes et al., 1999). Information on the number of herds used in ®eld trials, production
systems, herd size and information concerning respiratory diseases other than M. hyop-
neumoniae is given in Table 4. Due to missing information, these variables were not
included as covariates in the same analysis as the evaluation of lung lesions (Table 5).

Table 6
Descriptive results of the effect of Stellamune1 and Suvaxyn1 vaccines on the ADWG in the meta-analysis
(n ˆ 28 trials)

Vaccine Vaccinations Sample Number Mean S.D.

status of pigs size of trials ADWG (g)

Stellamune1 Vaccinated 2131 12 665 164

Non-vaccinated 1946 12 640 164
Suvaxyn1 Vaccinated 1818 16 637 48
Non-vaccinated 1871 16 616 48
272 C.S. Jensen et al. / Preventive Veterinary Medicine 54 (2002) 265±278

Fig. 1. Paired results on the ADWG for non-vaccinated and vaccinated pigs from the trials included in the meta-
analysis (n ˆ 28 trials).

Larger group sizes were used in ®eld trials with Stellamune1 compared to Suvaxyn1.
There was also a tendency towards a higher crude ADWG in both groups in the
Stellamune1 ®eld trials compared to Suvaxyn1 ®eld trials (Table 6).
The test of heterogeneity for ADWG among ®eld trials by vaccine status was not
signi®cant …P ˆ 0:90†, indicating homogeneity among ®eld trials, and thus the results
could be pooled (Fleiss, 1986). A funnel graph further demonstrated no publication bias
(Fig. 2), and the variable `publication quality' was also non-signi®cant in the analysis.
A similar linear relationship between ADWG for vaccinated pigs compared to non-
vaccinated pigs was seen for the two vaccines (P ˆ 0:86 for identical lines) (Fig. 1).
The ®nal model for ADWG included treatment …P < 0:001† and age during the study
…P < 0:001† (Table 7). The remaining variables (Table 1) were non-signi®cant.

Table 7
Results from the final model from the analysis of variance of ADWG in the meta-analysisa

Variable Level Estimateb Degree of freedom F-test P

(g)
Numerator Denumerator

Treatment Vaccine 1 592b 2 26 25.4 <0.001

Vaccine 2 590b
Control 569c
Age during study From birth to slaughter 409b 2 26 31.5 <0.001
From weaning to slaughter 624c
From around 30 kg to slaughter 718d
a
Levels for the same variable with the same letters are not significant different on a 5% significance level.
b
Estimate given as least square means.
 C.S. Jensen et al. / Preventive Veterinary Medicine 54 (2002) 265±278 273

