Response to: MiniMed 780G System Outperforms Other Automated Insulin Systems Due to Algorithm Design,Not Bias—Response to Inaccurate Allegations

Dear Editor,

We would like to thank Dr. van der Heuvel and colleagues for providing additional data validating the integrity of the CareLink Personal database. As academic researchers also conducting prospective and real-world trials on the MiniMed 780G, ¹ we do not believe that there exists a causal link between performance outcomes and selection bias for this system, and thus never made that claim in our recent manuscript. ² However, we feel that in any peer-reviewed publication, it is important to acknowledge the possibility of bias in how a sample was obtained. Indeed, in the publication from where data on real-world use of the MiniMed 780G was referenced, the methods are reported as follows ³ :

MiniMed 780G system data uploaded to CareLink personal software from August 27, 2020 to July 22, 2021 by individuals who provided consent for their data to be aggregated and who resided in countries (Belgium, Czech Republic, Denmark, Egypt, Finland, UK, Greece, Hungary, Iceland, Ireland, Israel, Italy, Luxemburg, The Netherlands, Poland, Portugal, Qatar, Romania, Slovakia, Slovenia, South Africa, Sweden, Switzerland and Turkey) where local data privacy regulation permitted data aggregation, were analysed.

The additional details provided in the response letter are much more informative in terms of data capture than what was previously published. It is reassuring to see that >95% of EMEA 780G users have CareLink accounts with >90% consenting to upload data and >98% automatically uploading data.

We would also like to reiterate to the reader our caveats from the article that “direct comparison across real-world datasets must be taken with caution.” We share with the authors excitement about the performance of the 780G system in recent studies throughout Europe and the United States. However, we would be more cautious about making claims of “superior performance” between different systems given differences in baseline characteristics of users in clinical practice and in study participants, trial design, and sensors used in evaluating glycemic status. We would also be cautious in making causal claims regarding the degree to which the control algorithm alone drives outcomes as opposed to the contributions of broader system design and clinical support structure.

In the end, it is our belief that automated insulin delivery is the standard of care for individuals living with type 1 diabetes. The real-world data of all systems highlight that initial findings from pivotal trials are echoed with clinical adoption of commercially available devices. Hopefully, these important studies will support greater coverage of devices and ensure access to this technology so more individuals can benefit from it.