The potential for improvement of outcomes by personalized insulin treatment of type 1 diabetes as assessed by analysis of single-patient data from a randomized controlled cross-over insulin trial.

Aims: The evidence for optimal insulin treatment in type 1 diabetes is mainly based on randomised controlled trials applying a parallel-group design. Such trials yield robust general results but crucial individual treatment effects cannot be extracted. We aimed to assess the potential for further improvement of outcomes by personalized insulin therapy by analyzing data from a cross-over trial at individual level.

Methods: Post hoc analysis of data from a two-year multicentre, prospective, randomised, open, blinded endpoint (PROBE) trial (the HypoAna trial). In a cross-over design 114 patients with type 1 diabetes and recurrent severe hypoglycemia were treated with basal-bolus therapy based on analog (detemir/aspart) or human (NPH/regular) insulin aiming at maintenance of baseline HbA1c levels. For each patient a superior outcome was defined as fewer events of severe hypoglycemia defined by need for third party treatment assistance or a more than 0.4% (4.4mmol/mol) lower HbA1c.

Results: Only one quarter had comparable outcome of the two treatments in terms of rate of severe hypoglycemia or HbA1c. Twice as many patients had superior outcome of analog-based as compared to human insulin-based insulin treatment. The rate of severe hypoglycemia with the superior treatment was lower compared to the rates obtained with analog insulin and with human insulin (0.67, 1.09, and 1.57 episode per patient-year, respectively (p<0.0001)).

Conclusions: Personalized insulin treatment of type 1 diabetes based on single-patient evidence may improve outcomes significantly compared to a general treatment approach.