Randomised controlled trial finds no evidence that a synbiotic improves health in assistance dog puppies

Abstract

Background

Prebiotics and probiotics (‘synbiotics’ when combined) are purported to be effective in the treatment of diarrhoea and potentially other diseases. However, it is unclear whether preventive use early in life can affect health outcomes for puppies.

Methods

A triple‐blind randomised controlled trial was conducted to assess the effectiveness of a synbiotic in reducing the occurrence and severity of diarrhoea, gastrointestinal disease and atopy. The synbiotic contained Enterococcus faecium, fructo‐oligosaccharides and gum arabic. Puppies were randomly allocated to one of two groups, one receiving the synbiotic and the other receiving a placebo from five to approximately 10 weeks of age. A further 116 puppies from 15 litters served as a non‐operative control group. Faecal scores were recorded during supplementation, and the incidence of gastrointestinal disease and atopy later in life was assessed.

Results

Fifty‐seven litters (419 puppies) were given the synbiotic and 57 litters (412 puppies) were given a placebo. Attrition was minimal. The diarrhoea status did not differ significantly among the groups, nor did groups differ in incidence of gastrointestinal disease or atopy.

Limitations

The non‐operative control group was not randomly allocated. The effectiveness of only one type of synbiotic was investigated, and the effects were predominantly assessed for a relatively short period early in life.

Conclusion

No benefits of administering this synbiotic early in life were observed.

INTRODUCTION

The importance of the gut microbiome in health and disease has received greater recognition in recent years. ¹ , ² It has been shown that, in the first few days after birth, the species richness of the gut microbiome of puppies is low; by eight weeks of age, it has significantly increased but remains lower than that of their dams. ³ This may play a role in the higher incidence of diarrhoea in young dogs compared to adults. ⁴ In some cases, diarrhoea can be severe enough to be life‐threatening. ⁵

Diarrhoea can be caused by a range of factors, one of which may be stress induced by kennelling or movement to new homes. ⁶ , ⁷ It is often associated with gastrointestinal (GI) pathogens, but many dogs with pathogens exhibit no clinical signs and vice versa. ⁸ One potential cause of diarrhoea may be dysbiosis, a detrimental alteration in the composition of the gut microbiome. ⁹ , ¹⁰ Some studies have reported positive effects of probiotic ⁶ , ¹¹ and prebiotic ¹² , ¹³ , ¹⁴ supplementation in dogs. Probiotics are microorganisms that confer health benefits when consumed in adequate amounts. They are often provided alongside prebiotics, which are dietary fibre that aid in the growth of beneficial gut microorganisms. Products that contain both probiotics and prebiotics are called ‘synbiotics’. The results on their efficacy have been mixed, ¹⁵ , ¹⁶ perhaps due to low sample sizes, but a study of 773 shelter dogs of mixed (unknown/unspecified) ages found that a synbiotic containing Enterococcus faecium was associated with a 31% lower occurrence of diarrhoea compared to a placebo over 14 days of supplementation. ¹⁷

It has also been suggested that the gut microbiome has a role in the development of atopy, ¹⁸ , ¹⁹ a disease common in dogs. ²⁰ , ²¹ Previous studies that have investigated probiotics as a treatment for atopy demonstrated reductions in severity for dogs treated with probiotics. ²⁰ , ²¹ Furthermore, faecal matter transplants used in dogs with atopy have shown the potential to reduce the severity and clinical signs of disease. ²² , ²³ A small number of studies have also administered probiotics to pregnant dams and young puppies to determine whether there is a preventative effect, although the sample sizes were small and there was no evidence that probiotics reduced the manifestation of atopy. ²⁴ , ²⁵ Signs of atopy can be severe, representing an important welfare concern and resulting in the withdrawal of dogs from assistance dog programmes. ²⁶ While the aetiology of atopy is complex, ²⁷ it is pertinent to investigate ways by which it can be prevented or mitigated.

The aim of the present study was to determine whether administering a synbiotic early in life can affect health outcomes for puppies. The main objectives were (1) to examine any differences in the occurrence of diarrhoea in puppies from the nest stage until approximately 10 weeks of age between those receiving the synbiotic and those receiving either a placebo or no supplement, (2) to determine whether there was any difference in the occurrence of GI disease during the first year of life between puppies that did and did not receive the synbiotic, and (3) to assess whether synbiotic provision affected the development of atopy later in life. These objectives were addressed using a prospective, triple‐blind, randomised, placebo‐controlled trial. It was hypothesised that there would be fewer cases of diarrhoea, other GI problems and atopy in the synbiotic group compared to the control group if the synbiotic was effective for this purpose.

MATERIALS AND METHODS

For details on the housing, feeding, treatment and monitoring, please refer to the Supporting Information.

