Do people prefer to take part in a clinical trial from home or come to site? A discrete choice experiment in type 2 diabetes mellitus

Abstract

Objectives

To determine what drives participation in clinical trials with decentralised elements and to estimate trial participation probabilities for trials with different degrees of decentralisation.

Design

Patient preference study using a discrete choice experiment.

Setting

Recruitment in primary, secondary, tertiary care and other settings in the Netherlands (NL), Austria (AT) and Germany (DE).

Participants

People with type 2 diabetes mellitus (T2DM) aged ≥18 years. A total of 787 people (NL n=276, AT n=265, DE n=246) participated in the study.

Main outcome measures

Preferences for participation in clinical trials with different options for location and type of contact with the study team, activities to perform by participant, use of digital technologies by participant, number of scheduled contacts, trial duration, known safety and efficacy of the drug.

Results

How much was known about the safety and efficacy of the drug was the most important element in the decision whether to participate in a clinical trial in all countries. The trial duration, location and type of contact with the study team, and number of scheduled contacts were other important elements. Participation probabilities for hypothetical trial scenarios differed between countries, with the highest rates for a decentralised trial involving video contact (NL: 89%; AT: 99%; DE: 84%).

Conclusions

People with T2DM prefer to take part in clinical trials with decentralised approaches. Information on preferences can help trialists and protocol developers to design and plan future trials that integrate patients’ needs and thus reduce barriers to participation.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• Preferences for participation in trials with different levels of decentralisation were elicited in three European countries by applying different recruitment methods in various settings, with both online and paper-based questionnaires in countries where digital access and use were expected to be lower.

• The selection of the attributes and levels for the discrete choice experiment was based on extensive qualitative research, discussions with various stakeholders, as well as input from patient and public involvement, in order to ensure all important aspects are covered.

• Conducted in people living with type 2 diabetes mellitus, who are often familiar with self-management and digital tools, this study offers insights from a population well-suited to decentralised clinical trials, yet future work is needed to assess preferences in other disease contexts.

• This research focused on drug interventional clinical trials and provides a basis for future studies in non-drug trials to explore whether and how preferences differ across other trial settings.

Introduction

Voluntary participation in a clinical trial is crucial for investigating the safety and efficacy of treatments. However, the recruitment of participants poses a challenge in many trials, and recruitment targets are often not met,^1–3 which could lead to premature trial termination,^{4 5} increased costs and sometimes insufficient data to answer the research question.¹ The reasons for recruitment failure are manifold, including a general lack of awareness about clinical trials, geographical obstacles or high demands on participants, such as a large number of site visits.^6–9

Decentralised clinical trial approaches (DCTs) allow participants to take part in a trial from their home or local area. This is made possible by relocating some or all of the trial activities close to the participant through the use of digital technologies and innovations in trial operations.^{10 11} DCTs may allow for easier integration of trial participation into everyday life, for example, through trial visits via video or phone call, by decreasing or eliminating travel time to the trial site and potentially reducing the burden of participation.^{10 12 13} Furthermore, DCTs could positively impact trial access and diversity of study populations, as trial participation is not tied to location.^14–17 However, due to at-home data collection by participants and the use of technologies for automated data collection, participants may have less scheduled contact moments with healthcare professionals (HCPs) and most DCTs require participants to have a basic digital understanding.

When it comes to the decision whether to participate in a clinical trial, there are several influencing factors such as expected health advantages, altruism, trust, trial duration and time requirements.^18–21 In a focus group study, seven main themes were identified as drivers for participation in DCTs: location, time investment, contact with HCPs, digital technologies, data collection, perceived risk of the intervention and motivation.²² To be able to inform future trial design and planning, quantification of the importance of these drivers for participation in trials is required to understand people’s preference. Such insights are essential for trialists aiming to meaningfully integrate patients’ needs into the design of DCTs, a priority that is strongly underscored by current regulatory and methodological guidance documents.^23–25 By doing so, this can positively impact patient centricity, minimise burden for participation and potentially enhance trial participation.

To this end, the aim of this study was to determine the preferences and trade-offs for participation in a clinical trial with decentralised elements and to further investigate the acceptance of different levels of decentralisation in hypothetical clinical trial scenarios. This was done by conducting a discrete choice experiment (DCE) in three European countries. The research was performed in people living with type 2 diabetes mellitus (T2DM), a chronic disease where people are often familiar with self-management at home and with the use of digital technologies, and thus a suitable therapeutic area for the conduct of DCTs.

