Maureen J. Lage, PhD
I read with great interest the article published in an April 2018 supplement titled “Generalizability of Glucagon-Like Peptide-1 Receptor Agonist Cardiovascular Outcome Trials Enrollment Criteria to the US Type 2 Diabetes Population” by Wittbrodt et al.1
Although the authors made an important contribution to the literature by examining the generalizability of cardiovascular (CV) outcome trials to the population of US adults with type 2 diabetes (T2D), I believe that their analysis contains an important omission. Specifically, the authors state that the Exenatide Study of Cardiovascular Event Lowering (EXSCEL) trial contains no eligibility criteria regarding CV events.1
However, although patients with any level of CV risk may be enrolled, the original protocol clearly states that “recruitment will be constrained such that 40% will not have had a prior CV event and 60% will have had a prior CV event…”2
This requirement was subsequently made more restrictive in the protocol amendment, which modified the inclusion and exclusion criteria such that “…approximately 30% will not have had a prior CV event and 70% will have had a prior event.”2
An examination of the characteristics of patients who actually enrolled in the trial revealed that this more restrictive requirement was used, with 73% of patients enrolled in EXSCEL having a prior CV event.3
Although the EXSCEL trial did include patients both with and without prior CV events, it is not true that the study had no eligibility requirement regarding CV events.
An examination of the 2013-2014 National Health and Nutrition Examination Survey (NHANES) data4
revealed that this omission has a significant impact on the findings presented in the authors’ research. Specifically, a replication of the Wittbrodt et al1
examination of the EXSCEL trial revealed that, ignoring the CV enrollment criteria, 50.5% of patients with T2D would be eligible for inclusion in the EXSCEL study (Table1,5,6 [part A and part B]
). This number is similar to the finding of the authors, using NHANES data from 2009-2010 and 2011-2012, which found that 47.2% of adults with T2D would be eligible for EXSCEL. However, within this 50.5% sample of individuals who fit the eligibility requirements examined by Wittbrodt et al, just 10.7% had a prior CV event (Table1,5,6
). When also incorporating the recruitment criteria that 60% of patients should have had a prior CV event, the 50.5% of potentially eligible patients is reduced to 17.9%. If, alternatively, one examines the amended protocol requirement that 70% of patients had a prior CV event, the proportion of eligible patients would be reduced to 15.3%. These results suggest that by omitting the CV inclusion criteria of the EXSCEL study, the authors have included an oversampling of patients with no prior CV event relative to the stated criteria of the EXSCEL trial. As such, the analysis substantially overstates the generalizability of EXSCEL to the overall US T2D population.