Study animals

Nine hundred and forty‐seven puppies from 129 litters (from 75 unique sires and 113 unique dams) born to a UK assistance dog programme between March 2016 and March 2018 were allocated to the study. All litters were reared in volunteer homes before travelling as a litter to the organisation's centre, where they remained for 7‒9 days before being transferred to their ‘Puppy Raising’ homes, as previously described by Moxon et al. ²⁸ Puppies were Labrador retrievers, golden retrievers or crosses of these breeds.

Power analysis was conducted using ClinCalc.com ²⁹ to run two‐sided Z‐tests of the difference in proportions between two independent groups with a dichotomous outcome (diarrhoea: yes/no), with an alpha and beta of 0.05. A minimum sample size of 344 puppies in each trial group in the centre was required for an estimated reduction in diarrhoea from 32% to 20% and 271 puppies in each group in Puppy Raising for an estimated reduction in diarrhoea from 17% to 7%. These values were decided based on known diarrhoea incidence in the organisation at the time of study design (unpublished data) and in collaboration with the organisation's dog health team to decide what would represent a clinically significant reduction.

Study protocol

Puppies were primarily allocated to one of two trial groups: group A, receiving a capsule containing the EU‐registered probiotic E. faecium (NCIMB 10415) 4b1707 at 2 × 10⁹ colony‐forming units and a prebiotic mixture of fructo‐oligosaccharide and acacia (gum arabic), and group B, receiving a placebo capsule containing maltodextrin. The capsules were identical and labelled ‘probiotic A’ and ‘probiotic B’ to prevent identification. For this prospective cohort study, the researcher, all staff and puppy raisers were blinded until the data analysis was complete, at which time blinding codes were provided by the supplier. All puppies from a litter were allocated within a few days of birth to either group A (n = 419) or group B (n = 412) on a random basis ³⁰ by a member of the research team with no prior knowledge of the litter. Puppies received one capsule daily commencing at 5 weeks and 1 day of age in the nest (the day following their second anthelmintic treatment) until 10 weeks of age. A small number of additional puppies (n = 116) were allocated to a non‐operative control group (group C) and received no capsules.

Monitoring faecal quality

Faecal quality was monitored from 4 weeks old in the nest until 3 weeks into Puppy Raising. A faecal scoring chart based on a five‐point puppy faecal scoring system ³¹ was provided to volunteers caring for bitches and puppies in the nest, staff at the centre and puppy raisers. Scores of ‘1’ or ‘2’ were considered to represent diarrhoea and whether puppies had diarrhoea was treated as a binary variable.

In the nest

Volunteers were provided with data collection forms (Supporting Information S1) and were asked to report whether there were any faeces that scored ‘1’ or ‘2’ during 4‒5 weeks and 5‒6 weeks of age. Data were captured for the litter and, where possible, for individual puppies.

In the centre

While in the centre, dog care staff recorded whether the litter had diarrhoea on each day on specially designed data capture forms (Supporting Information S2) by recording the score of the worst quality stool that had been observed that day using the faecal scoring system.

In Puppy Raising

During Puppy Raising, faecal quality scoring sheets were completed for each puppy by the puppy raiser (Supporting Information S3). Each faeces produced each day was scored using the same faecal scoring system as used previously. Faecal quality was recorded for 21 days.

Diarrhoea was considered ‘serious’ if it was (a) persistent, defined as 24+ hours of only stools scored as ‘1’ or ‘2’, or (b) repeated, defined as 3 or more days of at least one stool scored as ‘1’ or ‘2’ occurring each day. Persistent and repeated cases were combined into a category of ‘serious’ diarrhoea cases, which represented puppies that may have required further investigation, veterinary attention or treatment.

Some puppies that had diarrhoea in the centre entered Puppy Raising with a 10‐day course of the same synbiotic, entailing that some received additional synbiotic supplementation. Puppies were also given other synbiotic brands, antibiotics and/or other interventions, such as rehydration fluids. These were recorded on a weekly basis and reported.

GI health in the first year of life

All diagnosed health conditions experienced by individual dogs are recorded by the organisation. Electronic health records were examined retrospectively for episodes of GI disease in the first 12 months of life after data collection for faecal scores had stopped. All puppies that remained on the study during Puppy Raising to more than 11 months of age were included. Episodes of GI disease were considered distinct if there were more than 7 days between records.

Atopy later in life

Electronic health records were examined for cases of atopy for all puppies in the study from birth until the time of analysis (18 June 2024); this extended period was considered because atopy often arises after maturity. ³² Differences in the time that dogs remained on the programme were considered.

