A systematic review of risk communication in clinical trials: How does it influence decisions to participate and what are the best methods to improve understanding in a trial context?

BACKGROUND: Effective risk communication is challenging. Ensuring potential trial participants' understand 'risk' information presented to them is a key aspect of the informed consent process within clinical trials, yet minimal research has looked specifically at how to communicate probabilities to support decisions about trial participation. This study reports a systematic review of the literature focusing on presentation of probabilistic information or understanding of risk by potential trial participants.
METHODS: A search strategy for risk communication in clinical trials was designed and informed by systematic reviews of risk communication in treatment and screening contexts and supplemented with trial participation terms. Extracted data included study characteristics and the main interventions/findings of each study. Explanatory studies that investigated the methods for presenting probabilistic information within participant information leaflets for a clinical trial were included, as were interventions that focused on optimising understanding of probabilistic information within the context of a clinical trial.
RESULTS: The search strategy identified a total of 4931 studies. Nineteen papers were selected for full text screening, and seven studies included. All reported results from risk communication studies that aimed to support potential trial participants' decision making set within hypothetical trials. Five of these were randomised comparisons of risk communication interventions, and two were prospectively designed, non-randomised studies. Study interventions focused on probability presentation, risk framing and risk interpretation with a wide variety of interventions being evaluated and considerable heterogeneity in terms of outcomes assessed. Studies show conflicting findings when it comes to how best to present information, although numerical, particularly frequency formats and some visual aids appear to have promise.
CONCLUSIONS: The evidence base surrounding risk communication in clinical trials indicates that there is as yet no clear optimal method for improving participant understanding, or clear consensus on how it affects their willingness to participate. Further research into risk communication within trials is needed to help illuminate the mechanisms underlying risk perception and understanding and provide appropriate ways to present and communicate risk in a trial context so as to further promote informed choices about participation. A key focus for future research should be to investigate the potential for learning in the evidence on risk communication from treatment and screening decisions when applied to decisions about trial participation.

Introduction

Clinical trials are now widely accepted as the gold standard of evidence-based medicine for determining treatment effects [1]. The importance of recruiting adequately informed individuals to participate in clinical trials is paramount. However many studies have demonstrated that participants approached to take part and those consented to participate in trials have a limited understanding of key aspects of the trial [2]. One of the key areas to consider when presenting information to potential participants is the information on potential risks and benefits. The ethical principles for medical research involving human subjects, enshrined within the Declaration of Helsinki, state that ‘each potential subject must be adequately informed of the anticipated benefits and potential risks of the study’ [3]. This is echoed by guidelines for good clinical practice, which state that all the information provided to participants should include explanations of the ‘reasonably foreseeable risks or inconveniences’, expected benefits, and where there are no clinical benefits to the participants, they must be made aware of this’ [4]. However, mechanisms to operationalise the provision of such information are not provided in the guidance.

Risk communication can be defined as communication with individuals that addresses knowledge, perceptions, attitudes and behaviour related to risk, and risk itself can be defined as the probability that a hazard will give rise to harm [5, 6]. A correct understanding of risk therefore depends upon an accurate understanding of probabilities, a feat that is determined by several influencing factors, such as individual numeracy levels and cognitive abilities, but not least by the methods used to present probabalistic information [7]. There is a substantial amount of literature that focuses on risk communication with regard to public health messages, health behaviour, and treatment and screening decisions for patients [8-11]. Speigelhalter et al have shown that probabilities are ‘notoriously difficult to communicate effectively’ to lay audiences in various contexts, including health [12]. Yet minimal research has looked specifically at how to communicate probabilities within information provided to support decisions about trial participation (or not). In a trial context uncertainties relating to interventions will usually be greater purely by the nature of the trial endeavour—to generate evidence about benefit and harm.

Understanding, or more often mis-understanding, of risk information related to trials has been shown to influence decisions about participation in a range of trials, with those prepared to accept risk more likely to participate [13, 14]. Decisions about trial participation are inherently different from decisions about treatment. For example, one of the main influences on clinical trial participation is conditional altruism [13]. Conditional altruism is the concept that participation in the trial will benefit society but there must be a benefit (which is influenced by perception of risk) for self. Conditional altruism does not exist for decisions about treatment and as such it is important to understand how potential trial participants understand risk in a trial participation context. Additionally, trade-offs between risk and benefit in a trial involve layers of complexity in addition to those for treatment such as: loss of control over which treatment they receive; and potentially greater uncertainties, as often participants have to consider the risks and benefits of a minimum of two competing treatments. Existing studies in the domain of informed consent for clinical trials have repeatedly highlighted significant discrepancies between actual risk and participant interpretation of risk to themselves, or their child, in taking part in a trial [15, 16]. Participants frequently underestimate risks, leading them to believe that there would be little to no risk involved in trial participation. This pronounced lack of understanding strongly suggests the need for better communication about trial aims and design, particularly when it comes to the inherent risks, however small, that are almost always present in taking part in a clinical trial [15]. The intrinsic nature of trials means there is much unknown information and communicating probabilistic information in this context is more challenging as the layers of risk are greater, for example the risk of undertaking a trial as opposed to treatment, the outcome risks, and the risk of randomisation to a drug, procedure or placebo [17].

Preliminary findings from our group have shown that stakeholders have varied preferences about how probabilistic information relevant to trial participation (e.g. estimates of the likelihood of benefit and/or harm associated with trial interventions) is communicated [18]. In addition, a pilot study exploring decision support for trial participation decisions highlighted that patients’ preferences for risk information differed in a trial context compared to a treatment context [19]. Existing research on methods to present probabilistic information to improve patient understanding and decision making about treatment and screening decisions could provide valuable insights for enabling effective risk communication in the context of informed consent for trials [20]. Yet, surprisingly, the methods shown to be effective to improve understanding of probabilistic information are not routinely employed in participant information leaflets for trial participation [17].

A small number of studies have evaluated methods for presenting ‘risk’ in patient information leaflets for clinical trials. However, these studies have not been analysed together to allow judgements about optimal methods of presentation. This warrants further investigation both at the level of understanding and on the decision to participate (or not) in the trial. To address this, this study aimed to systematically review the literature focusing on presentation of probabilistic information within the informed consent process for trials. We focused our search on comparative effectiveness studies that tested interventions which varied the presentation of probabilistic information and the effects on potential trial participants’ understanding and/or the decision to participate.

Methods

Inclusion criteria

Evaluative studies using qualitative methods that investigated the methods for presenting probabilistic information to potential trial participants during the informed consent process for a trial were considered eligible. Specific study designs could include randomised controlled trials, case series, and prospective cohorts. Interventions that focused on optimising understanding (or another plausible outcome linked to decision making for trial participation) of probabilistic information within the context of a clinical trial were included. We chose to include studies of both real and hypothetical decisions about trial participation.

Exclusion criteria

Papers or articles that present findings on risk communication in a treatment or screening context or consider the decision to participate in research studies that are not RCTs were excluded. Studies investigating participants’ perceptions of receiving risk communication as part of the RCT decision process (which may include studies using methods such as interviews, focus groups and other methods) were not included.

Search methods for identification of studies

A search strategy for risk communication in clinical trials was designed in collaboration with a Senior Information Specialist (skilled in developing and running search strategies to identify relevant scientific literature) and informed by systematic reviews of risk communication in treatment and screening contexts and supplemented with trial participation terms. The search strategy is available on request. Four data bases were searched. Embase was searched from 1980 to 2019. Ovid MEDLINE(R) Epub Ahead of print, In-Process & other Non-Indexed Citations, Ovid MEDLINE(R) Daily and Ovid MEDLINE(R) was searched from 1946 to May 10th 2019. PsycINFO was searched from 1987 to May week 2 2019. Finally, CINAHL was searched from 1998 to 2019. No restrictions on language were imposed.

Screening and selection of studies

One author (MC) screened all articles identified within the database searches. Duplicate screening was carried out by one other author (KG) on a random sample (10%) of the search output. Papers were assessed at title and abstract level according to the eligibility criteria, and differences of opinion were resolved by discussion between MC and KG. Nineteen full text papers were identified for further investigation, and of these seven studies were deemed eligible for inclusion and progressed to data extraction procedures.

Data collection and analysis

The seven studies were summarised by study characteristics (see details below) and presented in tabular form. Due to the heterogeneous nature of the interventions and/or outcomes reported a meta-analysis was not appropriate. This review is therefore presented in a descriptive narrative form with studies grouped first by design of the embedded study (RCT, non-randomised) and then by content of intervention i.e., probability presentation, risk framing, risk intervention. This structured framework to present narrative findings has been recently proposed by Rowlands et al 2018 [21].

Data extraction and management

Data were extracted independently by two reviewers (MC & KG). The following summary features of the host trial (i.e. the trial the potential participants were being asked to consider participation in) for each study were summarised in table form: study design; study aim; author details; year and journal of publication; population demographics; sample size; phase of trial; intervention(s). Specific details on the intervention(s) being evaluated (i.e., risk communication tools), embedded study results and associated outcomes were extracted. These included: comparative methods of disseminating probabilistic information to potential trial participants using different communication tools/aids; mode of intervention delivery (i.e., paper, computer, verbal); study outcomes to be extracted; cognitive outcomes (i.e., potential trial participant comprehension of probabilistic information and subsequent risk perception); affective outcomes (i.e., participant preferences and/or satisfaction with communication method, and level of decisional conflict and concern); and behavioural outcomes linked to trial participation (i.e., willingness to participate in clinical trial).

