Memorandum

BMJ meta-analysis of AID: in 40 RCTs, outpatient closed loop increased time-in-range by 2.3 hours per day (+9.6%) – May 14, 2018

A new BMJ meta-analysis of 40 randomized controlled trials dating back to 2014 (n=1,027 participants) compares outcomes on closed-loop in the outpatient setting to open-loop pump therapy in type 1 diabetes. Overall, time-in-range (70-180 mg/dl) significantly increased by 9.6%-points, translating to 2.3 more hours in range per day with closed loop. As expected, observed differences were more dramatic overnight (as defined by individual studies), with participants spending 15.2%-points more time in-range in those periods. Time spent in hypoglycemia (<70 mg/dl) and in hyperglycemia (>180 mg/dl) also improved, with time <70 mg/dl decreasing by 1.5%-points with closed loop therapy (-21 minutes/day) and time >180 mg/dl decreasing by 8.5%-points (-2 hours/day). The figure below shows that only three studies trended toward favoring standard pump therapy, but none appear significant. Notably, the robust results were consistent in unsupervised, free-living studies, and across single- and dual-hormone systems.

Drs. Eleni Bekiari (Aristotle University of Thessaloniki), Hood Thabit (Cambridge), Roman Hovorka (Cambridge), et al. assessed 40 trials with varying parameters: age (17 in children or adolescents, 13 in adults only, 11 mixed), design (33 crossover, 7 parallel), duration (most <4 weeks), system (32 of 40 used insulin-only). We’d note that Medtronic’s 670G pivotal study was not amongst the selected trials, since it lacked a control group.

Results from this review add to evidence from a similar meta-analysis (n=27 RCTs) presented at ADA by University of Toronto’s Dr. Alanna Weisman. Dr. Weisman et al. determined closed loop therapy conferred a mean increase in time-in-range of 12.6%-points vs. conventional pump therapy (+3 hours per day). However, as pointed out by Dr. Bekiaria et al., the validity and clinical interpretation potential of Dr. Wesiman’s results were somewhat undermined by methodological decisions. This recent study includes a substantially larger pool of eligible studies and evaluated a broader variety of outcomes. It’s very encouraging to see such positive data, especially because it reports double the +5% time-in-range improvement seen in the 670G pivotal trial (67%->72%).

Once more commercial systems report pivotal results, it will be fascinating to better understand and compare outcomes – what’s a strong improvement in time-in-range with AID? How will systems compare? And of course, how will time-in-range gains compare with MDI/CGM/decision support?

  • Dr. Bekiari et al. noted significant variation in outcomes assessed in these studies, and recommended establishing a minimum set of agreed-upon measures to collect in every study. Of course, these agreed-upon measures were published for AID studies in Diabetes Care two years ago (Maahs et al.), and the field has since aligned on them more broadly (Diabetes Care 2017). We expect this will be less of a problem in the future.

  • Moving forward, the investigators advised exploration of closed loop therapy in type 2 diabetes, as well as its impact on quality of life and reducing patient burden – we agree! Cost-effectiveness should obviously be evaluated to increase reimbursement and adoption – e.g., how does a two-hour per day improvement in time-in-range translate to reduced healthcare costs? Are long-term outcomes studies the only way to analyze this?

--by Maeve Serino, Brian Levine, Adam Brown, and Kelly Close