How to improve participant compliance and retention in clinical trials：A Scoping Review


 Background: The reliability of results can be compromised when subsets of participants who remain on a study differ from those who drop out. Although recent studies have investigated strategies for improving compliance and retention , there has been very little investigation of process factors, regardless of the type of literature and the publication date. factors that may influence how and why compliance and retention may be successful (or not).Aim: To identify process factors that may influence participant compliance and retention in clinical trials, and potential strategies to improve the compliance and retention, from published studies, grey literature, and the reference lists of included articles to facilitate future design and implementation of trials.Methods : Six databases and two clinical trial registries were searched on November 29, 2019. Surveys, interviews, retrospective data analyses, theoretical research, reviews and clinical studies aiming at investigating factors and potential strategies were included. Data synthesis followed an iterative process to develop a list of process factors and potential retention strategies.Results were presented mainly as texts, including visualization using word cloud, and descriptive data.Results: 182 publications were included in this review, composed of 58 retrospective reflection of RCTs based on 1,132 clinical trials with 177804 participants, 27 implementation researches, 22 theoretical researches, 20 interviews, 18 surveys, 17 randomized controlled trials, 8 literature reviews, 5 systematic review and meta analysis, 3 mixed methods researches, 2 case report, 1 meta-ethnographic synthesis and 1 cohort study. We identified 70 process factors that may affect the compliance and retention . The most commonly addressed factors were age, education, economy, trust in clinical trials, supporting from surrounding people, safety concerns and effectiveness. We found 42 potential strategies to improve compliance and retention. Strategies reported most frequently were that researchers should pay attention to the changes on participants' psychological condition, try their best to build good relationships, provide some compensation and educate the participants about compliance and retention . Conclusion：These factors that may affect the compliance and retention can help researchers predict participants who are less likely to adhere and develop screening tools to find efficiently their suitable participants. These potential strategies that may improve the compliance and retention.

42 key themes of potential strategies were eventually included. According the scenarios for actual clinical research, the included potential strategies were divided into three phases: protocol design, recruitment and informed consent, trial implementation period ( Table 2).The potential strategies with highest frequencies were:(1)Researchers should pay attention to the changes on participants' psychological and building good relationships with participant(41.2%); (2)Money, gifts and other compensation can be provided(36.8%); (3)Researchers should carry out the education about compliance and retention during the study (including the knowledge of diseases, psychological and medication, trained in the broader context of medical ethics, cultural training)(35.7%); (4)Researchers should give full informed consent to the participant(safety concerns, effect, study protocol, the time they may take,etc)(29.6%);(5) Online reminder (phone, email, etc) should be provided (29.6%).

The analysis of process factors and improvement strategies based on diseases and population
According to the prognosis and course of the diseases, we divided the diseases into chronic non-fatal, chronic fetal, acute non-fetal, acute fetal ,and unclear.
The population was divided into adults, women, juveniles, seniors, pregnant women, special population, minority, homosexuality, healthcare providers, or Ebola front-line workers and "not reported". The division of the population is based on the original article. The process factors and improvement strategies were analyzed based on diseases and population(Shown in Figure 4-7). Data of different groups were not overlapped. We could nd that process factors and improvement strategies varied in different populations. At the same time, Appendix gures 1-10 showed the process factors and improvement strategies under the combination of different diseases and populations.
In pregnant women, the top two process factors were frequency of inspection(2), compensation(2); the top ve improvement strategies were money, gifts and other compensation(3), exible study time and location(3), recruiting full-time researchers (2), expanding the channels of recruitment(2), providing online reminder(2), carrying out education(2), recording 3-4 contact information of participants or their family(2), xed researchers and healers (2).
In some studies that did not report speci c populations, the top four process factors were: the relationship between researchers and participants (35), researchers' skills on communication (33), frequency of inspection (31), age (28), trust for the trial(28); the top ve improvement strategies were: paying attention to participants' psychological condition (34), full implementation of informed consent (30), money, gifts and other compensation (28), carrying out education (25), pre-compliance and retention assessment (23).
In acute non-fatal diseases, the top three process factors were gender(5), trust for the trial(4), age(4);the top ve improvement strategies were money, gifts and other compensation(3), paying attention to participants' psychological condition(3), handle adverse reactions timely(3), the drug package should be accompanied by detailed information(3), online data collection(3).
In some studies with unclear diseases, the top ve process factors were researchers' skills on communication (30), education (27), age (26), trust for the trial(26), support from surrounding people (26); the top ve improvement strategies were paying attention to participants' psychological condition (29), carrying out education (26), money, gifts and other compensation (24), full implementation of informed consent (22), establishing the table of follow-up (20).

