Introduction

Health research, and clinical trials which aim to influence and inform treatment decisions should recruit populations that match, as closely as possible, the populations who would be eligible for these same treatments if offered in routine care [1,2]. This should explicitly include groups considered ‘underserved’ due to underrepresentation in research, but who often experience a higher disease burden [2,3]. Despite recent efforts to improve recruitment and retention rates for research overall, groups traditionally underserved by health research, such as ethnic minority populations, low socioeconomic status groups and individuals with disabilities are still underrepresented in studies impacting external validity and increasing research waste [4–6]. The reasons driving underrepresentation of underserved groups in health research overall, include language barriers [7], financial issues [8], medical mistrust [9], experiences of racism and discrimination [10] and lack of awareness [11].

Previous research has reported that sociodemographic characteristics such as, age, gender, and ethnicity are predictors of whether an individual participates in a health research study [12,13]. A meta-analysis of 29 UK-based COVID-19 trials reported that 84.8% of participants were White, with Asian, Black and Mixed ethnic groups underrepresented, despite ethnic minority groups having an increased risk of severe disease and premature mortality from SARS-CoV-2 [14]. Additionally, all 30 UK-based COVID-19 vaccine trials (n = 118,912) underrepresented ethnic minority groups when compared to national population ethnicity data; differences were most marked in Black (1% 95% CI [0.6–1.5%] vs 3.3%) and Asian (5.8% 95% CI[4.4–7.6%] vs 7.5%) groups [14]. Despite the increased prevalence and severity of COVID-19, vaccine uptake in ethnic minority groups was also found to be lower than White ethnicities in the UK [15,16].

Aside from ethnicity, other sociodemographic factors may influence participants willingness to participate in health research. A study in a Danish population of employees (n = 50,806) found that women were more likely to participate than men, and participation increased with higher annual income and educational level [17]. Individuals with higher education levels may have higher health literacy, a greater understanding of what research entails, and a greater interest in research overall [18]. Lower educational levels are associated with difficulty in understanding the research process, including the informed consent processes [19,20].

Other factors such as having multiple long-term health conditions can also affect individuals’ willingness to participate in research. In an online survey study of 2150 participants based in the US, a greater proportion of individuals with two chronic diseases were more willing to participate in clinical research studies (53%) compared to those individuals with no chronic diseases (41%) [21]. This could be due to increased engagement with healthcare services, having a greater interest in improving their own health, and wanting to participate in research that could impact their quality of life [22].

Aside from the initial challenges in recruiting participants, there are also challenges in retaining participants. Poor retention rates of participants can potentially bias outcomes, negatively impacting the generalisability and validity of a study [23]. In the UK, trial methodologists identified clinical trial retention as one of the top three priority areas [24]. A review consisting of 6 European large, multicentre randomised controlled trials (n = 7612) concluded that barriers to retaining older participants in clinical trials included, high rates of comorbidities with high symptom burden, difficulties in accessing facilities, number of study visits and long durations for trial participation [25].

A meta-ethnographic synthesis found that participant’s decisions to discontinue trial participation included, being ‘too well’ to continue engagement with the trial, non-acceptance of a diagnosis amongst those who were newly diagnosed, or being too ill to engage fully with research processes [26].

While there is an existing body of evidence detailing factors associated with willingness to participate in research, less is known on willingness to consent to recontact studies (where existing research participants already consenting to a study are recruited to further data collection or future studies). Understanding the association between sociodemographic characteristics and willingness to re-participate in health research studies is essential for developing and sustaining of targeted, inclusive recruitment strategies.

Therefore, this study aims to address the following research question: What are the associations between sociodemographic characteristics (age, gender, ethnicity, migration status, education, employment) and number of health conditions, and participants willingness to consent to recontact studies. Furthermore, a secondary aim is to determine how sociodemographic characteristics and number of health conditions may influence participants’ consent for NHS data linkage and data sharing.