Results

Study selection and summary characteristics

The search strategy identified a total of 4931 studies. Full text papers for 19 potentially eligible studies were sourced, and following full text screening a further 12 studies were excluded from the review (Fig 1). The included seven studies all reported results from studies set within hypothetical randomised controlled trials [22-28]. To provide an example of how this embedded evaluation is operationalised, the studies asked participants to imagine they were being recruited into a clinical trial, provided brief information about the hypothetical trial (such as clinical population, intervention, comparator, outcomes, etc), then provided various formats of risk communication (such as verbal or numerical descriptors) followed by assessment of relevant outcomes.

Fig 1

The PRIMSA diagram details our search and selection process applied during the literature review.

The seven included studies had various designs: five were randomised comparisons of risk communication interventions considering participation in hypothetical RCTs; and two were prospective, non-randomised studies, one being a comparative cohort study (three groups) and the other a single cohort. The included studies spanned a range of clinical settings. Three of the included studies were trials in neurological settings, and the other four were within dermatology, cardiology, oncology and surgery. Only one study was set within a trial considering a non-drug intervention, where the other six were identified as trials testing drug-based interventions. All of the included studies had at least two arms as part of their hypothetical trial design. Six of the seven studies reported trials where an individual was considering consenting for themselves and one study included only parents who were considering participation for their child. Most studies were single centre, however two of the studies did not specify the number of centres involved. The number of participants in the embedded studies ranged from 50 to 4885 with a median of 240. (Table 1).

Table 1

Summary characteristics of hypothetical host trials.

Trial characteristic	Hypothetical RCTs (n = 7)
Clinical setting
Dermatology	1
Neurology (pain x 2, ALS)	3
Cardiology	1
Oncology	1
Surgical site infection	1
Types of interventions
Drug	6
Non-drug	1
Trial design
Cohort	1 (Sutherland)
Two groups	2 (Berry, Kim)
More than two groups	4 (Cheung, Schwartz, Tait, Treschan)
No of centres
Single centre	5
Multicentre	0
Unknown	2
No of participants
<500	5
500–1500	1
>1500	1
Median no of participants (range)	240 (50–4885)

The final seven studies were grouped according to the study design (i.e., RCT or prospective cohort) and the topic of the described intervention: ‘probability presentation’ (22, 23), ‘risk framing’ (24, 25, 26, 27), and ‘risk interpretation’ (28). The studies are presented alphabetically based on these similar characteristics under their category headings (Table 2). Further information on each study detailing intervention content, mode, and outcome are presented in Table 3.

Table 2

Catalogue of included studies by study design and real or hypothetical RCT setting.

Category	Risk communication study design	Real RCTs	Hypothetical RCTs
A	RCTs of interventions to explore risk communication in RCTs	N/A	BerryKimSchwartzTaitTreschan
B	Prospectively designed, non-randomised studies of interventions to explore risk communication within RCTs	N/A	CheungSutherland

Table 3

Types of intervention(s) reported in included studies.

Author/Date	Content	Mode	Outcome
RCTs of interventions to explore risk communication in RCTs
Probability presentation interventions
Berry & Hochhauser 2006	Intervention communicates risk for the ‘experimental’ drug onlyTwo experimental conditions: probability of side effects described using verbal descriptors, or verbal descriptors with associated numerical ranges. i.e. Common (EU equivalent = 1–10%) Uncommon (EU equivalent 0.1–1%) Rare (EU equivalent = 0.01%-0.1%)	Written four-page questionnaire booklet	Satisfaction with the information; perceived risk to health from taking the drug; perceived effectiveness of the drug; how beneficial for their health it would be if they took part in the trial; and how likely it was that they would participate. (assessed using a 6-point unipolar Likert rating scale) Estimation of the probability of their experiencing each side effect as a percentage, between 0% and 100%. Main reasons for taking part or not
Risk framing interventions
Kim et al 2015	Intervention communicates risk for the ‘experimental’ drug onlyOne of two statements: in ‘no guarantee’ group, the likelihood was described as ‘It is not guaranteed you will benefit’; in ‘control’ group likelihood described as ‘There is some but very small chance that you might benefit’	Online survey	Willingness to participate in trial on a 10-point scale, from ‘would not consider at all’ = 0, to ‘definitely would consider’ = 10. Likelihood that ALS would improve from being in this study, from 0% to 100%.
Schwartz & Hasnain 2002	Intervention communicates risk for the ‘experimental’ drug onlyInformation in each group’s consent form was identical except for the second paragraph describing the probable risks and benefits of the new drug.In the ‘gain’ group, benefits were framed in terms of gains, benefits were framed as losses in the ‘loss’ group, and in the ‘both’ group, both framings were presented i.e.:‘Anyone taking this drug has a small risk of a severe allergy that could result in death. Out of 100 people whose lives would likely be cut short by heart disease and begin taking the drug, we expect that 5 people will show no improvement and will go on to die from heart disease (loss), and 95 people will substantially improve their chance of survival and reduce their chance of death (gain)’	Paper based	Riskiness of participation in the clinical and riskiness of non-participation in the clinical trial on a category rating scale from 1 (not at all risky) to 10 (extremely risky) Willingness to participate in trial (yes or no)
Tait et al 2010	Intervention communicates risk for both interventions (drug)Three different risk/benefit message formats (text, tables or pictographs), and the presence or absence of a risk severity graphic.Risks (itching and slowed breathing) and benefit (pain relief) were communicated in one of the three different formats.Comparing risks and benefits between drugs A and B: absolute risk of occurrence for drug A was presented; information for drug B was presented as incremental risk increase or decrease.	Online survey	Verbatim understanding (the ability to correctly report the actual risk and benefit frequencies of drugs A and B) Gist understanding (the ability to identify the essential meaning about the observed differences between the risks and benefits of drugs A and B) Perceptions of the risks and benefits of drugs A and B e.g. ‘how worried would you be about your child experiencing pain after surgery?’. Also perceptions of frequency and severity of side effects, scored using 1–11 interval scales from e.g. ‘not at all likely/worried’ to ‘extremely likely/worried’ etc. Perceptions of the risk/benefit communication format
Treschan et al 2003	Intervention communicates risk for both the control and the treatment groupThree versions of study protocol: ‘control’ involved little if any risk or pain; ‘pain’ required additional procedures that were described as provoking considerable pain and discomfort; and ‘risk’ involved additional procedures that were described as inducing risk of injury	Study protocol and informed consent document	Willingness to participate (yes or no) Understanding of risks involved in participation i.e. asked to mark the statement they found most applicable: A. ‘Participation in this study is not associated with additional risks, discomfort or pain’, B. ‘Participation in this study is associated with additional risks, but does not cause any discomfort or pain’, and C. ‘Participation in this study is not associated with additional risks, but might cause discomfort or pain’. Factors that influenced willingness to consent
Prospectively designed, non-randomised studies of interventions to explore risk communication within RCTs
Probability presentation interventions
Cheung et al 2010	Intervention communicates risk for the new medication onlyThree formats of risk presentation: frequency, percentage and verbal descriptors. The verbal description followed the EU guideline on drug labelling; risk levels of ≤0.01%, >0.01% to 0.1%, >0.1% to 1%, >1% to 10%, and >10% were described as ‘very rare’, ‘rare’, ‘uncommon’, ‘common’ and ‘very common’ respectively.Card 1 showing information about side effects of a new medication for pain relief in one of 6 ways of risk presentation.Card 2 with the same risk information presented in all three formats (in the same sequence in severity)	Paper based	Willingness to participate in trial after card 1 presentation, and then willingness to participate in trial after card 2 presentation; A change in decision would indicate a potential problem in the initial format. Preference for risk communication Understanding of EU descriptors: which of the five (from ‘very rare’ to ‘very common’) best describe the frequency of 1 out of 40, 1 out of 4,000, 1 out of 5, 1 out of 200 and 1 out of 20,000.
Risk interpretation interventions
Sutherland et al 1990	Intervention communicates risk for the ‘experimental’ drug onlyPatients asked to underline statements in the consent form that were pertinent to making a decision about participating in the study.Three statements about the likelihood of certain events occurring were given; ‘itchy, red skin rashes are unlikely to occur’, ‘a particular type of cancer responds to radiation treatment in 10% of cases’, and ‘nausea and vomiting occurs in 45% of patients’	Paper based	Willingness to participate in trial Understanding of three statements describing probability of an event occurring. Preferences for the way potential benefits and risks or side effects of therapy are described Preference for verbal and/or numerical descriptors of probability

RCTs of interventions to explore risk communication in RCTs

Probability presentation interventions

One study was identified that used a randomised design to investigate different probability presentations in the context of risk communication in clinical trials. Berry & Hochhauser (2006), compared European Union (EU) verbal descriptors only versus verbal descriptors and their associated numerical values (e.g., Common (EU equivalent = 1–10%)) [22]. Participants were asked to imagine they had been approached to take part in a clinical trial and given a booklet detailing the possible side effects of a new drug (versus nothing) for a skin condition and were asked to complete a questionnaire (N = 96, 48 in each arm). When asked to rate on a scale from 1 to 6 (p = 0.03), those who received only verbal descriptors were significantly less satisfied with the information than those who also had the numerical values. Participants in the verbal descriptors only group also perceived the risk to health to be higher (p<0.0001) and the benefit to be lower (p = 0.03), and were significantly less likely to participate in the trial (p = 0.01). When asked to make probability estimates for experiencing side effects, the verbal only group estimated these approximately three times higher than the combined group. When asked to consider the main reason for participating in the trial, participants in both groups reported long term relief/possible cure and the main reason for not participating was fear of side effects. There were no significant differences between the reasons listed by the two groups.

