The OMACS-PIL study - a randomised controlled trial within the OMACS observational study

Background: There has been little research to investigate whether the appearance of paper Patient Information Leaflets (PILs) used to describe research studies to potential participants influences their decision to take part. Embedding a Study Within a Trial (SWAT) is an efficient way of answering this type of methodological question. We included a randomised SWAT within a large cohort study, Outcome monitoring After Cardiac Surgery (OMACS), to address this question. Methods: Potential participants for the OMACS study were randomised to receive one of three PILs, identical in content but with varying formatting and use of colour: PIL A (enhanced format), PIL B (hybrid format) and PIL C (standard format). Consent to OMACS was the primary outcome. Consent rates using the three different PIL formats were collected and compared. Qualitative feedback on the different formats was obtained from a Public and Patient Involvement (PPI) group. Results: For the SWAT, 1517 PILs were sent to potential participants and 640 (42%) consented to take part in OMACS. PIL B had the highest recruitment rate, with 45% of patients consenting to participation; 40% and 41% of patients consented to participation with PIL A and C respectively. Compared to PIL C, the consent rate was 4% higher with PIL B (45% versus 41%, 95% Confidence Interval (CI) -2% to +10%, p=0.16) and 1% lower with PIL A (40% versus 41%, 95% CI -7% to +5%, p=0.72). Conclusions: Consent rates were similar for all 3 PIL formats. PIL B is being used for the remainder of the host study and will be used to inform the design of PILs for other research studies as it was the preferred format of the PPI group.


Abstract
Background: There has been little research to investigate whether the appearance of paper Patient Information Leaflets (PILs) used to describe research studies to potential participants influences their decision to take part. Embedding a Study Within a Trial (SWAT) is an efficient way of answering this type of methodological question. We included a randomised SWAT within a large cohort study, Outcome monitoring After Cardiac Surgery (OMACS), to address this question. Methods: Potential participants for the OMACS study were randomised to receive one of three PILs, identical in content but with varying formatting and use of colour: PIL A (enhanced format), PIL B (hybrid format) and PIL C (standard format). Consent to OMACS was the primary outcome. Consent rates using the three different PIL formats were collected and compared. Qualitative feedback on the different formats was obtained from a Public and Patient Involvement (PPI) group. Results: For the SWAT, 1517 PILs were sent to potential participants and 640 (42%) consented to take part in OMACS. PIL B had the highest recruitment rate, with 45% of patients consenting to participation; 40% and 41% of patients consented to participation with PIL A and C respectively. Compared to PIL C, the consent rate was 4% higher with PIL B (45% versus 41%, 95% Confidence Interval (CI) -2% to +10%, p=0.16) and 1% lower with PIL A (40% versus 41%, 95% CI -7% to +5%, p=0.72). Conclusions: Consent rates were similar for all 3 PIL formats. PIL B is being used for the remainder of the host study and will be used to inform the design of PILs for other research studies as it was the preferred format of the PPI group. Background 4 Much research has been conducted on how to improve Patient Information Leaflets (PILs), the consent process and recruitment in Randomised Controlled Trials (RCTs) and to improve response rates to postal questionnaires and this is summarised in several reviews [1][2][3][4][5][6]. The main focus of this research with respect to PILs has been on the readability or length of PILs [7][8][9][10][11][12][13][14][15], use of different media such as audiovisual materials [5], or interactive electronic materials [16] rather than how the appearance of paper PILs can be improved. The outcomes have been patients' understanding of the study, or satisfaction with the informed consent procedure.
There has been less focus on whether recruitment rates can be improved, leading to a lack of evidence in this area. The conclusion of a Cochrane review [4] was that there is currently no clear evidence as to whether modifications to the information presented to participants improves recruitment to RCTs. This conclusion is consistent with that of previous reviews that reported that more research is required [1,2]. In addition, the design and delivery of information used to invite potential participants into a study has also been identified as a research priority by the Prioritising Recruitment in Randomised Trials study (PRioRiTy) to improve the process of how people are recruited to clinical trials. [17] With respect to the format of the PIL, Reinert and colleagues [11] performed a quantitative and qualitative analysis of study PILs and concluded that information presented to potential participants "needs to be well structured and designed in an appealing manner" . This study focused on how the appearance of the paper PIL might affect the decision to take part in a study, rather than the content. However, efforts were still made to make sure the content was readable and the length of the PIL kept to a minimum. This research was conducted as a Study Within a Trial (SWAT) by nesting it within a large cohort study, Outcome Monitoring After Cardiac Surgery (OMACS). The objective of the SWAT, OMACS-PIL, was to compare rates of consent using different formats of the host study PIL. Conducting a SWAT is an efficient way to gather evidence to address uncertainties about the best way to conduct clinical studies and allows methodological evidence to be collected alongside the aims of the main study. [18,19] In the context of this study, we wanted to make sure that any updates to our PIL were based on evidence. We hypothesised that an appealing appearance of the PIL might make it more likely for it to be read, which could in turn lead to an increased recruitment rate.

