A review of the data available from diabetes-related trials registered in ClinicalTrials.gov from 2007 to 2010 provides an important window on the current clinical research enterprise in this therapeutic area. Our descriptive analysis found that the majority of registered trials involve drug therapies rather than preventive or non-drug interventions. Trials appear to include relatively small numbers of patients, are primarily conducted at single sites and are of fairly short duration. Trials often exclude children and elderly participants, their global distribution does not correlate with regional disease prevalence, and only small numbers of trials have focused upon mortality or clinically significant cardiovascular complications.
The International Diabetes Federation (IDF) and the ADA emphasise diabetes prevention as a focus of future research [10, 11]. Previous trials have demonstrated that various lifestyle and pharmacological interventions may delay the onset of diabetes in high-risk persons [12–15]; however, additional study is needed to enhance the implementation of preventive strategies in practice and assess the utility of novel interventions. And despite convincing evidence that intensive glycaemic control minimises the onset and progression of complications [16, 17], a significant percentage of persons with diabetes have not achieved optimal glycaemic control [18, 19]. Further study regarding the translation of effective educational, preventive and therapeutic interventions in the community setting is also encouraged by the ADA . We have found that most diabetes-related studies in ClinicalTrials.gov focus on treatment (usually drug-related rather than behavioural), while only small percentages are primarily concerned with prevention, health services research, supportive care, diagnosis or screening. And although the ideal proportion of trials focused on prevention has not been established, the current trials portfolio, comprising studies with smaller sample sizes and shorter durations, appears to be inadequate for expanding and refining preventive efforts or translating effective care strategies into the community setting.
The IDF, ADA and others have emphasised the need for trials designed to compare the effects of therapies in diverse, high-risk and representative populations [10, 11, 20]. The prevalence of diabetes varies by global region and country  and by race/ethnicity . In addition, rates of complications including diabetic retinopathy, lower extremity amputation and end-stage renal disease vary among ethnic groups [22–25]. To achieve the greatest impact upon clinical care, trials should enrol patients representative of populations disproportionately affected by diabetes, including high-risk patients and those ≥65 years. A better understanding of responses to interventions among diverse individuals and groups may inform individualised treatments of greater effectiveness and tolerability [26, 27].
Race and ethnicity of trial populations are not required fields for registering studies with ClinicalTrials.gov; therefore, there is no readily available information for this category within the dataset. However, the location of trials within countries provides insight into the relationship between clinical trial activities and highly affected populations. Registration with ClinicalTrials.gov is not required for studies taking place outside US jurisdiction; nevertheless, approximately half of the trials registered in ClinicalTrials.gov did not have any US sites. Thus, although our dataset affords an incomplete view of trials activity worldwide, it still likely that it provides a reasonably accurate global perspective.
The IDF list of the ten locations most affected by diabetes includes multiple Middle Eastern countries in which the prevalence of diabetes among adults is approximately 20% . However, our analysis suggests that this region is minimally involved in diabetes-related trials. A comparison of trial activities in countries with the highest prevalence of diabetes among adults reveals over 500 (1,126) trials in the USA; however, we also noted state- and regional-level exceptions to this, as detailed in the ESM text and ESM Fig. 1. China, India and Mexico participated in 101–250 trials each; however, the Russian Federation (12.6 million persons affected) and Brazil (12.4 million affected) are involved in only 51–100 registered trials despite heavy disease burdens .
Trials registered in ClinicalTrials.gov are predominantly conducted in North America, Western Europe and a small number of countries in Asia. Notably, most of Africa is either uninvolved or minimally involved in registered studies. Thus, current trials appear unlikely to provide significant insight into the management of patients from many highly affected or under-studied areas.
The ClinicalTrials.gov database permits a review of the ages of participants sought for (or excluded from) trials. Although those aged 40–59 years constitute the largest number of persons affected by diabetes worldwide, older persons are at greatest risk of the disease. For example, 26.9% of US residents ≥65 years were estimated to have diabetes in 2010 . Our analysis found that persons >65 years were excluded from 30.8% of trials, and that the majority of trials excluded those aged >75 years. Thus, the current clinical research portfolio may not allow us to robustly address issues in older persons with diabetes.