Risk framing interventions

Four of the included studies employed randomised designs to explore risk framing in communication within clinical trials. Kim et al (2015) recruited 584 participants to investigate the language framing of benefit statements within a hypothetical trial for amyotrophic lateral sclerosis. An online survey administered one of two statements within a consent form to participants; either ‘there is some but very small chance that you might benefit’ (control group n = 290), or ‘it is not guaranteed you will benefit’ (intervention group n = 294) [24]. The intervention group had a slightly greater, but not significant, willingness to participate in the trial as scored on a 10 point scale (p = 0.11). However, the average estimate of the likelihood of their condition improving was significantly higher in the intervention group than in the control group (p<0.0001).

Schwartz & Hasnain (2002) explored the effects of gain and loss framing on risk perception and attitude by randomising 284 participants to one of three groups receiving a consent form about a trial for a new cholesterol lowering drug [25]. One group were given information where benefits were framed in terms of gains (e.g., ‘Out of 100 people whose lives would likely be cut short by heart disease and begin taking this drug, we expect that 95 will show substantial improvements in their chance of survival and 5 will show no improvement in survival’, n = 98), the ‘loss’ group received benefit information framed as losses (e.g.,. ‘Out of 100 people whose lives would likely be cut short by heart disease and begin taking this drug, we expect that 5 people will go on to die from heart disease, and 95 people will reduce their chance of death’, n = 93), and the third group were given information where both framings were presented (e.g.,. ‘Out of 100 people whose lives would likely be cut short by heart disease and begin taking this drug, we expect that 5 people will show no improvement and will go on to die from heart disease, and 95 people will substantially improve their chance of survival and reduce their chance of death’, n- = 93). The majority of participants (59%) chose to take part in the trial when outcomes were framed as losses, while only 35% of the ‘gain’ group chose to participate. When both framings were presented, 62% of participants chose to participate, making a similar choice to the ‘loss’ group. When it came to perceiving riskiness of participation, the ‘gain’ group were more likely to rate this as riskier than non-participation (66%) compared to the ‘loss’ group (55%). For the ‘both’ group, the results were again similar to the loss condition, with 52% reporting trial participation as riskier than not. Respondents in the gain condition rated participation as significantly riskier (on a 10 point scale) than those in the loss condition (p<0.05), and respondents in the loss condition rated non-participation as significantly riskier than those in the gain condition (p<0.05). There was a significant association between domain (gains vs loss) and relative riskiness of participation vs non-participation (p<0.05).

In the study by Tait et al [26] 4685 parents were asked to consider their child was being randomised into a trial testing two drugs for post-operative pain, one a standard treatment and the other proven in adults but not in children. The risks and benefits of the two drugs were presented in absolute terms with comparisons presented as incremental changes. Four scenarios that provided different risk/benefit trade-offs were developed and considered: one benefit and 2 risks (a minor and a major), which were varied for Drug B across each scenario but remained static for Drug A. There was one scenario with no trade off, where there was an increase in benefit as well as risk reduction (n = 1171), whereas the other three included a loss of benefit but gains in risk reduction (n = 1184, n = 1196, n = 1134). Overall the study showed that parents who received the ‘no trade off’ (i.e., improvements across benefit and risk) scenario had both improved gist (defined as ‘ability to identify the essential meaning about the observed differences’ and measured using 4 items where ≥3 correct answers were required, p<0.01) and verbatim understanding (defined as understanding or knowledge to ‘correctly report the actual risk and benefit frequencies’ and measured using 7 items where ≥5 correct answers were required, p<0.01). The no trade off scenario also enabled parents to correctly perceive the potential benefits as greater, risks as lower, (p<0.01) and to be more likely to agree to their child participating in the trial (measured using an 11 point scale) compared to the other three groups (p<0.01). Taken together these results suggested the no trade off scenario offering multiple gains resulted in a higher level of scrutiny compared to when only reductions in risk were presented.

Treschan et al (2003) randomised 148 participants to one of three versions of a study protocol to examine how understanding of risk and discomfort associated with a clinical trial influences patients’ decision to participate [27]. The proposed trial was comparing peri-operative oxygen (30% vs 80%) to reduce the risk of surgical site infections. The control group received a version of the protocol that stated there would be little if any risk or pain involved in participating (n = 47), the ‘pain’ group were told that there would be additional procedures that would cause considerable pain and discomfort (e.g., dressing of wounds, cannulation, blood samples, n = 51)), and for the ‘risk’ group procedures were described as having a high risk of injury (e.g.,. extra oxygen is dangerous, risks of cannulation, risk of blood samples, etc, n = 50). Participants in the control group were more willing to participate in the trial (64%), with significantly fewer consenting in the risky (26%) and painful (35%) groups (p<0.001). There were no significant differences in understanding of the level of risk or pain for the three groups (p = 0.884). Those who correctly understood the risk or pain described in the protocols were twice as likely to consent to participation in the trial (49% vs 24%, p = 0.003).

Prospectively designed, non-randomised studies of interventions to explore risk communication within RCTs

Of the two non-randomised papers that met the inclusion criteria, Cheung et al (2010) is the only study that investigated probability presentation within risk communication for clinical trials [23]. This study implemented a cognitive experiment (N = 240) and preference survey about risk within a hypothetical trial for pain medication for arthritis. The intervention used a factorial design to study the impact of three formats (frequency (n = 82), percentage (n = 80) and verbal descriptors (n = 78)) and two sequences on willingness to participate and likelihood to change one’s willingness after given additional information. Participants were presented with information in one of the six combinations. Participants were given a card that showed information about side effects of a new medication for pain relief in one of six ways of risk presentation, and then were asked whether they would be willing to take part in the trial. They were then presented with a second card, with the same risk information presented in all three formats being studied. A change in decision would indicate a potential problem in the initial format given to participants. There was no difference in willingness to participate in the trial across all presentations (p = 0.886), and there was also no difference in the likelihood of a participant changing their mind after being given the information in additional formats (p = 0.529). After reading card 2, the proportion of participants in each group showing a willingness to participate increased significantly (p<0.05). With regardto presentation preferences, 43% of participants preferred the frequency format, 32% preferred percentages, and 25% preferred the verbal descriptors.

Risk interpretation interventions

The remaining non-randomised study (Sutherland et al, 1990) explored risk interpretation within a consent form for a hypothetical drug trial for cancer [28]. All participants (N = 50) were given a consent form and asked to underline statements that were important to them in terms of making a decision about participating in the trial. They were also asked to indicate if their chosen statements were positive or negative. A questionnaire including preferences for probability descriptors (verbal or numerical) was also administered. Of those who refused to take part in the hypothetical trial, 70% noted only the potential for risk, 10% only for benefit, and the remaining 20% noted both risk and benefit information as important for their decision. Just 33% of those who ‘consented’ identified only risks, 27% noted only benefits, and 30% noted both risk and benefit. The remaining 10% identified neither as important to their decision. One third of participants were unable to identify the correct interpretation of the ‘unlikely’ verbal descriptor, and 54% gave an incorrect interpretation of ‘10% response rate’ meaning. When it came to preferences for benefit descriptors, 16% of patients preferred words, 34% numbers, 48% both and 2% other. For risk communication preference the results were very similar; 16% verbal, 28% numerical, 48% both and 2% other.

Discussion

The study is one of the first to systematically review the published evidence on methods for communicating risk to potential trial participants during the informed consent process. It has examined and summarised the existing evidence about how risk information is perceived by potential participants and highlights how these factors may influence decisions to participate in a clinical trial context. Only seven studies were identified that have investigated aspects of communicating risk information in a clinical trial setting. Whilst the majority of studies were randomised comparisons, we also identified 2 non-randomised evaluations. Given the heterogeneity of the interventions investigated in the included studies and the variability in outcomes reported, a meta-analysis of these studies was not possible. This work therefore highlights the need for the rigorous development and evaluation of interventions to improve the presentation and communication of risk information for potential trial participants.

One of the studies investigated probabilistic presentation methods and demonstrated that numerical formats appear to be better at communicating risk to potential trial participants, when compared to text [22]. Participants receiving verbal descriptors alone were less likely to consent to take part in a trial and were less satisfied with the information, perceiving risks of side effects to be much higher than participants receiving both numerical and verbal descriptors. Similar findings can be seen in a review on communicating with patients about evidence (for treatment decisions), which illustrated that patients have a better understanding of risk if probabilistic information is presented numerically rather than verbally [29]. It is worth considering that studies in a treatment setting have shown that using visual aids such as pictographs or bar charts to present event rates may aid accurate understanding of probabilities, and they can help reduce several biases including framing effects [30]. There are many variants of visual aids however, and how these are utilised and understood by potential trial participants warrants more investigation using the best practice examples from treatment decision making as a starter.

The second study (Cheung et al, 2010) looking at probabilistic presentation found no difference in willingness to participate between frequency, percentage and verbal conditions; however, it did find a strong preference for numerical presentations over verbal descriptors, particularly for frequency formats [23]. Research by Price et al (2007) found that frequency statements are generally better understood by participants compared to ratios or percentages [31]. An important finding from this study highlighted major errors in correctly matching EU descriptors of risk to associated frequencies, findings echoed by the other study which looked at risk interpretation showing a large proportion of participants were unable to correctly interpret verbal descriptors or percentage formats [28]. A number of studies have demonstrated that many lay persons are unable to understand basic aspects of probabilities that are essential to risk understanding, nor to comprehend the concept of risk in general [32, 33]. This poses a challenge to effective risk communication and demonstrates a need for improved methods for better informed consent within the context of clinical trials.

