| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
* School of Veterinary Medicine, University of Wisconsin, Madison 53706
Chr. Hansen’s Biosystems, Milwaukee, WI 53214
Spruce Haven Farm and Research Center, Union Springs, NY 13160
1 Corresponding author: groetzel{at}wisc.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
Key Words: dairy cow • direct-fed microbial • milk yield • milk component
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
Enterococcus faecium produces moderate amounts of lactic acid in the rumen. This could stimulate growth of lactic acid utilizers and stabilize ruminal pH. Nocek et al. (2002) fed a combination of Enterococcus faecium and Lactobacillus plantarum with Saccharomyces cerevisiae to lactating cows receiving a diet high in NFC. Cows fed this DFM combination to 1 x 105 cfu/mL of ruminal fluid had higher ruminal pH nadirs than cows fed to 1 x 106 or 1 x 107 cfu/mL.
A DFM combination providing a daily dose of 5 x 109 cfu of 2 strains of Enterococcus faecium and 5 x 109 cfu yeast has been evaluated in 2 controlled studies in transition cows. In both studies DFM were fed from about 21 d prepartum through 70 d postpartum. In the first study (Nocek et al., 2003), the DFM combination significantly reduced early lactation drop in ruminal pH and increased DMI, milk yield, and milk protein percentage in early lactation. Blood glucose and insulin concentrations were significantly higher, and blood NEFA concentrations were numerically lower for cows that received the DFM. In the second study (Nocek and Kautz, 2006), the DFM combination significantly enhanced ruminal digestibility of forage, increased milk yield, increased blood glucose, and decreased blood BHBA.
The objective of this study was to determine the effects of a supplemental DFM combination (2 strains of Enterococcus faecium and yeast) on the performance of transition dairy cows in a field trial on a commercial dairy.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
The study was conducted over a 29-wk period from October 2000 to April 2001. Cows were enrolled in 1 of 4 groups (2 placebo and 2 DFM-supplemented groups). Groups were enrolled consecutively. Sequence of treatment assignments was randomly initiated with the B treatment and followed the sequence B-A-B-A. The identity of the A and B treatments was not revealed to the dairy producer or to the investigators until the study was completed. All eligible cows that consumed the specified diets for at least 10 d before calving and at least 23 d after calving were enrolled in the study. Cows were moved into the prefresh group approximately 21 d before expected calving. Data were collected from each cow until 85 d after calving or until the cow was removed from the herd.
Treatments were switched after approximately 40 second-lactation cows were enrolled in a group. Many cows necessarily received both treatments before calving and therefore were not eligible for enrollment in the study. After several weeks, cows that had received only one diet could again be enrolled.
Cows within the 4 groups were divided into 2 study populations based on parity. The main experimental population consisted of 163 cows entering their second lactation (about 40 cows per group). Cows in the main study population were intensively monitored for precalving NEFA, BCS, milk production, milk components, estimated DM intake, and postcalving BHBA. A secondary population of 132 first-lactation and 71 second- or greater lactation cows was less intensively monitored for milk production and milk components only. The main and secondary populations of cows were enrolled in the study at the same time and received the same diets. The number of cows in the secondary study populations was not targeted and consisted of any eligible cows that consumed the proper diet during the same time period as the main study population.
Temperature and humidity were recorded every 10 min in the pre- and postfresh pens every day of the study using an automated recording device (Ryan Instruments, HAT Monitor, Redmond, WA). Average temperature and relative humidity values during the study for each cow were calculated based on the days spent in each pen.
On-farm personnel recorded pen moves daily, and a backup of the on-farm record system (Dairy Comp 305, Valley Agricultural Software, Tulare, CA) was created each day. This information was used to determine each cow’s daily pen location, dates of pen moves, and daily stocking density for each pen (number of cows in the pen that day divided by the number of free-stalls in the pen). The average stocking density experienced by each cow during the pre- and postfresh periods was calculated from the daily pen stocking densities and each cow’s daily pen location. Each pen contained 2 rows of free-stalls (except for the maternity pen, which was loose housing) and a single row of headlocks at the feed bunks. The mean number of headlocks in the pens was 11% greater than the mean number of free-stalls.
