Background: The prevalence and predictors of inaccurate self-monitored blood glucose (SMBG) diaries in type 2 diabetes are not well defined.
Methods: This was a cross-sectional study of adults with type 2 diabetes in a diabetes clinic at a large urban public hospital. We collected copies of SMBG diaries and downloaded data from their glucose meters if patients brought them to their clinic appointment. Trained interviewers used standard tests to assess literacy, depression, and cognitive function. The main outcome measure was accuracy of the glucose diary assessed by comparing reported values to meter memory readings and to results of hemoglobin A(1c).
Results: Blood glucose self-monitoring was either missing or misleading for 48% (55 of 115) because (1) patients brought neither meter nor the SMBG diary (n = 26) or (2) the diary was inaccurate (n = 29). An inaccurate glucose diary (by comparison with meter readings) was predicted by normotension (odd ratio 5.6, P = 0.02) and one measure of cognitive impairment, slow Digit Symbol Coding (odds ratio 2.2, P = 0.02). A patient's self-assessment of his or her diary's accuracy was unreliable (sensitivity 63%, specificity 56%).
Conclusions: SMBG diaries are frequently either not accurate or not brought to clinic visit. Some inaccuracy might be due to cognitive impairment. To achieve maximum benefit from glucose self-monitoring, glucose meter memory analysis is crucial before making therapy adjustments based on SMBG.