The Sutherland et al (1990) study found that the majority of non-consenters to the trial noted only the potential for risk in the provided information, whereas the information was interpreted very differently by consenters where a minority saw only risks, and many perceived benefits instead [28]. A qualitative study into patient decisions about taking part in an epilepsy treatment trial noted that participant decision making was most commonly influenced by their perception of harm and benefit [34]. Those who agreed to take part usually saw the risks involved as acceptable, in this case because of the ‘tried and tested’ nature of treatments. However, the non-consenters viewed participation as ‘an unknown quantity’ and defined the risks of being randomised to an unsuitable drug as being too high or not in their best interest [34].

When it came to studies looking at risk framing, the results were mixed. The study by Kim et al found no significant difference in willingness to participate in the trial, although participants in the intervention group (no guarantee for benefit statement) were much more likely to believe that their condition would improve [24]. When benefits were framed as losses participants were more likely to take part in the trial, and when benefits were presented as both losses and gains, participants seemed to respond similarly to the loss group, suggesting that loss framing had more impact on decision making than gain, where perceived risk was higher [25]. However, many of the statements used in this study were vague and uninformative, putting into question what understanding participants had in relation to these statements in addition to willingness to participate. Conversely, Treschan et al found that when outcomes were framed as gains the majority of participants were less likely to participate [27]. Earlier research by Tversky & Kahneman (1981) on framing and the psychology of choice demonstrated that framing outcomes in terms of gains does indeed generate risk-averse choices, which could translate to, for example, a decreased willingness to participate in a clinical trial [35]. A more recent study highlighted the introduction of potential bias in decision making about trial participation when the effects of language framing are not addressed [36]. This study explored whether presenting health care decisions as ‘opportunity’ rather than ‘choice’ biased individuals’ preferences in the context of trial participation for cancer treatment. They found that a ‘choice’ frame, where all treatment options are explicit, is less likely to bias preferences [36]. It is therefore of paramount importance that information given to participants include neutral statements, or at a minimum balanced statement about participation or not, so as not to unduly manipulate or ‘nudge’ decisions in ways that are not consistent with the individual’s values and preferences [18].

Five out of the seven studies included in this review only communicated risk information about the ‘experimental’ treatment [22–25, 28]. Two studies communicated risk information about both the intervention and its comparator or indeed both active interventions [26, 27]. Given that decision making about clinical trials is complex and requires trade-offs between both (or all) options and therefore presenting risk (and benefit) information on these options would be important to support fully informed choices. This should be acknowledged and explored in future studies.

Complex language and details included in participant information leaflets (PILs) and consent forms for trials can be difficult for some people to comprehend properly and may engender more confusion than understanding of trial processes, including risks [37]. An analysis of PILs used in clinical trials by Gillies et al (2011) found that: explaining trial processes; presenting probabilities; and expressing values, were consistently poor across all PILs when assessed using an informed consent evaluation instrument [17]. These information leaflets clearly need to be improved to encourage higher quality decision making when it comes to trial participation. It is also clear that potential trial participants continue to have significant deficits in their recall and understanding of trial related information, and that such information is often not presented in a comprehensive way that optimises participant understanding [38, 39]. The recent study by Gillies et al (2014) explored whether patient information leaflets (PILs) were able to effectively support decision making about trial participation [17]. They found that information that demonstrated support for good quality decision making in other contexts was lacking in PILs for UK clinical trials. In particular, the section on ‘presenting probabilities’ was almost always absent, despite its proven importance for supporting good quality decision making [17].

Whilst not a focus of this review it is important to point out that none of the included studies reported including patients or the public as partners in the research to identify what the content and/or presentation of the information should be for the studies. Also, no input was sought with regard to whether the outcomes being evaluated were appropriate and meaningful for patients faced with decisions about trial participation.

Lessons from effective risk communication in a treatment and/or screening context can provide examples of best practice that could be used for those developing PILs for patients considering clinical trial participation. A systematic review on risk communication published since has shown that visual aids, such as icon arrays and bar graphs, improved both understanding and satisfaction [40]. Interestingly, this review showed that presenting absolute risk reduction was better at maximising accuracy and less likely to influence decisions. The presentation of information on numbers needed to treat reduced understanding. This review also concluded that due to the quality and heterogeneity of included studies, it is not possible to determine a ‘best’ method for conveying probabilistic information [40]. However, whilst there might be a paucity of high quality evidence to support an unequivocal ‘best’ method there have been recommendations for guiding principles developed by several groups. The first, developed using an international consensus process involving researchers and patients, provided key considerations for presenting probabilities of outcomes [41]. These include:

Use event rates to specify the population and time period

Compare outcome probabilities using the same denominator, time period, and scale;

Describe uncertainty around probabilities;

Use visual diagrams;

Use multiple methods to view probabilities (words, numbers, diagrams);

Allow the patient to select a way of viewing the probabilities (words, numbers, diagrams);

Allow patient to view probabilities based on their own situation (e.g. age);

Place probabilities in context of other events;

Use both positive and negative frames (e.g. showing both survival and death rates).

An expert consensus group further developed these IPDAS items to develop a set of guiding principles and key messages which cover eleven components of risk communication and consider what information to present and how it should be presented within tools such as patient decision aids [42]. The guiding principles range from how best to present the chance an event will occur, to use of interactive web-based platforms for delivery, and narrative methods for communication [42]. A recent study published ‘good practice statements’ for the development of evidence-based information communicating the effects of healthcare interventions [43]. Many of these statements would be relevant for developing information related to risk communication to support decision about trial participation. For example: using numerical formats that are easy to understand; present both numbers and words; and report absolute effects [43]. In summary, whilst there may be a paucity of high quality evidence to underpin decisions about effective risk communication in clinical trial contexts, many of the good practice recommendations developed through empirical research provide sensible frameworks to promote informed choices, enable good quality decision making, and are unlikely to cause significant harm. As such, these guiding principles could also serve as a foundation on which to develop (and test) effective methods of risk communication within the context of clinical trials.

Strengths and limitations

The low number of studies included for review means it is difficult to confidently make far reaching recommendations based on the findings, and the heterogenous nature of the studies mean a meta-analysis was not feasible. The studies in our review included decisions about trial participation that were hypothetical which may limit the extent to which these findings are applicable to a real world setting. Understanding and assessing risk and risk communication is pertinent to the trial phase, as the magnitude of risk is much greater in earlier phases of clinical trials; however, only one of the studies stated the trial phase being investigated. This review is, however, the first to systematically investigate risk communication within a clinical trial context. With ever increasing numbers of trials, the importance of informed consent, and yet no consistent, evidence-based format for presenting probabilistic information in a clinical trial setting, this study supports the argument for effective future research within this area.

Conclusions

The evidence base surrounding risk communication in clinical trials indicates that there is as yet no clear optimal method for improving participant understanding, nor a clear consensus on how understanding affects willingness to participate, indicting a necessity for robust, high quality research in this area. Further research into risk communication during the informed consent process for trials, based on examples of best practice in other settings such as treatment and screening decision making, is needed to help illuminate the mechanisms underlying risk perception and understanding and provide appropriate ways to present and communicate risk in a trial context so as to further promote informed choices about participation.

PRISMA checklist.

(DOC)

Click here for additional data file.

Search strategy MEDLINE.

(DOCX)

Click here for additional data file.

30 Oct 2019

PONE-D-19-19791

A systematic review of risk communication in clinical trials: how does it influence decisions to participate and what are the best methods to improve understanding in a trial context?

PLOS ONE

Dear Dr. Gillies,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

==============================

The first reviewer argues that this manuscript is not suitable for publication since it cannot be regarded as a systematic review and PLOS ONE only accepts systematic reviews. At the same time, systematic reviews of ethical issues may differ from conventional medical systematic reviews. There are several methodogical ethical papers on the conduct of systematic reviews in ethics. Therefore, please revise the manuscript along these lines and clarify how the manuscript can be regarded as a systematic review.

==============================

We would appreciate receiving your revised manuscript by Dec 14 2019 11:59PM. When you are ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter.

To enhance the reproducibility of your results, we recommend that if applicable you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). This letter should be uploaded as separate file and labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. This file should be uploaded as separate file and labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. This file should be uploaded as separate file and labeled 'Manuscript'.

Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.

We look forward to receiving your revised manuscript.

Kind regards,

Rieke van der Graaf

Academic Editor

PLOS ONE

Journal Requirements:

1. When submitting your revision, we need you to address these additional requirements.

Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

http://www.journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and http://www.journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. Please include your tables as part of your main manuscript and remove the individual files. Please note that supplementary tables (should remain/ be uploaded) as separate "supporting information" files

3. In your Data Availability statement, you have not specified where the minimal data set underlying the results described in your manuscript can be found. PLOS defines a study's minimal data set as the underlying data used to reach the conclusions drawn in the manuscript and any additional data required to replicate the reported study findings in their entirety. All PLOS journals require that the minimal data set be made fully available. For more information about our data policy, please see http://journals.plos.org/plosone/s/data-availability.

Upon re-submitting your revised manuscript, please upload your study’s minimal underlying data set as either Supporting Information files or to a stable, public repository and include the relevant URLs, DOIs, or accession numbers within your revised cover letter. For a list of acceptable repositories, please see http://journals.plos.org/plosone/s/data-availability#loc-recommended-repositories. Any potentially identifying patient information must be fully anonymized.

