Clinical trials are needed to develop new medicines and other medical interventions. For the results from these trials to be valid, it is important that the patients who participate in clinical trials look like patients who will be treated with the intervention in the future. However, it is known that clinical trial participants are not always representative of the patients who are treated in clinical practice. For example, certain racial/ethnic groups may be underrepresented in clinical trials. Underrepresentation can occur when the criteria to participate in a clinical trial are strict. However, other factors may also influence the representativeness. For example, the location of the trial site and the study burden (e.g., how many visits are required to participate) may also influence the willingness and ability to participate.
Conducting clinical trials at participants’ homes may affect the representativeness of clinical trials. For example, because individuals who would not participate in a trial with on-site visits could participate in a trial that is conducted from the participant’s home. In the current study, we will describe whether participants from a trial conducted at participants’ homes look like real-world patients and describe whether this is different for participants who participated in a trial that is similar to the home trial, but was conducted on-site. CPRD data will be used to investigate whether the trial participants are representative of patients in the United Kingdom primary care.
Decentralized clinical trial approaches conducted at participants' homes have the promise of improving trial representativeness by allowing the inclusion of immobile participants or participants from rural areas and lowering the burden of trial participation. Broad trial representativeness is important from a health equity perspective and improves the generalizability of study results. However, there is limited data on the potential impact of decentralized trial approaches on representativeness. Therefore, this study aims to compare characteristics of participants from a fully decentralized trial (the ASCEND trial) and a comparable – in terms of clinical research question – clinical trial run at investigative sites (the POPADAD trial) to real-world patients. In this descriptive study, characteristics from primary care patients will be obtained from CPRD GOLD. Practice and patient level deprivation data will be used to determine socioeconomic status (Townsend).People with type 1 or type 2 diabetes without vascular disease (intended users) and patients who would have been eligible to participate in respectively the POPADAD and ASCEND trial will be identified by applying (the trial) eligibility criteria to the CPRD data for the trial recruitment periods and one year before: 1 November 1996 to 31 July 2001 and 1 June 2004 to 31 July 2011 for the POPADAD and ASCEND, respectively. Outcomes of interest include demographics (e.g., age and sex), medical history (e.g., smoking and diabetes duration), biomarkers (e.g. BMI,HbA1c),and medication use (e.g., insulin) as reported in the clinical trial publications. Differences and ratios (including participation to prevalence ratios) will be calculated for these characteristics. This study will add to the (limited) literature regarding decentralized trial approaches and representativeness. Additionally, the framework that is used in this study may be used to evaluate the representativeness of other (decentralized) trials in the future.
Primary outcomes
The proportion of participants and real world patients for categorical variables (baseline characteristics) and mean (SD) for continuous variables. We will describe the differences between the trial participants, intended users, and eligible patients for both the decentralized and the conventional clinical trial.
Secondary outcomes
Baseline characteristics that were not reported in both the decentralized trial and the conventional trial (but only in one of these trials) may be explored. That is, to compare whether the respective trial was representative of the real world patients without comparing the decentralized and conventional operational trial approaches.
The baseline characteristics include:
- Demographics:
Age; Biological sex/gender ; Race/ethnicity; Socioeconomic status (Townsend Index)
- Medical history:
Smoking status; Duration of diabetes; Type of diabetes/insulin use; Diabetic retinopathy; Hypertension
- Biomarkers:
Body mass index; Systolic blood pressure; Diastolic blood pressure; HbA1c; Total cholesterol; HDL cholesterol; LDL cholesterol
- Medication at randomization/index date:
Metformin; Sulphonylurea; Thiazolidinedione; other hypoglycemic agents; Angiotensin-converting enzyme inhibitor/angiotensin receptor blocker; Beta-blockers; Calcium channel blockers; Thiazide or related diuretics; Statins; NSAIDs; Proton pump inhibitors; Aspirin
Helga Gardarsdottir - Chief Investigator - Utrecht University
Patrick Souverein - Corresponding Applicant - Utrecht University
Amos de Jong - Collaborator - Utrecht University
Anthonius de Boer - Collaborator - Utrecht University
Mira Zuidgeest - Collaborator - University Medical Centre Utrecht
Yared Santa Ana Tellez - Collaborator - Utrecht University
Patient Level Townsend Index;Practice Level Townsend Index