Risk factors and epidemiological characteristics of new neonatal porcine diarrhoea syndrome in four Danish herds
© Kongsted et al.; licensee BioMed Central Ltd. 2014
Received: 1 March 2014
Accepted: 1 July 2014
Published: 10 July 2014
The epidemiology of New Neonatal Porcine Diarrhoea Syndrome (NNPDS) was studied in four selected herds. A total of 941 new born piglets in 86 litters were evaluated for five consecutive days. NNPDS is a newly emerged syndrome, characterized by diarrhoea within the first week of life, which is un-responsive to antibiotics and not associated with known pathogens. The aetiology behind the syndrome is unknown, and specific risk factors predisposing piglets to develop NNPDS also remain to be determined.
The study evaluated sow and piglet-level risk factors for developing NNPDS and described the epidemiologic characteristics within four herds previously diagnosed with the syndrome. NNPDS was defined as diarrhoea at any time-point during the second to fifth day of life.
NNPDS was observed in a total of 60% (range: 39%-89%) of first parity piglets and 36% (range: 19-65%) of piglets born by mature sows. In total of 26% of piglets had liquid faeces on the day of birth. Approximately half of these piglets developed NNPDS. In the majority of cases (50-70% of cases within herds) symptoms started on the second or third day of life. Piglets in Herd 1 had12.8 times higher probability of developing NNPDS than piglets in Herd 4. First parity piglets had a 4.1 higher probability of developing NNPDS than piglets born by mature sows. Birth weight and faecal consistency on the day of birth were minor risk factors, each significant within one herd.
The most important factors associated with NNPDS were herd of origin and sow-parity. The reason for one of the herds experiencing a considerably more severe outbreak than the others was not explained by factors addressed in this study.
The epidemiological pattern of diarrhoea varied a lot between herds; however, in all herds first parity piglets seemed predisposed. This association may be explained by an infectious background of the syndrome, but further studies are needed to explain this association.
Neonatal diarrhoea is a well-known disease complex in modern swine production influenced by individual, maternal and environmental factors. The aetiology in specific herd-cases may differ and is often incompletely diagnosed. Until recently, this complexity was not of major practical concern, since most problems could be controlled by vaccination or antibiotics. Around 2008, however, a new syndrome that did not respond to antibiotics or commercial vaccines seemed to emerge ( + Personal communications, S.E. Jorsal, National Veterinary Institute, Technical University of Denmark and B. Svensmark, Pig Research Centre, Danish Agriculture & Food Council, Denmark). The number of affected herds is unknown, but 80% of Danish swine practitioners report to experience these problems .
Currently, New neonatal porcine diarrhoea syndrome (NNPDS) refers to a clinical picture with piglets developing diarrhoea that is un-responsive to antibiotics within the first days of life. The suggested piglet-level case-definition is; Non-haemorrhagic diarrhoea during the first week of life without detection of known infectious pathogens, which is characterized by milk-filled stomachs and flaccid intestines at necropsy . This definition is based upon diagnostic examination of 101 Case and Control piglets from the four herds of the current study. Infectious agents which were evaluated and considered not to be involved in the syndrome included; Enterotoxigenic E. coli, Clostridium perfringens type A and C, Clostridium difficile, rotavirus A, coronavirus, Cystoisospora suis, Strongyloides ransomi, Giardia spp and Cryptosporidium spp.
Specific factors predisposing piglets to develop NNPDS remain to be discovered. Health problems in sows have previously been associated with diarrhoea in suckling pigs [3–5], but veterinary practitioners do not seem to associate the mastitis-metritis-agalactiae syndrome (MMA) or other sow health conditions with NNPDS . Practitioners report on an association with first parity sows, , but this experience needs scientific evaluation.
Insufficient prenatal nutrition or inadequate colostrum supplies are well-known risk factors for neonatal diarrhoea, thus clinical signs suggesting such problems need evaluation in outbreaks of NNPDS. A previous study showed that liquid faeces on the day of birth did not have any negative effects on piglets in these herds in terms of weight gain. Therefore, it was hypothesized that liquid faeces at birth might be a normal phenomenon, unrelated to the syndrome . One of the aims of the present study was to elaborate on this hypothesis, by evaluating if liquid faecal consistency on the day of birth was associated with diarrhoea on the subsequent days.
