Rickels et al. (1) report the results of a randomized crossover noninferiority study, making the firm assertion that intranasal glucagon was highly effective in treating insulin-induced hypoglycemia in type 1 diabetes. Unfortunately, the methods described in their article fail to support such a strong claim. We have several concerns about their ambitious conclusion. First, because of the controlled and artificial environment, this was not an effectiveness but an efficacy trial. Testing effectiveness requires carrying out the comparison in a “real-world setting” of severe hypoglycemia. Second, the design of the study does not seem appropriate considering that the intention was to prove intranasal glucagon was clinically equivalent to intramuscular glucagon. A noninferiority study does not prove that two treatments are equally effective. Therefore, a more accurate objective would be to demonstrate that a needle-free intranasal glucagon preparation is not worse than intramuscular glucagon for treatment of insulin-induced hypoglycemia. Third, the authors do not explicitly describe the reasoning to define their generous noninferiority boundary of 10%. Because the noninferiority boundary defines the maximum acceptable difference between intramuscular and intranasal glucagon before considering the intranasal glucagon treatment as worse than intramuscular, this is a critical issue. We suspect this noninferiority boundary was chosen arbitrarily because of the lack of sound data from previous studies. Clinical judgment suggests that tolerating 10% of a difference between the two treatments seems to be excessively generous considering the magnitude of the harm allowed. Most physicians will find it unacceptable to sacrifice 10% of the efficacy in exchange for a dubious benefit of additional simplicity in the preparation. Moreover, setting an inappropriate margin can cause a noninferiority test to misleadingly conclude an ineffective treatment to be effective. Fourth, we are concerned with the low power of the study considering that noninferiority trials are biased toward noninferiority if they are underpowered. For this reason, noninferiority trials should be designed with at least 90% of statistical power. Fifth, and probably most concerning, is the relaxed definition of the outcome. In this study, success was defined as an increase in plasma glucose to ≥70 mg/dL or ≥20 mg/dL from the glucose nadir after receiving glucagon. We are perplexed with the 30-min criteria used in the study by Rickels et al. to define success. If glucose levels are not restored quickly, prolonged hypoglycemia can become catastrophic (2). Prompt correction is so critical that “the rule of 15” emphasizes the need to correct hypoglycemia within the first 15 min. In Fig. 2 of the study (1), it is clear that before 30 min, the probability of success was lower for intranasal glucagon. As per the data in that figure, the probability for success at minute 15 for intramuscular glucagon was 86% as compared with 71% for the nasal glucagon. This difference seems clinically significant.
Given our concerns cited above, we request that the authors provide rationale for the study design and their interpretation of the data to support their claims of noninferiority.
Duality of Interest. No potential conflicts of interest relevant to this article were reported.
- © 2016 by the American Diabetes Association.
Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at http://www.diabetesjournals.org/content/license.