Detection limits of four antimicrobial residue screening tests for β-lactams in goat's milk

Article Outline

• Abstract
• Introduction
• Materials and Methods
  • Animals and Milk Samples
  • Antimicrobial Spiked Samples and Residue Screening Tests
  • Statistical Analysis
• Results and Discussion
• Conclusions
• Acknowledgments
• References
• Copyright

Abstract 

This study was conducted to compare the detection limits (DL) of several antibiotic residue screening tests with the maximum residue limits (MRL) authorized by the EU according to the guidance for the standardized evaluation of microbial inhibitor tests of the International Dairy Federation. Composite antibiotic-free milk samples from 30 primiparous Murciano-Granadina goats in good health condition were used to prepare test samples spiked with different concentrations of each antimicrobial. In total, 5,760 analytical determinations of 10 β-lactam antibiotics (penicillin-G, ampicillin, amoxicillin, cloxacillin, oxacillin, dicloxacillin, cefadroxyl, cefalexin, cefoperazone, and cefuroxime) were performed using 4 antibiotic residue screening tests: the brilliant black reduction test BRT AiM (AiM-Analytik in Milch Produktions-und Vertriebs GmbH, München, Germany), Delvotest MCS (DSM Food Specialties, Delft, the Netherlands), Eclipse 100 (ZEU-Inmunotec SL, Zaragoza, Spain), and the Copan Milk Test (CMT; Copan Italia SpA, Brescia, Italy). For each method, we estimated the detection limits of the antimicrobial agents using a logistic regression model. Using the CMT and Delvotest on samples spiked with the 8 antibiotics for which MRL were available, DL were at or below the MRL. The BRT test provided DL at or below the MRL for all of the agents except cefalexin, whereas the Eclipse 100 method failed to detect 4 antibiotics (ampicillin, amoxicillin, cloxacillin, and cefoperazone) at MRL or below. Logistic regression-determined levels of agreement were highest for the CMT method (98.6 to 100%) and lowest for Eclipse 100 (66.3 to 100%). In general, agreement levels indicated good correlation between observed results and those predicted by logistic regression. The lowest b values (closely related to test sensitivity) were recorded for the cephalosporins (0.074 to 0.430) and highest for penicillin G, ampicillin, and amoxicillin (11.270 to 11.504). Delvotest and CMT best fulfilled IDF criteria for the ideal test for detecting antibiotic residues in milk.

Key words: β-lactam, goat's milk, antibiotic residue, detection limit

Back to Article Outline

Introduction 

The possible presence of antimicrobial residues (AR) in milk poses a risk for consumers because these may cause allergic reactions, intestinal dysbiosis, or even the emergence of resistant bacterial populations (Allison, 1985; Dewdney et al., 1991). Besides consequences for consumers, the presence of AR in milk can cause problems for producers because residues may impair bacterial fermentation processes (Mourot and Loussouarn, 1981) when milk is used to produce fermented dairy products. This problem especially affects sheep and goat's milk because they are often used to produce cheese. Given possible repercussions on public health and the dairy industry, official limits exist for AR and European Union (EU) legislation (European Union, 1990) stipulates maximum residue limits (MRL) for defined antimicrobials. Producers and the dairy industry may also suffer economic losses if a false positive AR result is obtained in an uncontaminated milk sample, and in some countries this may even have legal consequences. It is therefore important that besides showing high specificity, an antimicrobial residue screening test (ARST) should be sufficiently sensitive to detect minimal amounts of antimicrobials but not so sensitive that false-positive results are easily produced.

Although β-lactam antimicrobials are widely used in dairy goats and other dairy ruminants and milk testing programs in dairy goat herds include the determination of several residues in bulk tank milk samples, there is little information in the literature on ARST in goats. Zeng et al. (1996) compared the Delvotest P and Penzyme tests with a standard reference test (BsDA) and concluded that these tests were sensitive and reliable for detecting AR in individual goat milk samples. These same authors (Zeng et al., 1998) also assessed the use of the SNAP test, Charm BsDA, Charm II, and LacTeck for ARST in goat's milk and found that the first 3 were sensitive and reliable for detecting AR but their high sensitivity could lead to false-positive results when AR were below the permitted, or tolerance, level. In contrast, the LacTeck B-L failed to detect tolerance levels of several antimicrobials, including penicillin G, in goat's milk. Given the devastating consequences of false positives (Cullor, 1992), in a previous study we assessed 9 different ARST commonly used in the United States for cow's milk on individual goat's milk samples (Contreras et al., 1997). These tests covered a variety of analytical principles for assaying β-lactam and ceftiofur residues and we were able to conclude that they were capable of adequately identifying goat's milk that was free of antimicrobial residues.