Methods

Study population and recruitment

Persons diagnosed with T2DM at least 1 year before the start of the survey, who were ≥18 years old and were residing in the Netherlands (NL), Austria (AT) or Germany (DE), were eligible to participate in the DCE survey. The country selection was based on differences in internet access and use,²⁶ driving distance to a hospital,²⁷ population density,²⁸ clinical trial activities,²⁹ the native language and the networks of the researchers. For instance, it is reported that internet access and use are higher in the NL compared with AT and DE,²⁶ and both the NL and DE have more regions where the entire population resides within a 15 min drive of a hospital compared with AT.²⁷ Recruitment was performed by HCPs, general staff and/or researchers in various settings, for example, pharmacies, general practices, hospitals, outpatient departments, diabetes/internal medicine specialist practices, clinical trial units, diabetes associations, diabetes support groups. Depending on the country, existing networks and databases were utilised and diabetes events and conferences were visited. Recruitment was conducted through direct HCP outreach, face to face or by letters, and was supported by flyers and posters in the above-mentioned settings, and posts on social media and relevant websites. A sample of around 300 participants was aimed for in each country, based on calculations using a general rule of thumb as per guidance considerations.^30–32 Participants did not receive any financial incentive for completing the survey.

Discrete choice experiment

A DCE was performed to elicit preferences regarding trial participation by means of a survey. The DCE was planned, conducted and analysed according to international guidance reports.^32–34 A DCE is a commonly used method to elicit preferences in healthcare.^{32 35–37} In a DCE, participants are asked to make choices on a series of hypothetical choice tasks. Each choice task comprises two or more alternatives, which are described by different characteristics, that is, attributes, and their varying degrees, that is, attribute levels.^{35 38 39} As participants choose the most preferred alternative for each choice task, this enables the calculation of utility estimates using statistical methods based on the underlying random utility theory.^{35 40}

Attributes and levels

The selection of attributes and levels for the DCE was derived from qualitative work. Drivers for participation in clinical trials with different decentralisation levels were identified in focus groups conducted with people living with T2DM.²² Each focus group participant individually ranked the final list of identified drivers according to their importance. This ranking was subsequently used by the researchers to prioritise the drivers to the six most important attributes for use in the DCE. In a next step, the research team defined the levels for the attributes based on the information gained from the focus groups and the literature. Additionally, discussions were held with various stakeholders in the three participating countries, that is, people living with T2DM, clinicians, ethicists, clinical trial managers, study nurses and researchers, to ensure the completeness and accuracy of the attributes and levels and to ascertain that the language and wording chosen was understood by the anticipated participants. Table 1 provides an overview of the final attributes and levels used in the DCE.

Table 1.

Attributes and levels of the discrete choice experiment

Attributes	Levels
Contact with the study team	Trial site at 30 min travel time
	Trial site at 90 min travel time
	Your home: study team who visits you, no travel time
	Your home: video contact, no travel time
	Your home: messages and chat functions, no travel time
	Combination trial site and your home, site at 30 min travel time
	Combination trial site and your home, site at 90 min travel time
Activities that you need to perform yourself	Take a drug, diary, questionnaire
	Take a drug, diary, questionnaire+simple measurements
	Take a drug, diary, questionnaire+simple measurements+complex activities
Use of digital technologies	No digital technologies
	Digital technologies for communication
	Digital technologies for communication and measurements
Number of scheduled contacts	Once a week
	Once a month
	Once every 3 months
	Once a year
Trial duration	Half a year
	1 year
	3 years
	5 years
Safety and efficacy of the drug	Drug is safe and works
	Drug appears to be safe and to work
	Little knowledge about safety and efficacy

DCE design and survey

For the DCE questions, a Bayesian D-efficient design³⁴ was generated using the Ngene software V.1.4.⁴¹ In the initial pilot studies, the priors for all levels were set to zero. The design included the restriction that trials that did not use digital technologies could not be fully decentralised and take place at home without study team visits. The design did not include any interactions. Pilot data from AT (n=157) was analysed with a conditional logit model and priors of the pilot design were updated accordingly to improve the efficiency of the design. The updated design was implemented in all surveys and data was collected in all three countries.

Two unlabelled alternatives (trial A and trial B), which included trial options ranging from no decentralisation to full decentralisation, and an opt-out option, were programmed for the DCE in the SurveyEngine platform. The opt-out option was included to reflect reality, where people also have the possibility of not participating in a trial.⁴⁰ Due to the high number of required choice tasks per participant as determined by the number of parameters in the utility function, a blocked design was generated to reduce the number of required choice tasks per participant to 16 (3 blocks) to decrease participant cognitive burden and fatigue. An example of a choice task is illustrated in figure 1.