Important: If there are ethical or legal restrictions to sharing your data publicly, please explain these restrictions in detail. Please see our guidelines for more information on what we consider unacceptable restrictions to publicly sharing data: http://journals.plos.org/plosone/s/data-availability#loc-unacceptable-data-access-restrictions. Note that it is not acceptable for the authors to be the sole named individuals responsible for ensuring data access.

We will update your Data Availability statement to reflect the information you provide in your cover letter.

4. Please amend the file type of your PRISMA checklist to a 'Supporting Information' file type.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Partly

Reviewer #2: Yes

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: N/A

Reviewer #2: N/A

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: The authors conduct a systematic review of studies examining different methods of presenting risk and benefit information in the context of clinical trial consents. While the initial search generated a large number of prospective papers, the inclusion criteria, which specifically required the studies to be examining communication in a clinical trial context, resulted in only 7 widely varying papers (all of which relate to hypothetical clinical trials decisions) being actually examined. The authors summarize the findings of these 7 papers and provide some additional context in the discussion.

While the basic structure and writing of the paper were acceptable, I have a number of conceptual concerns.

First, this paper is really more of a review of a systematic review. Yes, there was a systematic process for identifying papers and abstracting information. But, the papers are themselves so varied that no conclusions can be drawn except at the individual study level. In other words, no new insights are gained by the set of information beyond the understanding of best practices that existed previously. Given that this journal explicitly excludes review papers, I am thus skeptical that this paper is appropriate for publication here.

The main problem is in the definition of the problem to be studied: The authors restrict themselves to papers about clinical trial contexts. The logic for doing so is based on the idea that risk communication in a clinical trial context is somehow fundamentally different than other contexts. It therefore excludes many other, very similarly structured studies examining risk communication in other kinds of decisions. While systematic reviews are very valuable way of considering a range of papers that have studied similar questions, the current set of papers are not really "similar" in important ways. Thus, the criteria is both narrow (in focusing on clinical trials situations only) and very broad (in allowing in all kinds of risk communication of very different types).

I can't really complain about the analysis because the analysis is entirely individual paper focused. Each study is summarized, but there is no real integrative analysis. Because of its narrowness, little of the risk communication lessons derived from other work are discussed. This leaves readers with the mistaken impression that little is known about how to effectively communicate the types of risk information relevant to clinical trials. What is lacking (as this paper shows) is empirical studies showing that these lessons carry over to this context. The unfortunate reality is that the lack of relevant studies is because there is little incentive for researchers to do such conceptual replication studies when the answers seem so obvious.

Given the limitations of the data, i am also concerned by the length of the discussion section and the degree that it tries to draw conclusions from single studies (which is antithetical to the concept of a systematic review).

Reviewer #2: The paper reports a systematic review of studies addressing the effects of various methods to communicate the patients about the risks of participation in RCTs of treatments. They found 7 studies after considering almost 5000 titles. Of the 7, five were RCTs (of communicating about RCTs) and two had a different design (prospective, non randomized). Additionally, NONE of these studies were done with patients who actually considered participation in a clinical trial for their own disease. Rather they asked the participants to hypothetically pretend they had a disease and decide whether they would participate in a trial of a fictional treatment for the disease. This is a limitation of the field, not of the review.

Because the studies did not use the same interventions nor measure the same outcomes, it was not possible to summarize the effects of the communications quantitatively (meta analysis), so they did a narrative summary. What this means in practice is that they summarized in extensive detail the methods and findings of each study, in alphabetical order of authors, grouped by whether the study was an RCT or not.

My perception of the paper has two high level comments.

Comment 1. The descriptions of the studies have too many details, described “telegraphically”. That is, you almost have to go read the paper to understand what was done. An example is this extract from the description of the second study described:

“three different risk/benefit formats: text (e.g. risks and benefits described as

219 number out of 100 experiencing outcome); , tables (e.g. same as text based but presented in tabular

220 format); or pictographs (e.g. a matrix of 100 rectangular blocks representing 100 children then

221 coloured to represent presence (blue) or absence (grey) of outcome). The presence or absence of a

222 risk severity graphic was also investigated. This graphic was described as having increasing risk

223 represented by blocks of different sizes and colours (yellow through red), and an arrow depicting

224 where the risk or side effect fell on the scale accompanied by a description of risk as ‘mild’, ‘moderate,

225 or ‘severe’. Risks and benefits of two drugs for post-operative pain in children were communicated in

226 one of three formats and presented as either absolute risk or incremental risk to participants.”

While I am sure these statements accurately describe the elements of the study, it is difficult for the reader of this summary to get a sense of the design – was it 3 by 2 by 3 by 2, i.e., 36 different cells? Or 12?

Comment 2. The introduction presents “participation in an RCT” as just another area in which physicians might need to communicate about risks to patients, as if it were a routine drug or an operation. The reader might ask, what is new here, it is just a different domain in which risks must be communicated. Surely we’d expect that communication aids or approaches that work for other treatment decisions would similarly work here.

But participation in a randomized clinical trial actually has a degree of complexity, uncertainty, risk much larger than choosing a treatment. While the tradeoffs between alternative treatments involve risks and value tradeoffs, the tradeoffs between a study or not doing a study involves a) a loss of control, a LACK of being informed, and b) greater uncertainty about the benefits. I say greater uncertainty, because while treatment A has its own uncertainties (will it work? Will there be side effects?), a trial in which treatment A is one of the options to which one will be assigned has all those uncertainties, plus the uncertainties about the alternative (experimental) treatment whose uncertainties are even greater, plus the uncertainty about which of the two (or more) arms one will be assigned to. Quite possibly the communication aids or approaches that work for a decision among accepted treatments with (somewhat) known success probabilities would NOT work for this more complex situation. It would be helpful for motivating the reader to engage with the paper if this difference between communicating the risks of RCTs, versus simple (?) treatments, were stated explicitly. It makes the paper more interesting.

It would also be interesting if, in describing the studies, the review noted whether and how the papers acknowledged and addressed the added complexity of the participation decision. Did they communicate ONLY about the probabilities of treatment assignment? Or did they name the good and bad outcomes that could occur with either treatment? Did they explicitly state that the uncertainty about the benefits and side effects of the experimental treatment is greater than the uncertainty about the benefits and side effects of the currently accepted, comparison treatment? That is, what in particular was the “risk communication” about?

The paper is somewhat verbose; partly in the attempt to be very clear about what is being said, but it could be tightened.

Particular suggestions. (Some of these suggestions may be too minor, or wrong. But the reviewer instructions say that there is not going to be any copy editing on the part of the journal. Your $1500 is not sufficient to buy that service.)

Abstract, line 19. “context” should be “contexts”.

Line 27. When first encountered, it is difficult to understand what “studies set within hypothetical host trials” refers to. I have interpreted it as “studies in which risk information is communicated to support the participant’s hypothetical decision to participate in a fictional trial of a fictional treatment for a disease that the participant does not have”, but that is not what the words “set within a hypothetical host trial” means.

Line 31. “In term so” should be “in terms of”.

P 4, line 50. Expand “participants approached to take part and those consented to trials” to ““participants approached to take part and those who consented to participate in trials”

Line 51, participants.

Line 57, participant should be plural.

Line 66, 67, change “risk communication with regard to public health messages, health and treatment decisions for patients, and in screening contexts” to “risk communication with regard to public health messages, health [behavior?], and treatment and screening decisions for patients”. I guess “screening” may be done on people who are not technically “patients”, but so often it is done in primary care clinics.

Line 88, change “findings from work from our group has shown” to “findings from our group have shown”?

Line 91, change “highlighted patients preferences” to “highlighted that patients’ preferences”.

Lines 93 and 94, move “from treatment and screening decisions” to after “decision making” in the next line, and change “from” to “about”.

Last sentence before “Methods”. Participants should be possessive.

Methods. I am not sure what an “explanatory” study is. I don’t think it is an “exploratory” study (though that is used in next paragraph, but those studies were excluded). Does it mean, “a study of ways of explaining something to a patient”? If that is what it means, I don’t think the phrase “explanatory study,” without more elaboration, will be understood.

Style. “were included” was used three times in this one paragraph.

Lines 199 and 120, I don’t think it needs both “exploratory studies” and “that have explored”. Maybe drop the first.

Lines 124 and 125. What is a “senior information officer”? a modern librarian?

Line 129. Say May 10th is in 2019.

Line 138. “and progressed for data extraction procedures”? does than mean, “were followed by,” “led to”, “progressed to”?

Line 142. Given the immediately preceeding sentence, change “The seven studies deemed eligible and identified for data extraction were summarized” to “The seven selected studies were summarized”.

146. I’m not sure the Journal’s conventions. I would add commas around “i.e.” and “e.g.”, through out, “by content of intervention, i.e., probability” not “by content of intervention i.e. probability”

Data extraction and management.

Line 151. Data were, not data was.

Lines 151 and 152. Change “summary data was extracted regarding the host trial for each study and summarised in table form” to “summary features of the host trial for each study were summarised in table form”. Of course you had to extract it before summarizing it, but it need not be said.

Line 155. Space after the comma.

4 instances of i.e. without surrounding commas in this one paragraph.

Line 160. Change “satisfaction with communication method; and level of decisional conflict” to “satisfaction with communication method, and level of decisional conflict”

Lines 168 169. Here might be a place to give an example of how a study of communication about the risks of RCTs might be “set within a hypothetical host RCT”.