Most screening methods for AR detection in milk are based on inhibiting the growth of Geobacillus stearothermophilus var. calidolactis and have been developed and validated for use in cow's milk. Although some authors have assessed the use of screening kits in sheep's milk (Althaus et al., 2003a,b; Berruga et al., 2003; Molina et al., 2003; Montero et al., 2005; Linage et al., 2007), most of the information available is based on results obtained in cow's milk. The present study was designed to compare 4 commercial screening tests and determine their detection limits (DL) for 10 β-lactams in goat's milk.

Back to Article Outline

Materials and Methods 

Animals and Milk Samples 

Composite milk samples from 30 Murciano-Granadina goats in good health reared on an organic farm were used for the study. To ensure that the milk collected was antibiotic-free, we adapted the requirements of the guidance on the standardized evaluation of microbial inhibitor tests (IDF, 1999). Thus, the goats selected were primiparous, in midlactation and had never been given any form of antimicrobial treatment or feed supplements. Composite milk samples (200mL) from the 30 selected goats were obtained on the morning of milking and transported refrigerated to the laboratory. The samples were then mixed to obtain AR-free milk, divided into 100-mL aliquots and frozen (−20°C) until used to prepare the samples spiked with the different antimicrobial agents. The animals’ health status was good, including no clinical or subclinical IMI (including Mycoplasma spp.). To ensure mammary health (especially in terms of checking for the possible presence of β-lactamase-producing organisms), 1 and 2 d before selecting the animals, individual milk samples were obtained for bacteriological analysis according to the recommendations of the National Mastitis Council (Harmon et al., 1990). This involved collecting 10mL of foremilk aseptically from each udder half. Teats were scrubbed with cotton saturated with 70% ethanol and the first 3 streams of milk were discarded. Ten microliters of each sample were plated on blood agar plates (5% washed sheep erythrocytes). The plates were incubated aerobically at 37°C and examined at 24h and 48h. The presence of 500cfu/mL was taken to indicate an IMI. Similarly, milk samples were tested for Mycoplasma spp. 2 weeks before animal selection by inoculating the samples in modified PH medium broth followed by incubation in a high-humidity CO₂ (5%) atmosphere for 2 d, filtering and subpassaging in broth, and incubating for an additional 9 d. Thus, all the goats finally selected had tested negative for IMI, including Mycoplasma spp., and their milk SCC was lower than 256×10³.

Antimicrobial Spiked Samples and Residue Screening Tests 

Milk samples containing the following β-lactam antimicrobials were prepared following the recommendations of IDF (1999): penicillin-G, ampicillin, amoxicillin, cloxacillin, oxacillin, dicloxacillin, cefadroxyl, cefalexin, cefoperazone, and cefuroxime. These drugs were stored and handled according to the manufacturer's instructions before use. Drugs were dissolved (1mg/mL) in distilled water and 8 different dilutions of spiked milk samples prepared for each antimicrobial, including a blank (Table 1). All dilutions were prepared in 100-mL volumetric flasks on the day of analysis, to avoid possible errors caused by the instability of the solutions. For each antimicrobial, we tested 18 replicates of each dilution using the 4 ARST, to give a total of 5760 analytical determinations.

Table 1. List of the β-lactams used in this study (all supplied by Sigma, St. Louis, MO), their commercial reference numbers, and final concentrations in the spiked milk samples

