Introduction The association of knowledge against actual practice has not been properly investigated in the BIRDEM (a tertiary care hospital of BADAS). The aim of this study was to assess the level of knowledge about type 2 diabetes mellitus (T2DM) and to see the association between knowledge and action among T2DM patients.

Methods An analytical cross-sectional design. A total of 500 patients withT2DM (M 58%, age 49.4±8.8 years) were selected from the outpatient department. Knowledge score was categorised as poor (<45%), fair (45–60%) and good (>60%) on the basis of education criteria in our 10th and 12th grade public examinations.

Results The levels of knowledge were poor in 45%, fair in 47% and good in 9% of the study subjects. Only 12% subjects had good level of HbA1c. Among them the levels of knowledge were poor in 44%, fair in 48% and good in 8%. About 24% subjects belonged to fair HbA1c level and the levels of knowledge were poor in 39%, fair in 52% and good in 9% among these subjects. More than half of the subjects (64%) had poor glycaemic status. In this group the levels of knowledge were poor in 46%, fair in 45% and good in 9%. A significant negative association was found between total knowledge score and HbA1c level (r=−0.08, p=0.05).

Conclusion There is a knowledge-action gap in Bangladeshi T2DM subjects which may indicate largely ineffective education programs. Patient empowerment may facilitate better control of diabetes.