Latent class evaluation of a milk test, a urine test, and the fat-to-protein percentage ratio in milk to diagnose ketosis in dairy cows

In this study, 3 commonly used tests to diagnose ketosis were evaluated with a latent class model to avoid the assumption of an available perfect test. The 3 tests were the KetoLac BHB (Sanwa Kagaku Kenkyusho Co. Ltd., Nagoya, Japan) test strip that tests milk for β-hydroxybutyrate, the KetoStix (Bayer Diagnostics Europe Ltd., Dublin, Ireland) test strip that tests urine for acetoacetate, and the fat-to-protein percentage ratio (FPR) in milk. A total of 8,902 cows were included in the analysis. The cows were considered to be a random sample from the population of Danish dairy cattle under intensive management, thus representing a natural spectrum of ketosis as a disease. All cows had a recorded FPR between 7 and 21 d postpartum. The KetoLac BHB recordings were available from 2,257 cows and 6,645 cows had a KetoStix recording. The recordings were analyzed with a modified Hui-Walter model, in a Bayesian framework. The specificity of the KetoLac BHB test and the KetoStix test were both high [0.99 (0.97-0.99)], whereas the specificity of FPR was somewhat lower [0.79 (0.77-0.81)]. The best sensitivity was for the KetoStix test [0.78 (0.55-0.98)], followed by the FPR [0.63 (0.58-0.71)] and KetoLac BHB test [0.58 (0.35-0.93)].     

Short communication: A note on the correction for the effect of freezing on the outcome of pregnancy-associated glycoprotein measurement in blood and serum of cows

Early pregnancy detection is a measure of considerable economic relevance for dairy cattle breeders, and analysis of pregnancy-associated glycoprotein (PAG) values in blood is one of the methods implemented in practice. Starting from d 30 postconception, cows are considered to be pregnant at PAG levels of 2.0 ng of PAG/mL of blood and higher. However, little is known about preanalytic sources of errors that might affect PAG values. Based on blood samples from 65 dairy cows, the present study showed that freezing of samples, such as may be the case during shipping in wintertime, will lower PAG values considerably. Therefore, a Bland-Altman analysis was used to derive a correction factor. Overall, the mean differences (± standard deviation) between frozen and respective fresh samples was −5.5 ± 7.4 ng of PAG/mL of blood and 0.9 ± 6.1 ng of PAG/mL of serum. However, the Bland-Altman plot revealed a concentration-dependent effect of freezing on PAG values with higher variability and larger declines at higher PAG levels. Therefore, to minimize chances of false-negative results, different correction factors are suggested for different levels of PAG (e.g., based on the upper bound of the 95% confidence interval 0.67 for PAG levels between 2.0 and 3.9 ng of PAG/mL and 0.25 for PAG levels between 4.0 and 7.9 ng of PAG/mL). With these concentration-dependent correction factors, implementation into practice will be possible. The accuracy is adequate because no quantitative information but qualitative results (pregnant vs. nonpregnant) are required. However, due to larger chances of false-negative results, the application of the correction factor should only be a last resort if temperature exposure of a sample is unknown.     

Effects of resynchronization programs on pregnancy per artificial insemination, progesterone, and pregnancy-associated glycoproteins in plasma of lactating dairy cows

