Comparing patient generated blood glucose diary records with meter memory in type 2 diabetes

To assess agreement between meter and diary self monitoring of blood glucose (SMBG) records, over a year, in a sample of patients with type 2 diabetes.

Meter and diary records were available, for 95 individuals, who took part in the Efficacy of self monitoring of blood glucose in patients with newly diagnosed type 2 diabetes study. Pearson's correlation coefficient was used to explore the relationships between the types of error. Maximum likelihood estimation was used to explore changes over time through a structural equation modelling approach. Paired samples t-tests were used to determine if the presence of errors led to a significant difference between the mean diary and meter SMBG concentrations or coefficients of variation. Multiple regression was used to explore possible predictors of the error indices.

Mean over-reporting, under-reporting, concordance and overall reliability were 8.4%, 10.0%, 83.5% and 71.3%, respectively. The first week of monitoring had significantly more under-reporting, over-reporting and less concordance and overall reliability than subsequent weeks. The majority of concordance errors were not clinically significant. Those that were, tended to occur during the first three months of monitoring. Participants' at one trial site were significantly more likely to have recording errors than those at the largest site.

Error levels were similar to those described previously in type 1 diabetes and there was a suggestion of an initial learning curve for record keeping. For some individuals diary records would not be considered acceptable if held to the same standards as blood glucose meters.