Statistical analyses

Chi‐squared tests were used to check for deviations from the null hypothesis that puppies were equally allocated based on sex and breed. Data were described and reported for litters or puppies in each trial group (A, B or C), including the reasons for exclusion and the final study sample. The number of litters or puppies with missing data from each stage (nest, centre and Puppy Raising) were reported and the data were analysed accordingly. The occurrence of diarrhoea was compared between groups, as was the number of litters or puppies that received any treatment for diarrhoea. Litter data were examined for the nest and centre periods and individual puppy data for the nest and Puppy Raising periods. Statistical analyses were run using R (version 4.3.1) and were conducted blind to the trial group. Where mixed‐effects logistic regressions were run, models were fitted using a binomial error distribution with a logit link function in the package ‘lme4’, which uses maximum likelihood estimation. ³³ For all models, the function simulateResiduals() from the package ‘DHARMa’ ³⁴ was used to simulate residuals from the models for visual inspection and to test for uniformity of residuals, overdispersion and outliers; no deviations were detected. Pairwise comparisons were conducted using the package ‘emmeans’, ³⁵ allowing comparisons within and between groups while accounting for time effects and potential differences among groups due to chance. Both unadjusted and adjusted (using the Holm method) p‐values were reported. Note that the Holm method can sometimes yield identical p‐values within a family. The estimated marginal means from all models are reported in the Supporting Information.

Model estimates (coefficients) were exponentiated to aid interpretation, and these were reported alongside confidence intervals (CIs). In the case of logistic regression, the exponentiated estimate for the intercept is the odds of diarrhoea occurring in the reference group; this is the probability of diarrhoea occurring divided by the probability of diarrhoea not occurring. For comparator groups, the exponentiated estimate is the odds ratio (OR). An OR of 1, or where the CI overlaps 1, shows that there was no significant difference between groups. An OR significantly greater than 1 indicates that the odds of diarrhoea occurring were greater in the comparator group and an OR significantly less than 1 indicates the converse. For negative binomial models, exponentiating the estimates yields the mean count for the reference group (intercept) and the incidence rate ratio (IRR) for the comparator groups.

Faecal quality in the nest

The data were examined for 4‒5 weeks of age, representing the final week before the introduction of capsules, and for 5‒6 weeks of age, representing the first week of receiving capsules. Mixed‐effects binomial logistic regression analyses were used to assess whether diarrhoea in the nest at the litter and individual levels was related to trial group. An interaction between week (before vs. after) and trial group (A, B or C) was included, and litter identity was included as a random effect. For the individual‐level model, puppy identity was included as a nested random effect within litter identity. Pairwise comparisons testing the differences within and between trial groups before and after treatment were assessed using emmeans. ³⁶ A chi‐squared test was used to verify that litters with diarrhoea in trial groups A and B were not given therapeutic treatments in statistically different numbers.

Faecal quality at the litter level in the centre

The number of litters with and without diarrhoea in the centre was determined and a logistic regression was used to analyse differences among the trial groups. To control for the baseline level of diarrhoea, whether the litter experienced diarrhoea in the nest in the week prior to entering the centre was included as a covariate. The package emmeans was used to compare diarrhoea cases among trial groups. The same model was run but with whether or not litters received therapeutic treatments for diarrhoea as the response variable.

Faecal quality in Puppy Raising

The number of puppies that had ‘serious’ cases of diarrhoea in the first 3 weeks of Puppy Raising was compared between the three trial groups, with cases categorised separately for each week. An interaction between trial group (A, B or C) and week in Puppy Raising (1, 2 or 3) was included as a predictor in a mixed‐effects logistic regression along with whether the puppy's litter was recorded as having diarrhoea in the centre. Whether or not puppies had ‘serious’ diarrhoea was the response term. Puppy and litter were included as random effects. Pairwise comparisons of trial group and week were conducted using emmeans. As some puppies received therapeutic treatments for diarrhoea, further analyses were conducted to account for this, as well as previous diarrhoea status (see the Supporting Information).

GI health in the first year of life

To assess whether the number of episodes of GI disease differed among the trial groups, a mixed‐effects negative binomial regression was used with litter identity as a random effect. In addition, the number of puppies with health records relating to diarrhoea only was considered. Whether or not puppies had recorded instances of diarrhoea during the first 12 months was analysed using a mixed‐effects logistic regression, with trial group as a predictor and litter identity as a random effect. Pairwise comparisons to compare all trial groups were conducted using emmeans.

Atopy later in life

Whether or not dogs developed atopy at different frequencies according to trial group was assessed using a mixed‐effects logistic regression, with litter included as a random effect. The time for which each dog was able to be diagnosed was included as a covariate, but, because there may be simultaneity bias (i.e., dogs that develop atopy could be withdrawn from the programme earlier than dogs that do not develop atopy), a separate, basic model not including this adjustment was run in addition. Pairwise comparisons were conducted using emmeans.