The primary aim of the study was to investigate sow- and piglet-level risk factors associated with NNPDS. Furthermore, the epidemiological pattern of diarrhoea in terms of prevalence, timing, duration and tendency to cluster within litters was described within the separate herds of the study. Since day one was hypothesized not to be part of the syndrome and piglets were only evaluated until the fifth day of life, NNPDS was defined as diarrhoea at any point during the second to fifth day of life.
Description of herds
Characteristics of the four study-herds
Duration of problems
Since establishment (2 years)
Herd size (n sows)
Sows per farrowing room
SPF + AP12
1st parity litters (%)2
Recruitment of gilts
Liquid (residue-free)/Home made
Liquid (residue-free)/Home made
Liquid (residue-free)/Home made
Routine treatment of piglets4
Amoxicillin at birth
Amoxicillin at castration
Routine treatment of sows5
Oxytocin after farrowing
Oxytocin after farrowing
Data structure and important herd differences
Altogether, 989 piglets within 86 litters were included at birth. A total of 48 piglets were removed from the data because they were euthanized for necropsy with no history of NNPDS (n = 27), died (n = 20) prior to day five with no history of NNPDS or were hermaphroditic (n = 1). Thus, a total of 941 piglets (227, 245, 216 and 253 from Herds 1, 2, 3 and 4) were included in the analyses. Within Herds 1–4, 5, 10, 5 and 9 first parity sows and 17, 11, 16 and 13 mature sows (2nd-7th parity) were included.
In Herds 1–3, 10% of sows were treated with antibiotics and NSAIDs on day one, whereas this counted for 41% (78% of first parity sows and 15% of mature sows) in Herd 4. All sows in Herd 3 and Herd 4 were treated with oxytocin postpartum. In Herd 2 and Herd 4 a milk formula was given to piglets by drench for supportive care (approximately 10 piglets per herd). Despite the general rule of no antibiotic treatment prior to day 3, in Herd 4, a total of 13 piglets were treated with streptocillin on the second day of life due to arthritis. These piglets were kept in the study, since the treatment was considered of minor importance in the context.
Clinical findings in piglets on the first day of life
Day one clinical findings in 941 piglets from the four herds
Piglets (first parity/mature sows)
Birth weight (kg) Mean (sd)
First parity piglets 1
Piglets born by mature sows 1
Clinical appearance (parities combined)
Rough hair coat
Perineal staining 2
Liquid faeces 2
Clinical signs of failure to thrive 3
First parity piglets
Piglets born by mature sows
Clinical findings in sows on the day of parturition
Litter sizes, clinical registrations and medical treatment on the day of parturition in 86 sows within the four herds
n (first parity/Mature)
Litter size Mean (sd)
Litter size 1st parity 1
Litter size mature sows 1
Stillborn Mean (sd)
Occurrence of NNPDS (diarrhoea at day 2–5 of life)
Prevalence of liquid faecal consistency day one and NNPDS in first parity piglets and piglets born by mature sows in the four herds
First parity piglets (n)
Liquid faeces day one
NNPDS/Liquid faeces day one2
Piglets born by mature sows (n)
Liquid faeces day one
NNPDS/Liquid faeces day one2
Apart from the tendency to affect first parity litters most, no obvious pattern was seen across herds. In Herd 1, piglets with NNPDS seemed to be clustered in litters, whereas in Herd 3 and 4, piglets with NNPDS seemed to be more evenly distributed among litters. In Herd 2, a strong tendency for NNPDS to cluster in first parity litters was observed. Herd 4 stood out as the least affected herd – half of the litters in this herd were either unaffected or had a single piglet with NNPDS only.
Risk factors for NNPDS at sow level
In the separate parity models (step 1 of the statistical analysis), none of the sow-effects came out significant. Thus, neither litter size, stillborn piglets nor clinical disease in sows were significant risk factors for NNPDS.
Results of the overall generalized linear mixed model on NNPDS
Per 100 g
Day one faecal consistency
ICC 4 litter
Risk factors for NNPDS at piglet level
Results of herd-specific generalized linear mixed models on NNPDS
Per 100 g increase
ICC 2 litter
The inclusion of herds for this study was based upon a high prevalence of unexplained neonatal diarrhoea prior to investigation, and intensive diagnostic investigations had suggested that they were all suffering from the emerging syndrome, NNPDS . However, the low prevalence of diarrhoea during the study-period in Herd 4, suggested that this herd was in fact in remission at the point of the study (this was confirmed by follow-up interviews). Since many sows in this herd were medicated on the day of farrowing, the low prevalence of diarrhoea could be hypothetically linked with this. However, data did not support this theory and according to the herd-manager the rate of medication was not higher than in preceding periods with a high prevalence of diarrhoea.
