Table 4 Table of key recommendation and implementation considerations

Target population

Always consider underserved groups; socioeconomic disadvantage linked to inequalities across the board.

Difficult to know what the population should be.

Intersectionality of under-served groups means more than one group needs to be considered when designing trials.

Under-served groups by their very nature may not routinely access services where we recruit from e.g. primary care.

Assumptions from researchers can pose an issue. For example, the incorrect assumption that because you are an ethnic minority, you will experience socioeconomic disadvantage.

There may be a selection bias going on, and those experiencing socioeconomic disadvantage and lower education/ literacy level may be routinely left out of research/ not approached by staff due to this.

As a minimum you should aim to include ethnic groups in line with the national census.

Information about the population with a particular condition or disease is not readily available.

To make changes to include people that are usually excluded will take longer and require more effort.

Diverse PPI.

Lack of experience with diverse groups up to now; difficult to develop lasting relationships with communities for isolated research projects.

More preparation is needed for the meetings and communication.

You cannot be parachuted into a setting, a community group for example, and expect people to volunteer themselves. You need to build trust.

One interviewee reported using the INCLUDE Ethnicity Framework successfully to remove barriers and recruit the right proportions from different ethnicities.

Some PPI contributors can be quite experienced in research so people with less research experience should also be consulted.

Using existing PPI members is likely to involve people with higher literacy skills; need to include people with lower literacy.

Ensuring eligibility criteria do not exclude under-served groups (routinely missed groups = prisoners, learning difficulties, serious mental health).

Eligibility criteria are often linked to safety, especially in drug trials, or trials are focussed on a specific population that is dependent on the research question.

Some eligibility criteria can be difficult to identify, or misdiagnosed, for example most learning disabilities are not formally documented, making it hard to identify this group.

Refugees with asylum status may be easy to engage with, but those that are undocumented and often highly vulnerable are less likely to engage with services.

Implicit criteria needs to be considered e.g., recruitment methods that rely on reading and responding to a letter or email, being able to attend a particular clinic may exclude some groups.

Recruitment settings

Choose sites from diverse areas.

It can be more work for trials team to add sites that are not ‘research ready’.

Even when trials have sites in areas of ethnic and socioeconomic diversity such as Bradford and Leeds, most participants seem to be middle class and white.

Need to acknowledge the impact of working with sites new to research – may take longer and more training and monitoring.

Just going to a site does not necessarily mean the population taking part will be diverse, cannot be the only method used to improve inclusion.

Have more than one method of recruitment to capture the required population.

Will need increased costs and staffing.

Research teams are stretched enough as it is, and sites/trials often do not have dedicated trial staff to do this.

When using non-research staff to recruit, e.g. ward nurses, additional training in recruitment methods and trials is needed.

People often rely more on the consent conversations than on written information, which can be brought into training for recruiters.

Moving recruitment into the community.

This depends on the research question, and not possible in some circumstances, e.g. If it is a surgical trial, you usually need to recruit the people listed for surgery.

A lot of our research is NHS focussed, and so if people are not accessing NHS services, they will be missed.

Approaching people in a familiar environment helps increase trust.

Just turning up in the community and trying to recruit people does not work; it takes time to build trust and may not fit into trial timelines.

Community engagement efforts for clinical research, requires a long-term investment and may not have immediate outcomes.

Constraints of certain trials e.g. discharge from hospital following a diagnosis of heart failure means by its nature, the patient would need to be recruited in an acute hospital setting.

Building relationships with the community is difficult but there are examples, and we need to share them in the trial community – example of Talking Trials project in Cardiff as a way of building two way relationships.

Researchers could offer skills in research methods, evaluation and writing to help with bi—directional relationships.

Interventions

Use videos to explain the intervention.

A lot of our world is now on screen and in video format, and is almost expected.

Videos do not always work for those hard of hearing or with visual impairment.

Voiceovers in different languages to explain the patient journey were well-received in a maternity trial.

CTUs do seem to be using videos for patient information and dissemination, but not necessarily to help with the intervention delivery – this is dependent on the intervention.

Employ staff from underserved group to deliver intervention.

Dependent on the availability of staff.

We should not have to rely on specific individuals; this can be tokenistic and offensive.

Don’t just employ diverse staff—engage with community partners for wider engagement and co-production.

Provide cultural (or similar) awareness training to staff.

It may be a challenge to get all staff to engage with this, and to complete it.

