An optimised patient information sheet did not significantly increase recruitment or retention in a falls prevention study: an embedded randomised recruitment trial

Background Randomised controlled trials are generally regarded as the ‘gold standard’ experimental design to determine the effectiveness of an intervention. Unfortunately, many trials either fail to recruit sufficient numbers of participants, or recruitment takes longer than anticipated. The current embedded trial evaluates the effectiveness of optimised patient information sheets on recruitment of participants in a falls prevention trial. Methods A three-arm, embedded randomised methodology trial was conducted within the National Institute for Health Research-funded REducing Falls with ORthoses and a Multifaceted podiatry intervention (REFORM) cohort randomised controlled trial. Routine National Health Service podiatry patients over the age of 65 were randomised to receive either the control patient information sheet (PIS) for the host trial or one of two optimised versions, a bespoke user-tested PIS or a template-developed PIS. The primary outcome was the proportion of patients in each group who went on to be randomised to the host trial. Results Six thousand and nine hundred patients were randomised 1:1:1 into the embedded trial. A total of 193 (2.8%) went on to be randomised into the main REFORM trial (control n = 62, template-developed n = 68; bespoke user-tested n = 63). Information sheet allocation did not improve recruitment to the trial (odds ratios for the three pairwise comparisons: template vs control 1.10 (95% CI 0.77–1.56, p = 0.60); user-tested vs control 1.01 (95% CI 0.71–1.45, p = 0.94); and user-tested vs template 0.92 (95% CI 0.65–1.31, p = 0.65)). Conclusions This embedded methodology trial has demonstrated limited evidence as to the benefit of using optimised information materials on recruitment and retention rates in the REFORM study. Trial registration International Standard Randomised Controlled Trials Number registry, ISRCTN68240461. Registered on 01 July 2011. Electronic supplementary material The online version of this article (doi:10.1186/s13063-017-1797-7) contains supplementary material, which is available to authorized users.


Background
Randomised controlled trials (RCTs) are generally regarded as the 'gold standard' experimental design to determine the effectiveness of an intervention. Unfortunately, many trials either fail to recruit sufficient numbers of participants, or recruitment takes longer than anticipated [1]. Such difficulties not only have an impact on the power and external validity of the study's findings but may also increase the overall financial cost. Many trials implement methods to improve recruitment and/or retention such as the use of monetary incentives. There is, however, limited evidence as to the effectiveness of recruitment strategies in healthcare settings [2], and as a result, the UK Medical Research Council funded the Systematic Techniques for Assisting Recruitment to Trials (MRC START) research programme [3]. The aim of this programme was to "improve the evidence-base of recruitment to trials, enhance recruitment rates and make research more accessible to the public." One key objective of the project was to develop interventions to improve recruitment to trials and test them in embedded trials within ongoing 'host' trials. The initial focus was on testing optimised patient information sheets and multimedia resources.
Patient information sheets (PISs) are always given to potential trial participants along with verbal information as part of the informed consent process. A trial PIS has to be reviewed and approved by an ethics committee or an internal review body; nonetheless, there remain longstanding concerns regarding their length, complexity and the level of literacy skills required to understand the information, all of which may have a negative impact on recruitment [4]. Indeed, a recent sample of 20 PISs from recently completed or ongoing RCTs were found to be lengthy (mean word count 1853, standard deviation = 960) and lacking information important to making an informed decision about trial participation [5]. However, Brierley et al. found that reducing the length of the PIS did not influence recruitment and actually yielded more ineligible patients [6]. Therefore, simply producing a shorter PIS is ineffective, and the content and interpretability of the information presented must be retained. One possible approach to improve the quality and appearance of PISs is to develop their content through a process of re-writing (for a lay audience), re-organisation, enhancing their appearance through input from a graphic designer and involving user testing [7,8]. The use of a professionally designed information pack was seen to have a small positive effect on initial response rates relative to a 'standard' pack (2.7% difference in response rates, 95% confidence interval (CI) −0.06 to 5.5%, p = 0.06) in an RCT embedded in an Avon Longitudinal Study of Parents and Children (ALSPAC) cohort [9]. However, undertaking commercial user testing and incorporating graphic design input into PISs is costly, especially if the process needs to be completed for each trial; therefore, it has been suggested that a 'bank' of template PISs could be developed (which have been re-written and have undergone bespoke user testing and graphic design) in different populations. Researchers could select the most appropriate template on which to base their trial PIS according to their particular study population.
Embedding recruitment trials into ongoing trials is an efficient method to test recruitment strategies, and to help fill the knowledge gap with minimal resource use [10]. Table 1 presents a Consolidated Standards of Reporting Trials (CONSORT) checklist for reporting embedded recruitment trials. The REFORM study team were invited by the MRC START programme to embed a recruitment methodology trial within the REducing Falls with ORthoses and a Multifaceted podiatry intervention (REFORM) study. The REFORM study was selected, as the aims of the MRC START programme were compatible with both the need to maximise recruitment to the REFORM trial and the ethos of the York Trials Unit (YTU) to develop an understanding of effective recruitment interventions. The aim of the designed three-arm embedded trial was to determine if the number of participants recruited, randomised and retained to the REFORM trial could be improved by the use of both a bespoke usertested and a template-developed optimised PIS.