According to interviews with herd-managers, the prevalence of diarrhoea during the study-periods in Herds 2 and 3 was slightly lower than normal, but otherwise reflected their normal situations quite well. A lower prevalence in the study-period can be explained by elements in the study-design made in order to evaluate risk-factors and avoid excess mortality. Thus, restricting litter sizes to a minimum, prohibiting cross-fostering as well as excluding underweight piglets could explain a lower prevalence of diarrhoea than normal. The severe symptoms in Herd 1 matched recordings carried out before and after the study-period. Investigated sow and piglet level risk factors did not give any obvious explanation for this herd to stand out. As previously published, pathological and microbiological findings in piglets from this herd did not differ markedly from findings in the remaining herds . Furthermore, seasonal variations were unlikely to play a role, since Herd 1 was investigated during winter, which is the low season for neonatal diarrhoeas [3, 7].
The study intended to describe the epidemiological pattern of NNPDS in terms of prevalence, timing, duration and tendency to cluster within litters. However, important limitations of the study that are relevant in the interpretation of its results need to be mentioned. An obvious limitation of the study was the fact that piglets were only examined for five days. Thus, the definition on NNPDS used in the study was made on practical grounds and should not be interpreted as if NNPDS does not occur beyond the fifth day of life. In fact, 25-50% of piglets within herds started having symptoms on the fourth or fifth day of life, and some of them were probably diarrhoeic beyond the period of examination.
In Herds 1 and 2, the risk of developing diarrhoea was 6–7 fold increased when born by a first parity sow. In the other herds, the association with parity was weaker, but still significant. An association with young sows is in line with previous studies on suckling piglet and neonatal diarrhoea [3, 7, 8]. Different factors, such as lower levels of colostral antibodies , differences in milk composition  and stressful behaviour in first parity sows  may explain this association. Although no specific microorganism has been identified in the pathogenesis of NNPDS the overrepresentation of first parity litters may be due to lack of specific colostral immunity of a yet unknown infectious agent. It seems intriguing to interpret a tendency to cluster within litters (as seen in Herd 1 and 2) as an indication of the syndrome being of infectious nature. However, inborn (genetic or developmental) and environmental factors are also likely to cluster within litters, thus could also be a part of the explanation.
Obvious health problems in sows were rare and were not statistically associated with development of NNPDS. The lack of association between sow disease and NNPDS is interesting, since it might differentiate this syndrome from previously known neonatal diarrhoeas [3–5]. However, sows were only examined on the day of parturition and may have developed clinical symptoms later that were not taken into account. This study-design was chosen in order to be certain on cause-effect relationships (with piglets developing symptoms on different time-points, it seemed too difficult to evaluate the effect of clinical disease in sows during the whole study-period). In this study, the actual disease effects were probably integrated in the– highly variable – random effects of litters. Thus, the random litter effects probably represented a combination of undiagnosed disease, genetics and local environmental conditions as well as perhaps an infectious agent spreading within litters.
The fact that clinical signs of failure to thrive in piglets at day one were very infrequent (a total of 2% of piglets had protruding ribs) suggested that prenatal nutrition was generally adequate. Furthermore, since hollow flanks on the first day of life were not associated with the development of diarrhoea in these herds, symptoms seemed not to be caused by insufficient nutrition.
Potential associations between consistency of faeces on day one and the development of NNPDS were of major interest in this study, since a previous study suggested liquid faeces at birth to be a normal finding in these herds . The present study showed that many piglets (46%) having liquid faeces at birth did not develop NNPDS, and that consistency of faeces at birth was only a minor risk factor for developing NNPDS. The decision to of the study to draw a sharp line between day one liquid faeces and day two NNPDS may be problematic, since so many piglets (20-50% of piglets within herds) were found to experience the first symptoms of NNPDS on day 2. Some of these piglets (perhaps especially within Herd 2) probably experienced the first symptoms of NNPDS on day one. The decision to draw this sharp line was made on practical grounds, in order to be able to evaluate the hypothesis of liquid faeces on the day of birth being unrelated to the syndrome. Since the study did (weakly) associate liquid faeces day one with the development of NNPDS, future studies should not rule out that NNPDS sometimes starts on the day of birth.