We shouldn’t be too reductionist about cultural competency and awareness—researchers/ those consenting should always take an empathetic, friendly approach irrespective of the patient’s culture.

Cultural humility is the idea that self-awareness, self-reflection and supportive interactions should always be practiced, and in this context, that the recognition of diversity and power imbalance are very important in patient care—can also be known as ‘institutional accountability’[22].

Consider alternative routes for delivery, as different methods will exclude different people.

Need to consider digital poverty, rural participants and limited internet access, the two are interlinked.

Age may be a barrier to digital methods, and those from an older age group may be more reluctant and/or unable to engage in digital alternatives.

Trials need to test the same intervention and therefore researchers need to consider if the intervention changes when it is delivered by different routes.

Consent & communication

Use simple language (lay language and simple concepts) and videos.

Reluctance of researchers to challenge REC approved processes due to the fear of extending study timelines.

REC may not approve this because it may be difficult to include all the key information in a simple information sheet.

The word ‘trial’ can be difficult to explain to some people outside research, and people often associate it with a trial in law, which can be associated with punishment

The national reading age is around 11—12 years old.

If you explain to RECs that you are doing this to be more inclusive, and give the reasons, they are unlikely to say no.

It can be difficult getting recruiting staff to use the videos in recruitment settings, particularly, for example, in the emergency department.

The purpose of this is not necessarily to increase recruitment but to increase understanding, and support informed decision making.

Initially provide shorter information sheets (layered information).

May encounter issues with REC for not including key information, but it is fine to include these in addition to usual information sheets to aid the consent process.

It may be difficult to include all the key information in a shortened information sheet.

Sponsors and organisations need to have certain information given to participants for liability reasons, and therefore some text cannot be removed.

Employ staff from underserved group recruit.

Dependent on the availability of staff.

Considered valuable, though focus should be on people having interpersonal skills such as empathy, listening and communication.

Where there is not a diverse workforce, researchers could work with community ambassadors.

Recruitment is often out of our hands as we sit in an organisation, though we could try to advertise more widely.

We could work with community researchers or help to develop community researchers in the communities.

There is separate workforce work being done in the NHS and the NIHR.

Provide cultural (or similar) awareness training to staff.

Unless this is mandated, it is unlikely that the people who could most benefit from the training would do it.

An example of staff coming from different backgrounds and discussing their culture at a break time was given as a good way to share and learn from each other.

Focus should be on skills and competency, rather than just awareness.

Additional training in recruitment methods and processes for specific groups, such as people lacking capacity to consent, or people with learning disabilities may be required, as being aware of it is not enough.

Use translation and interpreters.

There was little experience of this in trials, with some reliance on NHS interpreters.

Also consider additional support for e.g., people with learning or communication difficulties.

Experience that funding is not in place for translation or interpretation and that this is even stricter for commercial studies.

Translation and interpretation is not standardised by the NIHR, and different CTUs use different providers.

One example of working with local authority interpreters and a suggestion to ask community groups what they do in regard to interpretation, as they may have experience.

Ensuring the translation is accurate can be difficult and costly, in some cases this can involve getting someone to ‘check’ a translation.

Examples of making videos in different languages to explain the trial, or clinical trials in general.

Interpretation is needed as well, a lot of people may not be able to read and therefore translation of documents is insufficient.

It is not clear how translations are actually being used in trials, or whether it is effective.

If translation and cultural adaptation is required for the intervention, researchers need to consider if the intervention is the same.

We can’t just translate a consent form or PIS, it has to be throughout the trial, including being able to report adverse events.

It is difficult to know which languages to include, particularly in trials where this cannot be identified in advance of inviting a patient.

There needs to genuine efforts made to provide translation and interpretation, so that it is not tokenistic.

Outcomes

Allow proxy completion where possible.

You want to hear the patient’s voice where possible, but proxy completion means people can still be included if they are unable to.

Self-completion of measures raises questions about the acceptable level of support needed and the boundary between self-completion and proxy assistance.

You don’t want people to complete outcomes for participants unnecessarily and should always try to include the patient voice.

There are only a limited number of proxy measures appropriate for certain populations because they haven't been developed.

Data collection should be around supporting the patient's voice where possible and where that's not possible then considering alternatives like proxy reported.

In some trials you might collect both knowing that some people may lose the ability to complete self-report in the trial and then you have some comparison.