Line 170, drop unnecessary words, in the context. Change “The seven included studies that explored interventions to explore risk communication during informed consent varied by design” to “The seven included studies had various designs”.

Line 172. Could “prospectively designed” be simply “prospective”?

P 9, line 179. “an” not “and”

Line 180, “included” is ambiguous. Does it mean one study was only of parents, or one study had not only patients deciding for themselves but it also included some parents deciding for children?

Lines 186-188. It was unclear to me at first that these were the authors of the studies you were summarizing; at first I thought they were references for the kinds of method used, e.g., probability presentation, risk framing, risk interpretation. Be more clear.

194. Hypothetical is unclear. Is it that the participant is hypothetically making a decision about a real study, or is hypothetically making a decision about participation in a hypothetical or fictional study? The word is too vague for the many possibilities.

194. But since you’ve already told us all the studies involve pretending to decide about a fictional study, if I interpret correctly, you do not need to repeat it every time you describe a study. It just confuses the reader.

Line 197. Is the reference a cite of an included study, or of the paper describing numerical values associated with words like “common” in an official EU document? Or of an earlier study of the EU prescribed communication language?

Line 199. Add “were” before “asked”.

P 10, lines 208 and 209. Put quotes around the terms common, uncommon, and rare.

Line 217. Drop “t”.

Line 219, and extra comma.

Line 225. 2 by three by 2 design? Or did each get info about both drugs in the same format?

Line 230. Move “(66.55) back to after “gist understanding”.

234. orphan “).”

239. Comma after “respectively)”?

239. Comma after graphics.

252. space after 8.3. comma after “scale”.

259 to 262. The instructions in this study were, ‘Out of 100 people whose lives would likely be cut short by heart disease and

260 begin taking this drug, we expect that 95 will show substantial improvements in their chance of

261 survival and 5 will show no improvement in survival’. Do you want to comment on how vague and uninformative that is?

268 to 271. “The majority of participants (59%) chose to take part in the trial when outcomes were framed as losses, however a greater majority of the ‘gain’ group (65%) declined to participate. When both framings were presented, 62% of participants chose to participate, making a similar choice to the ‘loss’ group.” I find the switch in reference to be confusing. Who cares the relative size of the majority? Keep it consistent: “The majority of participants (59%) chose to take part in the trial when outcomes were framed as losses, while only 35% of the ‘gain’ group chose to participate. When both framings were presented, 62% of participants chose to participate, making a similar choice to the ‘loss’ group.”

278 needs a period at the end.

313 to 317. This discusses inaccuracy of patient mapping of numbers onto EU definitions of probability words. To interpret this, one would need to know what all the options are. Are they missing by one category, or are they several off? But the larger question is, what is the relevance of this to communicating about the risks of participating in randomized clinical trials? This seems a general problem with any risk communication. Of course, it is part of the whole comprehension process.

Line 326. Change “for only benefit” to “only for benefit”?

Line 326. I’m not sure what “saw” means. Was this a study intervention, or a participant behavior, picking up on an implication that is present? I think it is the former. I think it is a situation of: show one of three formats; see how many consented. If that is what it is, then you are describing it backwards, in terms of causality. “of those who consented, X saw a format” is the reverse order. The interesting question is, “of those who were shown x format, who saw x format, what proportion consented? I think this needs to be restated.

Line 327. Not clear what it was “to see neither”

Line 339. “aspect” should be plural.

345. “identified” seems an unnecessary word.

354. “deemed”. Is this that the participants reported the pictographs as more helpful, or that the authors of the reviewed studies concluded based on their data that the pictographs were more helpful. Who is “deeming”.

355. Not clear what the distinction is between a pictograph and a severity graphic, why is the one good and the other bad?

In this discussion, I feel a need for the authors to more authoritatively criticize the reviewed studies, not just to summarize them. I wonder if it would be helpful to have a model of comprehension of risk communication, e.g., the required steps: a) comprehend the scale, b) use the info to interpret one’ situation and guide action. Of course, if they fail the first step the second step has little chance of happening – unless they get it from alternative clues like verbal summary of gist. Perhaps you get at that in line 371.

Line 363. Change “and verbal conditions, however it did find” to “and verbal conditions; however, it did find”

Line 371. Change “as well as comprehending the concept” to “nor to comprehend the concept”. Preserves parallel with the previous clause.

373 to 374. It would help if the manuscript more clearly separated the paragraphs. This is an example where it is hard to tell.

Line 374. Put “The” before “Sutherland”.

Line 379. “generally”? should that be “usually”? Mathematically, “generally” means always true.

382. I think it is “their best interest” rather than plural, but I’ve seen it both ways.

390. “when gains were framed as gains”? “when outcomes were framed as gains”.

403. sheets = leaflets. If you don’t say that, the reader does not know what “PIL” means.

Lines 406 407. This list is hard to interpret, the elements do not seem parallel, and it is hard to tell if “understanding of” applies only to “trial processes”, or to all three elements in the list.

416. Not sure the meaning of “its evidenced importance”. It’s evident importance (on the face of it), or its proven importance (proven by lots of previous studies, including the evidence reported by study 17)?

Line 426. I’m not sure if agreement should be “number… means” or “number of studies… mean”.

428.”hypothetical nature of the studies”. But these were real studies, actually done. Need a different way to talk about this. And it is ambiguous if you mean “the 7 studies we reviewed” or “the stimulus studies, presented to the participants in the studies”.

Line 432. I don’t think “however” obeys the same rules as “but”. So I think the punctuation should be not “earlier phases of clinical trials, however only one of the studies” but “earlier phases of clinical trials; however, only one of the studies”.

Line 433. I think however needs to be set off on both sides by commas.

Line 440. What does “it” refer to?

443. “most useful” compared to what, or out of what set of options for this particular situation? Or do you mean more useful in this situation than in other situations where it might be used?

444. I think it is “complementary” not “complimentary”.

In the reference list, words are capitalized inconsistently. Some titles have all the words capitalized (appropriate for book titles but not paper titles) while others have just the first word capitalized. Be consistent, and consistent with the journal standards.

Reference 25 needs a period between the title and the journal name.

Reference 38. Laura siminoff’s paper had only one page?

Figure 1. I note that you found 3 systematic reviews – how did they differ from the present systematic review? Should they be included in your literature review (not just in your characterization of your systematic review’s rejects).

Table 2. Why a date for Tait but not the others.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: Yes: Robert M. Hamm

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files to be viewed.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at figures@plos.org. Please note that Supporting Information files do not need this step.

19 Dec 2019

Please see uploaded 'cover letter' which contains a detailed response to reviewers comments.

Submitted filename: PONED1919791_Response_to_reviewers_121219.docx

Click here for additional data file.

4 Jul 2020

PONE-D-19-19791R1

A systematic review of risk communication in clinical trials: how does it influence decisions to participate and what are the best methods to improve understanding in a trial context?

PLOS ONE

Dear Dr. Gillies,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Generally the reviewers are happy with the review. Much of the problem focuses on the nature of the review. PLOS ONE accepts systematic reviews but not narrative reviews. However, reviewer #1 raises an important point that the pure systematic review that concludes that there isn't enough evidence is useful but not very helpful. On the other hand a more narrative review might be more helpful. I think that there may be a compromise here. Maintain the systematic review but have a more detailed analysis of the topic in the discussion. Thus, it is a systematic review with a more comprehensive analysis.

Please submit your revised manuscript by Aug 18 2020 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

We look forward to receiving your revised manuscript.

Kind regards,

Dermot Cox

Academic Editor

PLOS ONE

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: (No Response)

Reviewer #2: (No Response)

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Partly

Reviewer #2: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: N/A

Reviewer #2: I Don't Know

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: This revised manuscript reports a systematic review of studies examining different methods of presenting

risk and benefit information in the context of clinical trial consents. The authors required the studies to be examining communication in a clinical trial context, resulting in only 7 widely varying papers which are then summarized one at a time. The primary conclusion is that there is little extant guidance for presenting risk and benefit information to guide clinical trial participation decisions.

In my previous review, I expressed significant concerns about the authors' framing of the scope of this review. In particular, the restriction to studies whose scenarios / tasks could only be about clinical trial decisions results in a small, heterogeneous set of papers and does not consider the relevance of structurally similar papers addressing other contexts such as treatment or screening decisions. The resulting manuscript thus is more descriptive than integrative, and it conveys a primary message of "we don't know", which is literally true but practically false (we do know a lot that has been demonstrated to generalize across decision contexts).

In their response to my review, the authors reiterated that their work aligns to the process of a systematic review. Such was not actually my concern, and I agree that their work does in fact follow that process. My concern is that, in this context, a true systematic review is less valuable to the field and less useful to practitioners than a review that is more integrative.

The issue is actually simple: I fully agree that actual clinical trial participation decisions are different in some important ways from actual treatment decisions. I am quite skeptical that _hypothetical_ clinical trial decisions (of the type studied in the reviewed papers here) are actually that different from _hypothetical_ treatment decisions because many of the differences in real-world mindsets are gone, replaced by the common experience of reading hypothetical scenarios. More importantly, I fundamentally disagree with the idea that the process of _communicating_ _risks_ is so categorically different in clinical trial contexts that the exclusion of all other studies, and the resulting conclusion that little is known about how do risk communication well in this context, is justified. The authors do not provide a clear argument how the comprehension task, the interpretation task, and/or the integration tasks associated with receiving and using risk information in this context are different from those same tasks in, say, a decision support tool for treatment or screening decision making.