Naturally, the overall limitation of the study is the lacking definition on NNPDS. Thus, it is important to underline that the conclusions from this study may not apply to all cases of NNPDS, since they were drawn from findings in four herds only. However, the four herds were all thoroughly investigated in terms of possible infectious aetiologies, and none of them were diagnosed with any well-known agent to explain the symptoms. Therefore, it presently seems fair to consider the diarrhoeal outbreaks in these herds to represent NNPDS.
The prevalence and the duration of diarrhoea as well as the tendency of diarrhoea to cluster within litters differed much between herds diagnosed with NNPDS. In most cases, symptoms started on the second day of life, but in 25-50% of cases within herds symptoms started on the fourth or fifth day of life. The duration of diarrhoea was most often one to two days.
Herd of origin and sow-parity were the most important factors associated with the development of NNPDS, whereas birth weight and faecal consistency on the day of birth were less important risk factors. The study did not point out other sow-level risk factors than parity. The reason for the more severe outbreak of NNPDS in Herd 1 was not explained by factors addressed in this study.
The general hypothesis of liquid faeces day one to be unrelated to the syndrome did not hold true. However, taking all results together, it seems that a liquid consistency faeces at birth is sometimes a harmless phenomenon, unrelated to NNPDS.
The study did not evaluate factors associated with the high prolificacy of Danish genetics (such as longer duration of farrowing) or herd-factors associated with NNPDS. Further studies are needed to look into these aspects.
The study was conducted in accordance with the guidelines of the Danish Ministry of Justice with respect to animal experimentation and care of animals under study. According to Danish legislation this type of study does not require ethical approval.
Selection of herds
Four well-managed conventional herds were selected based on these criteria: 1) Persistent problems with diarrhoea during the first week of life with a poor response to antibiotic treatment, 2) Vaccination of sows against Escherichia coli and Clostridium perfringens type C, 3) Failure of preventive management interventions, 4) PRRS negative farrowing unit as demonstrated in blood samples tested by ELISA/IPT or PCR and 5) Negative results of routine diagnostic examination for enteritis in five piglets. All herds had a history of neonatal diarrhoea for a period of at least one year. Detailed interviews with herd owners, local veterinarians and feed consultants as well as preliminary herd visits were performed in order to exclude herds with obvious management related problems. Prior to the onset of investigations, herd-managers were instructed to carry out prevalence counts of diarrhoea in order to document a relatively high and constant prevalence of diarrhoea (rough estimates – data not shown). Furthermore, herds were instructed not to change any routines before, during or immediately after the study period. After the study periods, follow-up interviews were carried out in order to evaluate if the clinical picture had changed after leaving the herds. Herd 1, 2, 3 and 4 were investigated in January, March, May and July of 2011, respectively.
Diagnostic examinations on a total of 101 Case- and Control piglets from these herds ruled out that well-known infectious agents (enterotoxigenic E. coli, Clostridium perfringens type C, rotavirus A, coronavirus and Cystoisospora suis) could explain the aetiology of diarrhoea. Furthermore, previous studies suggested that neither Clostridium perfringens type A, Clostridium difficile, Strongyloides ransomi, Giardia spp nor Cryptosporidium spp was involved in the diarrhoeal outbreaks. Gross-pathologically, affected piglets were characterized by flaccidity of intestines with no mucosal lesions and milk-filled stomachs .
Study design and case-definition
The study was carried out as a cross-sectional study with follow-up during the first five days of piglets’ lives in four herds previously diagnosed with NNPDS . On each day of the study, rectal swabs were used to evaluate whether piglets were diarrhoeic (liquid or watery consistency of faeces) or not (creamy, firm and solid consistencies or absence of faeces on swab). NNPDS was defined as diarrhoea at some point during the second to fifth day of life (day one was hypothesized not to be part of the syndrome and piglets were only evaluated until the fifth day of life). Clinical explanatory variables in sows and piglets were registered on the day of parturition/birth (day one).