The TMR diets offered to the pre- and postfresh groups were sampled weekly at 10 locations along each feed bunk. Feed was sampled immediately after delivery, and samples were collected only from undisturbed portions of the bunk. The composite bunk sample was reduced after mixing into a smaller subsample. Analysis for standard nutrients was made on the subsample using wet chemistry procedures (Cumberland Valley Analytical Services, Maugansville, MD).
Dry matter intake was estimated for each pen by subtracting the estimated amount of feed refused from the amount of feed offered each day. Weights of feed offered and refused each day were obtained from the scales on the feed mixing wagon. Accuracy of the TMR mixer scales was verified monthly on a platform scale with the mixer empty, half-empty, and nearly full. Dry matter intake was adjusted for the number of primiparous cows in each pen using an intake adjustment factor of 0.908 for the prefresh period and 0.808 for the post-fresh period. Adjustment factors were derived from data contained in a dairy ration software model (National Research Council, 2001). Estimated DMI was calculated for each cow based on the days she was present in each pen.
Data Collection and Sampling
Daily milk production for each postpartum cow was recorded using computerized weigh meters in the parlor. Milk weights were collected starting on the third day after calving until 85 DIM. Composite milk samples were collected weekly from cows in the main study population up to 23 DIM during the 0400 h milking using an automated sampler (Milk Meter Sampling Device, WestfaliaSurge Inc., Naperville, IL). The WestphaliaSurge dealer calibrated the milk meters at the beginning of the study according to company calibration guidelines.
Additional milk samples from the main study population were collected monthly as part of the herd’s routine DHI testing program. Cows in the secondary study populations were sampled monthly for milk components through d 85. Milk samples from all study populations were refrigerated and analyzed for fat and protein content using infrared techniques by the Ag-Source Milk Analysis Laboratory in Menomonie, WI (MilkOScan 4000 Infrared Analyzer; Foss Technology, Hillerød, Denmark).
Blood samples were collected once weekly via the coccygeal vein from cows in the main study population from the time they entered the prefresh group until 23 DIM. Blood samples from prefresh cows were collected into heparinized tubes and immediately placed on ice. Tubes were later centrifuged at 2,500 x g for 10 min. Plasma was separated, immediately frozen, and later submitted for NEFA analysis. Blood from postfresh cows was collected into tubes without additive and allowed to clot at room temperature. Tubes were later centrifuged at 2,500 x g for 10 min and the serum separated and refrigerated. Serum was later analyzed for BHBA concentration. Blood analyses for BHBA and NEFA were conducted at a commercial laboratory (Marshfield Clinic Veterinary Diagnostic Services, Marshfield, WI) using a Roche Hitachi 911 chemistry analyzer (Roche Diagnostics Corp., Indianapolis, IN), following the procedures of Williamson et al. (1962) for BHBA analysis and the Wako NEFA-C test kit (Wako Chemicals USA, Richmond, VA) for NEFA analysis.
Cows in the main study population were evaluated for BCS using a 5-point scale (Edmonson et al., 1989). Prefresh cows were sampled and scored weekly. However, only the last sample prior to calving was recorded and used in the analysis. If a cow calved less than 48 h after the last prefresh period sample was collected, then her next to last score was recorded instead. Each cow’s body condition loss from her prepartum score to her last score (third week after calving) was calculated and recorded as her body condition score drop.
The proportion of cows above threshold concentrations for NEFA and BHBA were also recorded and evaluated. Thresholds concentrations were greater than 400 µM for NEFA and greater than 1,400 µM for BHBA (Oetzel, 2004).