4. Discussion

Many studies have shown that the vaccination of pigs against M. hyopneumoniae can
increase the ADWG in infected herds (e.g. Vraa-Andersen and Christensen, 1993; Charlier
et al., 1994; Lium et al., 1994; Scheidt et al., 1994; Wallgren et al., 1998; Maes et al., 1999).
Unfortunately, many publications fail to report whether the effect was statistically
signi®cant (e.g. Pejsak et al., 1992; Le Grand and Kobisch, 1996; Dalloli et al., 1998;
Hogg, 1991).
Meta-analysis increases the power of individual and relatively small studies by
combining results from independent studies. The increased power results in a higher
precision of the estimatesÐthereby decreasing the variance and more correctly pointing
out signi®cant results (Dickersin and Berlin, 1992).
A meta-analysis often uses the inverse variance of the outcome (effect measure) as a
weight (Fleiss, 1986). The inverse variance of the measurements in each trial is used so that
more precise trials (those with small variance) will be given a higher weight in the meta-
analysis (Armitage and Berry, 1987). However, reported ADWG variations were only
available in a few studies. As an alternative, the analysis was weighted by trial sample size.
The rationale for this is that ®eld trials with larger sample sizes tend to have a lower
variance, compared to ®eld trials with smaller sample sizes (Rosenthal, 1991).
To reduce bias, only randomized studies were included in the meta-analysis. Rando-
mization in the reported vaccine studies was mostly at the pig level (e.g. Hogg, 1991;
Dalloli et al., 1998; Maes et al., 1998c; Diekman et al., 1999), one study at litter level (Maes
et al., 1999) and one at pen level (De Jong et al., 1996). In two studies, the assignment of
pigs was based on `weeks' (e.g. pigs born in week 1 were vaccinated, pigs born in week 2
were non-vaccinated, etc.) (Dohoo and Montgomery, 1996; Pihl and Bñkbo, 1996). Some
studies did not state how the pigs were assigned, except that vaccinated and non-vaccinated
pigs were housed in the same facilities (e.g. Blagovi'c et al., 1992; Charlier et al., 1994;
Bilic et al., 1996; Karge et al., 1998). Only studies with randomization at the pig level were
included in the meta-analysis to avoid bias.
Different biases can occur when performing a meta-analysis. All available databases
were searched to avoid sampling bias. The trend towards publishing only positive
results (Rosenthal, 1979) can lead to publication bias. This is very dif®cult to assess
because it is often not possible to identify unpublished studies. A funnel graph can be
used to investigate publication bias, because publication bias will tend to skew the
pyramid shape by selectively excluding studies with small or non-signi®cant effects
(Everitt, 1995). No indication of publication bias was seen from the funnel graph (Fig. 2)
and the variable `publication quality' was non-signi®cant in the analysis of variance.
Seventy-eight percent of the studies were excluded because they did not ful®ll all of the
inclusion criteria. This might lead to exclusion bias. Most of the studies were excluded
for not reporting ®gures on the ADWG or group size, which were necessary for the
analysis.
There was a large variation between studies in the way that lung lesions were evaluated
(Table 5), and no direct standardization of the diagnostic methods was possible. For this
reason, the analyses were not based on the results of lung lesions, although this is a more
direct measure of vaccine effect. Only ®ve ®eld trials were performed with Hyoresp1
274 C.S. Jensen et al. / Preventive Veterinary Medicine 54 (2002) 265±278

Fig. 2. Funnel graph showing sample size against increase in ADWG between vaccinated and non-vaccinated
pigs for the studies included in the meta-analysis and comparing publication quality yes/no (n ˆ 28 trials).

which was excluded from the meta-analysis. This might be due to more recent marketing
(1998) of Hyoresp1 compared to the other two vaccines.
The effect of vaccination might be underestimated because only a partial vaccination of
the pig population was performed in the vaccine studies. However, in practice all pigs on a
farm are vaccinatedÐleading to greater herd immunity than in ®eld trials. In a study
estimating the cost of pneumonia (Straw et al., 1989), the ADWG decreased by 37 g for
every 10% of a pig's lungs affected by pneumonia. The results of own investigation
(Table 5) could not be directly related to those of Straw et al. (1989) due to the wide
variation among methods to characterize lung lesion among different studies.
In the present study, the average vaccination effect was a 21 g increase in ADWG
(Table 5). Because the prevalence of pneumonia in vaccination trials either differed or was
not reported, then the level of lung lesion reduction corresponding to a 21 g increase in
ADWG is unknown. The economic value of increases in ADWG might differ among herds,
but a 21 g increase might be too small to justify vaccination from an economic viewpoint.

5. Conclusion

In conclusion, the present meta-analysis of 28 published ®eld trials investigating the effect
of vaccinating pigs against M. hyopneumoniae showed that vaccinated pigs had a 21 g higher
ADWG compared to non-vaccinated pigs when adjusted for age during the study. Vaccine
type, vaccination schedule, housing system and publication quality had no signi®cant effect
on ADWG, and the funnel graph suggested that there was no publication bias.

Acknowledgements

The authors thank Dr. Margit Andreasen for discussions and comments on the manu-
script. The present study was ®nanced by Research Centre for the Management of Animal
 C.S. Jensen et al. / Preventive Veterinary Medicine 54 (2002) 265±278 275

Production and Health (CEPROS), Foulum, DK-8830 Tjele, as a part of the project;
``Animal health economics. Decision support systems for disease control in pig and cattle
herds'', CEP97-5/411-231.

