Randomisation and masking
Women were randomized in the intervention arm or usual care based on a list of randomly permuted blocks (block size of four to eight), distributed in opaque consecutively numbered sealed envelopes. Clinicians and participants had no access to the list, but were not masked to group allocation, which was made known after inclusion of the participant. The diabetologists responsible for the decision to prescribe insulin during follow-up were, as much as possible, kept blinded to the assignment.