Health outcomes (displaced abomasum, antibiotic treatments excluding treatment of mastitis, and removal from the herd) were recorded until 85 DIM for cows in the main study population. Health outcomes were determined and recorded by the producer. Antibiotic treatments due to clinical mastitis were not included in the analysis. The 85 DIM limit was chosen because it encompassed the time of peak milk yield for most cows. Supplementation of the DFM product stopped at about 23 DIM; therefore, any effects observed from 23 to 85 DIM would indicate a residual treatment effect.
Statistical Analysis and Experimental Design
Data were analyzed by ANOVA using a completely randomized design with subsampling. The experimental unit (i.e., the smallest unit to which the treatment was applied randomly) was the group of cows (n = 4 for each study population). The error term used for testing the treatment effect was treatment by group. Analyses were conducted using SAS Release 8.02 for Windows (SAS Institute, 1999).
Continuous Outcomes with Single Measures.
Effect of DFM on continuous outcomes with single measures (or outcomes calculated as single means) was determined using mixed models. Covariates available for inclusion in the final models were recorded on an individual cow basis and included previous lactation 305-d mature equivalent milk production, diet composition, temperature, humidity, stocking density, days spent in the prefresh group, and days spent in the maternity pen. Because there were up to 44 possible covariates per outcome, covariates were first screened for inclusion in the final models by evaluating their correlation to each outcome using the CORR procedure of SAS. Covariates with significant Pearson product-moment correlations (P < 0.05) were included in a backwards elimination model in the REG procedure of SAS. Covariates included (P < 0.10) in the regression model were made available for the final mixed model. Group was entered into the model as a random effect. The final mixed model, using the MIXED procedure of SAS, was determined by manual backward elimination of eligible covariates until all remaining covariates were significant in the model (P < 0.05).
Residual plots were visually evaluated as a test for the assumption of normal distribution of the data. Data for BHBA and NEFA were transformed (natural log) to reduce heteroscedasticity of the residual plots. The least squares means and standard errors for these outcomes were reported from model results using untransformed data.
Continuous Outcomes with Repeated Measures.
Effect of DFM on continuous outcomes with repeated measures (daily milk yield, milk fat test, milk fat yield, milk protein test, and milk protein yield) was determined using repeated measures analysis in mixed models as described by Littell et al. (1998). The model included auto-correlation using the spatial powers covariance structure in the MIXED procedure of SAS. Data available for this analysis included milk components that were measured on wk 1, 2, and 3 plus additional DHI tests before 85 DIM. All available daily milk weights from 3 to 85 DIM were also included in the analysis. Covariates based on DIM considered for inclusion in the mixed model for repeated measures in the main study population were the linear, squared, cubic, and quadratic effects of DIM, the inverse of the same, and all interactions of DIM variables with treatment. These were screened using a backwards elimination model in the REG procedure of SAS. Covariates eligible for inclusion in the mixed models were those included in the regression model (P < 0.10) plus previous lactation 305-d mature equivalent milk production, stocking density of the prefresh group, and days spent in the maternity pen. The model was reduced by manual backward elimination of eligible covariates until all were significant in the final model (P < 0.01 for outcomes with denominator degrees of freedom > 500, and P < 0.05 for all other outcomes). Some cubic transformations of DIM could not be included in the final models because they rendered the least-squares means nonestimable. Excluding these terms from the models resulted in negligible changes in P-values for treatment effects.
Repeated measures data from cows in the secondary populations were analyzed using similar methods, except that inverse measures of DIM were not considered, and all eligible covariates were entered into the mixed models without prior screening in a regression analysis. Because previous ME milk production was a key covariate for milk yield in the second-lactation cows and because first-lactation animals had no record of previous milk yield, their data were analyzed separately from the data for the second and greater lactation cows. Lactation and treatment by lactation were eligible covariates for models involving the second and greater lactation cows.
Residual plots from the repeated measures analyses were visually evaluated as a test for the assumption of normal distribution of the data. There was no evidence of heteroscedastity in the residual plots, and no data transformations were necessary.
Proportional Outcomes with Single Measures.
