Traversing the valley of glycemic control despair

The debate on glycemic control (GC) in critical care nears the 16-year mark since the initial landmark study. In this journal, Mesotten et al. [1], from the Leuven group of the original study, present the results of the LOGIC-2 multicenter trial of model-based GC versus standard nurse GC. In 1979 Kelley and Conner [2] published the “Emotional Cycle of Change” defining five emotional phases involved in any major change: 1) uninformed optimism; 2) informed pessimism; 3) the valley of despair—where many “opt out”; 4) informed optimism; and eventually 5) completion and success. We have certainly seen uninformed optimism at initial results [3], followed by many unsuccessful attempts to repeat them and the rise of (increasingly) informed pessimism [4–6]. The resulting confusion in the field led to rival camps of “believers” agreeing to disagree [7], and the valley of despair as many clinicians found GC unnecessary [8], despite strong associations between GC performance, such as time in band and reduced hypoglycemia, and clinical outcomes [9]. The results of the study by Mesotten et al, in the context of other very recent results, suggest GC has made it to step 4—informed optimism. Very recent analysis in this journal suggests the association between mortality and glycemic levels, safety, and variability is a function of the quality of GC and not of patient condition or outcome, indicating GC can play a major role in patient outcomes [10]. Other results in the journal have indicated achieving outcome benefits requires essentially all patients receive safe, effective GC [11]. Hence, we can begin to conclude that GC is important, yet very difficult, and thus how it is implemented may matter as much as whether it is implemented. In light of these points, the multicenter LOGIC study [1] offers two key insights into the main needs and the basis of future success in GC in critical care ... thus taking us to informed optimism. First is the need for repeatability across patients, best defined as the need for patient-specific “one method fits all” solutions, where many studies fail to achieve safe, effective control for (essentially) all patients [12]. Second, and most critical, is the need for repeatability across units, where many prior trials have failed at the hurdle of variable performance within, but especially across, units. The results by Mesotten et al. in [1] support both these needs.