Objectives were to develop a timed artificial insemination (TAI) resynchronization program to improve pregnancy per AI and to evaluate responses of circulating progesterone and pregnancy-associated glycoproteins in lactating cows. Cows (n = 1,578) were presynchronized with 2 injections of PGF_2α, given 14 d apart starting on d 45 ± 3 postpartum, followed by Ovsynch [2 injections of GnRH 7 d before and 56 h after injection of PGF_2α, TAI 16 h after second injection (d 0)]. The Resynch-treated cows received an intravaginal progesterone insert from d 18 to 25, GnRH on d 25, and pregnancy diagnosis on d 32, and nonpregnant cows received PGF_2α., GnRH 56 h later, and TAI 16 h later (d 35). The control cows were diagnosed for pregnancy on d 32 and nonpregnant cows received GnRH, PGF_2α 39 d after TAI, GnRH 56 h later, and TAI 16 h later (d 42). Pregnancy was reconfirmed on d 60 after AI. Ovarian structures were examined in a subset of cows at the time of GnRH and PGF_2α injections. Blood samples for analyses of progesterone and pregnancy-associated glycoproteins were collected every 2 d from d 18 to 30 in 100 cows, and collection continued weekly to d 60 for pregnant cows (n = 43). Preenrollment pregnancies per AI on d 32 did not differ for cows subsequently treated as Resynch (45.8%, n = 814) and control (45.9%, n = 764), and pregnancy losses on d 60 were 6.7 and 4.0%, respectively. Resynchronized service pregnancy per AI (36%, n = 441; 39.5%, n = 412) and pregnancy losses (6.3 and 6.7%) did not differ for Resynch and control treatments, respectively. Days open for pregnant cows after 2 TAI were less for the Resynch treatment than for the control treatment (96.2 ± 0.82 vs. 99.5 ± 0.83 d). Cows in the Resynch treatment had more large follicles at the time of GnRH. The number of corpora lutea did not differ between treatments at the time of PGF_2α. Plasma progesterone for pregnant cows was greater for Resynch cows than for control cows (18-60 d; 6.6 vs. 5.3 ng/mL), and plasma concentrations of progesterone on d 18 were greater for pregnant cows than for nonpregnant cows (5.3 vs. 4.3 ng/mL). Plasma pregnancy-associated glycoproteins during pregnancy were lower for cows in the Resynch treatment compared with control cows on d 39 (2.8 vs. 4.1 ng/mL) and 46 (1.3 vs. 3.0 ng/mL). Cows pregnant on d 32 that lost pregnancy by d 60 (n = 7) had lower plasma concentrations of pregnancy-associated glycoproteins on d 30 than cows that maintained pregnancy (n = 36; 2.9 vs. 5.0 ng/mL). Pregnancy-associated glycoproteins on d 30 (>0.33 ng/mL) were predictive of a positive d 32 pregnancy diagnosis (sensitivity = 100%; specificity = 90.6%). In conclusion, Resynch and control protocols had comparable pregnancy per AI for first and second TAI services, but pregnancy occurred 3.2 d earlier in the Resynch group because inseminations in the Resynch treatment began 7 d before those in the control treatment. Administration of an intravaginal progesterone insert, or GnRH, or both increased progesterone during pregnancy. Dynamics of pregnancy-associated glycoproteins were indicative of pregnancy status and pregnancy loss.     

Effect of interval to resynchronization of ovulation on fertility of lactating Holstein cows when using transrectal ultrasonography or a pregnancy-associated glycoprotein enzyme-linked immunosorbent assay to diagnose pregnancy status

The objective of this study was to compare 2 strategies for resynchronization of ovulation based on nonpregnant diagnoses using transrectal ultrasonography or a pregnancy-associated glycoprotein (PAG) ELISA. Lactating Holstein cows (n = 1,038) were submitted for first postpartum timed artificial insemination (TAI) using a Presynch + Ovsynch protocol. After the initial breeding, cows were randomly assigned to initiate resynchronization 25 d (D25) or 32 d (D32) later. Pregnancy status of cows initiating Resynch 25 d after TAI was determined 27 d after TAI by using a PAG ELISA, whereas pregnancy status of cows initiating Resynch 32 d after TAI was determined 39 d after TAI using transrectal ultrasonography. Cows diagnosed as not pregnant continued the Resynch protocol by receiving an injection of PGF_2α 7 d after the initial GnRH injection and a second GnRH injection 54 h after the PGF_2α injection. Cows in both treatments were inseminated approximately 16 h after the second GnRH injection. Blood samples for analysis of progesterone (P₄) were collected at the first GnRH injection of each Resynch protocol. Pregnancies per AI (P/AI) of nonpregnant cows initiating Resynch 25 vs. 32 d after first postpartum TAI did not differ 39 d after TAI and were 28.3 vs. 30.9% for D25 vs. D32 cows, respectively. Mean P₄ at the first GnRH injection of Resynch was greater for D32 than for D25 cows (3.67 ± 0.22 vs. 2.83 ± 0.22 ng/mL), indicating that the Resynch treatments were initiated at different stages of the estrous cycle. After blocking P₄ concentration into low (<1.0 ng/mL) or high (≥1.0 ng/mL) classes, P₄ class was not found to affect P/AI 39 d after TAI. Early resynchronization was not found to affect P/AI 39 d after TAI; however, early resynchronization did decrease days between inseminations and the interval from the initial nonpregnant diagnosis to conception. Earlier detection of nonpregnant cows using the PAG ELISA in conjunction with a TAI resynchronization program may improve the rate at which cows become pregnant in a dairy herd compared with transrectal ultrasonography conducted at a later stage after TAI.     