Less than 4% of registered trials targeted the enrolment of participants ≤18 years. This may be appropriate given the number of children affected by diabetes; however, the estimated 3% annual increase in incidence of type 1 diabetes may warrant greater representation . Furthermore, the increase in type 2 diabetes among adolescents, particularly noticeable in wealthier nations, is of considerable concern. It is unclear whether findings obtained from adults with diabetes are readily translatable to paediatric/adolescent populations. The inclusion of younger participants in diabetes trials is essential to ensure safe and effective clinical interventions for these groups, particularly given their risk of developing disease complications early in life. Current clinical trials do not appear to be appropriately positioned to address issues related to disease prevention or management in the young.
Organisations including the US Institute of Medicine have encouraged comparative effectiveness trials to comprehensively assess the benefits and risks of multiple therapeutic options for diabetes and other diseases [27, 28]. Trials comparing the safety, effectiveness and durability of the many glucose-lowering therapies now available will create a reliable evidence base for clinical care guidelines.
The majority of currently registered diabetes trials have a parallel intervention model; however, most have two treatment arms (54.7%) or one treatment arm (19.1%), leaving only a small percentage with three or more arms. Among trials with two or more arms, 51.5% include an active comparator. Furthermore, the relatively short duration typical of these trials may compromise our ability to ascertain the durability of therapeutic interventions or the effects of interventions upon long-term complications. The single-site nature and limited enrolment of most trials are likely to limit the conclusions drawn from their results. Therefore, the current set of trials may not contribute to meaningful changes in recommendations for care.
Diabetes care organisations worldwide have emphasised a need to minimise diabetes-related complications. Groups including the ADA have strategically prioritised investigations that will enhance our understanding of these complications, including cardiovascular disease . The relationship between glycaemic therapeutic targets, hypoglycaemia and cardiovascular complications remains inadequately understood and contentious, despite multiple recent outcomes studies [29–32]. In addition, scrutiny of the cardiovascular effects of individual glucose-lowering therapies has increased following concerns about rosiglitazone and other drugs in development , resulting in new FDA and EMA guidelines for evaluating the cardiovascular safety of new glucose-lowering agents [34, 35]. Of the 2,439 trials in the dataset with available outcomes descriptions listed, only 35 show a primary outcome related to mortality or clinically significant cardiovascular endpoints (e.g. myocardial infarction or stroke). Only small numbers of trials reported primary outcomes related to bone metabolism, malignancy or pancreatitis despite significant clinical interest in the relationship between these issues and glucose-lowering therapy or diabetes itself [36–39].
There are limitations to our ability to draw firm conclusions from the data available, many of which have been previously described . Although ClinicalTrials.gov encompasses a substantial proportion of clinical trials and an estimated 80% of studies in the WHO portal, it does not include all studies performed worldwide . Incorporating non-duplicate trials registered with other international databases would have provided a more complete global perspective; however, such an undertaking would require relatively intensive curation efforts to ensure that duplicate studies were removed and categories appropriately matched, and thus lies beyond the scope of the present work.
Requirements and methods for collecting information about trials have changed over time, and data completeness and quality are variable—an unsurprising finding, as the data collection was not initially designed to support aggregate analysis. Missing data, classification of data as ‘other’ in many circumstances, and non-standardised free-text descriptions also complicated our analysis, particularly when reviewing data related to funding sources and trial outcomes. Funding sources are also classified in a manner most relevant to US-based trials. We were able to identify the presence of a trial within countries and specific US states; however, the number of unique sites per country or state could not be determined, thus limiting our capacity to assess the proportion of trial activity in relation to the population density of interest within a given area. In addition, information about facilities that had not yet been activated when the database was downloaded or had been removed from the current study record is excluded from this investigation. Future refinements to data collection may permit a more complete and sophisticated analysis of trials characteristics.
With respect to the data analysis, the non-hierarchical MeSH classifications may categorise a condition in multiple locations, potentially leading to false positives upon querying for a specific condition. Endocrinology experts at two institutions annotated the database; however, this annotation has not yet been externally validated. In this initial overview, we did not examine whether various characteristics have changed over time and are thus unable to discern meaningful trends in the design or implementation of clinical trials.
In summary, this descriptive analysis of data from the ClinicalTrials.gov registry provides a broad overview of interventional clinical studies related to diabetes. Although many trials will provide valuable information upon completion, our review suggests that the current portfolio does not adequately address disease prevention, management or therapeutic safety. This information may be meaningful in the allocation of future research activities and resources.