Choose outcomes with simple language or translate them/ get interpreters.

Researchers tend to use outcomes that have been used in research before and are validated. These measures might not be validated in a different language.

Certain terms about health and illness do not always exist in different languages.

This must be planned for in advance, as it is difficult to implement during a trial.

Involve patients in discussions and choices around outcomes.

Focus on key outcomes with participants if burden is too high.

No barriers identified.

No barriers identified.

Field experts know some measures do not work and are being used as they were used in previous research. This can mean the secondary outcomes are more important to patients and so the outcomes to focus on should be discussed with PPI.

Follow-up

Arrange travel or upfront payment (not only reimburse).

Upfront payments are usually not possible, organisation payment systems are not fit for purpose.

Experience of not being reimbursed, reduces trust and people will not want to take part in research again.

Issues with payment through universities.

There are alternatives such as setting up payment cards.

Allow different methods of data collection.

Costs need to be included.

Staff can be stressed and under-resourced, and this could increase the workload for them.

Validated outcomes may have conditions around how they are presented.

There needs to be some consideration of the different methods being comparable.

Be flexible with times for clinic visits (out of hours).

Limited by the appointment systems and working hours of staff.

Research teams are stretched and often understaffed, so out of hours may not be an option.

Availability of staff out or hours can be an issue.

Consider home visits and more support.

No barriers identified.

People are already doing this, and no specific issues were reported.

Home visits take longer and are more costly so needs to be resourced appropriately.

Analysis

Undertake sub-group analysis (even if not powered).

Sub-group analysis is rarely powered and the relevant data is not routinely collected.

Suggestion to over-recruit under-served groups and do sub-group analysis on different categories.

Issues with subgroup analysis: be careful about singling underserved groups out—is it about the treatment effect, or acceptability of treatment offer/plan?

Consider if subgroup analysis is appropriate.

This has been requested by other patient groups: where this is been requested, trial teams should explain where this is appropriate and what information it can provide. i.e. that it is unlikely findings will be powered and should be exploratory.

Collect the relevant data.

Can be difficult to collect the correct data to measure socioeconomic status, there are several measures, and it is sensitive information.

Staff involved in recruitment reported that sometimes they do not have ethical approval to collect certain data.

It is difficult to collect screening data (for those not recruited), as you cannot collect personal data prior to consent.

There are several measures used for socioeconomic status: E.g. Deprivation index, postcode, social group via occupation, housing, income, education levels.

We try to collect the minimum amount of data needed– but we are using the data to assess generalisability.

Dissemination

Use videos and short communications.

Cost and resource at the end of a project might be limited.

Lack of identified staff to lead on this at the end of the study means it is not done.

CTUs do seem to be doing this now, usually through companies during, or shortly after, writing the results up.

Translation.

Cost and resource at the end of a project might be limited.

No standardised way of doing this across CTUs/ NIHR, which can mean staff don’t know where to start.

Additionally, there are often extra costs incurred that staff may not originally have considered e.g. having someone to proofread the original translation. This is not always offered in translation/ interpretation services.

There are some examples of translated dissemination videos.

Involve diverse PPI in dissemination plans.

Cost and resource at the end of a project might be limited.

No barriers identified.

No barriers identified.

Other

Sub-studies should ensure they include representation from people from underserved groups.

No barriers identified.

Not discussed.

Same issues with recruitment as for the whole trial.

It is difficult to know who to aim to include, particularly in smaller samples.

Process evaluations should consider how they look at differences between underserved groups, may not be powered but can help hypothesis generation.

No barriers identified.

Groups are not homogenous, there are complexities within ethnic minority groups (and other under-served groups).

May have similar issues with recruitment (if applicable) as for the whole trial.

Monitoring recruitment and retention by group throughout the trial.

Issues around data collection and data protection.

Restrictions on what kind of data you are collecting and analysing for the trial.

This should be done throughout a trial.

If this shows that we aren’t recruiting a representative sample, what do we do about it? Should we raise this to funders? It may slow or halt recruitment.

Having time to do this is a real issue.

Specific interviews or questions to try to understand reasons for withdrawal and if this differs by underserved group.

Needs to be asked in the right way.

Patients from minorities may be reluctant to discuss their health in detail, particularly if it is a taboo subject e.g., prostate cancer.

The need for better data reporting around dropout.

People have the right to withdraw without giving a reason and this needs to be respected.