Figure 1.

Example of a choice task presented to DCE participants. DCE, discrete choice experiment.

The DCE choice tasks were embedded in a survey, in which participants were first asked to give informed consent. This was followed by an introduction to the topic of DCTs as well as demographic and socioeconomic questions. Before starting with the DCE choice tasks, participants were introduced to each attribute and its corresponding levels by providing an explanation in simple and lay language. An example choice task was presented to facilitate understanding of the choice exercise. After the 16 choice scenarios, participants were asked which attribute mattered most to them and questions concerning disease characteristics, trial experience, digital access and use, health literacy,^42–44 recruitment into the DCE study and evaluation were presented. All answers to the questions were self-reported by the participants.

The survey, including the DCE, was translated into the respective languages and was administered to patients online via the SurveyEngine platform. Before going live, the survey was pretested in think aloud rounds with people living with T2DM to ensure comprehension, readability, understandability and wording of the text provided. The think aloud rounds were held in AT and the NL to cover both languages used. The feedback received was incorporated into the final version of the survey. A link to the final survey was generated and placed on the websites, allowing participants to access the survey online via desktop, smartphone or tablet. A printed version of the survey was also made available in AT and DE as both countries were expected to have lower internet access and use compared with the NL.²⁶ Data were collected between July 2023 and July 2024. The data from the printed surveys were manually entered into the SurveyEngine platform by two researchers (BL and JK) applying the four-eyes principle to ensure correctness of the data.

Patient and public involvement

This study is part of the IMI Trials@Home Project and the study was supported by the Trials@Home Patient Expert Panel (PEP) members. The PEP members were actively involved in the discussion process concerning the attributes and levels for the DCE. Further, people living with T2DM were involved in the pretesting of the survey in the form of think aloud rounds.

Data analyses

Data obtained from the DCE were analysed using a mixed multinomial logit (MIXL) model to allow preference heterogeneity.^{45 46} The data were analysed separately per country and only respondents who had at least one choice task completed were included in the analysis. The pilot study data were also included in the final analysis.

Effect coding was used for all attributes because it enables calculation of a coefficient for the omitted level per attribute.^{33 47 48} An alternative specific constant was included in all models to adjust for potential left-right bias (ie, participants being inclined to choose option A over option B). The MIXL model was run with 5000 Halton draws and with all parameters included as random. The p values of the coefficients of the SD parameters were checked for significance (ie, indicating preference heterogeneity). If one attribute level within an attribute was significant, the attribute was included as random and otherwise as fixed parameter. The utility equations of the final model used are provided in online supplemental appendix 1.

The beta coefficients of the mean parameters of the final model per country were interpreted and used for further calculations. If at least one level had a significant coefficient at the 5% level, the attribute was considered important in the decision to take part in a trial. The relative attribute importance was calculated by the maximum utility difference between the most and least favoured level per attribute. To compare importances, each utility difference was divided by the sum of all utility differences.⁴⁹

Trial uptake probabilities were calculated using the estimates of the individual beta coefficients to account for heterogeneity in preferences. All estimated trial scenarios were compared with the opt-out option using equation a.

Equation a:

$$P_i=\frac{e^{V {trial scenario}_i}}{\sum { e}^{V {trial scenario}_i}+{ e}^{V {opt-out}_i}}$$

Additionally, the number of respondents per country who always, never or sometimes opted out was calculated. Descriptive statistics were calculated for the other survey questions. Categorical variables were presented in frequencies and percentages, and continuous variables were presented with mean, SD, median, IQR, minimum and maximum, depending on the distribution. All data were analysed in Stata Version NOW/SE 18.5.

Results

Characteristics of survey participants

The participant characteristics are presented for each country in table 2. A total of 787 people living with T2DM consented and participated in the survey by filling in at least one choice task. Detailed information on the excluded participants is provided in the flow charts for each country in online supplemental appendix 2. In all countries, more male than female participants took part in the survey. The median age ranged from 60 to 66 years. Most frequently, participants reported having a higher secondary education and residing in rural areas. In the NL and AT, more participants stated to be retired than in paid work, whereas in DE this was reversed. Two-thirds of the participants from AT had previous experience in trial participation, whereas this percentage was much lower in the other countries. Internet access at home was almost always available, with the majority of participants being online every day or almost every day in the last 3 months. Health experience was reported as high. Additional results concerning the feedback of the participants to the survey evaluation are provided in online supplemental appendix 3.

Table 2.