RESULTS

Study animals

Thirty‐three puppies were removed from the study (Table 1). Data were included for 914 puppies; 798 were fed capsules (416 A and 382 B) and 116 were in the no‐capsule control group (C). These puppies were from 126 litters (57 A, 54 B and 15 C) produced by 73 unique sires and 110 unique dams. There was no difference among trial groups in the number of male and female puppies (χ ² = 0.484, D.F. = 2, p = 0.785; Table 2); however, there were differences in the number of puppies of different breeds or breed mixes allocated to each study group (χ ² = 92.7, D.F. = 10, p < 0.001; Table 2). Data were available for all three stages for 667 puppies (Table 3).

TABLE 1.

Reasons for removal of 33 puppies excluded from the study, including their study group.

Reason for removal	Frequency	A	B	C
Capsules not fed in the centre	15	0	15	0
Died in nest/centre	3	2	1	0
Out of protocol in nest/centre	9	1	8	0
Only nest data collected	6	0	6	0
Total	33	3	30	0

Note: A—group A (synbiotic group); B—group B (placebo group); C—group C (non‐operative control group).

TABLE 2.

Number of puppies in each trial group (A: synbiotic group, B: placebo group, C: non‐operative control group) by (a) sex and (b) breed.

	A	B	C	Total
(a) Sex
Female	214	189	61	464
Male	202	193	55	450
(b) Breed
G Ret	57	36	3	96
G Ret × G Ret* or G Ret × Lab*	37	50	0	87
G Ret × Lab	111	104	54	269
Lab	131	101	8	240
Lab × G Ret	55	50	24	129
Lab × G Ret* or Lab × Lab*	25	41	27	93

Note: Crossbreeds (×) are recorded as sire first. Asterisks (*) denote a crossbreed dam.

Abbreviations: G Ret, golden retriever; Lab, Labrador retriever.

TABLE 3.

Number of puppies in each trial group (A: synbiotic group, B: placebo group, C: non‐operative control group) with data available for each stage.

Data	Frequency	A	B	C
All stages	667	284	295	88
Nest and centre only	228	116	84	28
Nest and Puppy Raising only	3	0	3	0
Centre and Puppy Raising only	13	13	0	0
Centre only	3	3	0	0
Total	914	416	382	116

Faecal quality in the nest

Litter faecal quality

Data were collected from 914 puppies (400 A, 382 B and 116 C) from 124 litters (55 A, 54 B and 15 C). Fifty‐two litters (41.9%; 25 A, 20 B and 7 C) had diarrhoea reported in the nest at 4‒5 weeks of age (pre‐supplementation) and 64 litters (51.6%; 32 A, 23 B and 9 C) had diarrhoea reported at 5‒6 weeks of age (the first week of supplementation; Figure 1). Logistic regression analysis confirmed that the occurrence of diarrhoea was not significantly different among groups prior to the trial commencing (Supporting Information S4 and Table 4). The occurrence of diarrhoea did not significantly change in the week after supplementation with the capsules commenced, nor did it significantly differ among groups in this period (Table 4; B vs. A, OR = 0.47 [0.19‒1.17], p _unadjusted = 0.105, p _adjusted = 0.940; C vs. A, OR = 1.09 [0.28‒4.25], p _unadjusted = 0.901, p _adjusted = 1.000). The results relating to therapeutic treatments are given in the Supporting Information; no significant differences were found.

FIGURE 1.

Percentage of litters in trial groups A (synbiotic group), B (placebo group) and C (non‐operative control group) at 4‒5 weeks and 5‒6 weeks of age in the nest that did and did not have diarrhoea reported.

TABLE 4.

Pairwise comparisons from the mixed‐effects logistic regression assessing the effect of a synbiotic supplement on instances of diarrhoea in litters in the nest.

Comparison	Estimate (SE)	OR (CI)	Z	p‐Value	p‐Value (adjusted)
B/before versus A/before	‒0.41 (0.46)	0.66 (0.27‒1.63)	‒0.90	0.368	1.000
C/before versus A/before	0.06 (0.69)	1.06 (0.28‒4.08)	0.09	0.931	1.000
C/before versus B/before	0.47 (0.69)	1.61 (0.41‒6.25)	0.68	0.495	1.000
B/after versus A/after	‒0.75 (0.46)	0.47 (0.19‒1.17)	‒1.62	0.105	0.940
C/after versus A/after	0.09 (0.70)	1.09 (0.28‒4.25)	0.12	0.901	1.000
C/after versus B/after	0.84 (0.70)	2.31 (0.59‒9.13)	1.20	0.232	1.000
A/after versus A/before	0.61 (0.43)	1.85 (0.80‒4.25)	1.44	0.149	1.000
B/after versus B/before	0.28 (0.43)	1.32 (0.57‒3.06)	0.64	0.520	1.000
C/after versus C/before	0.64 (0.81)	1.90 (0.39‒9.23)	0.79	0.427	1.000

Note: Diarrhoea in the weeks before and after commencing the experimental prophylactic treatment was compared among trial groups (A: synbiotic group, B: placebo group, C: non‐operative control group).