Effect of DFM on proportional outcomes with single measures (NEFA concentration >400 µM, BHBA concentration >1,400 µM, removal from the herd, antibiotic treatment, and displaced abomasum) was determined using logistic regression in a mixed model (Glimmix macro of SAS). Covariates with Pearson product-moment correlations P < 0.05 were eligible for inclusion in the final logistic regression model. Group was considered a random effect in the model. Covariates were eliminated by manual backwards elimination until all were significant (P < 0.05) in the final model.
Criteria for Significance.
Only 2 degrees of freedom were available for testing treatment effects in all of the models. Significance was declared at P < 0.10 for treatment effects, unless otherwise noted, because of the few degrees of freedom. Values presented are least-squares means with the pooled standard error of the difference.
|
RESULTS AND DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
. Values for all variables did not differ (P > 0.10) by treatment. No maximum temperature exceeded the upper critical temperature of about 21°C for lactating dairy cows.
|
. Composition of prepartum diets for the placebo and DFM treatments generally did not differ (P > 0.10). The placebo treatment had significantly lower crude fat concentration, although both treatments were below the maximum recommendation of 4.5% dietary fat, DM basis (National Research Council, 2001). Samples of prepartum TMR containing the DFM product had significantly lower DCAD. All prepartum diets had relatively low DCAD due to the addition of anionic salts.
|
The cause of significant differences between the composition of diets containing the placebo or the DFM product was not investigated. Variation in feed ingredient composition is expected in the time period between sampling and analyzing a feed ingredient, so some differences between diets over time were expected.
Measurement of nutrient intakes was important because placebo and DFM-supplemented groups were not concurrent in this study design. Intake of all measured nutrients was estimated on an individual cow basis and included as potential covariates in the final models for treatment effects. Most nutrients were included as significant covariates in one or more of the final models (data not shown), indicating that for this field trial it was important to intensively collect information about diet and include it in the statistical analysis. There was no apparent pattern to the inclusion of nutrient intakes in the final models; this indicated that feed delivery and nutrient intakes were well controlled on the farm and did not overly influence study outcomes.
Prepartum Performance
Estimated DM intake and average blood NEFA concentration were not affected (P > 0.10) by DFM supplementation (Table 3). Groups receiving the DFM supplementation had numerically higher DMI and numerically lower but statistically nonsignificant NEFA concentrations. Nocek et al. (2003) also reported no effect of DFM supplementation on precalving NEFA.
|
All cows lost BCS (about 0.41 units) from the last score taken before calving to the third week of lactation (Table 3
). Treatment did not affect (P > 0.10) BCS loss.
Postpartum Performance
Average and peak milk yield, DIM at peak, milk components, estimated DM intake, and blood BHBA results from the main population of cows for the first 3 wk of lactation are presented in Table 3. No outcomes were affected (P > 0.10) by treatment. Postpartum outcomes were also evaluated separately for wk 1, 2, and 3 (data not shown). No outcomes were affected by treatment on wk 1, 2, or 3.
Repeated measures analyses indicated that milk fat percentage was not affected by DFM supplementation for the main population of cows from 3 to 85 DIM (Table 4) and for all the second- and greater lactation cows (Table 5). Milk fat percentage was higher (P < 0.10) for first-lactation cows that received the DFM product (Table 5) and particularly in very early lactation (Figure 1). The interaction terms treatment x DIM and treatment x DIM2 were included (P < 0.05) in the final model for milk fat percentage in first-lactation cows.
|
|
|
Nocek et al. (2003) reported numerically but not significantly increased milk fat percentage when the DFM product was supplemented. The magnitude of the increase was greatest in the first week of lactation, which is consistent with the results of the current field study. In contrast, DFM supplementation significantly decreased milk fat percentage in another trial (Nocek and Kautz, 2006). These results indicate an inconsistent effect of the DFM product on milk fat percentage. Other factors or interactions may be more important determinants of milk fat percentage in early lactation.
