Abstract
This study was aimed at assessing the accuracy of the blood glucose meter currently
in use in comparison with the reference test. A cross-sectional analytical study design
was used. The sample population comprised diabetic patients from the diabetic clinic at
Maesai Hospital, Chiang Rai Province during the month of September 2007. Blood glucose
levels, sex, and age of the samples were collected. The statistical methods used were
descriptive, Pearson’s correlation, Spearman’s rho correlation, Wilcoxon’s signed rank test, paired t-test, and univariate linear regression. The results of the study showed that,
out of 139 patients, 69.6 percent were female, with the average age being 54.3 ± 12.5 years.
It was found that the average blood glucose levels were not statistically significant.
Spearman’s rho coefficient was 0.96 (p<0.001). Since the difference in the results from
both tests was higher when the blood glucose level was higher, subgroup analysis was
conducted by separating the blood glucose level into two groups: less than 300 mg/dL
and 300 mg/dL or over. The results revealed that in the former group, the total number of
samples was 130. The average of the blood glucose levels was not statistically significant.
Spearman’s rho coefficient was 0.95 (p<0.001). The results of the analysis of the latter
group showed that there were 8 patients whose average blood glucose levels were not
statistically significant. Pearson’s correlation coefficient was 0.76 (p=0.03). The univariate
linear regression model in this part was the blood glucose level from the reference test =
15.9+blood glucose level from the blood glucose meter. Moreover, the clinical differences
(±20%) from both subgroups were less than 5 percent.
