Lung Function Testing Prediction Equations Response: Do They Fit All?

Vanhorebeek and colleagues1 recently addressed the controversies surrounding insulin treatment of hyperglycemia in the ICU. The authors reviewed single-center studies2,3 that demonstrated benefits of tight glucose control, and a multicenter study4 that showed no improvement with insulin use. Failure to achieve glucose targets was proposed to explain the negative multicenter findings.1 This plausible hypothesis begs the question: how did the single-center and multicenter studies compare in achieving target glucose? Vanhoorebeek and colleagues1 stated that 70% of treated patients in the Leuven studies2,3 had mean daily blood glucose levels 110 mg/dL, compared with a median glucose level of 118 mg/dL in the Glucontrol study.1 The Leuven studies2,3 reported 6:00 am mean glucose levels, but if the mean daily blood glucose level was calculated from these values alone, meaningful comparison of glucose control in the different studies requires the assumptions that glucose does not vary with time of day and is distributed normally. We recently determined that these two assumptions are incorrect; in our ICU, glucose levels varied in an ultradian pattern, peaking twice every 24 h, and had a positively skewed distribution.5 These new findings complicate the comparison of the reported data undermining the conclusion that failure to achieve target glucose levels underlies the failure of the multicenter study.1 For clinicians and investigators attempting to replicate the impressive results of the Leuven studies, it is essential to know what proportion of all glucose measurements were in range for patients receiving insulin in those studies because 6:00 am values are not representative of mean daily glucose levels.5 We agree with Vanhorbeek et al1 that achieving the glucose target range is crucial when investigating the efficacy of insulin infusion protocols, but the objective comparison of these studies requires that the glucose values achieved be clearly reported and include multiple time points.