Accuracy of a pregnancy-associated glycoprotein ELISA to determine pregnancy status of lactating dairy cows twenty-seven days after timed artificial insemination

To determine the accuracy of a pregnancy-associated glycoprotein (PAG) ELISA in identifying pregnancy status 27 d after timed artificial insemination (TAI), blood samples were collected from lactating Holstein cows (n = 1,079) 27 d after their first, second, and third postpartum TAI services. Pregnancy diagnosis by transrectal ultrasonography (TU) was performed immediately after blood sample collection, and pregnancy outcomes by TU served as a standard to test the accuracy of the PAG ELISA. Pregnancy outcomes based on the PAG ELISA and TU that agreed were considered correct, whereas the pregnancy status of cows in which pregnancy outcomes between PAG and TU disagreed were reassessed by TU 5 d later. The accuracy of pregnancy diagnosis was less than expected when using TU 27 d after TAI (93.7 to 97.8%), especially when pregnancy outcomes were based on visualization of chorioallantoic fluid and a corpus luteum but when an embryo was not visualized. The accuracy of PAG ELISA outcomes 27 d after TAI was 93.7, 95.4, and 96.2% for first, second, and third postpartum TAI services, respectively. Statistical agreement (kappa) between TU and the PAG ELISA 27 d after TAI was 0.87 to 0.90. Pregnancy outcomes based on the PAG ELISA had a high negative predictive value, indicating that the probability of incorrectly administering PGF_2α to pregnant cows would be low if this test were implemented on a commercial dairy.     

Short communication: The effect of storage conditions and storage duration on milk ELISA results for pregnancy diagnosis

The objective of this study was to evaluate the effect of storage temperature and time from sample collection to analysis on test classification of a commercially available ELISA for diagnosis of pregnancy using the measurement of pregnancy-associated glycoproteins (PAG) in milk samples from dairy cows. Few studies have evaluated the effects of sample handling on milk PAG results. Using a repeated-measures study design, we evaluated sample storage at 5 temperatures: 37°C, 22°C, 4°C, ?20°C, or ?80°C. Sample aliquots from 45 cows (20 with a pregnant test result, 10 open, and 15 recheck) were stored for 4, 7, 14, 28, 60, 90, or 365 d. The measured PAG level was influenced by storage duration and condition. Samples stored for 365 d had a slightly increased PAG level, whereas samples stored for all other durations showed a slight decline in PAG level compared with the initial result. The reason for an increase in PAG level following long-term storage is not known. This will not affect dairy producers using the test but may be important in samples stored for research applications. The changes in PAG level were small and within the expected variation for this test. Fewer than 6% of samples changed in classification and, as expected, they were samples near the test interpretation cut-points.     

Effect of early or late resynchronization based on different methods of pregnancy diagnosis on reproductive performance of dairy cows

The aim of this study was to compare the reproductive performance of dairy cows subjected to early (ER) or late (LR) resynchronization programs after nonpregnancy diagnoses based on either pregnancy-associated glycoproteins (PAG) ELISA or transrectal palpation, respectively. In addition, the accuracy of the PAG ELISA for early pregnancy diagnosis was assessed. Lactating Holstein cows were subjected to a Presynch-Ovsynch protocol with timed artificial insemination (AI) performed between 61 and 74 DIM. On the day of the first postpartum AI, 1,093 cows were blocked by parity and assigned randomly to treatments; however, because of attrition, 452 ER and 520 LR cows were considered for the statistical analyses. After the first postpartum AI, cows were observed daily for signs of estrus and inseminated on the same day of detected estrus. Cows from ER that were not reinseminated in estrus received the first GnRH injection of the Ovsynch protocol for resynchronization 2 d before pregnancy diagnosis. On d 28 after the previous AI (d 27 to 34), pregnancy status was determined by PAG ELISA, and nonpregnant cows continued on the Ovsynch protocol for reinsemination. Pregnant cows had pregnancy status reconfirmed on d 46 after AI (d 35 to 52) by transrectal palpation, and those that lost the pregnancies were resynchronized. Cows assigned to LR had pregnancy diagnosed by transrectal palpation on d 46 after AI (d 35 to 52) and nonpregnant cows were resynchronized with the Ovsynch protocol. Blood was sampled on d 28 after AI (d 27 to 34) from cows in both treatments that had not been reinseminated on estrus and again on d 46 after AI (d 35 to 52) for assessment of PAG ELISA to determine the accuracy of the test. Cows were subjected to treatments for 72 d after the first insemination. Pregnancy per AI (P/AI) at first postpartum timed AI did not differ between treatments and averaged 28.9%. The proportion of nonpregnant cows that were resynchronized and received timed AI was greater for ER than for LR (30.0 vs. 7.6%). Cows in ER had a shorter interval between inseminations when inseminated following spontaneous estrus (21.7 ± 1.1 vs. 27.8 ± 0.8 d) or after timed AI (35.3 ± 1.2 vs. 55.2 ± 1.4 d). Nevertheless, the ER did not affect the rate of pregnancy (adjusted hazard ratio = 1.23; 95% confidence interval = 0.94 to 1.61) or the median days postpartum to pregnancy (ER = 132 vs. LR = 140). A total of 2,129 PAG ELISA were evaluated. Overall, sensitivity, specificity, and positive and negative predictive values averaged 95.1, 89.0, 90.1, and 94.5%, respectively, and the accuracy was 92.1%. In conclusion, PAG ELISA for early diagnosis of pregnancy had acceptable accuracy, but early resynchronization after nonpregnancy diagnosis with PAG ELISA did not improve the rate of pregnancy or reduce days open in dairy cows continuously observed for estrus.     