Characteristics of survey participants

	NL (n=276)	AT (n=265)	DE (n=246)
Demographic characteristics
Female	107 (38.8%)	58 (21.9%)	102 (41.5%)
Age in years (median (min-max))	66 (21–87)	63 (35–81)	60 (30–87)
Living area
Urban area	60 (21.7%)	86 (32.5%)	31 (12.6%)
Suburban area	89 (32.2%)	51 (19.2%)	57 (23.2%)
Rural area	127 (46.0%)	128 (48.3%)	158 (64.2%)
Highest level of education*		n=264	n=244
No education or primary education	1 (0.4%)	6 (2.3%)	3 (1.2%)
Lower secondary education	72 (26.1%)	56 (21.2%)	48 (19.7%)
Higher secondary education	125 (45.3%)	148 (56.1%)	119 (48.8%)
Tertiary education	76 (27.5%)	51 (19.3%)	65 (26.6%)
Rather not say	2 (0.7%)	3 (1.1%)	9 (3.7%)
Occupational situation		n=262
In paid work	86 (31.2%)	50 (19.1%)	112 (45.5%)
Self-employed	15 (5.4%)	16 (6.1%)	9 (3.7%)
Unemployed	2 (0.7%)	16 (6.1%)	8 (3.3%)
Not working due to illness or disability	21 (7.6%)	14 (5.3%)	17 (6.9%)
In retirement	131 (47.5%)	160 (61.1%)	85 (34.6%)
Rather not say	–	–	3 (1.2%)
Others†	21 (7.6%)	6 (2.3%)	12 (4.9%)
Previous trial participation	n=205	n=229	n=197
Yes	39 (19.0%)	140 (61.1%)	23 (11.7%)
No	166 (81.0%)	88 (38.4%)	165 (83.8%)
Rather not say	–	1 (0.4%)	9 (4.6%)
Digital access and use
Access to the internet at home	n=202	n=228	n=196
Yes	202 (100.0%)	226 (99.1%)	191 (97.4%)
No	–	2 (0.9%)	5 (2.6%)
Average internet use for private purposes in the last 3 months	n=199	n=222	n=185
Every day or almost every day	186 (93.5%)	204 (91.9%)	163 (88.1%)
At least once a week (but not every day)	10 (5.0%)	13 (5.9%)	17 (9.2%)
Less than once a week	3 (1.5%)	5 (2.3%)	5 (2.7%)
Health literacy	n=203	n=229	n=195
Inadequate	4 (2.0%)	11 (4.8%)	15 (7.7%)
Adequate	199 (98.0%)	218 (95.2%)	180 (92.3%)

*Additional information concerning the level of education: no education or primary education (up to approximately 6 years); lower secondary education (up to approximately 9 years); higher secondary education (up to approximately 12 years); tertiary education (bachelor’s degree or higher).

†Other occupations listed by respondents: NL: fulfilling domestic tasks and care responsibilities (n=6), community service (n=5), being retired but still working/employed (n=3), caregiver (n=2), partly work disability and partly employed (n=1), director/owner (n=1), stopped in care after 45 years (n=1), on-call duty in healthcare (n=1), a pupil, student, in training (n=1); AT: sick leave due to surgery (n=1), retired (n=1), unable to work due to health reasons (n=1), retired and working in own one-man business (n=1), marginal employment (n=1), community service (n=1); DE: fulfilling domestic tasks and care responsibilities (n=5), community service (n=2), helping in the family business (unpaid) (n=1), ill since 1 year (n=1), reduced earning capacity retirement (n=1), passive phase partial retirement (n=1), missing entry (n=1).

AT, Austria; DE, Germany; NL, The Netherlands.

Preferences for participation in clinical trials

The results of the MIXL model are presented in tables 3 and 4; the detailed table including confidence intervals is provided in online supplemental appendix 4. The attributes ‘contact with the study team’, ‘number of scheduled contacts’, ‘trial duration’, ‘safety and efficacy of the drug’ significantly contributed to the choice of trial participation in all three countries. For AT, the attribute ‘activities yourself’ and for DE the attribute ‘use of digital technologies’ did not significantly influence the choice, while this was the case in the other countries.

Table 3.