References

Anon., 1998. Annual Report, Danish Veterinary Laboratory.

Armitage, P., Berry, G., 1987. Statistical Methods in Medical Research, 2nd Edition. Blackwell Scientific
Publications, London.
Bili'c, V., Lipej, Z., Simpraga, B., Sudri'c, F., 1992. Usage of RespiSure vaccine in Croatia. In: Proceedings of
the 12th IPVS Congress, Hague, The Netherlands, p. 326.
Bili'c, V., Lipej, Z., Zuri'c, M., Stigli'c, N., 1994. Immunoprophylaxis of mycoplasmal pneumonia in fattened
pigs. In: Poomvises, P., Ingkaninun, P. (Eds.), Proceedings of the 13th IPVS Congress, Bangkok, Thailand,
p. 193.
Bilic, V., Lipej, Z., Valpotic, I., Habrun, B., Humski, A., Njari, B., 1996. Mycoplasmal pneumonia in pigs in
Croatia: first evaluation of a vaccine in fattening pigs. Acta Vet. Hung. 44 (3), 287±293.
Blagovi'c, S., Fluksek, V., Lausin, M., Stigli'c, N., Cazin, P., 1992. Clinical evaluation of the protective
capabilities of an adjuvanted M. hyopneumoniae vaccine (RespiSure) in swine. In: Proceedings of the 12th
IPVS Congress, Hague, The Netherlands, p. 327.
Busse, F.W., BoÈhne, I., 1996. Vaccination against enzootic pneumonia (EP) or mycoplasma induced respiratory
disease (MIRD) in two breeding farms. In: Proceedings of the 14th IPVS Congress, Bologna, Italy, p. 234.
Charlier, P., Jambers, B., Martinod, S., Legrand, A., 1994. Efficacy of Stellamune mycoplasma in European field
trials. In: Poomvises, P., Ingkaninun, P. (Eds.), Proceedings of the 13th IPVS Congress, Bangkok, Thailand,
p. 136.
Chiu, Y.T., Weng, C.N., Liu, C.I., 1992. Protective efficacy of Mycoplasma hyopneumoniae vaccine against
mycoplasmal pneumonia combined with Pasteurella multocidae infection. In: Proceedings of the 12th IPVS
Congress, Hague, The Netherlands, p. 323.
Christensen, G., Vraa-Andersen, L., 1993. Forebyggelse af almindelig lungesyge. Dan Dyrlaegeforen Medlemsb
76 (5), 177±183.
Christensen, G., Sùrensen, V., Mousing, J., 1999. Diseases of the respiratory system. In: Straw, B.E., D'Allaire,
S., Menegeling, W.L., Taylor, D.J. (Eds.), Proceedings of the Eighth Symposium on Diseases of Swine. Iowa
State University Press, Ames, IA, pp. 913±940.
Dalloli, C., Geroldi, S., Guadaginini, P.F., 1998. Efficacy of Mycoplasma hyopneumoniae vaccine (Suvaxyn M.
Hyo) under field conditions. In: Done, S., Thomson, J., Varley, M. (Eds.), Proceedings of the 15th IPVS
Congress, Birmingham, England, p. 156.
Dayalu, K.I., 1994. Beneficial effects of a Mycoplasma hyopneumoniae vaccination (Stellamune) as evaluated
under both experimental and field conditions. In: Proceedings of the 13th IPVS Congress, Bangkok,
Thailand, p. 137.
Dayalu, K.I., Keich, R.L., Charlier, P., Martinod, S., 1992. Evaluation of the beneficial effects of a Mycoplasma
hyopneumoniae vaccine (RespisureTM)Ðresults from controlled and field studies. In: Proceedings of the
12th IPVS Congress, Hague, The Netherlands, p. 302.
De Jong, M.F., Jedema, E.J., Sampimom, O., 1996. Mycoplasma hyopneumoniae vaccination in 10-week-old
piglets. Results of a field trial. In: Proceedings of the 14th IPVS Congress, Bologna, Italy, p. 220.
Dickersin, K., Berlin, J.A., 1992. Meta-analysis: state-of-the-science. Epidemiol. Rev. 14, 154±176.
Diekman, M.A., Grant, A.L., Kelly, D.T., Pusaateri, A.E., Wagner, J.R., Scheidt, A.B., 1994. Reproductive and
lean growth performance of gilts reared in all-in/all-out facilities: effect of vaccination against Mycoplasma
hyopneumoniae and exposure to ammonia. J. Anim. Sci. 72 (2), 30.
Diekman, M.A., Scheidt, A.B., Grant, A.L., Kelly, D.T., Sutton, A.L., Martin, T.G., Cline, T.R., 1999. Effect of
vaccination against Mycoplasma hyopneumoniae on health, growth, and pubertal status of gilts exposed to
moderate ammonia concentrations in all-in-all-out versus continuous-flow system, swine health. Swine
Health Prod. 7 (2), 55±61.
276 C.S. Jensen et al. / Preventive Veterinary Medicine 54 (2002) 265±278

