Improving assent in health research: a rapid systematic review.

Background Enrolment in a research study requires the participant’s informed consent. In the case of minors, informed consent of the respective legal guardian is obtained in conjunction with informed assent of the underage p articipant. Since comprehension of the information provided may be limited, effective interventions to improve understanding should be identified. Thus, it is the objective of this study to review quantitative studies that tested interventions to improve the understanding of information provided during assent processes in health research. The studied population consisted of minors that participated or were willing to participate in research. The primary outcome was the level of comprehension after intervention. Methods A systematic search was conducted in eleven databases including regional databases: PubMed, Web of Science, ERIC, PsycINFO, CINAHL, POPLINE, AIM, LILACS, WPRIM, IMSEAR, and IMEMR and included references from inception of the database until July 2018 except PubMed which spanned the period from May 2013 to July 2018. Search terms focused on Informed Consent/Assent, Minors, and Comprehension. To complement the search, reference lists of retrieved publications were additionally searched. We included all quantitative studies that were conducted in minors, tested an intervention, covered assent processes in health research, and assessed comprehension. One reviewer screened titles, abstracts, and full-texts to determine eligibility and collected data on study design, population, intervention, methods, outcome, and for critical appraisal. Interventions comprised enhanced paper forms, interspersed questions, multimedia format, and others. Results Out of 7089 studies initially identified, 19 studies comprising 2805 participants and conducted in seven countries were included in the review. Fourteen studies (74 %) tested an intervention against control and ten (53 %) were randomized controlled trials. Heterogeneous methodology as well as incomplete outcome and statistical reporting impaired the reliability of the collected data. Positive effects were suggested for use of enhanced paper forms, interspersed questions, use of pie charts, and organizational factors. Conclusions Improving assent in health research is an under-researched area with little reliable evidence. While some interventions are proposed to improve understanding in assent processes, further investigation is necessary to be able to give evidence-based recommendations. Trial registration PROSPERO ID: 106808.


Background
Out of 26 917 clinical trials that have been registered on ClinicalTrials.gov in 2019, more than 4 700 included individuals under the age of 18 years [1]. Similar to adults, adolescents and children have a right and interest to participate in health research to ultimately benefit from its outcome. Since minors may have difficulties balancing risks and benefits, they are considered a vulnerable population. Thus, special requirements have to be met to include them in research.
Based on the principle of respect for persons, involvement in a study requires the participant's informed consent [2]. In the case of minors, obtaining informed consent of the respective legal guardian in conjunction with informed assent of the underage participant is required. Usually, the minor's decision prevails. The Council for International Organizations of Medical Sciences (CIOMS) provides guidelines regarding the consent/ assent process in matters of content and comprehension. Among other things, the information should cover the study's aims, procedures, anticipated benefits and potential risks as well as the voluntariness of participation and the right to withdraw at any times [2]. It is also essential, however, to ensure that the potential participant sufficiently understands the information provided.
One report demonstrated that the comprehension of study details by minors often is unsatisfactory with about 50 % not remembering that their treatment was considered research a few months after enrolment in the studies [3]. Although there are publications that provide guidance in assent processes, the recommendations often lack evidence [4]. However, different and novel ways of communicating information to minors have lately been under investigation.
This review summarizes published data from quantitative studies examining assent processes to identify interventions that promote the highest level of understanding among minors in health research of the information provided. Thus, it is the aim of this paper to provide guidance to future researchers on how to develop more effective assent processes.

Methods
This rapid systematic review was registered in PROS-PERO 2018 (ID: 106808) and follows the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement (Fig. S1).

Eligibility criteria
We included studies that evaluated factors and interventions in assent processes for minors in health research in regard of their impact on understanding. However, studies were excluded if 1) the study did not include children or adolescents (< 18 years); 2) the tested assent form was not for health research; 3) no intervention was tested; 4) the produced data was purely qualitative or narrative or 5) comprehension was not tested. There were no restrictions in respect of language.
The search strategies were designed with input from an expert librarian to cover publications about Informed Consent/Assent, Minors, and Comprehension (Table  S1). The searches were conducted between 26/07/2018 and 03/08/2018. Since a prior review covered earlier publications, the search in PubMed was restricted to papers published since May 2013 [5]. For other databases, the search was not limited in time. Additional records were identified by perusing references of retrieved publications.