Ethics approval
This trial was embedded within the National Institute for Health Research (NIHR)-funded REFORM study [11], which aims to evaluate the clinical and cost effectiveness of a podiatry intervention for the prevention of falls in older people. Ethical approval for the REFORM study was given by NRES East of England -Cambridge East Research Ethics Committee and the University of York, Department of Health Sciences Research Governance Committee. Ethical approval for the PIS embedded methodology trial was given via a substantial amendment to the same committees. The embedded methodology trial is registered as a substudy to REFORM (ISRCTN68240461).

Participant recruitment
REFORM is a cohort RCT [12] in which patients were initially recruited to an observational cohort before potentially being randomised into an RCT. Electronic medical notes at participating National Health Service (NHS) podiatry clinics were searched to identify communitydwelling patients eligible for the REFORM observational cohort (i.e. those who were over the age of 65 and had attended a routine podiatry appointment within the past 6 months). Potentially eligible participants were mailed an invitation pack (letter of invitation, one of the three  Mechanism used to implement the random allocation sequence (such as sequentially numbered containers), describing any steps taken to conceal the sequence until interventions were assigned Mechanism used in the embedded recruitment trial to implement the random allocation sequence (such as sequentially numbered containers), describing any steps taken to conceal the sequence until interventions were assigned Implementation 10 Who generated the random allocation sequence, who enrolled participants and who assigned participants to interventions Who generated the random allocation sequence(s), who enrolled participants and who assigned participants to embedded recruitment interventions Blinding 11a If done, who was blinded after assignment to interventions (for example, participants, care providers, those assessing outcomes) and how If done, who was blinded after assignment to embedded recruitment interventions (for example, participants, care providers, those assessing outcomes) and how For each primary and secondary outcome, results for each group in the embedded recruitment trial, and the estimated effect size and its precision (such as 95% confidence interval) PISs being evaluated in this embedded trial, consent form, screening questionnaire and prepaid envelope) asking whether they would like to participate in the RE-FORM study. The information and consent process covered all aspects of the cohort and trial, and so only one PIS was seen by participants whether they went on to be randomised into the REFORM trial or not. Participants who returned their completed consent and screening form to the YTU by post were assessed by researchers at the YTU for eligibility. Participants were ineligible if they reported neuropathy, dementia or another neurological condition such as Parkinson's or Alzheimer's disease; were unable to walk household distances without the help of a walking aid; or had a lower limb amputation or were unwilling to attend their local podiatry clinic. Eligible participants were sent a baseline questionnaire and monthly falls calendars. Participants who returned a baseline questionnaire and at least one monthly falls calendar (recording falls in the previous month) were included in the observational cohort. Those cohort participants who had either had one fall in the past 12 months or one fall in the past 24 months requiring hospital attention, or who reported a fear of falling on the baseline questionnaire (worried about falling at least some of the time in the past 4 weeks) were then eligible for randomisation to the REFORM trial. During the course of the trial, cohort participants who went on to have a fall could then become eligible for the RE-FORM trial. In order to minimise the delay between randomisation and the podiatrist seeing the participants in At the point at which this embedded trial was nested within the host trial, the next podiatry clinics due to begin recruitment were Harrogate, Leeds, Scarborough, Selby and Sheffield. All patients due to be sent a RE-FORM invitation pack from these clinics were randomised to one of the three arms in the embedded methodology trial which determined which format of PIS they were sent.

Randomisation and blinding
This was an embedded, individually randomised, recruitment trial. A list of patients was generated and ordered by NHS number. Each patient was assigned a unique identification number. An independent data manager at the YTU, who was not involved in the recruitment of participants, generated the allocation sequence for the embedded methodology trial electronically. Randomisation was stratified by centre, using a single large block per site, which corresponded to its total sample size. Participants were allocated 1:1:1 to receive, in their invitation pack, either (1) the control PIS and control invitation letter; or (2) an optimised version of the PIS and invitation letter developed through bespoke user testing; or (3) an optimised template-developed PIS and the original invitation letter. Patients were then sent the allocated invitation pack by members of the research team based at the University of York. The researchers, patients and podiatrists were blind to the allocation. Patients who received the information sheet were unaware that they had been randomised to receive different information leaflets.