Dohoo, I.R., Montgomery, M.E., 1996. A field trial to evaluate a Mycoplasma hyopneumoniae vaccine: effects
on lung lesions and growth rates in swine. Can. Vet. J. 37, 299±302.
Everitt, B.S., 1995. The Cambridge Dictionary of Statistics in the Medical Sciences. Cambridge University
Press, Cambridge.
Fleiss, J.L., 1986. Analysis of data from multiclinical trials. Contr. Clin. Trials 7, 267±275.
Guadagnini, P.F., Alborali, L.G., Meloni, S., Martelli, P., 1998. Herd health status in finishers; lung lesions and
productive performance after Mycoplasma hyopneumoniae vaccination. In: Done, S., Thomson, J., Varley,
M. (Eds.), Proceedings of the 15th IPVS Congress, Birmingham, England, p. 156.
Harrison, R., Bryant, D., 1999. Field performance using a Mycoplasma±Atrophic rhinitis±Pasteurella±Erysipelas
combination Bacterin/toxoid: ProSystem 1B*P*M*E. AASP, pp. 207±208.
Hogg, A., 1991. Response to Mycoplasma hyopneumoniae. AASP, pp. 23±24.
Jayappa, H., Lehr, C., Tracewell, C., Wasmoen, T., 1998. Prevention of Diseases in Swine with A Mycoplasma
hyopneumoniae and Haemophilus parasuis Combination Vaccine. AASP, pp. 251±254.
Karge, I., SchroÈder, L., Elsen, R., 1998. Vaccination of young piglets against Mycoplasma hyopneumoniaeÐ
evaluation of economic parameters. In: Done, S., Thomson, J., Varley, M. (Eds.), Proceedings of the 15th
IPVS Congress, Birmingham, England, p. 148.
Kobisch, M., LabbeÂ, A., Morvan, P., Carolet, R., 1994. Evaluation of a Mycoplasma hyopneumonia vaccine in
pigs experimentally infected with Mycoplasma hyopneumoniae and Pasteurella multocida. In: Poomvises, P.,
Ingkaninun, P. (Eds.), Proceedings of the 13th IPVS Congress, Bangkok, Thailand, p. 194.
Kyriakis, S.C., Alexopoulos, C., Viemmas, J., Sarris, K., Lekkas, S., Koutsoviti-Papadopoulous, M., Saoulidis,
K., 1999. Field experience on performance improvement after vaccination with HYORESP in Mycoplasma
hyopneumoniae infected commercial pig unit. Eur. Focus, 21.
Le Grand, A., Kobisch, M., 1996. Enzootic pneumonia: comparison of the effect of pulse medication and
vaccination. In: Proceedings of the 14th IPVS Congress, Bologna, Italy, p. 223.
Lium, B., Lund, A., Skomsùy, A., 1994. A field study on vaccination against Mycoplasma hyopneumoniae
infection in pigs. In: Poomvises, P., Ingkaninun, P. (Eds.), Proceedings of the 13th IPVS Congress, Bangkok,
Thailand, p. 191.
Maes, D., Verdonck, M., Castryck, F., Vrijens, B., Verebek, W., Viaene, J., de Kruif, A., 1998a. The effect of