Assessment of studies
One reviewer screened all titles and/or abstracts and assessed full-texts to determine eligibility. If no full-text was available, authors were contacted to gain access. In case of queries about the potential eligibility of a study, these were discussed with at least one of the other reviewers and a joint solution was found.

Data collection, synthesis and critical appraisal
All included studies were read and data were extracted by one reviewer. For data collection and assessment of risk of bias, a form was created based on the Data Extraction and Assessment Template by The Cochrane Public Health Group [6]. The obtained information included data on study design, population, intervention, methods, outcome and critical appraisal, among other things. As preferred outcome, mean of overall correct answers in a post-intervention comprehension test and respective statistical appraisal were obtained; if possible, absolute values where transformed to relative values. Furthermore, studies were clustered according to the intervention tested: enhanced paper forms, interspersed questions, multimedia format, and others. Levels of evidence are as follows: (1) randomized controlled trial (RCT); (2) controlled trial without randomization or prospective comparative cohort trial; (3) case-control study or retrospective cohort study; (4) case series with or without intervention or cross-sectional study or study without control [7]. In case of queries about the data collection and appraisal process, these were discussed with at least one of the other reviewers and a joint solution was found.

Results
The primary search produced 7063 reports which were complemented by 26 reports obtained from other sources. After removal of duplicates and screening of titles and/or abstracts, 225 potentially relevant publications were identified and the full-texts were screened for eligibility. Application of exclusion criteria resulted in 19 studies with 2805 participants that were included in the analysis (Fig. 1).

Study characteristics
The nineteen included studies were conducted in seven different countries (Fig. 2) and comprised ten (53 %) RCTs of which six (32 %) failed to clearly state the method employed for group allocation. Assessors were stated to be blinded only in two studies (11 %). Overall, fourteen studies (74 %) compared an intervention to a control and thirteen (68 %) used a standard assent process/form as control. A large variety of methods were used to assess understanding ranging from written questionnaires with multiple choice or open-ended questions in most studies to interviews and observations. Two studies (11 %) included less than five participants. Assent processes in real research settings were covered by seven studies (37 %) while the others (63 %) used hypothetical or simulated study protocols. In total, six studies (32 %) reported their outcomes incompletely. (Table 1)

Interventions
Nine studies investigated the effects of using enhanced paper forms during the assent process. Enhanced forms included those with simplified text, illustrations, supplemental information, and narrative approaches. Six studies tested an enhanced assent form against the respective standard form as a control. Five of the six studies stated to use randomization for group allocation. Three of these studies found that the enhanced form resulted in significantly better understanding than the standard form, while one study found the opposite being the case [9,17,21,23]. However, the one study describing the standard form to be more effective was the only one where the intervention and the standard form did not cover the same content, but were used for two different clinical trials [9]. A non-randomized study found the enhanced form to significantly improve understanding in adolescent patients suffering substance use disorder but to have no effect in the control group of healthy adolescents [14]. Another randomized study tested block text format against questions and answers (Q&A) as well as story format, but failed to provide statistical calculations for the means of correct answers. Instead, they only stated that the highest portion of participants that answered all questions correctly was in the story group [11]. Three additional studies found that using everyday language with graphs and Q&A format, an illustrated booklet, and a comic strip in the assent process generally led to good understanding of the research details, but the interventions were not tested against a standard or control format [16,18,20].
Three studies investigated the effects of using questions that assessed comprehension being interspersed during the assent process. All three studies compared the standard form with the same form plus probing questions during the process. Two studies used randomization for group allocation. Two studies found the understanding to be improved with interspersed questions during the assent process. However, one study considered the effect not to be significant with p = 0.055 [12], while the other study used the same questions for probing during the process and assessment of understanding after the process [15]. The third study described better understanding of study purpose and benefits with probing questions but failed to provide the respective data and statistics in the report [8]. Five studies investigated the effects of using multimedia formats during the assent process. Three studies were RCTs that tested a multimedia format against the standard form. One study found the multimedia approach to significantly increase comprehension levels compared to the standard form, while another study found no significant difference [12,22]. Another RCT tested a multimedia format with interactive, interspersed exercises against the standard form and found the multimedia format to be significantly better in improving understanding in the adolescent participants [25]. Two additional studies tested a non-fiction narrative on a touch computer and a dialogue by avatars in a low number of participants without a control condition [19,24]. Interestingly, one study included two three-year-old children who could already be successfully introduced to certain aspects of research [19].
Three more studies investigated the effects of other interventions during the assent process. One RCT showed that adolescents above the age of 15 years demonstrated higher levels of comprehension when the assent process for them and their parents was conducted separately, while younger children showed no differences [10]. In a cross-sectional study, several different methods to explain probabilities were tested against each other, demonstrating that illustrations as pie charts were easiest to understand for children, followed by verbal labels, percentages, proportions as words, and proportions as notation [26]. An observational study demonstrated in interviews that understanding of study details was better if the recruitment and assent process took place more than seven days after the respective diagnosis. However, participants from several different clinical trials were included in the study and not controlled for their allocation which might have influenced the observed effect [13].