Genetic and nongenetic profiling of milk pregnancy-associated glycoproteins in Holstein cattle

Pregnancy-associated glycoproteins (PAG) are secreted by the trophoblast and are detectable in maternal circulation around the time of attachment of the fetal placenta, as well as in blood and milk throughout gestation. The understanding of the genetic mechanisms controlling PAG levels can confer advantages for livestock breeding programs given the precocity and the ease of obtaining this phenotype from routine pregnancy diagnosis. The aim of this study was to characterize PAG levels by estimating genetic parameters and correlations with other dairy traits, and to identify genomic regions and candidate genes associated with PAG levels in milk. The PAG data consisted of pregnancy diagnoses using commercial assays from 2012 to 2017, and genotype data consisted of 54,123 SNP markers for 2,352 individuals (embryos and dams). The model included contemporary group (herd, year, and season) and embryo age as fixed effects, and random embryonic (direct) and maternal (indirect) genetic effects. Using genomic data, the estimated heritability for direct and maternal genetic effects (± standard deviations) were 0.23 ± 0.05 and 0.11 ± 0.05, respectively. The genetic correlation between these effects was almost zero (0.001 ± 0.06). A preliminary analysis revealed low correlations between milk PAG levels and other dairy traits. The genome-wide association analysis was performed using 2 approaches: single-marker and single-step using all markers. Four genomic regions with direct genetic effects were detected on Bos taurus autosome (BTA) 6, BTA7, BTA19, and BTA29 of the embryonic genome. The BTA29 locus was within the bovine PAG gene cluster, suggesting a cis-regulatory quantitative trait locus on the PAG expression. However, other associations, without an obvious link to PAG expression, could be related to the transportation of PAG and their concentration in milk. Only 1 region from the maternal genome, on BTA4, had a significant indirect effect, where WNT2 is a candidate gene related to placenta vascularization and gestation health. Collectively, our results suggest a moderate genetic control of milk PAG levels from both maternal and fetal genomes, but larger studies are needed to fully evaluate the usefulness of milk PAG in the genetic evaluation of fetal growth and cow fertility.     

Estimating test characteristics of somatic cell count to detect Staphylococcus aureus-infected dairy goats using latent class analysis

The aim of this study was to estimate test properties of composite somatic cell count (cSCC) to detect subclinically Staphylococcus aureus-infected dairy goats. Staphylococcus aureus is the most prevalent major pathogen in goats, and responsible for the majority of clinical mastitis cases. Therefore, a diagnostic tool that detects subclinical Staph. aureus infections may be useful in decreasing the number of clinical cases. We collected samples from 384 animals in 4 herds for bacteriological culture and cSCC on 3 occasions in lactation: once in early lactation, once around peak lactation, and once in late lactation. Latent class models were used to estimate test properties of cSCC and bacteriological culture in the absence of a gold standard reference test under the assumption that both tests detect Staph. aureus intramammary infection. Estimates for test properties of cSCC in early lactation at a cut-off value of 1,500 × 10³ cells/mL were 0.90 for sensitivity and 0.95 for specificity, making cSCC a useful screening tool for detection of Staph. aureus. An effect of lactation stage was observed, causing an increased sensitivity and decreased specificity in late lactation. The sensitivity of bacteriological culture was estimated to be very low in the latent class models and the models suggested that the true prevalence of Staph. aureus in dairy goat herds is much higher than what is commonly reported based on bacteriological culture. This implies that intramammary infection by Staph. aureus may be an underestimated problem in dairy goat herds, and that cSCC can be used to diagnose infected animals.     