Control group
The 'control' PIS was the original PIS developed for the REFORM trial. It was written in accordance with the standard National Research Ethics Service template available at the time at which the study was set up. It underwent several revisions by the research team and was five pages of A4 paper in length. The invitation letter was one A4 page on NHS trust headed paper. The control PIS is shown in Additional file 1.

Intervention group: 'bespoke user-tested' PIS
The 'bespoke user-tested' PIS was developed in the following way: it was re-written for a lay audience (led by author PK), re-organised into eight subsections, had graphic design input and underwent user testing. The graphic design and user testing were both undertaken by commercial companies. Staff members who had considerable experience in writing patient material and a graphic designer revised the content and layout of the control PIS. A contents list was placed on the front page along with a list of key points about the study, with the main text divided into eight sections. Sentences and paragraphs were shortened and bullet points used for lists. The PIS then underwent three rounds of user testing, each involving 10 members of the public with limited podiatry-or trial-specific knowledge. During the usertesting process, the users (a group demographically similar to the REFORM trial target population but with no history of falls) were asked to find key pieces of information in the PIS relating to: the nature and purpose of the trial; the process and meaning of consent; and trial procedures and safety issues. Questions were arranged in such a way to ensure they did not match the order within the PIS. The final version of the PIS was printed on one sheet of double-sided A3 paper and folded to form a booklet. The control one-page invitation letter was revised and underwent user testing alongside the PIS, resulting in shortened text and the addition of bullet points to clarify what to do if the participant wished to take part. The user-tested PIS is shown in Additional file 2.

Intervention group: 'template-developed' PIS
The 'template-developed' PIS was written by three research fellows with more than 12 years' experience of recruiting patients to a range of randomised trials and who were currently working on the REFORM study. The content, layout and style were revised using a previously bespoke usertested PIS designed for an earlier trial conducted in a similar aged population [13]. In this process, the control PIS text was shortened and divided into seven sections, a contents list included and bullet points added to clarify what the participant should do if he/she wished to take part. A green colour scheme was used to match the trial logo. The PIS was reviewed and revised further by two other experienced researchers. The Public and Patient Involvement (PPI) group reviewed this PIS and gave feedback about its readability, layout, font size and content; however, no graphic design input or formal user testing was undertaken. The revised version was printed on one sheet of double-sided A3 paper and folded to form a booklet. Participants in the templatedeveloped PIS group were sent the same patient invitation letter as the control group. The template-developed PIS is shown in Additional file 3.

Primary outcome
The primary outcome was the proportion of patients in each group who went on to be randomised into the REFORM trial.

Secondary outcomes
Secondary outcomes were as follows: Proportion of patients in each group who were recruited into the REFORM cohort Proportion of patients retained in the trial at 3 months post randomisation defined as returning at least the first 3 months' worth of falls calendars from the date of randomisation.

Sample size
As is usual with an embedded trial within a trial, no formal power calculation was undertaken for this embedded methodology trial, but the MRC START programme set out to include trials which planned to approach enough potential participants to allow 400 potential recruits to receive each recruitment intervention. Five centres were involved in the substudy; thus, the sample was constrained to only include participants eligible for the REFORM study who were due to be mailed a recruitment pack from these centres.

Statistical analysis
The proportion of participants who: returned a consent form; were recruited into the cohort; were randomised into the main trial; and were retained in the trial is presented for the three groups. Odds ratios (ORs) and 95% confidence intervals (CIs) for each of the three pairwise comparisons were obtained from mixed logistic regression models adjusting for PIS allocation as a fixed effect and trial centre as a random effect. Analysis was conducted in Stata v13, using two-sided tests at the 5% significance level and based on the principles of intention to treat.
To account for the participants in the embedded trial of the newsletter who overlapped with participants in this trial, sensitivity analyses were conducted including a covariate for receiving a newsletter in all logistic regressions.