Inclusion of sows and piglets
In each herd, approximately 20 sows (6–8 per day) from one farrowing batch were selected on the day of parturition. All selected sows were situated in the same farrowing section. In herds predominantly experiencing problems in first parity litters (Herd 2 and 4), first parity sows were given high priority in the inclusion procedure. Otherwise, the sows that first finished farrowing on the major farrowing days were selected. At selection, litters were standardized to11 (Herd 1 and Herd 3) or 12 piglets (Herd 2 and Herd 4). Piglets to stay in the litters were selected by simple random sampling among littermates having a birth weight above 800 grams. Smaller piglets were excluded, since they were not expected to be able to survive among large litter-mates. All included piglets were kept in their original litters during the whole study period in order to be able to recognize sow-effects. During the study, selected piglets were euthanized for diagnostic purposes. Piglets euthanized without symptoms (Control piglets for the Case–Control study) were removed from data. Data used in the description of duration of symptoms only included piglets still present in the herds on day five of life.
Treatments during the study period
No preventive antibiotic medication of sows was given. Injection of oxytocin postpartum was accepted if recommended by the local veterinarian. Medical treatment of sows was carried out according to individual herd routines and involved antibiotics and NSAIDs. Decisions to treat were made by the herd-staff and not based on clinical registrations made in the study.
As a general rule, antibiotic treatment of piglets was not accepted within the first three days of life. Later in the study period, clinical diseases were treated according to individual herd routines. Non-antibiotic oral supplements to piglets were allowed during the whole study period.
Clinical examinations on day one
For each sow, parity, litter size and number of stillborn were registered. All sows were clinically examined between 5 and 20 hours postpartum. The clinical examination included assessment of udder, legs and vulva and registration of rectal temperature. Mastitis was recorded when one or more udder sections were firm, red or sore at palpation. Leg problems were recorded when sows were unwilling to bear equal weight on all legs or evaded palpation of legs or hooves. An excess of unclear or foul-smelling discharge from vulva was recorded as vulva discharge.
Piglets were weighed. The presence of hollow flanks, protruding ribs, rough hair coats, dehydration (lack of skin-elasticity and sunken eye-balls), skin-abrasions on fore-knees and faecal staining of perineum was dichotomously recorded. Faecal staining was assessed within a diameter of one cm around anus. By use of rectal swabs, consistency of faeces was evaluated as either normal (creamy, firm and solid consistencies or absence of faeces on swab) or liquid (liquid or watery consistencies).
All procedures in the herds were carried out by the same person.
Description of risk factors evaluated in the study
Gilts: >15 piglets, Sows: >18 piglets
Gilts: <16 piglets, Sows: < 19 piglets
Mastitis and/or temp > 39.5°C and/or leg problems and/or vulva discharge3
None of the above
Piglet-related factors 4
Watery or liquid consistency of rectal contents
Creamy, firm or solid consistency of rectal contents and if no faeces on swab
Area behind ribs turned inwards
Area behind ribs followed the line of the ribs
Hair coat appeared dull
Hair coat did not appear dull
Models were fit in R  using the lme4 package . Model reduction was carried out using stepwise backwards elimination, removing variables with p > 0.05. Confounding was assessed by taking out and re-entering variables into the final models one by one and check for biologically important changes of estimates. Interaction terms were not included due to problems with complete separation, caused by sparse data. The linearity of birth weight at the log odds scale for NNPDS was assessed by transforming birth weight into a categorical variable based on quartiles and then verify that a decreasing trend of the estimates for the levels of the categorical variable was observed.
where βSS is the Subject Specific regression coefficient and δ2litter is the litter variance. The constant 0.346 is an approximation of the residual variance .
Population averaged, rather than cluster specific estimates of OR’s were used, since the study aimed at drawing general rather than litter-specific conclusions.
Pairwise post-hoc comparisons within significant variables with more than two levels were carried out using the lsmeans package in R .
Since data was limited, modelling was performed using a three-step procedure in order to be able to identify all associations of practical interest.
Step 1: Two separate models – one for first parity piglets and one for piglets born by mature (2nd-7th parity) sows -were run to identify overall sow-related risk-factors for NNPDS. The separation into two models was done in order to be able to recognize possible differences in sow-effects in first parity vs. mature sows. Furthermore, since first parity sows had smaller litters, different cut-offs were needed in the evaluation of the effect of litter size. In each model, litter size was inserted dichotomously with cut-off at the mean litter size in question (15 and 18, respectively). Herds were not included in these models, in order to avoid overlooking any subtle sow-effects.
Step 2: If the results of the parity-specific models in step 2 were alike, an overall model including parity and herd-effects was generated.
Step 3: Separate models for each herd were used to evaluate risk factors and random litter effects within the separate herds. These models were furthermore used to properly evaluate piglet-level risk-factors (due to the previously mentioned inability to include interaction terms in Step 2).