vaccination against Mycoplasma hyopneumoniae in pig herds with an all-in/all-out production system. In:
Done, S., Thomson, J., Varley, M. (Eds.), Proceedings of the 15th IPVS Congress, Birmingham, England, p. 152.
Maes, D., Verdonck, M., Castryck, F., Miry, C., Gielis, D., Vrijens, B., de Kruif, A., 1998b. The effect of
vaccination against Mycoplasma hyopneumoniae in pig herds with a continuous production system. In:
Done, S., Thomson, J., Varley, M. (Eds.), 15th IPVS Congress, Birmingham, England, p. 153.
Maes, D., Deluyker, H., Verdonck, M., Castryck, F., Miry, C., Lein, A., Vrijens, B., De Kruif, A., 1998c. The
effect of vaccination against Mycoplasma hyopneumoniae in pig herds with a continuous production system.
J. Vet. Med. Ser. B 45, 495±505.
Maes, D., Deluyker, D., Verdonck, M., Castryck, F., Miry, C., Vrijens, B., Verebek, W., Viaene, J., de Kruif, A.,
1999. Affect of vaccination against Mycoplasma hyopneumoniae in pig herds with all-in/all-out production
system. Vaccine 17, 1024±1034.
Martinod, S., 1996. Protection against Mycoplasma hyopneumoniae and Actinobacillus pleuropneumoniae
infections using a mycoplasma inactive vaccine RespiSure under field conditions. In: Proceedings of the 14th
IPVS Congress, Bologna, Italy, p. 221.
Martinon, O., Tiberghien, M.P., Reynaud, G., Blanchet, M., Brun, A., Milward, F., 1998a. Efficacy of
Mycoplasma hyopneumoniae Bacterin (Hyoresp) in challenge studies. In: Done, S., Thomson, J., Varley, M.
(Eds.), Proceedings of the 15th IPVS Congress, Birmingham, England, p. 157.
Martinon, O., Tiberghien, M.P., Reynaud, G., Blanchet, M., Brun, A., Milward, F., 1998b. Efficacy of a `on shot'
schedule of Mycoplasma hyopneumoniae Bacterin (Hyoresp). In: Done, S., Thomson, J., Varley, M. (Eds.),
Proceedings of the 15th IPVS Congress, Birmingham, England, p. 284.
Martinon, O., Tiberghien, M.P., Reynaud, G., Blanchet, M., Brun, A., Milward, F., 1998c. Duration of immunity
of a Mycoplasma hyopneumoniae Bacterin (Hyoresp). In: Done, S., Thomson, J., Varley, M. (Eds.),
Proceedings of the 15th IPVS, Birmingham, England, p. 285.
Miller, S.K., Ross, R.F., Erickson, B.Z., Gerber, D., Schultz, R.H., Chladek, D.W., 1992. Response of pigs
vaccinated with Mycoplasma hyopneumoniae vaccines to challenge with virulent Mycoplasma hyopneu-
moniae. In: Proceedings of the IPVS Congress, Hague, The Netherlands, p. 324.
 C.S. Jensen et al. / Preventive Veterinary Medicine 54 (2002) 265±278 277

