Adults with early-onset type 2 diabetes (aged 18–39 years) are severely underrepresented in diabetes clinical research trials

Aims/hypothesis Early-onset adult type 2 diabetes (diagnosed between ages 18 and 39 years) is increasingly prevalent and associated with poor long-term outcomes. We hypothesised that individuals with early-onset adult type 2 diabetes were underrepresented in the prominent research trials that underpin type 2 diabetes management guidelines. Methods We reviewed the mean age of the study populations recruited to 90 prominent trials in type 2 diabetes, including 37 cardio-renal outcomes trials across a range of pharmacological, non-pharmacological and multifactorial interventions, 28 trials from the phase III programmes of three representative glucose-lowering therapies used routinely in clinical practice (empagliflozin, liraglutide and sitagliptin) and 25 prominent trials of diabetes self-management education and support or intensive lifestyle interventions (diet or supervised exercise training). We then estimated the number of individuals within these trials who were aged between 18 and 39 years. Results Across all 90 trials, the mean age of 268,978 participants was 63 years (range 51–69 years in individual trials). In 73 trials (81%), <5% of participants were estimated to be aged 18–39 years, despite this age group representing ~15–20% of the adult type 2 diabetes population. Twenty-nine of these trials (32%; total 164,953 participants) excluded individuals below 40 years of age altogether. Conclusions/interpretation Guidelines for early-onset adult type 2 diabetes are extrapolated predominantly from evidence in older individuals. Strategies to support the participation of individuals with early-onset adult type 2 diabetes in future research are imperative to ensure guidelines for these high-risk individuals are evidence-based. Electronic supplementary material The online version of this article (10.1007/s00125-020-05174-9) contains peer-reviewed but unedited supplementary material, which is available to authorised users.


Study Selection
Firstly, we searched for published manuscripts (from inception to 27 th September 2019) reporting demographic data of the study populations recruited to cardio-renal outcomes trials in adults with type 2 diabetes. This included both ongoing (but fully recruited) and completed trials. To do this, we: i. Searched the online database, PubMed, using search terms related to type 2 diabetes, mortality and cardio-renal outcomes.
ii. Examined the reference lists of review articles and meta-analyses identified by our search.
iii. Examined the reference list of the ADA-EASD consensus report for the management of hyperglycaemia in adults with type 2 diabetes [1].
Secondly, we identified studies contained within the Phase III research programmes for empagliflozin and liraglutide through a targeted search for trials in the "EMPA-REG" and "LEAD" programmes, respectively, and for manufacturer-funded trials examining the efficacy of sitagliptin on improving glycaemic control in isolation or alongside alternative glucoselowering therapies (GLTs). We pre-specified trials of empagliflozin, liraglutide and sitagliptin as they: a) Are representative of their class of therapy. b) Are the most commonly prescribed, were the first licensed worldwide or had the earliest cardio-renal outcomes data. c) Have Phase III research programmes that are representative of those required for modern licencing of pharmacological GLTs in type 2 diabetes.
Thirdly, prominent trials of diabetes self-management education and support (DSMES) and intensive lifestyle interventions (diet and/or structured exercise training) were identified by review of the ADA-EASD consensus report for management of hyperglycaemia in adults with type 2 diabetes [1], and subsequent examination of relevant reference lists. Specifically, we selected six DSMES trials that were highlighted within a recent narrative review as those examining the most well-established DSMES programmes in type 2 diabetes over the past 10 years [2], and supplemented these with trials recruiting more than 500 participants identified in two systematic reviews of DSMES on glycaemic control and all-cause mortality, respectively [3,4]. One trial was excluded because it recruited participants with type 1 or type 2 diabetes [5], whilst another was excluded because it examined both DSMES and, in a subset, concurrent pharmacological therapy [6].
We selected three dietary intervention trials which were cited directly within the ADA-EASD consensus report [1], examining the impact of meal replacement therapy, a Mediterranean-style diet and the "Dietary Approaches to Stop Hypertension" (DASH) diet. We included these trials as they were specifically in type 2 diabetes and did not include cited trials that recruited participants with overweight or obesity with or without type 2 diabetes. We also selected representative trials of low-fat, low-carbohydrate, low-glycaemic index, and high-protein diets from two systematic reviews cited within the ADA-EASD consensus report [7,8]. A large RCT of the commercially-available Weight Watchers programme was identified [9], but it did not report the age of recruited population and was thus excluded.
To explore studies of supervised exercise training, we selected a large RCT which was cited directly in the ADA-EASD consensus report [1], supplemented by a further five trials which examined the impact of exercise training in more than 100 adults with type 2 diabetes via five systematic reviews that were also cited [10][11][12][13][14]. The large "Early Actid" trial (more than 500 participants) was selected from a systematic review of pedometer use to support light-intensity (walking) physical activity in type 2 diabetes [15]. However, this trial examined the effect of dietary counselling with or without pedometer use and was subsequently categorised as a DSMES trial within our analyses.

Data Extraction and Analysis
We extracted data from selected manuscripts on the number of participants recruited (n), the age eligibility criteria, and the age of the recruited population. The latter were extracted as mean and standard deviation (SD) or median and interquartile range (IQR), as reported. SD was calculated from standard error of the mean (SE) or 95 % confidence intervals where required [16]. The pooled mean age of individuals recruited to various groups of studies (e.g. all studies collectively, cardio-renal outcomes trials alone, Phase III trials alone) were summarised as a weighted mean accounting for differences in study sample size. All trials allowed inclusion of adults aged ≥18 years, except one, which allowed those aged ≥21 years. All trials also reported age data as mean and standard deviation, thus allowing the estimation of the proportion of individuals aged 18-39 years.