Veterinary Medicine Teaching and Research Center, 18830 Road 112, Tulare, CA 93274.
Veterinary Medicine Teaching and Research Center, 18830 Road 112, Tulare, CA 93274; Department of Population Health and Reproduction, School of Veterinary Medicine, University of California, Davis 95616.
J Dairy Sci. 2019 Oct;102(10):9401-9408. doi: 10.3168/jds.2018-15688. Epub 2019 Aug 1.
The objective was to evaluate the precision and accuracy of 6 handheld glucose meters, designed for human use [Accu-Chek Aviva Plus (AC), Roche Diabetes Care, Mannheim, Germany; Aga Matrix (AM), AgaMatrix Inc., Salem, NH; Contour Next (CT), Bayer HealthCare LLC, Leverkusen, Germany; FreeStyle Precision Neo (FS), Abbott Diabetes Care Ltd., Alameda, CA; Nova Max Plus (NM), Nova Biomedical Corporation, Waltham, MA; and Precision Xtra (PX), Abbott Diabetes Care Ltd., Witney, UK] to measure blood glucose concentration in dairy cows. Blood samples from Jersey and Jersey × Holstein crossbreed cows (n = 97 for all; except CT, n = 71) were collected and analyzed in triplicate using the 6 handheld glucose meters evaluated. Plasma glucose was also measured with the laboratory reference method (hexokinase glucose-6-phosphate dehydrogenase). Based on the intra-assay coefficient of variation (CV), precision varied across handheld glucose meters: AC (2.2%), CT (4.0%), PX (4.7%), FS (5.6%), AM (6.2%), and NM (6.7%). Lin's concordance correlation coefficients between handheld glucose meters and the reference method were 0.75 for FS, 0.74 for PX, 0.62 for AC, 0.55 for CT, 0.53 for NM, and 0.48 for AM. Based on Passing-Bablok regression, the AM and PX meters showed bias in the measurements of blood glucose. Bland-Altman plots indicated a negative bias (FS = -0.25 mmol/L; CT = -0.60 mmol/L) or a positive bias (AM = 0.29 mmol/L; PX = 0.33 mmol/L; NM = 0.52 mmol/L; AC = 0.65 mmol/L) between handheld glucose meters and the reference method. All handheld glucose meters evaluated had wide limits of agreement (LoA) ranging from -0.18 to 1.47 mmol/L (AC, narrowest LoA) to -1.25 to 1.82 mmol/L (AM, widest LoA). Bias was the major contributor to the total observed error (TE), accounting for 81.5% of the TE in AC, 72.0% in CT, 64.9% in AM, 61.1% in NM, 57.8% in PX, and 56.2% in FS. Overall, although some handheld meters (AC, CT, and PX) showed satisfactory precision, none were accurate measuring glucose. Future studies should evaluate whether incorporating algorithms designed for cattle can improve accuracy and precision of handheld glucose meters.
目的是评估 6 种手持式血糖仪（Accu-Chek Aviva Plus [AC]、罗氏糖尿病护理公司、德国曼海姆；Aga Matrix [AM]、AgaMatrix 公司、新罕布什尔州塞勒姆；Contour Next [CT]、拜耳健康护理有限责任公司、德国勒沃库森；FreeStyle Precision Neo [FS]、雅培糖尿病护理有限公司、加利福尼亚州阿拉米达；Nova Max Plus [NM]、Nova 生物医学公司、马萨诸塞州沃尔瑟姆；和 Precision Xtra [PX]、雅培糖尿病护理有限公司、英国威特尼）测量奶牛血糖浓度的精密度和准确度。从泽西和泽西 × 荷斯坦杂交奶牛（所有奶牛均为 97 头；除 CT 外，均为 71 头）采集血液样本，并使用评估的 6 种手持式血糖仪进行三次重复测量。血浆葡萄糖也使用实验室参考方法（己糖激酶葡萄糖-6-磷酸脱氢酶）进行测量。基于室内变异系数（CV），手持式血糖仪的精密度存在差异：AC（2.2%）、CT（4.0%）、PX（4.7%）、FS（5.6%）、AM（6.2%）和 NM（6.7%）。Lin 一致性相关系数显示，FS 与参考方法之间的相关性为 0.75，PX 为 0.74，AC 为 0.62，CT 为 0.55，NM 为 0.53，AM 为 0.48。基于 Passing-Bablok 回归，AM 和 PX 血糖仪在血糖测量中存在偏差。Bland-Altman 图表明，与参考方法相比，FS（-0.25 mmol/L）、CT（-0.60 mmol/L）存在负偏倚，而 AM（0.29 mmol/L）、PX（0.33 mmol/L）、NM（0.52 mmol/L）和 AC（0.65 mmol/L）存在正偏倚。评估的所有手持式血糖仪均具有较宽的允许误差（LoA）范围，从-0.18 至 1.47 mmol/L（AC，最窄的 LoA）至-1.25 至 1.82 mmol/L（AM，最宽的 LoA）。偏差是总观测误差（TE）的主要贡献者，AC 占 81.5%，CT 占 72.0%，AM 占 64.9%，NM 占 61.1%，PX 占 57.8%，FS 占 56.2%。总体而言，尽管一些手持式血糖仪（AC、CT 和 PX）显示出令人满意的精密度，但没有一种血糖仪能够准确测量血糖。未来的研究应评估是否可以通过为奶牛设计算法来提高手持式血糖仪的精密度和准确度。