The screening tests assessed were: the brilliant black reduction test BRT AiM (BRT; AiM-Analytik in Milch Produktions-und Vertriebs GmbH, München, Germany), the Delvotest MCS (Delvotest; DSM Food Specialties, Delft, the Netherlands), the Eclipse 100 (Eclipse; ZEU-Inmunotec SL, Zaragoza, Spain) and the Copan Milk Test (CMT; Copan Italia SpA, Brescia, Italy). These tests are all based on detecting inhibition of the growth of spores of Geobacillus stearothermophilus var. calidolactis C (Gsc). The test medium is a mixture of nutrients, test bacteria (Gsc), a colored indicator (brilliant black for BRT and bromocresol purple for the other ARST) and other supplements. When the milk sample contains no residues or these are under the detection limits, the spores germinate, grow and their metabolic activity makes the indicator change color. In the presence of antimicrobial residues, no bacterial activity will be detected and the indicator color will remain unchanged (blue or purple). The color change can be visually or photometrically assessed. All ARST were carried out in 96-well microtiter plates according to the manufacturer's instructions. Briefly, the spiked milk samples (100μL each) were inoculated in the wells containing culture medium for Gsc and a pH indicator, including a negative and a positive control. The plates were then carefully sealed and placed in a water bath and incubated at 64±1°C for 3h. Results were interpreted visually for CMT and photometrically for the other 3 tests. A spectrophotometer was used at measuring and reference wavelengths of 450 and 620nm for the BRT and 590 and 650nm for Eclipse. The DelvoScan Reader was used for Delvotest. Copan Milk Test results were separately interpreted visually by 2 technicians and to ensure the validity of the results from the 2 technicians, the agreement between their results was estimated using the kappa value (Thrusfield, 1995).

Statistical Analysis 

The DL of each test was determined for each antimicrobial agent by logistic regression using the Logistic procedure of SAS (SAS Institute, 1998). The response observed was considered binary and classified as “positive,” “dubious,” or “negative.” The logistic regression model used was

where logit = linear logistic model, that is, ln[P_ij/(1-P_ij)]; P_ij = probability of “negative” versus “positive plus dubious” results; AC = antimicrobial concentration; a = intercept; b = slope; and ɛ_ij = residual error. Agreement coefficients were estimated as rank correlation between observed and predicted results (Althaus et al., 2003b). The DL was estimated as the concentration at which 95% of results were positive (IDF, 1999).

Back to Article Outline

Results and Discussion 

Tables 2, 3, 4, and 5 summarize the results of the logistic regression model in terms of relative frequencies of each ARST result and the DL value recorded. For 8 out of the 10 antibiotics for which EU MRL were available, CMT and Delvotest showed DL at or below the MRL. The BRT results indicated detection limits at or below the MRL for these 8 antibiotics except cefalexin, whereas Eclipse failed to detect 4 antibiotics (ampicillin, amoxicillin, cloxacillin and cefoperazone) at levels corresponding to MRL or below. Logistic regression agreement levels were highest for CMT (98.6 to 100%) and lowest for Eclipse (66.3 to 100%) indicating overall good correlation between observed results and those predicted by logistic regression. The coefficient b (slope) of the logistic regression line is closely related to the sensitivity of a test to a given antimicrobial agent. Using all 4 ARST, lowest b values were obtained for the cephalosporins (0.074 to 0.430) and highest values for penicillin G, ampicillin, and amoxicillin (11.270 to 11.504). Figures 1, 2, 3, and 4 show the DL results obtained for the 4 ARST assessed in relation to MRL represented as the detection pattern suggested by the IDF (1999). Thus, the ideal test should be capable of detecting 1× MRL of all antimicrobials of concern, which is indicated in the figures by the middle circle. The DL of the test under study is indicated by the circle in bold. It may be observed that Delvotest and CMT best fulfilled IDF criteria for the ideal AR test.

Table 2. Logistic regression model variables and detection limits of the β-lactams obtained using Delvotest MSC (DSM Food Specialties, Delft, the Netherlands) in goat's milk considering a binary response variable

1MRL = maximum residue limits.

Table 3. Logistic regression model variables and detection limits of the β-lactams obtained using the Brilliant Reduction Test (BRT AiM, AiM-Analytik in Milch Produktions-und Vertriebs GmbH, München, Germany) in goat's milk considering a binary response variable

1MRL = maximum residue limits.

Table 4. Logistic regression model variables and detection limits of the β-lactams obtained using Eclipse 100 (ZEU-Inmunotec SL, Zaragoza, Spain) in goat's milk considering a binary response variable

1MRL = maximum residue limits.

Table 5. Logistic regression model variables and detection limits of the β-lactams obtained using the Copan Milk Test (Copan Italia SpA, Brescia, Italy) in goat's milk considering a binary response variable

1MRL = maximum residue limits.

• 
  • View Large Image
  • Download to PowerPoint

• Figure 1. 

  Detection pattern obtained for Delvotest MSC (DSM Food Specialties, Delft, the Netherlands) used to detect residues of the 8 β-lactams. Detection limits are indicated as n-fold the EU maximum residue limits expressed in μg/kg (inner circles ≥10×; middle circles 1×; outer circles 0.1×).

