Dr L Meneghini

Correspondence to: Dr L Meneghini, University of Miami Miller School of Medicine, Miami, FL, USA; lmeneghi@med.miami.edu

STUDY DESIGN

Design:

cluster randomised controlled trial (PREDICTIVE 303* study).

Allocation:

unclear allocation concealment.†

Blinding:

unblinded.†

STUDY QUESTION

Setting:

1083 primary care practices in the USA.

Patients:

5604 patients ⩾18 years of age (mean age 59 y, 52% men) who had type 2 diabetes, haemoglobin (Hb) A_1c concentration ⩽12%, and body mass index (BMI) ⩽45 kg/m². Exclusion criteria were concomitant medication changes that would interfere with glucose metabolism, receipt of glucose-lowering medication that were not indicated with insulin, proliferative retinopathy or maculopathy, hypoglycaemia unawareness or recurrent major hypoglycaemia, pregnancy, lactation, or serious illness.

Intervention:

303 Algorithm (n = 2794) or standard of care (n = 2825). All patients were given insulin detemir once daily as add-on therapy to any other glucose-lowering regimen or replacement of previous basal insulin. Patients in the algorithm group …