The authors have made several relatively minor revisions that acknowledge that "lessons from effective risk communication in a treatment and/or screening context can provide examples of best practice that could be used for those developing PILs for patients considering clinical trial participation" and that future research could use "the best practice examples from treatment decision making as a starter" They also now refer to one of a number of relevant risk communication reviews to begin orienting readers to that literature. I thank the authors for these steps, as they do mitigate my concerns somewhat. But, I am not sure that they are enough to offset the primary conclusions (as still stated in the abstract) that "there is as yet no clear optimal method for improving participant understanding," despite lots of evidence from other non-included studies that certain approaches are likely to be more effective than others.

My preference would be for an option that I suspect the authors (and perhaps the journal) would reject: To create a kind of hybrid paper, one that does the systematic identification of papers and notes the dearth of relevant empirical evidence (which, to me, is the primary value of this paper) but then complements it with a much more detailed and somewhat speculative discussion of _specific_ principles of good risk communication evident in the broader literature that are relevant in clinical trial contexts. This type of paper could be quite valuable to many audiences in ways that I fear that the current manuscript will not be.

In net, I don't have much in terms of small comments. The authors have been generally responsive to the detailed comments provided by other reviewers, and they have taken some steps towards addressing my own comments. I just remain quite concerned that the takeaway message of this paper is not particularly helpful (and may be harmful) in guiding improvements in risk communications in clinical trial documents.

Reviewer #2: I read the review with better comprehension this time, thanks for working to organize and clarify it.

There are some verbal expressions that confuse me. It may be that the UK and the US are “two countries divided by a common language”. One is the characterization of the studies as having been “nested in” or “hosted in” or even “within” a hypothetical trial. From reading the paper the situation is just that the studies were done with a cover story or situation (a trial for a disease that the participant does not actually have) and the information about the situation is varied. The “hosting” and “nesting” and “within” convey that there is something bigger going on, say a study with multiple interviews over time, and that an incidental add on is that on one occasion among the 4 questionnaires filled out that day is the booklet that contains the study of risk communication. But I don’t find any evidence for this.

Is this part of the way people in the UK talk is it an idiosyncratic choice of words just this once? If the latter, it is confusing for this reader and it seems to add unnecessary work to the task of reading the review. There was some complaint about this last round, and the promise in the author’s response that it had been fixed, but it does not seem to have been addressed.

On the presentation of the reviewed studies’ findings, in the Results section. I note that when comparisons are made, the statistic presented is the p value. Only a few times was there an effect presented (X% versis Y%). Quite possibly those numbers were present in the previous version and have been removed because of the complaint of “too much unreadable information”. So I need to be careful here.

I know that the social science statisticians, such as the American Psychological Society (to which the authors need not have any loyalty), ask for an effect, a degree of freedom or N, as well as the p value. And then there are the new standards of presenting the effect estimate and confidence intervals. But these are for primary studies. For reviews, however, the same general ideas are still relevant for the reader to understand how much weight to put on a study.

Degrees of freedom. While the text need not have a df next to every p, how about when the study is first described, saying how many people were in it, in each group? Then we can know if it was 10 or 1000, and form opinions accordingly.

Effect size. Of course there is the issue of statistical significance compared with clinical or practical significance. If we knew the N was huge and the effect was p = .05, we might suspect the effect is actually small and of little practical importance (in the absence of the text giving the effect size). But really, for at least half of the “(p < .xxx)” statements, providing a simple idea of what the compared means were might be a great help. If probabilities, we know what those scales mean. Sometimes if it really would help understand the effect, you might need to explain the scale.

Particular suggestions.

Abstract. “participants” needs to be plural possessive in this: “studies that aimed to support potential trial participants decision”.

P 4. First para, I suggest changing “should include explanations of the ‘reasonably foreseeable risks or inconveniences’ and expected benefits, and where there are no clinical benefits to the participants, they must be made aware of this” to “should include explanations of the ‘reasonably foreseeable risks or inconveniences’, expected benefits, and where there are no clinical benefits to the participants, they must be made aware of this”

P 5, line 78. “Decision” should be plural.

I suggest changing “Conditional altruism is the concept that participation in the trial will benefit society but there is a need for a benefit (which is influenced by perception of risk) for self to be tied in” to “Conditional altruism is the concept that participation in the trial will benefit society but there must be a benefit (which is influenced by perception of risk) for self”.

Suggestion: change clumsy punctuation from “trade-offs between risk and benefit in a trial involve; layers of complexity in addition to those for treatment such as: loss of control over which treatment they receive; and potentially greater uncertainties as often participants have to consider the risks and benefits of a minimum of two competing treatments” to “trade-offs between risk and benefit in a trial involve layers of complexity in addition to those for treatment, such as loss of control over which treatment they receive and potentially greater uncertainties, as often participants have to consider the risks and benefits of a minimum of two competing treatments”

P 6, line 101. Add “that” after “highlighted”.

Line 121 “within information presented” These words seem redundant. Can they simply be deleted?

P7, line 126. Add “studies of” after “include”.

Lines 130-131. What does this mean? I don’t understand what kinds of studies were excluded. “More eStudies investigating participants perceptions of receiving risk communication…”

Is an “eStudy” a thing? Why exclude it?

Also, “participants” needs to be plural possessive.

I understand that you consider a “Senior Information Officer” to be distinct from a “librarian”, as stated in response letter. But simply leaving it as such does not solve the problem that for people in different university systems, there is no such thing. Could you explain it in a phrase? Is it like a police officer, the “fact police”. That would be interesting. In my area, “public information officers” are public relations flacks, and then there are “information systems security officers”, but I would not go to either one for help with a library search.

P 9. Line 183. “…the studies asked participants to imagine they were being recruited into a host clinical trial, provided brief information about the host trial (such as clinical population, intervention, comparator, outcomes, etc), then provided various formats of risk communication (such as verbal or numerical descriptors) and assessed relevant outcomes.” “assessed relevant outcomes” is ambiguous. Is this part of the participant’s task (to assess relevant outcomes that are described), or is “assessed” an adjective describing a part of the materials that the participants are shown; where someone else has assessed the outcomes (probably verbally, versus numerically) and the participant looks at it and does some other judgment (e.g, whether to sign up for the trial). I think it is the latter. If so, changing “assessed” to “assessments of” would make it clearer.

Line 186. An example of my problem with “nested in”. How about “in the context of” or even just “about”?

In 190, again, “set within a trial” just means that the context of the vignette was a non drug intervention rather than a drug.

P 10. Line 198, explanation of structure of upcoming review. Again, “embedded” is a confusing word that could simply be dropped!

I did not know whether the two groupings --first design, then content of intervention -- were nested (in the statistical sense) or crossed. It turns out that the “review” section groups them by design with intervention nested within; And then the discussion covers them again, but this time by intervention (all designs). The material is readable and accessible, when actually reading the two sections. This “heads up” could be less confusing if it dispensed with the “first, then” and just said the “The final seven studies were grouped according to the study design (RCT or prospective cohort) and the topic of the described intervention: ‘probability presentation’ (22, 23 ), ‘risk framing’ (24, 25, 26, 27), and ‘risk interpretation’ (28).”

Lines 216 to 219. There are three instances where an extra “9” has been inserted.

Line 216. The list might be changed from “…also perceived the risk to health to be higher (9p<0.0001), benefit to be lower (9 p=0.03), and were…” to ““…also perceived the risk to health to be higher (9p<0.0001) and the benefit to be lower (9 p=0.03), and were…”

Line 232. There is an extra “,” in the “p <” parenthesis. I think this indicates that you went through and removed them all. Sorry for my contradictory advice.

P 12, line 260. Could you add a comma after “pain” in “trial testing two drugs for post-operative pain one a standard treatment and the other”?

Line 261. There is an orphan “t”.

Line 265. Comma after “risk reduction”

Line 266. Change “parent show” to “parents who”. Autocorrect sometimes errs.

Line 271. Add “to be” before “more likely to agree”.

The Tait study seemed to manipulate whether disadvantageous information was presented or hidden. Some of the other studies manipulate how disadvantageous (and advantageous) info is described, framed, but they present essentially the same in every condition. Although you touch on this briefly in the discussion (pp 17-18), it seems to be a very important distinction between the kinds of manipulation in these studies. The implications of the findings are different. When you find the (obvious) that hiding the disadvantages makes people more likely to choose an option, that informs the researcher/sponsor’s decision whether to lie and obfuscate in a propagandistic marketing manner. When you find that the same information, framed differently, influences option choice, that informs the researcher/sponsor’s decision whether to “nudge” in a libertarian paternalistic manner https://en.wikipedia.org/wiki/Libertarian_paternalism.

P 14, line 299. “Participants were presented with information in one of six combinations in a pre-specified order.” First, you might put “the” before “six,” since you have already described the 2 (order) by 3 (content) design. But I thought that each person got only one condition? What is the function of “in a prespecified order” in this sentence? If it is about the two sequences, you have already incorporated it. Is there a different order? Of course there is a “procedure” that has a lot of steps in order: meet, qualify, consent, instruct, and so on. Usually that need not be said. Could you drop “in a pre-specified order”?

Line 308. You could drop the conversational “When it came to presentation preferences” and the reader would still know exactly what you are talking about.

Line 318 add “and” between “risk” and “benefit”.

Line 319. Add “noted only” after 27%.