Moore, C., Daigneault, J., Hoover, T.C., 1997. Efficacy of Different Strategies of Mycoplasma Vaccination As
Measured by Serology. AASP, pp. 135±136.
Morrow, W.E.M., Iglesias, G., Stanislaw, C., Stepphenson, A., Erickson, G., 1994. Effect of a mycoplasma
vaccine on average daily gain in swine. Swine Health Prod. 2 (6), 13±18.
MunÄoz, A., PallareÂs, F.J., Ramis, G., 1996a. Study of the effect of vaccination with Stellamune mycoplasma1 on
enzootic pneumoniaÐrelated clinical and productive parameters under field conditions. In: Proceedings of
the 14th IPVS Congress, Bologna, Italy, p. 229.
MunÄoz, A., PallareÂs, F.J., Ramis, G., 1996b. Use of a general linear model to evaluate the effect of vaccination of
swine with Stellamune mycoplasma1 on zootechnical performance and financial gain. In: Proceedings of the
14th IPVS Congress, Bologna, Italy, p. 222.
Murphy, D.A., van Alstine, W.G., Clark, L.K., Albregts, S., Knox, K., 1993. Aerosol vaccination of pigs against
Mycoplasma hyopneumoniae infection. Am. J. Vet. Res. 54 (11), 1874±1880.
Nash, W.A., 1996. Mycoplasma hyopneumoniae vaccination and disease control. Pig. J. 37, 78±94.
Okada, M., Okonogi, H., Ono, M., Sakano, T., Sato, S., 1998. Evaluation of Mycoplasma hyopneumoniae
vaccine in pigs under experimental and field condition. In: Done, S., Thomson, J., Varley, M. (Eds.),
Proceedings of the 15th IPVS Congress, Birmingham, England, p. 290.
Pejsak, Z., Tarasiuk, K., Cazin, P., 1992. Vaccination against mycoplasmal pneumoniaÐa field study on
RespiSure. In: Proceedings of the 12th IPVS, Hague, The Netherlands, p. 325.
Petersen, G.R., Weiss, D., 1992. Vaccination against mycoplasma hyopneumoniae in swine. Pig. Vet. J. 28,
35±39.
Petersen, G., Weiss, D., Egan, J., Korshus, J., Peters, R., Miron, M., 1991. Response to Mycoplasma
hyopneumoniaeÐvaccination in nursering piglets. AASP, pp. 17±21.
Pihl, K., Bñkbo, P., 1996. Vaccination mod mycoplasma-lungesyge hos svin. Vet. Inform. 6, 3±5.
Reynaud, G., Brun, A., Milward, F., Lacoste, F., Vandeputte, J., 1992. Clinical results obtained with an
inactivated vaccine against porcine mycoplasmosis. In: Proceedings of the 12th IPVS Congress, Hague, The
Netherlands, p. 303.
Reynaud, G., Martinon, O., Chouvet, C., Brun, A., Milward, F., 1998. Clinical field trial with Mycoplasma
hyopneumoniae bacterin (hyoresp). In: Done, S., Thomson, J., Varley, M. (Eds.), Proceedings of the 15th
IPVS Congress, Birmingham, England, p. 150.
Rosenthal, R., 1979. The ``fiel drawer problem'' and tolerance for null results. Psychol. Bull. 86 (3), 638±641.
Rosenthal, R., 1991. Meta-analytic Procedures for Social Research. Saga Publication.
Ross, R.F., 1999. Mycoplasmal diseases. In. Straw, B.E., D'Allaire, S., Menegeling, W.L., Taylor, D.J. (Eds.),
Proceedings of the Eighth Diseases of Swine. Iowa State University Press, Ames, IA, pp. 495±510.
SAS Institute, 1996. SAS/STAT Software, Changes and Enhancement, Version 6.11. SAS Institute, Inc., Cary,
NC, 1996. ISBN 1555442749.
Scheidt, A.B., Mayrose, V.B., Van Alstine, W.G., Clark, L.K., Cline, T.R., Einstein, M.E., 1993. The Benefit to
Cost Ratio of Vaccinating Pigs for Mycoplasmal Pneumonia in a Swine herd Affected by Enzootic
Pneumonia. AASP, pp. 29±30.
Scheidt, A.B., Mayrose, V.B., Van Alstine, W.G., Clark, L.K., Cline, T.R., Einstein, M.E., 1994. The effects of
vaccinating pigs for mycoplasmal pneumonia in a swine herd affected by enzootic pneumonia. Swine Health
Prod. 2 (1), 7±11.
Sheldrake, R.F., Gardener, I.A., Saunders, M.M., Romalis, L.F., 1991. Intraperitonal vaccination of pigs to
control Mycoplasma hyopneumoniae. RVTSA 51, 285±291.
Sheldrake, R.F., Romalis, L.F., Saunders, M.M., 1993. Serum and mucosal antibody responses against
Mycoplasma hyopneumoniae following intraperitonal vaccination and challenge of pigs with M
hyopneumoniae. RVTSA 55, 371±376.
Straw, B.E., Tuovinen, V.K., Bigras-Poulin, M., 1989. Estimation of the cost of pneumonia in swine herds.
JAVMA 12, 1702±1706.
Stripkovits, L., Miller, D.J.S., 1993. Comparative Studies of the Efficacy of Mycoplasma Hyopneumonia
Vaccine (RespiSure), Lincomycin and Tiamulin Against Experimental M. hyopneumoniae Infection of
Piglets. AASP, pp. 17±26.
Thacker, B., Boettcher, T., Anderson, T., Thacker, E., Young, T., 1998a. The influence of passive immunity on
serological responses to Mycoplasma hyopneumoniae vaccine. In: Done, S., Thomson, J., Varley, M. (Eds.),
Proceedings of the 15th IPVS Congress, Birmingham, England, p. 154.
278 C.S. Jensen et al. / Preventive Veterinary Medicine 54 (2002) 265±278