Discussion
Summary of results 182 publications were included in this review. We reviewed 58 retrospective researches from RCT based on 1,132 clinical trials with 177804 participants, and identi ed 70 factors, which were grouped into individual, disease, environment, protocol and investigators aspects, in uencing participant compliance and retention; and 42 potential strategies, which could be adopted in three phases (protocol design, recruitment and informed consent, the trial implementation period) to improve compliance and retention. There were unique factors and strategies among different population and disease conditions. Most prevalent factors were participant age, gender, education, economic condition, researchers' communication skill, support from surrounding people and trust to the trial.
Most proposed strategies were paying attention to participants' phycological condition, giving gifts or compensation, recording 3-4 contacts, expanding recruitment channels, carrying out participant education, exible time and location and providing online reminder.

Strengths And Limitations
In past systematic map of digital tools for recruitment and retention we found that process factors were hardly studied [191] , even though these may be important for understanding how and why strategies for improving retention work (or do not work). As far as we know, this is the rst attempt to summarize the factors and the potential strategies of compliance and retention, regardless of the type of literature and the publication date. The results were from 177804 people ( including pregnant women, minor, the elderly, college students, addicts, etc), 62 diseases (including cancer, HIV, cancer, diabetes, respiratory diseases, cardiovascular diseases, hypertension, parkinson, alcohol addiction, cocaine dependence, etc ) and 22 countries (high income: the United States, Korea, Canada, Spain, etc; middle income: China, India, etc; lower income: Gambia)(Appendix table 1). The ndings are more universal. The impact of different process factors and improvement strategies may be different in different diseases or populations. However, most previous studies have ignored them.
Compared with previous literature [63,65] , we analyzed the process factors and improvement strategies according to the characteristics of different diseases and populations to show their respective special factors and strategies. Scoping review methods and word cloud were used in our study. We used systematic and extensive approach to identify and analyze literature.However,the diverse factors and potential strategies reported in the included studies make it di cult to draw an overall and exact conclusion. The included studies were mainly re ections of conducting RCTs. At present, there was still a lack of risk bias evaluation tools for these types of research. So various types of study design limit our assessment of the risk of bias for included articles. [191] Relationship with previous relevant works The types of previous relevant works included systematic review(5) and traditional literature review (8). Most studies are only conducted on speci c diseases(HIV, stroke, schizophrenia, smoke, Cataract, Amyotrophic lateral sclerosis, Arthritis, Alzheimer). The purpose of systematic review and meta analysis was to test the effect of potential strategies from some trials to improve the compliance and retention. The design of compliance and retention studies includes observational studies, cohort studies, and randomized controlled trials, etc. Variations in research design and lack of overlapping outcome variables often lead to failure to complete systematic reviews. [192] But the in uencing factors of compliance and retention and theoretical perspectives were not provided in depth. The traditional literature review described the factors and potential strategies from other researches. But it is possible to miss some articles because the lack of systematic search. Compared with previous studies, the factors and potential strategies provided in our study are more comprehensive. We provide factors and methods for different diseases, populations and interventions. Their conclusions were summarized and analyzed in our study by scoping review. By scoping review, we systematically searched, selected, and integrated existing information to draw a picture of the status and association of compliance and retention studies. In addition, their conclusions were classi ed and displayed from these studies according to the different stages and characteristics of clinical trials.
The factors identi ed in our study can help researchers predict participants who are less likely to adhere, and develop screening tools to nd e ciently their suitable participants. The better researchers understand these factors, the better it will be to improve participant compliance and retention. The potential strategies for improving compliance and retention we found were generally targeting at the factors we identi ed. However, the factors were not su ciently covered by the potential strategies we found. There is a need to develop more comprehensive potential strategies to address the factors we already know.
In addition, the disagreement between studies demonstrated that these factors may play different roles in different studies, sometimes positive, sometimes negative, and sometimes no effect. For example, age was a commonly reported factor in uencing compliance and retention in the clinical trials, but it was reported that young people may increase or decrease participant compliance and retention. Therefore, most of the identi ed factors have been presented in a neutral form in our study. Researchers should well-thought-out according to their speci c trial context.
The potential strategies mostly mentioned were that researchers should pay attention to the changes on participants' psychological condition, and try their best to build good relationships. Providing compensation (such as gifts, money, postcard) and educating the participants about compliance and retention during the study (including the knowledge of diseases, psychology and medication), full implementation of informed consent and reminding participants online (phone, email, etc) were also mentioned by many. Potential strategies still need to be comprehensively developed. Furthermore, adopting potential strategies are often resource demanding. Researchers need to realize this and try to prepare it when planning the trial. Cost-bene t analysis may be required.
In different population, women, juveniles, seniors, and people with physical or mental disorders may need more support from external conditions.In addition to objective conditions such as age and economics, they also need the support of people around them and researchers.Their psychology may be more sensitive and need to be paid more attention. They may also need more exible research time and location in order to better complete the research. Culture, language and the efforts of researchers play an important role in in uencing the compliance and retention of minorities. In different diseases, the severity of the diseases had an important effect in the compliance and retention of patient in acute fatal diseases. Patients who are overly ill may not tolerate the treatment so that they may withdraw or become less adherent. The trust for the trials was very important for acute non-fatal diseases. In chronic non-fatal and fetal diseases,the prolonged course of the diseases may cause them to need more support from people around them and researchers. In chronic fatal diseases, full informed consent is essential.
Qualitative methods could be used to make an in-depth exploration to the factors and potential strategies in the future research. More researches are also needed to nd effective potential strategies to encourage the participants to return the sites for follow-up. Implementation research (including, but not limited to, randomized controlled trials) can be used to verify the effectiveness of these included potential strategies by evaluating speci c outcomes, such as compliance and retention rate, compliance and retention improvement rate, retention rate, etc.