Technical note: A rapid enzyme-linked immunosorbent assay blood test for pregnancy in dairy and beef cattle

The ruminant trophoblast produces pregnancy-associated glycoproteins (PAG) that can be detected in the blood of pregnant animals. The objective was to determine the accuracy of a rapid ELISA PAG-based test for the purpose of pregnancy detection in cattle. Blood was sampled from dairy cattle (539 Holstein cows, 173 Holstein heifers, 73 Guernsey cows, 22 Guernsey heifers, and 12 Jersey heifers) and crossbred beef cattle (145 cows and 46 heifers) that were ≥25 d after insemination (range = 25 to 45 d for dairy and 29 to 56 d for beef). Cattle were examined by ultrasonography for detection of pregnancy within 2 d of blood collection. Whole blood or plasma was incubated in a polystyrene tube coated with a monoclonal PAG antibody for 15 min. The tubes were then washed and subjected to sequential incubations with a biotinylated polyclonal PAG antibody (15 min, followed by wash), a horseradish peroxidase-streptavidin solution (15 min, followed by wash), and a peroxidase substrate. Tubes were visually assessed for color after 15 min (clear solution = PAG negative, not pregnant; blue solution = PAG positive, pregnant). Total assay time was approximately 90 min. The ultrasound examination was used as the standard for pregnancy diagnosis. The sensitivity (99.8 ± 0.2%), specificity (91.7 ± 1.4%), and negative predictive value (99.7 ± 0.3%) for the PAG test used in dairy cattle were similar for different breeds and for cows and heifers. The positive predictive value for the test was greater in dairy heifers than in dairy cows (96.5 ± 1.4% vs. 90.5 ± 1.7%, respectively). In beef cattle, the sensitivity (100%), specificity (92.3 ± 3.0%), positive predictive value (95.0 ± 2.0%), and negative predictive value (100%) for the PAG test were similar for cows and heifers. The accuracy of the test was not different for dairy and beef cattle. In conclusion, the rapid ELISA pregnancy test based on PAG was highly sensitive and specific for pregnancy detection in dairy and beef cattle.     

基于潜在类别模型的留守儿童食品营养研究

运用潜在类别模型研究食品营养缺乏儿童中一个特殊的群体——留守儿童的食品营养状况,分析个人遗传、后天的食品营养摄入、母亲的照料及基本的医疗卫生服务对留守儿童食品营养状况的影响,经过研究分析,识别出食品营养不良和食品营养状况良好两类留守儿童,其中食品营养不良儿童比例较高,达到69.4%。两类留守儿童最大的差别在于是否有母亲照顾、后天的食品营养摄入情况和基本的医疗卫生服务情况,而非个人遗传因素,据此为改善留守儿童食品营养状况提供可靠的理论依据。     

海蜇糖蛋白JGP-III2的结构及其免疫活性研究

从海蜇中获得一种新的糖蛋白JGP-III2,其糖含量为11.87%,蛋白含量为87.74%,分子量为39.5 k Da。JGP-III2中含有8种单糖分别为甘露糖、2-氨基-D-葡萄糖、氨基半乳糖、乳糖、葡萄糖、半乳糖、木糖、岩藻糖,含量比例为:6.88:5.13:5.19:70.78:1.47:2.09:6.59:1.85;氨基酸含量为85.95%,甘氨酸、缬氨酸和谷氨酸所占比例较高;JGP-III2含有O-糖苷键和N-糖苷键,与海蜇体内获得的蛋白(gi|156215071)具有同源性。通过建立免疫低下小鼠模型,对JGP-III2的免疫活性及免疫机理进行研究,结果表明:与模型对照组比,JGP-III2能显著提高免疫低下小鼠的脾脏指数,胸腺指数,巨噬细胞吞噬能力,迟发性变态反应,血清溶血素水平及抗体生成细胞数,增强机体免疫能力。JGP-III2能促进Th1类细胞因子m RNA的表达,抑制Th2类细胞因子的m RNA的表达,对模型组中的\"Th1/Th2漂移\"现象有恢复作用。推测JGP-III2可能是通过调节机体Th1/Th2的平衡来提高机体免疫活性。     

Evaluation and use of three cowside tests for detection of subclinical ketosis in early postpartum cows