Preferences for clinical trial participation based on the mixed multinomial logit model: mean preference weights

Attributes		The Netherlands (n=276)		Austria (n=265)		Germany (n=246)
		Coeff	SE	Coeff	SE	Coeff	SE
Mean
Contact with the study team	Trial site: 30 min travel (ref)	−0.07	x	−0.38	x	0.07	x
	Trial site: 90 min travel	−0.47***	0.12	0.31**	0.11	−0.81***	0.12
	Home visit: no travel	0.14	0.14	−0.38**	0.11	−0.22	0.14
	Home: video contact, no travel	0.56***	0.15	0.24	0.14	0.70***	0.14
	Home: messages, no travel	0.20	0.16	0.05	0.13	0.56***	0.14
	Trial site and home: 30 min travel	0.39**	0.12	−0.09	0.09	0.42***	0.11
	Trial site and home: 90 min travel	−0.75***	0.14	0.25*	0.12	−0.72***	0.14
Activities yourself	Basic (ref)	−0.23	x	−0.04	x	−0.17	x
	Basic+simple measurements	0.05	0.06	0.05	0.05	−0.10	0.06
	Basic+simple measurements+complex activities	0.18**	0.07	−0.01	0.05	0.27***	0.06
Use of digital technologies	No technologies (ref)	−0.13	x	−0.21	x	−0.09	x
	Technologies for communication	−0.07	0.06	0.13**	0.05	0.04	0.06
	Technologies for communication+measurements	0.20**	0.06	0.08	0.05	0.05	0.06
Number of scheduled contacts	Once a week (ref)	−0.73	x	−0.40	x	−0.58	x
	Once a month	−0.04	0.08	0.24***	0.06	0.11	0.07
	Once every 3 months	0.40***	0.07	0.28***	0.07	0.31***	0.08
	Once a year	0.37***	0.07	−0.12	0.06	0.16*	0.08
Trial duration	half a year (ref)	0.15	x	0.17	x	0.26	x
	1 year	0.45***	0.08	0.37***	0.06	0.30***	0.08
	3 years	−0.07	0.08	−0.08	0.06	−0.11	0.07
	5 years	−0.53***	0.10	−0.46***	0.08	−0.45***	0.09
Safety and efficacy	Drug is safe and works (ref)	0.87	x	0.47	x	0.99	x
	Drug appears to be safe and to work	0.49***	0.07	0.23***	0.05	0.09	0.07
	Little knowledge about safety and efficacy	−1.36***	0.13	−0.70***	0.08	−1.08***	0.13
Opt-out		0.36	0.22	−3.52***	0.32	0.18	0.29
Alternative specific constant A		−0.02	0.10	0.01	0.08	0.17	0.10

*p<0.05, **p<0.01, ***p<0.001.

Coeff, coefficient; ref, reference level; SE, standard error.

Table 4.

Preferences for clinical trial participation based on the mixed multinomial logit model: standard deviation of preference weights

Attributes		The Netherlands (n=276)		Austria (n=265)		Germany (n=246)
		Coeff	SE	Coeff	SE	Coeff	SE
SD
Contact with the study team	Trial site: 30 min travel (ref)	x	x	x	x	x	x
	Trial site: 90 min travel	−1.14***	0.17	1.25***	0.13	0.89***	0.17
	Home visit: no travel	1.37***	0.14	1.09***	0.14	1.28***	0.15
	Home: video contact, no travel	1.39***	0.17	1.55***	0.18	1.04***	0.16
	Home: messages, no travel	1.59***	0.17	1.50***	0.17	1.26***	0.17
	Trial site and home: 30 min travel	−0.75***	0.17	0.42*	0.16	−0.47*	0.24
	Trial site and home: 90 min travel	1.08***	0.16	1.20***	0.14	0.96***	0.17
Activities yourself	Basic (ref)	x	x	NA		NA
	Basic+simple measurements	0.13	0.08	NA		NA
	Basic+simple measurements+complex activities	0.40***	0.08	NA		NA
Use of digital technologies	No technologies (ref)	x	x	NA		x	x
	Technologies for communication	0.17	0.17	NA		0.35***	0.09
	Technologies for communication+measurements	0.32**	0.10	NA		0.31**	0.11
Number of scheduled contacts	Once a week (ref)	NA		x	x	x	x
	Once a month	NA		0.26**	0.10	0.00	0.21
	Once every 3 months	NA		0.56***	0.09	0.21	0.30
	Once a year	NA		−0.18	0.34	0.48***	0.11
Trial duration	half a year (ref)	x	x	x	x	x	x
	1 year	−0.02	0.12	0.12	0.14	−0,29	0.15
	3 years	−0.34*	0.17	−0.32*	0.13	0.02	0.16
	5 years	0.77***	0.10	0.79***	0.09	−0.60***	0.09
Safety and efficacy	Drug is safe and works (ref)	x	x	x	x	x	x
	Drug appears to be safe and to work	0.27*	0.12	0.18	0.12	−0.37**	0.11
	Little knowledge about safety and efficacy	1.32***	0.11	0.89***	0.07	1.33***	0.12
Opt-out		4.98***	0.36	3.96***	0.39	3.99***	0.34
Alternative specific constant A		1.09***	0.12	0.85***	0.10	0.83***	0.11

*p<0.05, **p<0.01, ***p<0.001.