Thacker, E.L., Thacker, B.J., Boettcher, T.B., Jayappa, H., 1998b. Comparison of antibody production,
lymphocyte stimulation, and protection induced by four commercial Mycoplasma hyopneumoniae bacterins.
Swine Health Prod. 6 (3), 107±112.
Thacker, E.L., Thacker, B.J., Jayappa, H., 1998c. Evaluation of humeral and cellular immune response and
protection induced in pigs by five commercial mycoplasma vaccines. In: Done, S., Thomson, J., Varley, M.
(Eds.), Proceedings of the 15th IPVS Congress, Birmingham, England, p. 151.
Thacker, E.L., Halbur, P.G., Thacker, B.J., 1999. Effect of Vaccination on Dual Infection with Mycoplasma
hyopneumoniae and PRRV. AASP, pp. 375±377.
Trayer, T., 1994. Performance of Pigs Vaccinated with RespiSure in All-in/All-out Housing. AASP, pp. 23±25.
Uhlenhopp, E.K., 1991. Suvaxyn Respired MH: Findings From the Field. AASP, pp. 123±131.
Vraa-Andersen, L., 1994a. Respiratory diseases in Danish slaughter swine application of epidemiological
methods. Ph.D. Thesis. KVL, Denmark.
Vraa-Andersen, L., Christensen, G., 1993. Forebyggelse af almindelig lungesyge. Dan Dyrlaegeforen Medlemsb
76 (4), 129±132.
Vraa-Andersen, L., Christensen, G., Kuiper, R., 1994b. Vaccine efficiency trial with Suvaxyn M. hyo in
Denmark. In: Poomvises, P., Ingkaninun, P. (Eds.), Proceedings of the 13th IPVS, Bangkok, Thailand, p. 192.
Wallgren, P., LoÈfstedt, M., Heldmer, E., 1998. Strategic vaccination against Mycoplasma hyopneumoniae to
avoid marketing contagious animals from multiplying herds. In: Done, S., Thomson, J., Varley, M. (Eds.),
Proceedings of the 15th IPVS Congress, Birmingham, England, p. 149.
Weiss, D.L., Peterson, G.R., 1992. Mycoplasma hyopneumoniae Bacterin field efficacy study. In: Proceedings of
the 12th IPVS Congress, Hague, The Netherlands, p. 305.
Weng, C.N., Tzan, Y.L., Liu, S.D., Lin, S.Y., Lee, C.J., 1992. Protective effects of an oral microencapsulated
Mycoplasma hyopneumoniae vaccine against experimental infected pigs. RVTSA 53, 42±46.