Supplementation with the DFM product increased (P < 0.10) milk protein percentage from 3 to 85 DIM for the main study population (Table 4) and for the second and greater lactation cows (Table 5). Increases in milk protein percentage were greatest in the first weeks of lactation (Figure 2). The interaction terms treatment x DIM and treatment x DIM2 were included (P < 0.01) in the final models for milk protein percentage.
|
Nocek at al. (2003) also found increased milk protein percentage when the DFM product was supplemented, and the increase was of a greater magnitude (about 0.15 percentage units). The improvement in milk protein percentage was not present immediately after calving but increased with DIM. Nocek and Kautz (2006) reported numerically increased milk protein (0.01 percentage units) in cows receiving the DFM product. These cows had considerably increased milk yield, which could explain why the increase in milk protein percentage was not greater. Improved rumen function and increased microbial protein yield could account for the consistent findings of improved milk protein percentage or milk protein yield in cows receiving the DFM product.
Cows supplemented with DFM had fewer antibiotic treatments (P < 0.10) than cows receiving the placebo (Table 6). Reasons for antibiotic treatment of the DFM-supplemented cows were metritis (13 cows), lameness (5 cows), displaced abomasum (1 cow), and unknown infection (1 cow). Reasons for antibiotic treatments in the cows receiving the placebo were metritis (21 cows), lameness (3 cows), displaced abomasum (1 cow), and unknown infection (10 cows). These results should be interpreted with caution. All diagnoses and treatment decisions were made by the producer, and detailed reasons for antibiotic treatments were not recorded. Other proportional outcomes were not affected by DFM. More detailed health data and larger sample sizes would be needed to clarify the effect of DFM on the general health of dairy cows during early lactation.
|
through 5). Previous studies with the same DFM product documented significant increases in milk yield of 4.6% (Nocek et al., 2003) and 6.0% (Nocek and Kautz, 2006). Sample size in the current study was very small and lacked statistical power to declare significance for similar numerical increases. Diets in this trial averaged about 29.3% NFC in the prepartum groups and about 32.4% NFC (DM basis) in the postfresh groups. These diets may not have been high enough in NFC to cause a ruminal acidosis problem, which the DFM product was designed to avert. General guidelines for providing adequate fiber in the diet and maintaining optimal DMI include total NDF between 25 to 35%, maintaining a minimum of 18% forage NDF, and feeding 33 to 40% NFC (Valadares Filho et al., 2000).
Field Study Design
Nonregulatory feeding trials in commercial settings can be useful in evaluating new nutritional supplements. Field studies can estimate the frequency of success under a wide range of conditions and suggest management and environmental factors that impact such success. Results can be used to determine potential economics impacts of a new practice and serve as a basis to make changes in protocols on real farms. Field studies are generally not appropriate for elucidating mechanisms or modes of action (St-Pierre and Jones, 1999).
Field studies with an on-off design are appealing because they can be conducted in commercial dairies without splitting transition cows into concurrent treatment and control groups. However, it can be practically challenging to acquire sufficient experimental units in an on-off field study. This limitation may be somewhat offset by reduced variance among experimental units, because each data point is a group mean (St-Pierre and Jones, 1999).
Another inherent difficulty in field studies is lack of control over management decisions. This study was constrained by management factors that were outside of the control of the investigators. The initial study design for this study called for the enrollment of 8 treatment groups of approximately 40 cows each. Power estimates using bootstrap simulation techniques showed that a difference of 1.7 kg of daily milk could be declared significant (P < 0.05 and a power estimate of 0.80) with this sample size. However, our final sample size was only 4 groups of cows.
Several factors contributed to the smaller than expected number of treatment groups enrolled in the study. The study was designed to take advantage of the cooler seasons so that heat stress would not confound results. However, weather conditions still caused difficulties in this study. In December 2000, extremely cold temperatures (Figure 3) caused freezing of the manure flush system. Passageways were covered with ice, making it difficult for the cows to walk. We observed cows struggling to reach their stalls and staying in them for long periods of time once they reached them. To avoid bias caused by these unusual weather conditions, the group of cows enrolled in the study during this time period was removed from the analysis.
|
The identity of the treatments was revealed after the study ended. The farm managers had correctly assumed that treatment A contained the DFM product. Statistical analysis of herd data then confirmed that the number of antibiotic treatments was significantly higher when the placebo was fed (Table 6). This illustrates a common difficulty in field studies—if the product tested is obviously better than the placebo, it is difficult (and in some circumstances unethical) to continue the study.
