Detection accuracy of three glucose meters estimated by capillary blood glucose measurements compared with venous blood evaluated by the diabetes unit of the Hospital Evangélico de Curitiba, Brazil

Abstract

Objective: To compare capillary blood glucose measurements between three different glucose meters and with the serum glucose values of inpatients at the diabetes unit of Hospital Universiterio Evangelico de Curitiba, Brazil. Materials and methods: A total of 132 non-intensive care unit patients admitted for medical and surgical pathologies were evaluated. All patients reported a previous diagnosis of diabetes mellitus, were under 60 years of age, had no hematocrit alterations, remained hemodynamically stable during the time of data collection, and were given no ascorbic acid, acetaminophen, dopamine, or mannitol during follow-up. Capillary and serum blood glucose samples were collected simultaneously by finger-stick and venipuncture 2 hours after lunch, by the same observer, who was blinded to the serum glucose results. First, between July and November 2009, capillary glucose levels were measured using the blood glucose meters OneTouch SureStep((R)) and MediSense Optium((R)). Between November 2009 and February 2010, capillary blood glucose levels were measured on the glucose meters OneTouch SureStep and Optium Xceed((R)). The capillary glucose readings were analyzed between meters and also in relation to the serum blood glucose values by the t-test for paired samples and the Mood two-sample test. Results: The patients' mean age was 50.45 years. The blood glucose means obtained using the meters OneTouch SureStep, MediSense Optium, and Optium Xceed were, respectively, 183.87 mg/dL, 178.49 mg/dL, and 192.73 mg/dL, and the mean for the serum glucose values was 174.58 mg/dL. A significant difference was found between the capillary measurements taken by the glucose meters and the serum glucose measurements (P < 0.05), and no significant interdevice difference was found. After stratification of the serum blood glucose values into two groups, below and above 180 mg/dL, the variance found for the glucose meter OneTouch SureStep was statistically greater (P = 0.03) in relation to the serum glucose levels above 180 mg/dL, which was not the case with the glucose meters MediSense Optium (P = 0.06) and Optium Xceed (P = 0.12). The percentage of capillary blood glucose values showing a variation of less than 20% compared with serum values was 64.94% for OneTouch SureStep, 47.83% for Medisense Optium, and 51.61% for Optium Xceed, when serum glucose was greater than 75 mg/dL. Conclusion: The glucose meters tested showed an adequate interdevice correlation in their capillary glucose readings, in addition to correlating with the serum glucose values (ie, if a blood glucose reading is high or low in one test, it is likely to be respectively high or low in another). The means for the capillary blood glucose readings, however, were significantly different from the mean serum glucose. When serum glucose was above 180 mg/dL, there was a greater variance in the capillary measurements on the glucose meter OneTouch SureStep, with less correlation with the serum blood glucose (P < 0.05), which did not occur significantly with the two other glucose meters. On the other hand, OneTouch SureStep had the highest accuracy in relation to serum glucose when the whole sample of serum glucose values above 75 mg/dL was analyzed, considering a variation of less than 20% in the measurements. The three glucose meters provide readings that correlate with the serum glucose values of hospitalized patients. However, one should bear in mind that capillary measurements quite often show more than a 20% variation in relation to serum glucose values, and caution should be exercised in interpreting the readings when serum glucose levels are elevated.