Randomised to REFORM trial
Eligible, consenting participants were then sent a baseline questionnaire (control n = 145 (6.3%); templatedeveloped n = 158 (6.9%); bespoke user-tested n = 160 (7.0%)). A baseline form and at least one falls calendar were returned by 134 (5.8%) individuals in the control group, 149 (6.5%) in the template-developed group and 149 (6.5%) in the bespoke user-tested group. These participants were recruited to the cohort of potentially eligible participants.
Of these cohort patients, 161 were immediately eligible for randomisation into REFORM, and of these, 122 could be randomised as there was sufficient capacity in the clinics to see them. A further 71 patients in the cohort became eligible for the trial as they subsequently reported a fall and were able to be seen in clinic. Therefore, a total of 193 of the 6900 potential participants (2.8%) went on to be randomised into the main RE-FORM trial (control n = 62 (2.7%, 95% CI 2.0-3.4%); template-developed n = 68 (3.0%, 95% CI 2.3-3.6%); bespoke user-tested n = 63 (2.7%, 95% CI 2.1-3.4%)). The difference in percentages and their 95% CIs are as follows: for template-developed vs controls there was a 0.3 percentage point difference (95% CI −0.7 to 1.2, p = 0.60); bespoke user-tested vs control 0.0 (95% CI −0.9 to 1.0, p = 0.93); and bespoke user-tested vs templatedeveloped −0.2 (95% CI −1.2 to 0.7, p = 0.66).

Sensitivity analyses
Three hundred and thirty four participants also received a newsletter in their recruitment pack (control n = 120 (5.2%); template-developed n = 112 (4.9%); bespoke usertested n = 102 (4.4%)). In the sensitivity analyses, only a negligible difference in the estimates, their associated 95% CIs and p values was observed.

Discussion
We have evaluated the effectiveness of two optimised patient information sheets (PISs) relative to the standard control version developed using the NRES template PIS to increase recruitment to the REFORM trial. Previous studies investigating the method of delivering paperbased information to patients have found that reducing the length of the PIS or supplementing the PIS with a booklet on clinical trials has little or no impact on trial recruitment, but that the use of a professionally designed information pack may be effective [6,9,14]. Alternative strategies could be evaluated in future embedded trials. Bower et al. [10] have identified three areas which merit further investigation. These include training site staff, communication with patients and incentives. The key finding of this study is that the use of enhanced PISs did not significantly increase recruitment or retention. The overall recruitment rate to the REFORM trial (2.8%) and the REFORM cohort (6.3%) was low, so it may be that there was insufficient power to detect a difference. The FES-I scored from 7-28, higher score indicates greater concern about the possibility of falling; FAI scored from 15-60, higher score indicates greater activity MRC START group will undertake a meta-analysis of six embedded trials evaluating enhanced information sheets which will have more power to provide evidence of effectiveness. Small differences in recruitment rates could make a large difference to trials similar to REFORM, which required a sample size of nearly 900 participants. For example, a 3% uptake would require 30,000 invitations, but if uptake to the study increased to 4%, then 7500 fewer recruitment packs would be required, saving up to £35,000 in packing and postage costs alone. This would be a significant saving, even after incorporating the cost of professional graphic design and bespoke user testing. In this study, there were no statistically significant differences between the bespoke user-tested and template-developed information sheets, so it may be beneficial to develop a bank of template information sheets rather than each trial having to take account of the additional time and expense of professional user testing. Alternatively, since many trials include PPI, a more structured approach to reviewing and developing the information sheet could be evaluated and then, if effective, adopted within PPI groups. One reason why there may have been little evidence of a difference is that the 'standard' PIS has been developed by experienced researchers, and consequently it may be difficult to achieve a significantly improved information sheet. It could also be that the optimised PISs increased the interpretably and clarity of the provided information, which deterred some patients from responding. Unlike the other host trials in the START programme, REFORM used a cohort randomised controlled trial design. Participants were recruited first to the observational cohort, and then eligible participants were randomised to the REFORM trial. Therefore, optimised PISs were more likely to have an effect on recruitment to the observational cohort than randomisation into the REFORM trial. As there was a lack of effect on recruitment to the observational cohort, it was unlikely that there would be an impact on recruitment or retention in the trial.
There are some potential limitations to this study. First, the results of this study are applicable only to those participants over the age of 65 years attending routine podiatry clinics. Further studies should substantiate the study results in other populations. Second, due to the nature of the intervention, it was not possible to blind administrative staff who mailed out the recruitment packs, but it is unlikely that allocation subversion could have taken place. Whilst participants were aware of which PIS they received, they were unaware that an embedded trial was being conducted and that recruitment to the REFORM study was being monitored. Finally, it was not possible to evaluate whether the enhanced information sheets affected the speed of recruitment, as podiatry clinics were limited to the number of participants they could see at any one time.

Conclusions
Optimised participant information sheets are one of a number of recruitment interventions amenable to testing through embedded trial methodology. Whilst the results of this study suggest limited benefits to enhancing patient information sheets, the findings add to the body of evidence around the effectiveness of recruitment strategies and may potentially help save time and money in future trials. The results of this study and the other embedded studies for the MRC START programme are to be incorporated into a meta-analysis which will provide more robust evidence as to the effectiveness of these types of recruitment strategies.