“…Based on selection quality of cases, 37.5% of the studies accompanied case definition adequately with laboratory diagnosis/validation (item 1a) 42 , 44 , 49 – 51 , 56 while 68.8% 41 – 43 , 45 , 47 , 52 , 54 , 57 – 59 relied on hospital records for case definition (item 1b); in term of representativeness of the cases, 93.8% 42 – 45 , 47 , 50 – 52 , 54 , 57 – 61 represent representative series of cases (item 2a), 6.1% 41 had or did not stated potential for selection biases (item 2b). On selection of controls, 18.8% of the studies 51 , 54 , 58 were based on community controls (item 3a), while 68.8% studies were hospital-based controls 42 – 44 , 47 , 50 , 52 , 57 – 61 (item 3b). All the studies (100%) defined controls based on no history of DM 12 , 41 – 45 , 47 , 49 – 52 , 54 , 57 – 61 (item 4a).…”