The Main Study (OMACS)
The OMACS study is an ongoing cohort study aiming to gain consent from cardiac surgery patients at our institution (University Hospitals Bristol NHS Foundation Trust) to collect and use quality of life data, plus routinely collected data on their operation and subsequent recovery to be used in a range of research studies.
Patients are approached for consent by post 3 months after their operation. The inclusion criteria for OMACS are adult participants over the age of 18 years who have undergone cardiac surgery at our institution, and around 1000 patients per year are approached to take part. The only patients not eligible to participate are prisoners, patients without the mental capacity to give consent and patients whose main residence is outside the United Kingdom. OMACS is ongoing and the results of the SWAT will be applied to future recruitment to OMACS. This study was chosen as it uses postal consent, where the appearance of the PIL may have more impact than face-to-face consent, plus the large target population meant that the SWAT could be completed quickly.

6
The SWAT (OMACS-PIL) OMACS -PIL was an RCT to investigate whether recruitment to the main OMACS study was affected by the format of the information provided to the potential participant. Three different information leaflets were prepared, each with the same content but using different styles and formatting. PIL A (enhanced format) and PIL B (hybrid format) were produced with a specialised graphics package called InDesign (Adobe Systems Incorporated) and PIL C (standard format) was produced using Microsoft Word (Microsoft Corporation). PIL A was a full colour tri-fold leaflet, PIL B was a double-sided A4 sheet using columns to break the text up and some colour on headings, and PIL C was a black and white A4 double-sided sheet with "standard" formatting. Examples of each format can be seen in Figure 1. Potential participants were randomly assigned to receive either PIL A, B or C in the ratio of 1:1:1. The randomisation list was prepared by the study statistician using Stata version 15.1 (StataCorp LP, College Station, Tex) and uploaded to the study database. The allocation was linked to the participant record and printed on to the footer of each mail-merged invite letter to allow the correct PIL to be attached. Invitation letters for the main study are prepared for mailing as a 'monthly batch' and to avoid making mistakes in including the correct PIL, the letters were printed in 3 separate batches each corresponding to a single PIL allocation. Potential participants were not told about the SWAT so were effectively blinded to their allocation. Members of the study team were not blinded but had no direct contact with potential participants and so were unable to influence consent rates.

The Study Population
The study population for OMACS-PIL comprised patients eligible for the OMACS study

Outcome
The outcome measure was consent to OMACS in each of the groups receiving PIL A, B or C.

Sample size
Assuming a consent rate of 70% with PIL C (based on historic data from a similar system which ran over several years [20]), a sample size of 1590 (530 per group) would give 90% power to detect a 10% difference in consent rate between any pair of PIL formats, with an overall significance level of 5% (with a Bonferroni correction for the three comparisons). This was the number of patients to be sent a PIL, rather than the number of patients required to consent. We chose a 10% difference as a compromise between: a) the large increase in return of postal questionnaires seen with some measures such as use of coloured ink (odds ratio 1.39, 95% CI 1.16 to 1.67) as reported in the Cochrane review of measures to increase responses to postal questionnaires; [6] and b) the conclusion of the Cochrane review about increasing participation on clinical trials that there is currently no clear evidence as to whether modifications to the information presented to participants improves recruitment to RCTs [4].

Statistical Analysis
The consent rates from the three different PIL formats were described and the differences in consent rate between pairs of PILs were modelled using logistic regression using Stata version 15.1 (StataCorp LP, College Station, Tex). Odds ratios with corresponding 95% confidence intervals have been presented for the 8 comparisons between PIL groups with PIL C as the reference category. Analysis took place at the end of the recruitment period for the SWAT.