• 
  • View Large Image
  • Download to PowerPoint

• Figure 2. 

  Detection pattern obtained for BRT AiM (AiM-Analytik in Milch Produktions-und Vertriebs GmbH, München, Germany) used to detect residues of 8 β-lactams. Detection limits are indicated as n-fold the EU maximum residue limits expressed in μg/kg (inner circles ≥10×; middle circles 1×; outer circles 0.1×).

• 
  • View Large Image
  • Download to PowerPoint

• Figure 3. 

  Detection pattern obtained for Eclipse 100 (ZEU-Inmunotec SL, Zaragoza, Spain) used to detect residues of 8 β-lactams. Detection limits are indicated as n-fold the EU maximum residue limits expressed in μg/kg (inner circles ≥10×; middle circles 1×; outer circles 0.1×).

• 
  • View Large Image
  • Download to PowerPoint

• Figure 4. 

  Detection pattern obtained for the Copan Milk Test (CMT; Copan Italia SpA, Brescia, Italy) used to detect residues of 8 β-lactams. Detection limits are indicated as n-fold the EU maximum residue limits expressed in μg/kg (inner circles ≥10×; middle circles 1×; outer circles 0.1×).

Delvotest is a simple, rapid, and economic ARST, and because of these features it is commonly used to detect antimicrobial residues in cow's milk. Despite scarce data on the use of this test to detect antibiotic residues in goat's milk, it is presently widely used for this purpose (Zeng et al., 1996). In separate assessments of Delvotest-P by Zeng et al. (1996) and Contreras et al. (1997) using individual goat's milk samples, no relationship was detected between the SCC and the results of the antibiotic detection test, despite contradictory results obtained in terms of false-positive results. Thus, Zeng et al. (1996) obtained 7% false-positive results whereas we observed no false positives (Contreras et al., 1997). Using Delvotest in sheep's milk, Althaus et al. (2003a) recorded 2.3% false-positive results and observed lower proportions of total solids and higher SCC in these false positive samples. Recently, Stead et al. (2008) validated the Delvotest SP-NT in cow's milk and no false positives or negatives were obtained. These authors also determined that this test could be used to AR screen a wide range of milk types in terms of fat content or SCC including goat's milk, with the exception of milk of low pH (spoiled milk).

The detection limits of Delvotest in cow's milk have been visually assessed by several authors. Le Breton et al. (2007) compared visual and scanner readings, concluding that the automated reader (scanner) is a useful and convenient tool. Our results indicate that Delvotest MSC is useful for detecting β-lactam residues in goat's milk because it was able to detect residues of all the β-lactams included in our study below or at the MRL (Table 2 and Figure 1). The DL for ampicillin and amoxicillin were in line with their EU MRL whereas those determined for the remaining agents were below the MRL. These results are in agreement with those reported by Althaus et al. (2003b) who estimated the Delvotest SP detection limits in sheep's milk of the photometric measurement of 4 β-lactams (penicillin G, amoxicillin, ampicillin, and cloxacillin). Although in both this study and our study residues of these 4 antibiotics below EU MRL were detected, our results were closest to the MRL (2μg/kg in our study versus 1μg/kg for penicillin G; 4μg/kg versus 2μg/kg for ampicillin; 4μg/kg versus 3μg/kg for amoxicillin; and 23μg/kg versus 18μg/kg for cloxacillin).

The BRT is another test based on microbiological inhibition widely used for the detection of AR in milk and has been marketed by different manufacturers under several trade names. Its DL for antibiotic residues has been widely established in cow's milk and there are a few reports available for sheep's milk, whereas no information exists for goat's milk. Using the BRT in sheep's milk, Althaus et al. (2003a) observed 3.7% false-positive results, and a high SCC was correlated with a false-positive result. Molina et al. (2003) determined 7 β-lactams at EU MRL in sheep's milk and only penicillin G residues below the MRL were detected. Our results indicate that the BRT is accurate at detecting β-lactam residues in goat's milk at or below EU MRL, with the exception of cefalexin (Table 3 and Figure 2). The other 7 β-lactams with established MRL were detected at or below this limit. The amoxicillin DL (4μg/kg) coincided with the MRL, whereas DL of the 6 remaining antibiotics were below the corresponding MRL: penicillin G (2μg/kg), ampicillin (3μg/kg), cloxacillin (27μg/kg), dicloxacillin (8μg/kg), oxacillin (8μg/kg), and cefoperazone (44μg/kg). Only the DL of cefalexin (336μg/kg) was higher than the MRL. According to these results and contrary to the results of the Delvotest in sheep's milk reported by Molina et al. (2003), the BRT is likely to be better at detecting β-lactam residues at or below the DL in goat's milk than in sheep's milk. Thus, in our study the BRT detected 7 β-lactams at or below the MRL and only its DL for cefalexin was higher than the official limit, whereas Molina et al. (2003) found that the BRT in sheep's milk was only capable of detecting penicillin G at levels below the EU MRL. For the other 6 β-lactams studied by these authors, the levels detected were above the EU MRL; 3 being close to the MRL (amoxicillin, ampicillin, and ceftiofur) and the remaining 3 well beyond the maximum limit (cloxacillin, cephalexin, and cefoperazone).