Coeff, coefficient; NA, not applicable; ref, reference level; SE, standard error.

In AT, people had a strong preference for participating in a trial over opting out. A priori to changing the attribute levels, people in the NL and DE had a preference for opting out, that is, not taking part in the trial options presented. This was also reflected in the actual number of respondents who never (NL: 34%, AT: 62%, DE: 32%), sometimes (NL: 49%, AT: 37%, DE: 54%) or always (NL: 18%, AT: 2%, DE: 15%) opted out.

While in the NL and DE, both respondent groups preferred contact with the study team from home including video consultation over contact modes requiring longer travel, participants in AT preferred contact with the study team on-site and at home including a 90 min journey. In all countries, synchronous contact (video) was preferred over asynchronous contact (messages) for remote interaction. In the NL and DE, participants favoured complex trial activities to conduct on their own, while in AT, trial activities did not influence the choice of trial participation. In the NL, people preferred digital technologies for communication and measurement purposes, whereas in AT, people favoured digital technologies for communication only. In DE, this attribute did not impact people’s choice. All respondents preferred to have a scheduled contact once every 3 months over other options. In all countries, a shorter trial duration was preferred over a longer one, and a drug with a more established safety profile was favoured over a less advanced one, as shown in table 3.

Significant preference heterogeneity describes the variation in how different individuals value the attributes presented to them and is indicated by significantly different SD parameters. It was present for the attributes ‘contact with the study team’, ‘trial duration’ and ‘safety and efficacy of the drug’ in all three countries. Further, significant preference heterogeneity was identified for the ‘activities yourself’ attribute in the NL, for the ‘use of digital technologies’ attribute in the NL and DE, and the ‘number of scheduled contacts’ attribute in AT and DE.

Relative attribute importance

The relative attribute importance, see figure 2, shows that relative to the other attributes, knowledge about the ‘safety and efficacy of the drug’ was the most important attribute in all countries and thus the strongest driver in the choice whether to participate in a trial. While in both the NL and DE, ‘contact with the study team’ followed by ‘number of scheduled contacts’ were the next most important attributes, in AT this was ‘trial duration’, followed by equal percentages for ‘contact with the study team’ and ‘number of scheduled contacts’. The least important attribute relative to the others was ‘use of digital technologies’ in the NL and in DE, while in AT this was ‘activities that you need to perform yourself’.

Figure 2.

Relative attribute importance per country based on the mixed multinomial logit model. AT, Austria; DE, Germany; NL, The Netherlands.

Trial uptake scenarios

To investigate the acceptance of different trial designs, three hypothetical clinical trial scenarios were created, modelled as a conventional clinical trial, a hybrid trial and two DCTs (figure 3).

Figure 3.

Acceptance of different trial scenarios. AT, Austria; DCT, decentralised clinical trial; DE, Germany; NL, The Netherlands.

The baseline clinical trial was defined as a trial taking place at the trial site including 30 min travel, participants having basic measurements to do themselves and using digital technologies for communication, with a scheduled contact once a month, a trial duration of 1 year and the drug that is being investigated appears to be safe and to work, modelled after a traditional clinical trial setting. For the hybrid and decentralised trial scenarios, the following attribute levels were kept equal: a scheduled contact once every 3 months, a trial duration of 1 year and a drug that appears to be safe and to work. A DCT with video contact at home, basic, simple and complex activities and digital technologies for communication and measurements had the highest estimated probability of participation compared with all other trial scenarios (NL: 88.8%, AT: 99.2%, DE: 84.3%). A hybrid trial that included a combination of trial site and home with 30 min travel time, basic and simple measurements and digital technologies for communication yielded lower acceptance compared with the DCT with video contact, but higher acceptance compared with the DCT with home visit and the baseline trial (NL: 81.1%, AT: 99.1%, DE: 72.9%). The DCT with home visit, which included basic and simple measurements and digital technologies for communication, resulted in an uptake of 76.7% in the NL, 98.9% in AT and 60.0% in DE, which was higher than the uptake of the baseline trial. The estimated probability of participating in the baseline trial compared with opting out was similar in the NL (56.2%) and DE (57.7%), but comparatively higher in AT (98.7%).