Public and patient involvement (PPI)
In order to get some qualitative information about the PILs, the PILs were presented to the Bristol Biomedical Research Unit Cardiovascular PPI group which is made up of representatives who have had cardiac surgery, to elicit their preferences for the format and appearance of the PILs and the reasons for their preferences. The PPI group were not participating in OMACS as they had had cardiac surgery before the study was started. The group was formed after the start of the SWAT and so did not contribute to the design or content of the PILs, and their observations were obtained whilst the SWAT was taking place. The PPI group had 5 members and the information was obtained by the PPI coordinator who was independent of the study.
The OMACS Chief Investigator was present to explain the study.

Results
For the duration of the SWAT, 1517 invitation letters and PILs were sent to eligible cardiac surgery patients. The numbers of each format of PIL sent were: Format A 505, Format B 506 and Format C 506. The age and sex of the target population is described in Table 1. The mean age was 64.9 years (standard deviation 13.4), and 1079/1517, 71.1% were male. The average age and proportion of male patients was very similar across the 3 groups (see Table 1). participants were randomised to receive one of eight combinations of three interventions: a prior-notification postcard or no contact, a standard or professionally designed consent pack, and a phone or postal reminder [21]. Of these interventions, the most effective was the reminder phone call with a 6.4% higher response rate (95% CI +2.3 to +10.6%; P = 0.002). The professionally designed consent pack had some impact, increasing response rates by 2.7% (95% CI: −0.06% to 5.5%; P = 0.06), but the prior notification postcard had no effect. Phone call reminders can have significant resource implications, hence our decision not to 11 include this as an intervention.
The MRC START programme has also looked at the format of the paper PIL, but this was part of an 'enhanced' PIL where the wording was refined after several rounds of user testing [12,13]. This approach has resource implications and of the work published so far there have been only marginal improvements in recruitment [12,15]. The Healthlines studies showed modest increases in a positive response to an invitation to participate in the study when using the enhanced PIL that, like OMACS-PIL, were not statistically significant (19% versus 16%, difference of 2.9%, 95% CI = -1.1% to 6.9% in the Healthlines Depression study, n = 1364, and 24.0% versus 21.9%, difference of 2.0%, 95% CI -4.3% to 8.4% in the Healthlines CVD study, n = 671) [12]. In Early CDT Lung Cancer Scotland (ECLS), the proportion of patients who positively responded to the invitation was 224/1136 (19.7%) in the intervention group (optimised PIL) and 205/1126 (18.2%) in the control group (difference of 1.5%, 95% CI = -1.7% to 4.7%) [15]. In addition to these MRC Start studies, the REFORM study concluded that there was limited evidence as to the benefit of using optimised information materials on recruitment and retention rates [14] and the recently updated Cochrane review of methods to enhance recruitment to RCTs, concluded that 'Using a tailored, user-testing approach to develop participant information leaflets makes little or no difference to recruitment.' [22]. There has been a Cochrane review of methods to increase response rates to postal questionnaires [6] but this focused on the design of the questionnaire or the method of postage (e.g. first class post) and not on the PIL provided to participants.
Resource implications for developing electronic media and the limited existing evidence that it can improve recruitment [5] were the reasons a multi-media approach was not explored in OMACS-PIL.
One area for concern in OMACS is that the overall response rate was much lower than expected. The expected response rate was based on similar work involving postal consent of cardiac surgery patients (Long Term Monitoring, LTM), which had been as high as 80% [20]. There are a few differences between the studies: the timing of approach for consent was at 3 months versus 12  A strength of OMACS-PIL is that the intervention could be implemented in any study and does not require rounds of user acceptability testing to see a positive improvement in recruitment rates as required by other projects (e.g. MRC START) or extra resources as required with telephone contact or using a multimedia approach.
This means that it can be implemented with minimal or no impact on the study budget. OMACS-PIL was also randomised, and the participants were unaware of the SWAT and so were effectively 'blinded' to the interventions. However, based on the results of OMACS-PIL and the other similar studies, any improvement in consent rates are likely to be small.

Conclusions
The 'hybrid' PIL (PIL B) was preferred by the PPI group and resulted in a slightly higher consent rate to the host study than PIL A (enhanced format) or PIL C (standard format), although this difference was not statistically significant. PIL A, which was the most expensive to produce, was the least preferred option for the PPI group and did not improve consent rates. PIL B is now being used as the PIL for the remainder of the OMACS study, and the findings from this study are being used to 14 inform the design of PILs for interventional trials. The work also highlighted the importance of involving PPI groups in the design of information aimed at potential study participants.  Figure 1 Appearance of the PILs PIL A -tri-fold coloured leaflet produced using a graphic design packa

Figures
19 Figure 2 Consent rates by PIL type