Conclusions
We identi ed 70 factors, which were grouped into individual, disease, environment, protocol and investigators aspects, in uencing participant compliance and retention; and 42 potential strategies, which could be adopted in three phases (protocol design, recruitment and informed consent, the trial implementation period) to improve compliance and retention. The process factors and improvement strategies of different diseases and populations are speci c. Women, adolescents, seniors, and patients with mental or physical disorders may need more support from relatives, friends and researchers than adults. The long process of treatment also requires them to give more support to patients with chronic diseases. These factors can help researchers predict participants who are less likely to adhere and develop screening tools to nd e ciently their suitable participants. Before using these potential strategies, researchers should well-thought-out according to budget, the range of diseases, population, the design of their research and their usual follow-up procedures.

Memory
People who have good memory may have higher compliance and retention [32,62,66,78,92,107,115,119,144,189] 10/182 The level of pressure T Physique The better the physique, the higher the compliance and retention [16,36,37,48,82,107,131,177,189] 9/182 Desire for pregnancy

If p th
Legal Patients with legal concepts may have higher compliance and retention [12,66,80,90,117,141,171] 7/182 Mobility P Social position People who have high social position may have higher compliance and retention [12,49,120,123,157,158] . 6/182 Use of illicit drugs P r

Different study sites
Due to factors such as the treatment environment or the investigator, the compliance and retention of patients in different research sites also varies [14,20,38,40,56,101,115,118,121,144,161,166,169,177,182] 15/182 Feeling to be cured P w The way of inspection Patients who are dissatisfied with the way of inspection have worse compliance and retention [29,46,79,80,92,93,125,134] 8/182 The level of study H tr

Figure 4
The process factors in different populations Figure 5 The process factors in different diseases The improvement strategies in different populations The improvement strategies in different diseases

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.