Abbreviations: CI, confidence interval; OR, odds ratio; SE, standard error.

Puppy faecal quality

Data were recorded and included for individual puppy faecal quality for 823 of the 898 puppies at 4‒5 weeks of age (383 A, 346 B and 94 C) and for 830 puppies at 5‒6 weeks of age (391 A, 342 B and 97 C).

Two hundred and sixteen puppies (26.2%; 97 A, 96 B and 23 C) had diarrhoea reported in the nest at 4‒5 weeks of age and 221 had diarrhoea at 5‒6 weeks of age (26.6%; 108 A, 81 B and 32 C; Figure 2). Seventy‐four puppies had diarrhoea reported at both 4‒5 weeks and 5‒6 weeks of age. The probability of puppies having diarrhoea did not differ significantly among trial groups in the week before the intervention started, and was not significantly different among groups in the week after commencing the capsules (B vs. A, OR = 0.50 [0.15‒1.65], p _unadjusted = 0.256, p _adjusted = 1.000; C vs. A, OR = 2.18 [0.39‒12.06], p _unadjusted = 0.373, p _adjusted = 1.000), nor was it significantly different before versus after capsules commenced in any of the three trial groups (Supporting Information S5 and Table 5).

FIGURE 2.

Percentage of puppies in trial groups A (synbiotic group), B (placebo group) and C (non‐operative control group) at 4‒5 weeks and 5‒6 weeks of age in the nest that did and did not have diarrhoea reported.

TABLE 5.

Pairwise comparisons from the mixed‐effects logistic regression assessing the effect of a synbiotic supplement on instances of diarrhoea in puppies in the nest.

Comparison	Estimate (SE)	Exp. estimate (CI)	Z	p‐Value	p‐Value (adjusted)
B/before versus A/before	‒0.18 (0.61)	0.84 (0.25‒2.75)	‒0.30	0.767	1.000
C/before versus A/before	0.27 (0.89)	1.31 (0.23‒7.49)	0.30	0.765	1.000
C/before versus B/before	0.45 (0.91)	1.56 (0.26‒9.36)	0.49	0.625	1.000
B/after versus A/after	‒0.69 (0.61)	0.50 (0.15‒1.65)	‒1.14	0.256	1.000
C/after versus A/after	0.78 (0.87)	2.18 (0.39‒12.06)	0.89	0.373	1.000
C/after versus B/after	1.47 (0.90)	4.35 (0.75‒25.43)	1.63	0.103	0.820
A/after versus A/before	0.24 (0.22)	1.27 (0.83‒1.96)	1.10	0.270	1.000
B/after versus B/before	‒0.27 (0.23)	0.76 (0.48‒1.20)	‒1.17	0.243	1.000
C/after versus C/before	0.75 (0.43)	2.12 (0.92‒4.93)	1.75	0.080	0.717

Note: Diarrhoea in the weeks before and after commencing the experimental prophylactic treatment was compared among trial groups (A: synbiotic group, B: placebo group, C: non‐operative control group).

Abbreviations: CI, confidence interval; SE, standard error.

Faecal quality in the centre

Data were included for 911 puppies: 795 fed capsules and 116 controls (416 A, 379 B and 116 C) from 125 litters (57 A, 53 B and 15 C). Ninety‐eight litters (77.4%; 46 A, 43 B and 9 C) had some diarrhoea recorded in the centre (Figures 3 and 4), 50 of these (51.0%; 26 A, 18 B and 6 C) had also been noted with diarrhoea between 5 and 6 weeks of age in the nest. The data for weeks 5‒6 were missing for two litters. There was no significant difference among trial groups in the number of litters that had diarrhoea reported at the centre (Supporting Information S6 and Table 6; B vs. A, OR = 1.09 [0.41‒2.84], p = 0.867; C vs. A, OR = 0.37 [0.11‒1.28], p _unadjusted = 0.117, p _adjusted = 0.286), and whether or not litters had diarrhoea reported in the previous week had no significant effect on the number of litters with diarrhoea reported at the centre (OR = 1.07 [0.45–2.57], p = 0.879). There was no significant difference among the groups in whether or not litters were given therapeutic treatments for diarrhoea (Supporting Information S7).

FIGURE 3.

Percentage of litters in trial groups A (synbiotic group), B (placebo group) and C (non‐operative control group) that had diarrhoea reported while at the centre.

FIGURE 4.

Heatmaps of each litter's worst faecal score (yellow = 1, orange = 2, blue = 3+, grey = no data) on each day in the centre according to trial group (synbiotic group [top] and placebo group [bottom]). Each row represents one individual litter, ordered by severity.

TABLE 6.

Pairwise comparisons from the logistic regression assessing the effect of the trial group (A: synbiotic group, B: placebo group, C: non‐operative control group) and whether or not litters had diarrhoea in the nest during the week prior on the number of recorded instances of diarrhoea in litters at the centre.