P 18, line 408, change punctuation and one word to “An analysis of PILs used in clinical trials by Gillies et al (2011) found that explaining trial processes, presenting probabilities, and expressing values were consistently poor across all PILs when assessed using an informed consent evaluation instrument”.

P 19. Lines 448-449, change “indicates that there is as yet not clear optimal method for improving participant understanding, or clear consensus on how” to “indicates that there is as yet not a clear optimal method for improving participant understanding, nor a clear consensus on how”

Line 451, change “within trials” to “about trials” or “during trials”? “within” evokes my ANOVA framework.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: Yes: Robert M. Hamm

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

14 Sep 2020

Please see uploaded response letter which contains a detailed response to reviewers comments.

Submitted filename: PONED1919791_Response to reviewers_140920.docx

Click here for additional data file.

14 Oct 2020

PONE-D-19-19791R2

A systematic review of risk communication in clinical trials: how does it influence decisions to participate and what are the best methods to improve understanding in a trial context?

PLOS ONE

Dear Dr. Gillies,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

The reviewers feel that the paper is much improved but they still have a few minor issues that need to be addressed.

Please submit your revised manuscript by Nov 28 2020 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

We look forward to receiving your revised manuscript.

Kind regards,

Dermot Cox

Academic Editor

PLOS ONE

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: (No Response)

Reviewer #2: (No Response)

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Partly

Reviewer #2: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: N/A

Reviewer #2: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: I thank the authors for adding the section towards the end of the paper alluding to the fact that other risk communication guidance does exist that plausibly informs what can be done in the clinical trial contexts.

One concern: The authors need to be careful in their language to draw distinctions between what is known about risk communication in clinical trials (this work) vs. risk communication in general, and within the latter, between risk communication intended to motivate vs. risk communication intended to inform. This latter distinction is often conflated in the literature, with problematic results.

For example, the authors cite a NICE report (reference 40). But their presentation notes that the outcome used to guide this review was "effective strategies for improving health behavior outcomes" -- in other words persuasion. Such is confirmed later on when it is noted that a strategy "was more persuasive for adoption of certain behaviors." This criteria is precisely NOT what one wants to use to evaluate strategies for use in clinical trial contexts where persuasion would be ethically inappropriate. Thus, I recommend removal of this paper and its recommendations from this analysis. References 41/42/43, by contrast, focus on more analogous medical risk communication contexts and are more appropriately included.

On a related note, later in that section, the authors summarize stating "In summary, whilst there may be a paucity of high quality evidence to underpin decisions about effective risk communication many of the good practice recommendations developed through empirical research provide sensible frameworks to promote informed choices, enable good quality decision making, and are unlikely to cause significant harm." This first phrase needs to be modified to focus on clinical trial contexts, as I would not characterize the general risk communication field as having a "paucity of high quality evidence", only evidence that aligns tightly with the clinical trial context. Thus, rewording to state "decisions about effective risk communication IN CLINICAL TRIAL CONTEXTS, many of the ..." would more accurately reflect the state of the field.

Reviewer #2: Thanks for careful edits and explanations. I have only minor wording suggestions.

Line 42, “potential for learning from the evidence on risk communication from treatment and screening,” change “from” to “in”.

Line 132 or 133, need a space between the sentence.

Line 166, “the host trial (i.e. the trial the potential participants were begin asked to consider participation in)”, that should be “being” not “begin”.

Lines 2016 to 218 “Those who received only verbal descriptors were significantly less satisfied

217 with the information than those who also had the numerical values when asked to rate on a scale from

218 1 to 6 (p=0.03).”

This is slightly ambiguous. Are the 1 to 6 numbers the “numerical values”? Of course not. But it would be clearer rearranged to “When asked to rate on a scale from 1 to 6 (p=0.03), those who received only verbal descriptors were significantly less satisfied with the information than those who also had the numerical values.”

Lines 286-287. I suggest changing “and the ‘risk’ group involved procedures

287 that were described as having a high risk of injury” to “and for the ‘risk’ group procedures

287 were described as having a high risk of injury”

Line 401. “decisions in ways that are not consistent with the individuals values and preferences”. Individuals should be possessive. I’m not sure if singular or plural.

439 and 440. Change “whilst the evidence on the other strategies was less clear it was not considered to be harmful.” To “whilst the evidence on the other strategies was less clear, they were not considered to be harmful.”

441. “non-numeric, visual, and statistical formats the evidence as inconsistent with regard to how” was, not as.

447-448. “Interestingly, this review showed that presenting absolute risk

448 reduction were better at maximising accuracy” was, not were.

475 “high quality evidence to underpin decisions about effective risk communication many of the good” put a comma after “communication”.

482-483 “The low number of studies included for review means it is difficult to confidently make far reaching

483 recommendations based on the findings” Change “as yet not” to “as yet no” or to “as yet not a”.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: Yes: Robert M Hamm

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

16 Oct 2020

Reviewer #1:

1 I thank the authors for adding the section towards the end of the paper alluding to the fact that other risk communication guidance does exist that plausibly informs what can be done in the clinical trial contexts.

One concern: The authors need to be careful in their language to draw distinctions between what is known about risk communication in clinical trials (this work) vs. risk communication in general, and within the latter, between risk communication intended to motivate vs. risk communication intended to inform. This latter distinction is often conflated in the literature, with problematic results.

For example, the authors cite a NICE report (reference 40). But their presentation notes that the outcome used to guide this review was "effective strategies for improving health behavior outcomes" -- in other words persuasion. Such is confirmed later on when it is noted that a strategy "was more persuasive for adoption of certain behaviors." This criteria is precisely NOT what one wants to use to evaluate strategies for use in clinical trial contexts where persuasion would be ethically inappropriate. Thus, I recommend removal of this paper and its recommendations from this analysis. References 41/42/43, by contrast, focus on more analogous medical risk communication contexts and are more appropriately included.

Response

The reviewer raises a very important point abut the different intended purposes of various forms of risk communication. On refection we also agree that including the NICE report we identified is misleading as it has a basic assumption that the intended purpose of the communication is to persuade or modify behaviour. As the reviewer highlights, this of course may not be appropriate when considering decision about participation in a clinical trial. Therefore we have removed this section of text. See edits page 19 lines 435-447.

2 On a related note, later in that section, the authors summarize stating "In summary, whilst there may be a paucity of high quality evidence to underpin decisions about effective risk communication many of the good practice recommendations developed through empirical research provide sensible frameworks to promote informed choices, enable good quality decision making, and are unlikely to cause significant harm." This first phrase needs to be modified to focus on clinical trial contexts, as I would not characterize the general risk communication field as having a "paucity of high quality evidence", only evidence that aligns tightly with the clinical trial context. Thus, rewording to state "decisions about effective risk communication IN CLINICAL TRIAL CONTEXTS, many of the ..." would more accurately reflect the state of the field.

Response

Edited accordingly. See page 20 line 477.

Reviewer #2

3 Line 42, “potential for learning from the evidence on risk communication from treatment and screening,” change “from” to “in”.

Response

Edited accordingly

4 Line 132 or 133, need a space between the sentence.

Response

Edited accordingly

5 Line 166, “the host trial (i.e. the trial the potential participants were begin asked to consider participation in)”, that should be “being” not “begin”.

Response

Edited accordingly

6 Lines 2016 to 218 “Those who received only verbal descriptors were significantly less satisfied

217 with the information than those who also had the numerical values when asked to rate on a scale from

218 1 to 6 (p=0.03).”

This is slightly ambiguous. Are the 1 to 6 numbers the “numerical values”? Of course not. But it would be clearer rearranged to “When asked to rate on a scale from 1 to 6 (p=0.03), those who received only verbal descriptors were significantly less satisfied with the information than those who also had the numerical values.”

Response

Edited accordingly

7 Lines 286-287. I suggest changing “and the ‘risk’ group involved procedures

287 that were described as having a high risk of injury” to “and for the ‘risk’ group procedures

287 were described as having a high risk of injury”

Response

Edited accordingly

8 Line 401. “decisions in ways that are not consistent with the individuals values and preferences”. Individuals should be possessive. I’m not sure if singular or plural.

9 439 and 440. Change “whilst the evidence on the other strategies was less clear it was not considered to be harmful.” To “whilst the evidence on the other strategies was less clear, they were not considered to be harmful.”

Response

In response to reviewer 1’s comment this section has now been deleted.

10 441. “non-numeric, visual, and statistical formats the evidence as inconsistent with regard to how” was, not as.

Response

Edited accordingly

11 447-448. “Interestingly, this review showed that presenting absolute risk

448 reduction were better at maximising accuracy” was, not were.

Response

Edited accordingly

12 475 “high quality evidence to underpin decisions about effective risk communication many of the good” put a comma after “communication”.

Response

Edited accordingly

13 482-483 “The low number of studies included for review means it is difficult to confidently make far reaching

483 recommendations based on the findings” Change “as yet not” to “as yet no” or to “as yet not a”.

Response

Edited accordingly. However, we believe the section the reviewer recommended editing was on line 496 not lines 482-483.

30 Oct 2020

A systematic review of risk communication in clinical trials: how does it influence decisions to participate and what are the best methods to improve understanding in a trial context?

PONE-D-19-19791R3

Dear Dr. Gillies,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Dermot Cox

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

5 Nov 2020

PONE-D-19-19791R3

A systematic review of risk communication in clinical trials: how does it influence decisions to participate and what are the best methods to improve understanding in a trial context?

Dear Dr. Gillies:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Dermot Cox

Academic Editor

PLOS ONE