For all clinical trial scenarios, changes in the attribute levels had a high impact on the participation probability in the NL and DE, but had almost no influence on the probability in AT as people generally had a strong preference for trial participation. The magnitude of the change in participation probability when changing the respective attribute levels for the baseline clinical trial is presented in figure 4 for each country. Changing one trial element, for example, replacing ‘contact at the trial site including a 30 min journey’ with ‘video contact at home’, can lead to 15% more people in the NL and 16% more people in DE being willing to participate in the trial.

Figure 4.

Changes in participation probability from the baseline clinical trial when changing the level of a certain attribute. Baseline clinical trial=trial that takes place at the trial site, where participants have 30 min travel time, have basic activities to do themselves, use digital technologies for communication, have once a month a scheduled contact moment, with a trial duration of 1 year and the drug that is being investigated appears to be safe and to work. The baseline clinical trial had the following participation probability relative to the opt-out per country: Netherlands (NL)=56.2%, Austria (AT)=98.7%, Germany (DE)=57.7%.

Discussion

This study aimed to investigate the preferences and trade-offs for participation in clinical trials in view of decentralisation and to explore the acceptance of decentralised trial elements in hypothetical trial scenarios in people living with T2DM. The type of contact with the study team and number of scheduled contacts, both related to decentralisation, were important trial elements that drive the decision to participate in a clinical trial. Overall, people prefer to participate in trials in which they can (partly) participate from home instead of fully site-based clinical trials. The highest willingness to participate was found for a trial with trial visits through video contact compared with other trial scenarios.

General trial factors, not related to decentralisation of trials, such as knowledge about the safety and efficacy of the drug and trial duration, also had a high influence on trial participation. Participation in trials in which more information about the safety and efficacy of the drug is known, that is, later phase 3 and 4 clinical trials,⁵⁰ was clearly preferred in our study. This type of late-phase trial is also increasingly recognised as well-suited for decentralisation.^{51 52} In contrast, a trial in which a drug is tested on humans for the first time, that is, a phase 1 clinical trial,⁵⁰ is less appropriate for the decentralisation of clinical trials.^{51 52} Participation in such early phase trials can cause harm,⁵³ which may also explain why participants in our study, who are people living with T2DM presented with hypothetical trial options, preferred not to participate in trials with limited information on the drug’s safety. Such preferences may be different in other indications, for example, in life-threatening diseases or when no treatment alternatives are available. For example, it is reported that oncology patients are motivated to participate in early phase trials by the hope of health benefits and for altruistic purposes.^{54 55}

Higher acceptance of hybrid and decentralised trial scenarios compared with a conventional site-based clinical trial scenario was found in all countries in our study, showing that the decentralisation of a clinical trial can increase the willingness to participate. Similarly, surveys reported that participants favoured hybrid and fully remote trials over conventional trials.^56–58 In our study, the location and type of contact were important drivers in all countries. While the Dutch and German participants favoured hybrid and decentralised trial options with video contact and no or short travel, Austrians preferred on-site or hybrid options involving a longer travel. This difference may be explained by geographical characteristics. Given that 48% of the Austrian participants lived in rural areas, it could be assumed that they answered the choice tasks having their current place of residence in mind, as Austrians generally have a longer travel to the nearest hospital as well as a lower population density compared with the NL and DE.^{27 28} Also, most of the Austrian participants were recruited through a database of a clinical trials unit. A positive experience with this clinical trials unit may have influenced their preference towards travel to that unit instead of a more local hypothetical trial site. Further, while reduced (face-to-face) contact moments in DCTs are thought to be challenging,^{59 60} for example, in developing a relationship between participants and HCPs that could benefit retention and compliance,^61–63 participants in our study mostly preferred a contact moment once every 3 months. Our results thus indicate that contact with the study team is important, also from the finding that remote synchronous contact (video) is preferred over remote asynchronous contact (messages), but it does not need to take place that often, which supports DCT approaches with data collection by participants outside of trial visits.