Comparison	Estimate (SE)	OR (CI)	Z	p‐Value	p‐Value (adjusted)
Group B versus group A	0.08 (0.49)	1.09 (0.41–2.84)	0.17	0.867	0.867
Group C versus group A	‒0.98 (0.63)	0.37 (0.11–1.28)	‒1.57	0.117	0.286
Group C versus group B	‒1.06 (0.64)	0.35 (0.10‒1.20)	‒1.67	0.095	0.286

Note: N = 123.

Abbreviations: CI, confidence interval; OR, odds ratio; SE, standard error.

Faecal quality in Puppy Raising

Data were included for 683 puppies (297 A, 298 B and 88 C) from 122 litters (54 A, 53 B and 14 C) in Puppy Raising. A small number of puppies had missing diarrhoea data: one (group B) for week 1, two (1 B and 1 C) for week 2 and 10 (6 A, 2 B and 2 C) for week 3. Three puppies (all in group B) had missing litter diarrhoea data in the centre. In week 1, 16.8% of puppies had diarrhoea categorised as ‘serious’. This reduced to 10.9% in week 2 and 7.3% in week 3.

The occurrence of ‘serious’ diarrhoea decreased for all trial groups over the course of the 3 weeks (Supporting Information S8); between weeks 1 and 3, these differences were significant for groups B and C even after p‐value adjustment (Table 7 and Figure 5). There was no significant difference in the occurrence of ‘serious’ diarrhoea between trial groups at any of the three time points (Table 7; week 1: B vs. A, OR = 1.43 [0.80‒2.57], p _unadjusted = 0.230, p _adjusted = 1.000; week 2: B vs. A, OR = 1.31 [0.68‒2.53], p _unadjusted = 0.416, p _adjusted = 1.000; week 3: B vs. A, OR = 0.80 [0.38‒1.68], p _unadjusted = 0.558, p _adjusted = 1.000). Whether or not puppies came from litters that had diarrhoea in the centre did not have a significant effect on the probability of ‘serious’ diarrhoea occurring over the 3 weeks in Puppy Raising (OR = 1.32 [0.76–2.32], p = 0.328).

TABLE 7.

Pairwise comparisons from the mixed‐effects logistic regression model of the occurrence of ‘serious’ diarrhoea in puppies in the three trial groups (A: synbiotic group, B: placebo group, C: non‐operative control group) during the first 3 weeks of Puppy Raising.

Comparison	Estimate (SE)	OR (CI)	Z	p‐Value	p‐Value (adjusted)
Week 1: group B versus group A	0.36 (0.30)	1.43 (0.80‒2.57)	1.20	0.230	1.000
Week 1: group C versus group A	0.35 (0.44)	1.41 (0.60‒3.35)	0.79	0.431	1.000
Week 1: group C versus group B	‒0.01 (0.44)	0.99 (0.42‒2.32)	‒0.03	0.979	1.000
Week 2: group B versus group A	0.27 (0.34)	1.31 (0.68‒2.53)	0.81	0.416	1.000
Week 2: group C versus group A	‒0.27 (0.55)	0.76 (0.26‒2.22)	‒0.50	0.618	1.000
Week 2: group C versus group B	‒0.54 (0.54)	0.58 (0.20‒1.68)	‒1.01	0.314	1.000
Week 3: group B versus group A	‒0.22 (0.38)	0.80 (0.38‒1.68)	‒0.59	0.558	1.000
Week 3: group C versus group A	‒1.09 (0.72)	0.34 (0.08‒1.37)	‒1.52	0.129	1.000
Week 3: group C versus group B	‒0.87 (0.72)	0.42 (0.10‒1.73)	‒1.20	0.230	1.000
Group A: week 2 versus week 1	‒0.54 (0.28)	0.58 (0.33‒1.01)	‒1.93	0.054	0.651
Group A: week 3 versus week 1	‒0.75 (0.30)	0.47 (0.27‒0.85)	‒2.52	0.012	0.186
Group A: week 3 versus week 2	‒0.20 (0.31)	0.82 (0.44‒1.51)	‒0.65	0.516	1.000
Group B: week 2 versus week 1	‒0.63 (0.26)	0.53 (0.32‒0.89)	‒2.39	0.017	0.253
Group B: week 3 versus week 1	‒1.32 (0.30)	0.27 (0.15‒0.48)	‒4.37	<0.001	<0.001
Group B: week 3 versus week 2	‒0.70 (0.32)	0.50 (0.27‒0.93)	‒2.21	0.027	0.384
Group C: week 2 versus week 1	‒1.16 (0.54)	0.31 (0.11‒0.90)	‒2.16	0.031	0.397
Group C: week 3 versus week 1	‒2.18 (0.71)	0.11 (0.03‒0.45)	‒3.09	0.002	0.035
Group C: week 3 versus week 2	‒1.02 (0.76)	0.36 (0.08‒1.59)	‒1.35	0.178	1.000

FIGURE 5.