The objective was to evaluate the performance of 3 cowside diagnostic tests for detection of subclinical ketosis, defined as a serum beta-hydroxybutyrate (BHBA) concentration >or=1400 micromol/L. On 16 d over a 5-mo period, samples of serum, milk, and urine were collected on a large dairy facility from cows of all parities between 2 and 15 DIM. The sample proportion of subclinical ketosis was 7.6% (n = 859 samples from 545 cows). The KetoCheck powder (Great States Animal Health, St. Joseph, MO) detecting acetoacetate in milk samples was very specific (99%) but poorly sensitive (41%). Respective sensitivities and specificities of the Ketostix strip detecting acetoacetate in urine samples (Bayer Corporation, Elkhart, IN) were 78 and 96% with a cut-off point of "small", or 49 and 99% with a cut-off of "moderate." The KetoTest strip (Sanwa Kagaku Kenkyusho Co. Ltd., Nagoya, Japan) using milk samples had a sensitivity and specificity of 73 and 96% with a cut-off of 100 micromol of BHBA/L or 27 and 99% with a cut-off of 200 micromol of BHBA/L. On average, use of the Ketostix at the "small" cut-off point or the KetoTest at 100 micromol/L would result in no more than 3 or 4 false positives per 100 cows screened, with prevalence levels ranging from 5 to 30%, whereas the number of false negatives would range from one false negative at 5% prevalence to 7 or 8 false negatives at 30% prevalence. Either the Ketostix or KetoTest strips would provide acceptable results for screening individual cows on commercial dairies to detect subclinical ketosis. Over this prevalence range, the KetoCheck powder test would have limited application as a screening test. Despite only one false positive per 100 animals screened, false negatives resulting from screening with the KetoCheck test would be too frequent, ranging from 3 false negatives at 5% prevalence to 18 at 30% prevalence in a population of 100 tested cows. Finally, given their relative imprecision, use of any of these individual cowside tests to estimate herd prevalence must be done cautiously, especially when only a small number of animals are sampled.     

Comparison of 2 electronic cowside tests to detect subclinical ketosis in dairy cows and the influence of the temperature and type of blood sample on the test results

The objective of this study was to determine the suitability of 2 electronic hand-held devices [FreeStyle Precision (FSP), Abbott GmbH & Co. KG, Wiesbaden, Germany and GlucoMen LX Plus (GLX), A. Menarini GmbH, Vienna, Austria] for measuring β-hydroxybutyrate (BHBA) in dairy cows. Three experiments were conducted to evaluate (1) the diagnostic performance of the devices, (2) the effect of the type of blood sample, and (3) the influence of the ambient temperature on the determined results. A total of 415 blood samples from lactating Holstein and Simmental cows were collected and analyzed with both devices (whole blood) and in a laboratory (serum). Correlation coefficients between whole-blood and serum BHBA concentrations were highly significant, with 94% for the FSP and 80% for the GLX device. Based on thresholds for subclinical ketosis of 1.2 and 1.4 mmol of BHBA/L, results obtained with the hand-held devices were evaluated by receiver operating characteristics analyses. This resulted in adjusted thresholds of 1.2 and 1.4 mmol/L for the FSP and 1.1 and 1.3 mmol/L for the GLX device. Applying these thresholds, sensitivities were 98 and 100% for the FSP and 80 and 86% for the GLX device, respectively. Corresponding specificities were 90 and 97% for the FSP and 87 and 96% for the GLX device, respectively. Additionally, concentrations of BHBA were tested with both devices in whole blood, EDTA-added whole blood, and in their resulting serum and plasma, collected from 65 animals. Determined BHBA concentrations were similar within each device for whole and EDTA-added blood, and in serum and plasma, but differed between whole blood and serum and between EDTA-added blood and plasma. Blood samples with low (0.4 mmol/L), medium (1.1 mmol/L), and high (1.6 mmol/L) BHBA concentrations were stored between +5 to +32°C and analyzed repeatedly at temperature levels differing by 4°C. Additionally, devices and test strips were stored at equal conditions and used for measurement procedures. Storage temperature of the devices and test strips did not influence the differences between the results of the laboratory and the devices, whereas the temperature of the blood samples caused significant differences. Although the level of agreement between the laboratory and the GLX device was lower than for the laboratory and the FSP device, both devices are useful tools for monitoring subclinical ketosis in dairy cows. Due to their effects on the determined results, the type and temperature of the tested sample should be considered.