Can Diabetes Management Be Done Effectively Through the Cloud?

In the current issue of the journal, Hsu et al. ¹ report how they investigated in a feasibility study whether information and communication technologies (ICT) through a cloud-based diabetes management program could be effective in helping improvement of glucose control in patients with type 2 diabetes who initiated insulin therapy. Although limited to a small number of cases, the randomized control design of the study and the impressive benefit obtained on A1c improvement in the patients who followed the cloud-based program support the interest of integrating ICT in diabetes care management.

The usefulness of frequent contacts between patients with diabetes and the diabetes care team is not a new idea and was fully integrated in the concept of intensive insulin therapy tested in the Diabetes Control and Complications Trial (DCCT) dedicated to patients with type 1 diabetes. ² Among noncommunicable chronic diseases, diabetes mellitus is indeed likely the most burdensome in matters of patient involvement and participation in the therapy, not only for observance but also for making daily decisions in the adjustment of the therapy. Because any patient decision includes a risk of mistake, a trend for a conservative behavior is common once no major glucose excursions and discomfort appear with used insulin doses. However, this attitude cannot guarantee to reach targeted near-normal glucose levels. The post-DCCT follow-up while patients previously in the intensive management group entered in a more relaxed management process has well illustrated the reduced efficacy in glucose control as shown by an average +0.8% A1c increase. ³

Initiation of insulin in patients with type 2 diabetes is a crucial time for acquiring new skills in therapy because using the appropriate insulin doses is tightly associated with the improvement of glucose control. ⁴ Beside technical support, human support is also essential at this period because acceptance of insulin therapy is often fragile. ⁵ The benefit of a scheduled plan of interactions between the healthcare team and the patient, both face-to-face and through phone calls, at insulin initiation has been underscored by Swinnen and DeVries, ⁶ who investigated from a Medline systematic search how predetermined contacts with the diabetes care team influenced A1c outcomes. Both visit frequency and number of phone calls appeared as significant independent predictors of A1c improvement. A1c reduction was significantly associated with the insulin dose, but this association was no longer significant when adjusted on contact frequency, supporting the role of human interactions in the decision process of dose adjustment. Of note is that this benefit of human support on A1c outcomes was not present at initiation of dipeptidyl peptidase-4 inhibitors, suggesting that beyond support to observance a support to decision making about the treatment was involved for insulin therapy. Outside of clinical studies, a clear limit in the implementation of such a support by the healthcare team comes from its availability related to human resources.

The considerable development of ICT during the last decade has raised the question of their applicability in the healthcare field. Many applications for personal computers or, more recently, smartphones or tablets have been developed by healthcare professionals or “geek” patients with diabetes in order to provide help in various aspects of diabetes therapy, including diet, physical exercise, and insulin management. A lot of them are available for downloading from the Internet, but few of them have been scientifically assessed in terms of measurable outcomes on diabetes control.

A systematic review on computer-based diabetes self-management interventions for adults with type 2 diabetes concluded there was a small beneficial effect on glucose control but that the effect was larger in the mobile phone subgroup. ⁷

Although not dedicated to type 2 diabetes, the DIABEO smartphone-based system that has been developed for patients with type 1 diabetes using multiple daily insulin injections or insulin pumps has shown promising effects. ⁸ This system includes two components: (1) smartphone-embedded algorithms for insulin bolus computing according to premeal blood glucose level, carbohydrate/insulin ratio, and insulin sensitivity factor and for advising on basal insulin adjustments based on fasting blood glucose levels and (2) smartphone-based transmission to a remote server of blood glucose values and insulin doses, allowing visualization of data on the healthcare professional's personal computer through a connection to the server. The patient's physician may then call the patient to suggest insulin dose adjustments. Both services have been investigated in a randomized control study during 6 months. The patients who tested the insulin dose advisor only showed an average 0.67% reduction of A1c compared with controls who did not use DIABEO software, but this reduction did not reach statistical significance on mean A1c data, whereas the patients who also used the data transmission followed by short teleconsultations obtained an average 0.91% A1c reduction, which was significant on mean A1c data.

Interestingly, a post hoc analysis investigated the specific usefulness of each service in low and high users of the software application for dose advising. ⁹ Although high users more deeply improved their A1c levels, no significant benefit resulted from teleconsultations. Low users reduced their A1c levels less, but teleconsultations tended to provide a higher benefit on A1c levels. These data suggest the existence of at least two subtypes of patients: one who benefits more from technical advising support and one who benefits more from motivational support provided by the phone calls.

The experience reported by Hsu et al. ¹ tested a more sophisticated cloud-based program that included enhanced features of interactions between the patients and the healthcare professionals. Indeed, interactions included suggested changes of insulin dose by the healthcare professionals based on transmitted blood glucose data as well as virtual visits, asynchronous text messages, and recommendations for collaborative decisions. Hence, this cloud-based program combined complementary actions on direct dose adjustments and on the promotion of patient involvement in the therapy. Results were very positive as shown both by A1c improvement and by higher patient satisfaction. Thanks to the easy interactions though the cloud-based system, a tight link could be created between the patients and the healthcare professionals as indicated by the average 130.2 messages per patient (i.e., a degree of interactions that could not be possible through the conventional face-to-face visits or even previous reported experiences of telemedicine such as DIABEO). Although limited to a small number of patients, this pilot experience seems to pave the way toward an innovative and more effective way of managing diabetes at important cornerstones such as initiation of insulin therapy.

The following question is about the sustainability of cloud-based interactions between healthcare professionals and patients in the long term in a chronic disease. Indeed, a chronic disease is characterized by long periods of rather quiet times and iterative sudden events of destabilization at high risk of acute complications and hospitalizations. These characteristics of chronic diseases ask for the further development of cloud-based applications that can identify acute high-risk situations and induce remote intensified support from the healthcare professionals. The implementation of this new paradigm for long-term “connected medicine” implies the ability to identify “biomarkers” associated with a risk of serious glucose deviations and the development of algorithms that will drive a decision strategy aiming at quickly restored glucose control in order to prevent acute complications and hospitalizations. New technologies for diabetes management such as continuous glucose monitoring and closed-loop insulin delivery are expected to promote the development of cloud-based diabetes management through innovative ecosystems. ^10,11 The subsequent question is then no longer “Can Diabetes Management Be Done Effectively Through the Cloud?” but “How?”