Heatmaps indicating whether each puppy in the three trial groups (A = synbiotic group, B = placebo group, C = non‐operative control group) experienced ‘serious diarrhoea’ (yellow = yes, blue = no, black = no data) in each week of Puppy Raising. Each row represents one individual.

The number of puppies that received therapeutic treatments according to whether or not they had serious diarrhoea is shown in Supporting Information S9 and the type of treatments they had received is shown in Supporting Information S10. When considering the number of puppies that received therapeutic treatments (aside from the synbiotic) as the response term, no significant differences were observed among the trial groups (Supporting Information S11 and S12). However, the use of treatments did significantly reduce in the second week of Puppy Raising compared to the first week in all groups and remained lower in the third week (Supporting Information S12).

There was no significant difference among the trial groups in the probability of diarrhoea occurring according to previous diarrhoea status in any of the 3 weeks of Puppy Raising (Supporting Information S13 and S14).

GI health in the first year of life

Seventeen puppies (8 A, 8 B and 1 C) were withdrawn from Puppy Raising at or before 11 months of age and were excluded from analysis. Six hundred and sixty‐five puppies remained in Puppy Raising and had data available for 12 months and one had data available for more than 11 months but less than 12 months. Therefore, to examine any differences in GI health between the study groups, data from 666 puppies (289 A, 290 B and 87 C) were included.

GI disease was noted in 102 (35.3%) group A puppies, 111 (38.3%) group B puppies and 22 (25.3%) group C puppies. For puppies that were noted to have GI disease, the total number of distinct cases of disease ranged from 1 to 10. There was no significant difference in the number of GI episodes experienced among trial groups (B vs. A, IRR = 1.16 [0.84–1.60], p _unadjusted = 0.362, p _adjusted = 0.724; C vs. A, IRR = 0.81 [0.48–1.37)], p _unadjusted = 0.427, p _adjusted = 0.724; Table 8, Figure 6 and Supporting Information S15). There was also no significant difference among trial groups when focusing on whether or not puppies had health records related to diarrhoea only (Supporting Information S16 and S17).

TABLE 8.

Pairwise comparisons from the mixed‐effects negative binomial regression assessing the number of episodes of gastrointestinal disease in the three trial groups (A: synbiotic group, B: placebo group, C: non‐operative control group).

Comparison	Estimate (SE)	IRR (CI)	Z	p‐Value	p‐Value (adjusted)
Group B versus group A	0.15 (0.16)	1.16 (0.84‒1.60)	0.91	0.362	0.724
Group C versus group A	‒0.21 (0.27)	0.81 (0.48‒1.37)	‒0.79	0.427	0.724
Group C versus group B	‒0.36 (0.27)	0.70 (0.42‒1.17)	‒1.36	0.174	0.522

Note: N = 666 (121 litters).

Abbreviations: CI, confidence interval; IRR, incidence rate ratio; SE, standard error.

FIGURE 6.

Number of distinct episodes of gastrointestinal disease during Puppy Raising according to trial group (A = synbiotic group, B = placebo group, C = non‐operative control group).

Atopy later in life

Thirty‐eight (12.8%) dogs in group A, 42 (14.1%) dogs in group B and eight (9.1%) dogs in group C developed atopy. These figures were not significantly different whether the model was adjusted to control for time in the programme or not (all p‐values > 0.2; Table 9 and Supporting Information S18), and any effect of the time covariate was also not significant (OR = 0.96 [0.87–1.05], p = 0.362).

TABLE 9.

Pairwise comparisons from the logistic regression assessing whether there are differences in the number of dogs developing atopy over their lifespan according to trial group (A: synbiotic group, B: placebo group, C: non‐operative control group), showing (a) the basic model and (b) the adjusted model controlling for time in the programme.

Comparison	Estimate (SE)	OR (CI)	Z	p‐Value	p‐Value (adjusted)
(a) Basic model
Group B versus group A	0.13 (0.30)	1.14 (0.64‒2.03)	0.44	0.662	0.803
Group C versus group A	‒0.41 (0.49)	0.67 (0.26‒1.73)	‒0.84	0.401	0.803
Group C versus group B	‒0.54 (0.48)	0.58 (0.23‒1.51)	‒1.11	0.267	0.802
(b) Adjusted model
Group B versus group A	0.11 (0.29)	1.12 (0.63‒1.99)	0.38	0.708	0.717
Group C versus group A	‒0.46 (0.49)	0.63 (0.24‒1.64)	‒0.94	0.347	0.717
Group C versus group B	‒0.57 (0.48)	0.57 (0.22‒1.46)	‒1.18	0.239	0.717

Abbreviations: CI, confidence interval; OR, odds ratio; SE, standard error.

