BACKGROUND: Self-monitoring blood glucose (SMBG) is a key tool for therapeutic decision and structured patient education. Despite there is evidence that SMBG improves metabolic control both in insulin-treated and in non insulin treated patients with type 2 diabetes (T2DM), the prescription and execution of SMBG is suboptimal and heterogeneous. 
AIM: AMD Annals initiative aims to improve quality of diabetes care in Italy. This analysis assessed the use of SMBG in patients with T2DM to evaluate frequency of SMBG, metabolic control, and hypoglycemia in the different therapeutic schemes including or not insulin. 
METHOD: A specific software has been used for the standardized extraction of the information contained in the electronic medical records. Extracted data included clinical data, therapies and all SMBG values downloaded on EMR by different glucose meters routinely used by the patients. Sample was constituted by T2DM patients with at least one HbA1c value during the years 2014 and 2015 and with at least 1 available SMBG value measured in the 90 days before the HbA1c test. 
RESULTS: Overall 21 centers and 27,768 observations were included in the analysis. Frequency of SMBG, levels of fasting and post-prandial blood glucose (FBG and PPG), and values below 60 mg/dl were evaluated in the following treatment schemes: metformin only (N=3188); secretagogues only (N=685); metformin + secretagogues (N=2929); metformin + DPP-IV inhibitor (N=612); metformin + secretagogues + DPP-IV (N=3132); metformin + secretagogues + acarbose (N=337); GLP1RA + metformin + secretagogue (N=375); GLP1RA + metformin (N=516); basal insulin + metformin + secretagogues (N=3486); basal + short acting insulin + metformin (N=1666); basal + short acting insulin (N=6842). SMBG was frequently recommended also to patients not treated with insulin: 45.7% of observations referred to patients not treated with insulin. Frequency of SMBG was suboptimal in all treatment schemes; even patients treated with schemes including insulin monitored their glucose less than two times per day. Furthermore, 39.5% of SMBG tests could not be univocally classified as FBG or post-prandial glucose (PPG). In all treatment schemes, of those values which could be univocally classified, over 95% of SMBG tests were FBG values and less than 5% were PPG values. Pre-breakfast FBG values represented about 50% of all available FBG values in all treatment schemes. The frequency of SMBG does not substantially differed among the most common schemes with oral agents, irrespective of the use of secretagogues. The average FBG during three months was over 130 mg/dl in 38% to 84% of the cases in the different schemes, while average PPG was over 140 mg/dl in 36% to 73% of the cases. Even when looking at the most common therapeutic approaches, it emerged that substantial proportions of cases have elevated FBG and PPG, even though average HbA1c levels were often acceptable. The use of therapeutic schemes including secretagogues was associated with a two to three-fold increased risk of glycemic values <70 mg/dl as compared to OHA schemes without secretagogues. The use of insulin markedly increased the risk of hypoglycemia, with a relevant difference between basal insulin+OHA and basal+short insulins with or without OHA. The same trends were also found as for glycemic values <50 mg/dl. 
DISCUSSION: Frequency of SMBG is suboptimal both in patients treated and not treated with insulin. More structured SMBG schemes should be used and not only FBG but also PPG should be monitored for phenotyping patients. Irrespective of therapeutic scheme, many patients show unsatisfactory levels of blood glucose even when HbA1c levels are acceptable. These data will be used in the context of AMD Annals initiative for benchmarking initiatives to increase the culture and the appropriate use of SMBG.

Self-monitoring blood glucose in type 2 diabetes: a continuous improvement initiative