In contrast to often voiced concerns regarding shifting burden to participants in DCTs and digital literacy requirements,^{13 52 60} in our study the attributes concerning performing trial activities yourself and use of digital technologies were least influencing the decision of trial participation. Participants had a slight preference for performing measurements themselves and using innovative technologies. If fit-for-purpose technologies and easy-to-perform activities are available for data collection by participants, this seems both preferred by and feasible for participants in clinical trials. Barge & Floody indicate that participants’ perceived burden of deploying digital technologies presumably has an influence on whether participants want to use them or not.⁵⁶ However, actual trial experiences of participants may differ from their perceived burden, since, for example, the ease of use of the digital devices deployed in trials is relevant to participants.^{64 65} Lack of digital skills and missing digital infrastructure can be an obstacle to taking part in DCTs, as is often reported in the literature.^{52 66–68} In our study, the age ranged from 21 to 87 years, with a median of 60–66 years, and the majority of participants had access to the internet and used it daily or almost daily, implying a high level of digital literacy. As both digital access and use are reported to be lower in AT and DE compared with the NL, a paper version of the survey was provided in AT and DE in our study to minimise selection based on internet access and use. We found that our study samples are similar to that of the general population with regards to internet access and use,²⁶ which means that the study samples are representative of the population for these important population characteristics. Our study shows that general assumptions about digital access and use, for example, in relation to age, should not be made as they may not be valid, but rather the digital access and use of the intended trial population should be assessed during the design phase of a trial. Notwithstanding, adequate digital training measures should be offered and digital infrastructure should be provided in DCTs, as recommended.24 52 56 66 69

Another observed country difference was that while in the NL and DE, whether to participate or opt out was determined by different trial scenarios, a strong preference for trial participation over opting out was recorded for AT. Since 61% of Austrian respondents had previous trial experience and most participants were recruited via a database of a clinical trials unit, we assume that a positive prior trial experience might underly this preference. In addition, differences could also be explained by other characteristics of participants, such as the higher number of retirees in AT, who presumably have more time to travel to the site and take part in a trial than those in employment. Given the preference heterogeneity identified in this study, and observed differences among participant groups with regard to trial participation,^{21 70} it is expected that population characteristics have an influence on preferences for trial participation. Thus, additional research is needed to explain the identified heterogeneity in preferences by investigating what preferences different groups of people have. This will be done as a next step in this study by conducting a latent class analysis.

The current results are limited to preferences from people living with T2DM. Although other chronic diseases may have similar preferences, further research is needed to determine preferences for participation in clinical trials with decentralised elements in other disease areas. For example, patients who are less accustomed to self-managing their condition and using digital technologies may have different preferences, particularly regarding their willingness to perform measurements at home or engage with digital tools in clinical trials. Furthermore, preferences might also differ for other types of clinical trials that do not involve drugs, since the attribute concerning drug safety and efficacy becomes redundant, potentially shifting the importance placed on other attributes. In any case, prior comprehensive qualitative research²² should be performed that would form the basis for the quantification of those preferences in other indications, which is a clear strength of the current study. Further strengths were the inclusion of three European countries and the recruitment of people living with T2DM in different settings.

The knowledge gained from our study contributes substantially to the understanding of patients’ views on participation in trials from home. Our results are particularly important to inform the design and planning of future trials based on patients’ preferences, as is also strongly recommended in regulatory and other guidance documents.^23–25 By incorporating the identified patients’ preferences for decentralised elements, barriers for participation could be reduced, which eventually could have a positive impact on the recruitment and decrease participant burden in clinical trials.

Conclusions

The results of this study in a large number of people living with T2DM support the view that the likelihood to participate in clinical trials is increased with decentralised approaches. The safety and efficacy profile of the investigational drug was the most influential factor in deciding whether to participate in a trial. However, when the safety and efficacy profile was the same across different trial scenarios, acceptance rates were higher for decentralised trials than for a conventional site-based one, supporting the use of decentralisation in clinical trials. A broader adoption of these approaches could facilitate participation by patients who otherwise might be excluded due to geographic, mobility or time constraints. Future trials can benefit from these insights on preferences for trial participation to better address patients’ needs, which may improve recruitment, decrease burden of participation and increase diversity in clinical trials. In order to better tailor decentralisation strategies to different disease stages and types of clinical trials, more research is needed to investigate how preferences differ within groups of patients and to confirm the findings in other therapeutic areas and types of clinical trials.

Data availability statement

The preference data of this study are available from the corresponding author on reasonable request.

Ethics statements

Patient consent for publication

Not applicable.

Ethics approval

The study was conducted in accordance with the principles of the Declaration of Helsinki, good scientific practice and the EU General Data Protection Regulation. The study complied with ethical requirements. Although our research concerns medical scientific research, participants are not subjected to procedures or required to follow rules of behaviour and no burdensome or intimate questions were asked. Therefore, in the NL, evaluation by a medical ethics committee was not required and waived. However, before the utilisation of the Utrecht Pharmacy Practice network for Education and Research (UPPER), the study was approved by the UPPER Institutional Review Board (UPF2312). In AT, the study was approved by the ethics committee of the Medical University of Graz (EK-No. 35-375 ex 22/23), and in DE, it was approved by the ethics committee of the Medical Association of Hessen (EK-No. 2023-3381-evBO). Informed consent was obtained from all participants at the start of the survey.