The results obtained for the Eclipse 100 test indicate that half the β-lactams studied were detected at levels below the MRL and the other half were close to, but above, this level (Table 4 and Figure 3). Thus, DL below the MRL were recorded for penicillin G (3μg/kg), dicloxacillin (27μg/kg), oxacillin (11μg/kg), and cefalexin (96μg/kg) and DL just above the MRL were observed for ampicillin (5μg/kg), amoxicillin (5μg/kg), and cloxacillin (42μg/kg). The DL determined for cefoperazone (97μg/kg) was double the MRL. The Eclipse test was developed for use in cow and sheep's milk. Montero et al. (2005) used this test in sheep's milk to establish the DL of 6 β-lactams with defined MRL and only oxacillin (28μg/kg) was detected at a level below the limit. Penicillin G (5μg/kg) and cefalexin (115μg/kg) were detected just beyond the limit and the DL established for amoxicillin (7μg/kg), cloxacillin (68μg/kg), and cefoperazone (110μg/kg) were almost twice the official limits. The variation observed here in cloxacillin residue detection (agreement was 66.3%) should be noted, because cloxacillin is widely used for dry treatment in cows as well as goats and sheep. When used to test sheep's milk, samples are diluted to avoid the effect of excess fat. We did not dilute our samples because the fat content of goat's milk is relatively lower. The Eclipse test was the least sensitive for detecting β-lactam residues in goat's milk and would need to be improved for AR screening, as suggested by Montero et al. (2005) for sheep's milk.

Contrary to the other ARST examined, there are no published data on the use of the CMT to test the milk of small ruminants. Because this test was visually read, we checked to ensure that there was agreement between the 2 involved technicians and the agreement was complete (kappa coefficient = 1). Using this test to determine the 8 β-lactams with available MLR, residue levels were always below the MRL (Table 5 and Figure 4). Test sensitivity was high, because cloxacillin and cefoperazone detection limits were close to MRL but the other DL were found to be lower than the MRL, and more than twice this limit for dicloxacillin, oxacillin and cefalexin. Similar results have been reported by Le Breton et al. (2007) for the β-lactams penicillin G, cloxacillin, and cefalexin in cow's milk. This high sensitivity has the benefit of guaranteeing the safety of milk for consumers, but will have the drawback of giving rise to many false positives with the consequent economic losses for producers.

Back to Article Outline

Conclusions 

Most of tests used were able to detect β-lactam residues at levels 1× the EU stipulated MRL or below. The Delvotest and CMT were the tests that best fulfilled IDF criteria for the ideal AR test. Despite one of the tests being unable to detect levels of 3 of the antimicrobials below the MRL (ampicillin, amoxicillin, and cloxacillin), our results indicate that under the analytical conditions specified by the IDF, the screening methods examined here for detecting β-lactam residues can be reliably used in goat's milk.

Back to Article Outline

Acknowledgments 

This study was funded by project AGL2006–03105GAN awarded by the Dirección General de Investigación (Spanish Ministry of Science and Education) and project 05693/PI/07, by the Fundación SENECA (Agencia Regional de Ciencia y Tecnología de la Región de Murcia, España).