The initial power estimates also did not account for the unexpectedly large variation in individual cow milk yields. The placebo group had a standard deviation of 8.2 kg for milk yield from 3 to 23 DIM. Our statistical power estimates (using data from previous studies) used 6.8 kg as the standard deviation for daily milk yield. The larger standard deviation meant that 20 groups (instead of 8) would have been needed to detect a milk yield response of 1.7 kg.
Intensive collection of covariate data (individual cow data, environmental data, and nutrient intake data) is essential for on-off study designs. Covariates included in the final models for this study were inspected for their extent and pattern of inclusion. Milk yield in the previous lactation was the key covariate for the main study population (data not shown). This covariate was highly significant (P < 0.001) in almost every measure of milk yield evaluated. It was also included in most of the milk fat and protein yield final models. Using the final repeated measures model for milk yield from 3 to 85 DIM for the second and greater lactation cows, an increase of 500 kg of previous ME milk was associated with an increase in daily milk yield of 0.8 kg (about 1.8%). The direction of this outcome was expected, but its magnitude was unexpectedly large and illustrates the importance of including previous milk yield in the analysis of this field study.
Body condition score prior to calving was also an important covariate in this study (data not shown). Increasing body condition score prior to calving was consistently associated (P < 0.05) with increased body condition drop after calving, increased blood BHBA concentration, and increased milk yield. Increasing dry cow body condition score 0.25 units increased milk yield (from 3 to 85 DIM) by 0.8 kg for the main study population and increased blood BHBA concentration by about 100 µM. These results were expected, and the inclusion of BCS as a covariate in this field study was important.
Days spent in the maternity pen, where most cows were moved several days before expected calving, were included in many of the final models and had negative implications (data not shown). Increasing days in the maternity pen increased prefresh blood NEFA concentrations, increased postfresh BHBA concentrations, decreased milk protein content, decreased milk protein yield, and increased risk for removal from the herd by 85 DIM (P < 0.05). Based on the final model for average blood BHBA concentration, increasing time in the maternity pen by 2 d increased BHBA by 50 µM. These results indicate that it is important to include pen movement in the data collected from a field study and that prolonged stays in a maternity pen prior to calving may negatively affect dairy cow health and production.
Stocking density of the prefresh pen was also an important covariate in the statistical analysis of this field trial. Increased stocking densities prior to calving decreased (P < 0.05) milk yield after calving. The final model for 3 to 85 DIM milk yield for the first lactation cows showed that a 10% increase in prefresh stocking density resulted in a 0.7-kg decrease in daily milk yield. These results indicate that pen stocking density should be recorded or controlled in field studies and that increasing stocking densities may adversely affect subsequent milk production.
Actual days prior to calving for the last NEFA sample and actual DIM for the first milk component test were highly significant in the final models for NEFA and wk-1 milk components, respectively. These results indicate that actual days relative to calving should be recorded and included as covariates in the statistical analyses of field studies with weekly sample collection. Many outcomes change rapidly around calving, and it is not sufficient to assume that they are equal if collected within the same week.
Interpretation of the covariate data from this study should be approached with caution because the observed effects are associative and do not establish causation. The study was designed to prospectively control supplementation with the DFM product; all other effects were evaluated retrospectively.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
|
ACKNOWLEDGEMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
Received for publication July 28, 2006. Accepted for publication November 6, 2006.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
—). The direct-fed microbial product was Probios TC (Chr. Hansen, Milwaukee, WI), which contained 5 x 109 cfu of bacteria (2 specific Enterococcus faecium strains) and 5 x 109 cfu of yeast. Pooled SE = 0.04%.