We would like to thank herd-owners and their staff for help and cooperation. Furthermore, we would like to thank Amalie Bækbo, Mikkel Haugegaard, Lars Jensen and Camilla Bjørn Olesen for practical assistance in the herds.
- Kongsted H: Spædgrisediarré i Danmark Anno 2013 (Danish). 2013, http://vsp.lf.dk/~/media/Files/PDF%20-%20Publikationer/Erfaringer%202013/Erfaring_1320_Sp%C3%A6dgrisediarre_i_Danmark_anno_2013.ashx, (Danish).Google Scholar
- Kongsted H, Jonach B, Haugegaard S, Angen O, Jorsal SE, Kokotovic B, Larsen LE, Jensen TK, Nielsen JP: Microbiological, pathological and histological findings in four Danish pig herds affected by a new neonatal diarrhoea syndrome. BMC Vet Res. 2013, 9: 206-10.1186/1746-6148-9-206.PubMed CentralView ArticlePubMedGoogle Scholar
- Svensmark B, Jorsal SE, Nielsen K, Willeberg P: Epidemiological studies of piglet diarrhoea in intensively managed Danish sow herds. I. Pre-weaning diarrhoea. Acta Vet Scand. 1989, 30 (1): 43-53.PubMedGoogle Scholar
- Wittum TE, Dewey CE, Hurd HS, Dargatz DA, Hill GW: Herd- and litter-level factors associated with the incidence of diarrhea morbidity and mortality in piglets 1 to 3 days of age. Swine Health Prod. 1995, 3 (3): 99-104.Google Scholar
- Lingaas F: Epidemiological and genetical studies in Norwegian pig herds. IV. Breed effects, recurrence of disease, and relationship between disease and some performance traits. Acta Vet Scand. 1991, 32 (1): 107-114.PubMedGoogle Scholar
- Kongsted H, Stege H, Toft N, Nielsen JP: The effect of New Neonatal porcine Diarrhoea Syndrome on average daily gain and mortality in four Danish pig herds. BMC Vet Res. 2014, 10: 90-10.1186/1746-6148-10-90.PubMed CentralView ArticlePubMedGoogle Scholar
- Lingaas F, Ronningen K: Epidemiological and genetical studies in Norwegian pig herds. II. Overall disease incidence and seasonal variation. Acta Vet Scand. 1991, 32 (1): 89-96.PubMedGoogle Scholar
- Sialelli J, Lautrou Y, Oswald I, Quiniou N: Peut-on établir une relation entre les charactéristiques de la trui et de sa portée et l’apparition des diarrhées néonathales? [abstract]. Journées Recherche Porcine. 2009, 41: 167-172.Google Scholar
- Fairbrother JM, Gyles CL: Colibacillosis. Diseases of Swine. 10th edition. Edited by: Zimmerman JJ, Karriker LA, Ramirez A, Schwartz KJ, Stevenson GW. Iowa, USA: Wiley-Blackwell; 2012:723-749.Google Scholar
- Beyer M, Jentsch W, Kuhla S, Wittenburg H, Kreienbring F, Scholze H, Rudolph PE, Metges CC: Effects of dietary energy intake during gestation and lactation on milk yield and composition of first, second and fourth parity sows. Arch Anim Nutr. 2007, 61 (6): 452-466. 10.1080/17450390701563433.View ArticlePubMedGoogle Scholar
- Pedersen LJ, Jensen T: Effects of late introduction of sows to two farrowing environments on the progress of farrowing and maternal behavior. J Anim Sci. 2008, 86 (10): 2730-2737. 10.2527/jas.2007-0749.View ArticlePubMedGoogle Scholar
- CoreTeam R: R: A Language and Environment for Statistical Computing. Vienna, Austria: R foundation for statistical computing; 2013. http://www.rproject.org/.Google Scholar
- Bates D, Maechler M, Bolker B: lme4: Linear Mixed-Effects Models Using S4classes. 2013, http://www.cran.r-project.org/package=lme4.Google Scholar
- Dohoo I, Martin W, Stryhn H: Veterinary Epidemiologic Research: 2nd ed. Charlottetown, Prince Edward Island, Canada: VER Inc; 2009.Google Scholar
- Lenth RV: lsmeans: Least-Squares Means. 2013, http://www.cran.r-project.org/package=lsmeans.Google Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.