Methods

Results

A total of 6147 participants were included in the analysis (mean age 51.9 years, ± 17.12). Of these participants, 62.3% were female and 60.8% of the sample identified as White. 28.1% of participants reported a diagnosis of two or more long term health conditions. Table 1 provides an overview of participant characteristics included in the analysis.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Sociodemographic characteristics of participants included in the analysis.

https://doi.org/10.1371/journal.pone.0335986.t001

Re-contact and NHS data linkage

Overall, 54% of participants (n = 3291) consented to be re-contacted regarding opportunities to participate in further health research. Furthermore, 56% (n = 3454) of participants consented to having their data linked to their medical records. Results showed that older age was associated with a higher likelihood of consenting to be re-contacted for future research studies (PR = 1.08, 95% CI [1.05–1.11], p < 0.001) and data linkage to NHS records (PR = 1.07, 95% CI [1.04–1.10], p < 0.001). Participants with higher education levels (A levels or above compared to GCSEs) showed a 10% increased likelihood of consenting to re-contact and data linkage (PR = 1.10, 95% CI [0.98–1.22], p = 0.12; see Table 2 for re-contact and Table 3 for data linkage).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Participant’s willingness to be recontacted about participating in future health research: results of the modified Poisson regression analyses.

https://doi.org/10.1371/journal.pone.0335986.t002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Participant’s willingness to consent to data linkage to medical records: results of the modified Poisson regression analyses.

https://doi.org/10.1371/journal.pone.0335986.t003

The likelihood of both re-contact and data linkage was significantly lower among participants of South Asian ethnicity compared to those of White ethnicity (Re-contact: PR = 0.79, 95% CI [0.71–0.88], p < 0.001; Data linkage: PR = 0.72, 95% CI [0.64–0.82], p < 0.001). Additionally, individuals with multiple long-term conditions compared to those with one, were associated with higher likelihood of consenting to re-contact (PR = 1.14, 95% CI [1.04–1.23], p = 0.002) and data linkage (PR = 1.13, 95% CI [1.04–1.22], p = 0.003).

There was no significant association between whether participants were born in the UK or not and data linkage (PR = 1.06, 95% CI [0.94–1.20], p = 0.316; see Table 2 for re-contact and Table 3 for data linkage).

Data sharing

74% of participants (n = 4530) consented to their data being shared with NHS, universities and other organisations.

There was no significant association found between data sharing to NHS, universities and other organisations and age (PR = 1.02, 95% CI [1.00–1.05], p = 0.100. The likelihood of willingness to consent to sharing data was lower in females than males (PR = 1.00, 95% CI [0.93–1.06], p = 0.96) and South Asian participants compared to white participants (PR = 0.85, 95% CI [0.77–0.94], p = 0.002). Participants with higher education levels (A levels or above compared to GCSEs) showed a 13% increased likelihood of consenting to data sharing (PR = 1.13, 95% CI [1.02–1.25], p = 0.015). An increased number of long-term conditions compared to just one was associated with higher likelihood of consenting to data sharing, albeit not statistically significant (PR = 1.01, 95% CI [0.99–1.04], p = 0.257). Individuals born outside of the UK were slightly less willing to consent to data sharing compared to their UK-born counterparts (PR = 0.93, 95% CI [0.84–1.03], p = 0.150; see Table 4 for data sharing).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Participant’s willingness to consent to data sharing to universities. NHS and other organisations: results of the modified Poisson regression analyses.

https://doi.org/10.1371/journal.pone.0335986.t004

Discussion

Our analysis showed that, of people taking part in a large cross-sectional questionnaire study, those from older age groups, those with a higher education level and those with more than one long-term health condition were more likely to give consent to be recontacted for future health research, and to grant access for linkage of their data to health records, and for their data to be shared with other universities and research organisations.

Previous research from the US reporting on willingness to be recontacted to participate in research found an association between self-identified White ethnicity and higher educational attainment with increased likelihood to consent [30]. Black/African American participants expressed a lower willingness to participate in research (56%) than White participants (70%). Furthermore, participants who were educated ‘up to some high school’ had lowest levels of willingness (51%) compared to those who had doctoral training (76%) [30]. These findings from a US population mirror results from this secondary data analysis. Participants with a higher level of education and higher level of health literacy had a higher level of interest in participating in health research [12].

The ATHENA COVID-19 study in Queensland, Australia aimed to recruit a cohort of patients with linked health information who were willing to be recontacted to participate in future research studies, including a long COVID study [28]. Among these 1155 patients whose contact details were available, 85% reached a consent decision. 73% of patient agreed to recontact and 69% agreed to data extraction, whereby their data would be extracted from their health practice and linked into other datasets for ethically approved studies. Altogether, 335 general practices took part and primary care data was successfully extracted with participants consent, highlighting the effectiveness of consent to recontact for the facilitation of recruitment to future studies [28].