Discussion
This review includes nineteen studies of which twelve have not yet been covered by Grootens-Wiegers et al. in a former systematic review [5]. We made an effort to cover a broad spectrum by inclusion of many regional databases and the literature search was designed to particularly imbed literature from many different cultural backgrounds. Unfortunately, only one study from a lowand middle-income country met the inclusion criteria for this review, while all other included studies come from Organization for Economic Co-operation and Development (OECD) countries (Fig. 2).
In general, reliable data published on this subject was scarce. At study level, we identified several factors that limit the power of the presented results. These include the low number of RCTs and the failure to sufficiently report the group allocation processes. The nature of the assent process impeded blinding of group allocation, whereas the possible blinding of assessors was undertaken only in two trials. In five studies, interventions were not tested against control, at all. Additionally, many of the comprised studies featured incomplete outcome reporting. This included especially the failure to provide mean and standard deviation values as well as the lack of statistical analysis.
The format of a rapid systematic review was chosen to provide high-level evidence for health researchers that    work with minors in a timely manner. This decision comes with limitations at review-level: only one author conducted the principal literature search and data collection. This may have resulted in incomplete retrieval of identified research. However, we made an additional effort to improve the quality of this review by inclusion of at least one other reviewer in any case of doubts during the process of literature screening, data collection, and critical appraisal. Unfortunately, the limitations at study-level impeded additional (meta-)analyses of the presented interventions.
To differentiate individual opinions and views from verifiable results, we decided to exclude qualitative studies from this review. Nevertheless, new ways to communicate information in research have been tested in qualitative studies, as well. Dockett et al. report how one child emphasized the importance of illustrations in information forms: 'I just read the pictures.' [27] Another report described the process of involving children in the development of information and assent forms. The children exclusively used active voice and named all function owners [28]. However, their effects on comprehension still needs to be assessed.

Conclusions
This report on a rapid systematic review includes nineteen studies that investigated factors in research assent processes in order to improve comprehension in underage participants. Unfortunately, available data on this topic proved to be rare and several major limitations restrict the power of the findings, so that we did not attain our initial goal to be able to provide researchers explicit evidence-based recommendations.
Nevertheless, positive impact on children's and adolescents' comprehension of research information was suggested for enhanced paper forms (e.g. by simplified text or illustrations), for the use of interspersed questions, for assent processes that are conducted separately from parents for adolescents older than 15 years, for the use of pie charts to communicate probabilities, and if trial recruitment took place more than seven days after diagnosis. The positive effect of simplified language, illustrations, and narrative approaches in enhanced paper forms may not be surprising given that presentations using various visualizations are generally supposed to be easier to understand and to follow [29]. And just like repetition is a widely accepted tool to study and understand any topic, the shown benefit from interspersed questions that require participants to double-check their own comprehension is quite intuitive. Like the use of pie charts, whenever probabilities are meant to be conveyed, both techniques are easily included in any kind of information sheets.
Younger children might profit from elements that show no impact in older ones and vice versa. Adolescents might feel a greater responsibility for their decisions. In line with that, Hein et al. claimed that children from the age of twelve may already be capable of giving consent instead of assent [30]. Therefore, future research on this topic should consider testing different interventions in different age groups. On the whole, the area of assent remains a largely under-researched issue. Further research and standardization of measures still remain necessary to be able to give stronger evidence-based recommendations.