DISCUSSION

This study examined the effectiveness of a commercially available synbiotic for the prophylactic treatment of diarrhoea in assistance dog puppies during their movements from the nest to new homes. Faeces were scored at the individual or litter level in the nest from 4 weeks of age, at the litter level in the centre and at the individual level for the first 3 weeks of Puppy Raising. In Puppy Raising, whether or not puppies had ‘serious’ diarrhoea was considered in order to discount sporadic and clinically insignificant cases. At each stage, no significant differences in the occurrence of diarrhoea were found between the experimental treatment group and the two control groups. In addition, no significant differences were found among groups in the probability of dogs having health records for diarrhoea later in Puppy Raising, in the number of instances of GI disease experienced, or whether or not dogs developed atopy.

This study has some limitations. First, puppies were not randomly allocated to the non‐operative control group. Second, the number of puppies with diarrhoea in the centre could not be determined because the data were collected at the litter level; it may be the case that most puppies in an affected litter did not have had diarrhoea themselves. Third, faecal scores were recorded by a large range of different staff and volunteers and scoring may not have been consistent among them. Fourth, in Puppy Raising, where data were available at individual level, only the occurrence of serious diarrhoea was considered, as this was deemed to be a much greater health and welfare concern than infrequent diarrhoea. While this should not represent an issue since the classification was decided pre hoc, it has a limitation in being a somewhat arbitrary classification. Despite these limitations, the study has some important strengths in being well powered and triple blind.

Although the primary aim of this study was to examine the use of a synbiotic as a prophylactic treatment, some insight was gained into its use as a therapeutic treatment. When considering only dogs that had ‘serious’ diarrhoea, there was no significant difference among trial groups in whether these dogs had serious diarrhoea during the following week in Puppy Raising. In addition, there was no significant difference in the probability of dogs in each group being given therapeutic treatments for diarrhoea either at the centre or at each week of Puppy Raising; puppies would only be given therapeutic treatments if their diarrhoea was deemed serious enough to warrant this intervention. These results suggest that the synbiotic was not effective as a therapeutic treatment, as, if it were, decreases in diarrhoea and/or the use of further therapeutic treatments would be expected in the experimental group compared to the control groups. This finding is in line with the European Network for Optimization of Veterinary Antimicrobial Therapy guidelines, which do not presently make any recommendations regarding the use of probiotics in cases of acute diarrhoea in dogs. ³⁷

While the results of this study show no evidence in support of providing the synbiotic as a prophylactic treatment for diarrhoea, other GI disease or atopy, the effectiveness of other synbiotics cannot be ruled out. Furthermore, this study's scope only encompassed very young puppies in a well‐managed breeding programme fed high‐quality diets with good animal husbandry; therefore, inferences cannot be made about the effectiveness of synbiotics for older dogs, dogs housed in different environments, on poorer‐quality diets, or perhaps for dogs with specific medical conditions. This study also cannot preclude beneficial effects of other bacterial species that are in use as probiotics, such as Bifidobacterium animalis, ⁶ Lactobacillus species ¹¹ or products containing multiple bacterial species. ³⁸ , ³⁹

The results may not be surprising considering that two relatively recent systematic reviews of studies of probiotic and synbiotic supplementation in dogs found limited evidence for their effect in treating GI disease ¹⁵ , ⁴⁰ and another randomised controlled trial, albeit with a smaller sample size, found no effect. ¹⁶ However, probiotics are the most common clinical treatment for diarrhoea in dogs in the UK, ⁴¹ and some more recent studies have demonstrated some beneficial effects in terms of improved faecal quality and duration of diarrhoea, ⁴² , ⁴³ although the risk of publication bias should be considered when examining the available literature. ⁴⁴

In addition to the findings relating to diarrhoea, the lack of association between receiving a synbiotic and a diagnosis of atopy later in life in the present study reflects the findings of a recent systematic review and meta‐analysis on the effect of probiotics as a treatment for atopic dermatitis, which found limited evidence in support of their use. ²⁰ Without first considering whether individual puppies with diarrhoea and atopy have dysbiosis, it is not clear whether any kind of biotic treatment would be appropriate; therefore, ascertaining whether and when the microbiome is a causal factor in the variation in the occurrence of disease would be beneficial.

AUTHOR CONTRIBUTIONS

Rachel Moxon and Gary England planned the experiment. Rachel Moxon and Madeleine Goumas conducted the analyses and wrote the first draft. Rebecca Hunt collated the data. Rachel Moxon, Madeleine Goumas, Rebecca Hunt and Gary England reviewed and edited the manuscript and gave approval for publication.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.

ETHICS STATEMENT

The study received ethical approval from the University of Nottingham School of Veterinary Medicine and Science, Ethical Review Committee (ethical review number: 1640 151214) on 15 December 2015.

Supporting information

Supporting information

Supporting information

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.