Furthermore, participation in health research, including clinical trials often requires the ability to take on and withhold complex information relating to health conditions, new treatments, and the research process itself [7,29]. As a result, populations face greater barriers to participation, are often characterised by lower education levels [30,31], lower levels of written and spoken English [32,33], lower levels of health literacy [18] time constraints and caring responsibilities [34]. It is important to emphasise that researchers and institutions are responsible for addressing barriers relating to underrepresentation of underserved groups. Researchers must take active approaches to engage underserved groups through recommended strategies such as, tailored communication, translated materials and flexible data collection methods [29].

Regarding long-term health conditions, this study indicated that in all three regression models, individuals with a higher number of MLTCs were more willing to re-participate in research, have their data linked and data shared. This finding aligns with prior research focussing on rates of recruitment and retention [20]. It is important to note that our findings are directly applicable to individuals with MLTCs who have previously taken part in health research. Participants with an increased number of health conditions may be more willing to consistently participate in research due to perceiving participation as an important opportunity to contribute to advancements which improve their health outcomes, quality of life and access to new treatments [31].

Our analysis provides an insight into the characteristics of participants in health research and is beneficial for researchers to produce targeted interventions for specific groups. A major strength of using the MELS dataset was that it included a substantial proportion of those from ethnic minority groups (29.1%). In comparison to the UK census data (2021), approximately 25.5% of the UK population belongs to an ethnic minority group [33]. In relation to regional ethnic diversity, in the East Midlands, 20.5% of the population identified as belonging to an ethnic minority group [34]. The MELS dataset overrepresented ethnic minority groups, who are typically underserved in research studies. Also, this study examined a number of different variables (age, gender, socioeconomic status, multiple long-term conditions) in relation to participants willingness to consent to taking part in future studies, data linkage and data sharing.

Despite a large sample size, categories such as education and employment were collapsed to increase the statistical power, which may have obscured important differences in the analysis process. In the MELS dataset, there was missing data relating to participant’s age, ethnicity and migration status which may have impacted the validity of results. Moreover, there was limited data regarding participants postcodes so Index of Multiple Deprivation could not be included in the analysis. Given sufficient data, this would have provided further information on socioeconomic status overall, alongside education and employment and association with re-contact, data linkage and data sharing.

Despite efforts to ensure accessibility for underserved groups, it is certain that selection bias will have occurred in the recruitment the MELS study. Rather than regarding this as a weakness in the methods used in this analysis, our results will have more relevance to those planning and conducting recontact studies in the future, recruiting from existing cohorts, which will likely have similar selection bias. The population included in our analysis is one which has already given consent to participate in an observational questionnaire study, which matches the population to which our results will apply, giving a more accurate and reliable outcome estimate for consent and data linkage rates.The findings from our analysis provide valuable insights and may help inform the development of effective recruitment and retention strategies, specifically for the conduct of large-scale recontact studies associated with UK Biobank [35] and ‘Our Future Health,’ [36] which aim to engage diverse populations for the purpose of longitudinal research. Building on the results of our analysis, further work is necessary to explore specific factors such as trust, experiences with healthcare, perceived benefits, which may influence a participant’s decision to participate in multiple research studies. It is important to acknowledge that results from this study relate to health research participation more generally. We did not have data available to investigate the effects of study type on willingness to participate in future research. It is possible that rates of re-contact may differ according to study type and demand on participants (time, effort, inconvenience).

Having an understanding of these factors could provide valuable suggestions for researchers when designing health research studies, including clinical trials. This would contribute towards effective recruitment and retention of participants from a range of diverse groups in research. Moreover, exploring the barriers and enablers which influence a participant’s decision to re-participate and follow a trial through to the endpoint would be advantageous for the development of inclusive recruitment and retention strategies.

In conclusion, our analysis shows that differential rates in recruitment and retention in research are also evident in willingness to participate in further research, willingness for linkage of research data with medical records, and data sharing with universities and research organisations. Given the growing use of large biomedical databases for identification of eligible participants for clinical trials, there is a need to embed recruitment and retention strategies which address differential rates in participation, retention and willingness to consent to maximise the societal and health benefits of diverse research participation.

Supporting information