Back to Article Outline

References 

1. Allison JRD. Antibiotic residues in milk. Br. Vet. J. 1985;141:9–16
   • View In Article
   • MEDLINE
   • CrossRef
2. Althaus R, Torres A, Peris C, Beltrán M, Fernández N, Molina MP. Accuracy of BRT and Delvotest microbial inhibition tests as affected by composition of ewe's milk. J. Food Prot. 2003;66:473–478
   • View In Article
3. Althaus RL, Torres A, Montero A, Balasch S, Molina MP. Detection limits of antimicrobials in ewe milk by Delvotest photometric measurements. J. Dairy Sci. 2003;86:457–463
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (155 KB)
   • CrossRef
4. Berruga MI, Yamaki M, Althaus RL, Molina MP, Molina A. Performances of antibiotic screening tests in determining the persistence of penicillin residues in ewe's milk. J. Food Prot. 2003;66:2097–2102
   • View In Article
5. Contreras A, Paape MJ, Di Carlo AL, Miller RH, Rainard P. Evaluation of selected antibiotic residue screening tests for milk from individual goats. J. Dairy Sci. 1997;80:1113–1118
   • View In Article
   • Abstract
   • Full-Text PDF (249 KB)
   • CrossRef
6. Cullor JS. Cowside testing for antibiotic residues: Problems and solution. In: Proc. Natl. Mastitis Counc., Arlington, VA. Natl. Arlington, VA: Mastitis Counc. Inc.; 1992;p. 153
   • View In Article
7. Dewdney JM, Maes L, Raynaud JP, Blanc F, Scheid JP, Jackson T, et al. Risk assessment of antibiotic residues of beta-lactams and macrolides in food-products with regard to their immunoallergic potential. Food Chem. Toxicol. 1991;29:477–483
   • View In Article
   • MEDLINE
   • CrossRef
8. European Union. 1990. Council Regulation (EEC) No. 2377/90 of 26 June 1990 laying down a Community procedure for the establishment of maximum residue limits of veterinary medical products in foodstuffs of animal origin. Off. J. L 224:1–8.
   • View In Article
9. Harmon RJ, Eberhart RJ, Jasper DE, Langlois BE, Wilson RA. Microbiological Procedures for the Diagnosis of Bovine Udder Infection. 3rd ed.. Arlington VA: Natl. Mastitis Counc. Inc.; 1990;
   • View In Article
10. International Dairy Federation. Guidance for the standardized evaluation of microbial inhibitor tests. Brussels, Belgium: Int. Dairy Fed.; 1999;FIL-IDF Standard No. 183
    • View In Article
11. Le Breton MH, Savoy-Perroud MC, Diserens JM. Validation and comparison of the Copan Milk Test and Delvotest SP-NT for the detection of antimicrobials in milk. Anal. Chim. Acta. 2007;586:280–283
    • View In Article
    • CrossRef
12. Linage B, Gonzalo C, Carriedo JA, Asensio JA, Blanco MA, De La Fuente LF, et al. Performance of blue-yellow screening test for antimicrobial detection in ovine milk. J. Dairy Sci. 2007;90:5374–5379
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (131 KB)
    • CrossRef
13. Molina MP, Althaus RL, Molina A, Torres A, Fernandez N. Antimicrobial agent detection in ewe milk by the microbial inhibitor test (brilliant black reduction test–BRT-AiM). Int. Dairy J. 2003;13:821–826
    • View In Article
14. Montero A, Althaus RL, Molina A, Berruga I, Molina MP. Detection of antimicrobial agents by a specific microbiological method (Eclipse 100) for ewe milk. Small Rumin. Res. 2005;57:229–237
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (125 KB)
    • CrossRef
15. Mourot D, Loussouarn S. Sensibilité des ferments lactiques aux antibiotiques utilisés en médicine vétérinaire. Rec. Med. Vet. (Kiev). 1981;157:175–177
    • View In Article
16. SAS Institute. 1998. SAS/STAT Software. Changes and Enhancements Through Release 6.11. SAS Inst. Inc., Cary, NC.
    • View In Article
17. Stead SL, Ashwin H, Richmond SF, Sharman M, Langeveld PC, Barendse JP, et al. Evaluation and validation according to international standards of the Delvotest (R) SP-NT screening assay for antimicrobial drugs in milk. Int. Dairy J. 2008;18:3–11
    • View In Article
18. Thrusfield M. Veterinary Epidemiology. 2nd ed. Oxford: Blackwell Science Ltd.; 1995;
    • View In Article
19. Zeng SS, Escobar EN, Brown Crowder I. Evaluation of screening tests for detection of antibiotic residues in goat milk. Small Rumin. Res. 1996;21:155–160
    • View In Article
    • Abstract
    • Full-Text PDF (559 KB)
    • CrossRef
20. Zeng SS, Hart S, Escobar EN, Tesfai K. Validation of antibiotic residue tests for dairy goats. J. Food Prot. 1998;61:344–349
    • View In Article