Recruitment
Participants will be a large, randomly selected representative sample of the adult Scottish general public. These participants will be randomly selected from a list containing 1.2 million members of the adult Scottish general public. Participants will be posted a brief questionnaire, together with information about, and a link to the UK Organ Donor Register. We will be receiving informed consent from all participants. However, we were concerned that if we were to require all participants to return signed consent forms by post, it is likely that people who are more likely to register as an organ donor would be more likely to participate in our study. This would create an unacceptable bias that could render our results scientifically meaningless. We are, therefore, using an 'opt-out' approach in which participants are required to contact us if they do not want to take part in the study. We will make it very clear that people do not have to participate if they do not wish to, and we have made it as easy as possible to withdraw either by; (a) ticking a form and returning it to us in the stamped addressed envelope that we provide to all, (b) email or (c) telephone. We believe that our approach is justified as; (a) no harm will come to the participants, (b) our research cannot be practically carried out if we had to receive written informed consent from all participants, (c) the potential benefits to the NHS (i.e. determining effective methods of increasing organ donor registrations) outweigh the cost, and (d) NHSBT will be conducting the search of the organ donor register, not us, we will never have access to individual identifiable results, and we will never know whether a specific individual has registered or not, only NHSBT will have this information, and crucially, they would have it regardless of whether this study was conducted or not.
Based on our pilot data [11], we estimate that in order to identify a significant effect at the conventional .05 alpha level, at a power of .80, we would need 565 completed questionnaires per condition. As with any questionnaire based study there are likely to be many people who do not complete our survey. The response rate for this type of survey ranges from 23% to 37% [37]. Based on the conservative lower estimate (23%), we calculate that we would need to distribute 2,460 questionnaires per condition in order to gain a sufficient number of completed surveys. Furthermore, we also need to oversample to account for the proportion of the adult Scottish general public that are already registered donors. Adjusting for this, we need to distribute 3,630 surveys per condition in order to gain a sufficient number of completed surveys. Because there are four conditions in this study (for details, see below), we need to distribute a total of 14,520 questionnaires. The logistics of this national field based study will be conducted by Perspektiv, a market research company (http://perspektivred.co.uk). It should be noted that in the present study we improve on O'Carroll et al. [11] by including a no questionnaire control arm. We do not have pilot data that would allow us to conduct power calculations for this condition. However, based on the questionnaire control arm in O'Carroll and colleagues [11], we estimate that the above sample size estimates should be more than sufficient.
Exclusion criteria
We are primarily interested in whether a simple anticipated regret intervention increases organ donor registration. We will, therefore, exclude people who are already registered organ donors from the main analyses. We will also exclude participants who actively withdraw from the study.
Design and materials
We will utilise a between-groups, prospective randomised controlled design (i.e. participants will be randomly allocated into one arm of the study). In line with previous research [11, 12, 18, 34], we will manipulate anticipated regret by altering the questions that the participants complete. Participants will be randomly allocated into one of four conditions: no questionnaire control (NQC), questionnaire control (QC), theory of planned behaviour (TPB) questionnaire, and anticipated regret (AR) questionnaire (for an overview, see Figure 1). The latter three conditions are similar to O'Carroll and colleagues [11].
No questionnaire control (NQC)
This extension of O'Carroll and colleagues [11] has been added to determine whether simply being contacted increases organ donor registration. This arm will simply receive a letter, donor registration form and questions collecting demographic information (e.g., date-of-birth, gender, occupation and postcode for socio-economic status estimation).
Questionnaire control (QC)
This arm will receive the same materials as the NQC arm plus a questionnaire measuring their emotions and non-cognitive affective attitudes towards organ donation [10–12]. This questionnaire will measure 5 emotions and affective attitudes: the ick and jinx factors, bodily integrity, medical mistrust, and perceived benefit. The ick factor will be measured using three items (e.g. "The thought of organ donation makes me uncomfortable"). Three items will be used to measure jinx (e.g. "The surest way to bring about my own death is to make plans for it like signing an organ donor card"). Two items will measure bodily integrity (e.g. "The body should be kept whole for burial"). There will be four medical mistrust items (e.g. "If I sign an organ donor card, doctors might not try so hard to save my life"). Perceived benefit will be measured with four items (e.g. "Organ donation helps to bring meaning to the death of a loved one"). All items will be rated on a 7-point Likert scale (1 = strongly disagree, 7 = strongly agree).
Non-donors will also be asked to complete two questions measuring their intentions to register as an organ donor in the future (e.g. "I will definitely register as an organ donor in the next few months"; 1 = strongly disagree, 7 = strongly agree). We will also include filler questions to ensure that the number of items in this arm is identical to the TPB and AR arms.
Theory of planned behavior (TPB) questionnaire
The TPB arm will complete the same materials as the QC arm, plus additional items measuring attitudes, perceived control, and subjective norms. Attitudes will be measured with two items (e.g. "I support the idea of organ donation for transplantation purposes"). Two subjective norm items will be included (e.g. "Most people who are important to me think I should register as an organ donor in the next few months"). The attitudes and subjective norm items will be rated on 7-point Likert scales (1 = strongly disagree, 7 = strongly agree). Three items will measure perceived control (e.g. "How much control do you have over registering as an organ donor in the next few months?"; 1 = no control, 7 = complete control). Once again, non-donors will also be asked to rate their intentions to register as a donor, using the items described above. We will also include filler questions to ensure that the number of items in this arm is identical to that of the QC and AR arms.
Anticipated regret (AR) questionnaire
This arm will complete the same indices as the TPB arm, plus two items measuring anticipated regret: "If I did not register as an organ donor in the next few months I would feel regret" (1 = definitely no, 7 = definitely yes) and "If I did not register as an organ donor in the next few months, I would later wish I had" (1 = strongly disagree, 7 = strongly agree). Non-donors will also be asked to complete the intention indices described above.
Organ donor registration
Participants in all four arms of the study will be told how to register as an organ donor online, and by telephone, post and text message. All participants will also receive a NHSBT organ donor registration form.
Outcomes
Our primary outcome variable is verified organ donor registration within 6 months of our postal intervention. Crucially, NHSBT has agreed to collaborate with us and perform a secure and confidential search of their organ donor register 6 months following our brief postal intervention. This search will tell us whether or not the participant is a registered organ donor and, if applicable, when they registered. NHSBT will conduct this anonymised search for all participants who have not opted out of the study, regardless of whether or not they returned the questionnaire. This will allow us to determine whether there is any bias caused by people being more willing to complete the questionnaire in certain conditions.
Our second outcome is intentions to become an organ donor in the future. If our anticipated regret intervention is successful we will test whether the increase in registration is due to participants having greater intentions to register as an organ donor. This outcome will be measured in our questionnaires, using the indices described above.
Analyses
Our primary analysis will be a logistic regression predicting donor status (registered vs. not registered) to explore the proportion of respondents who have registered as organ donors 6 months later as a function of the 4 arms. In this analysis we will control for any potential between-arm differences in age, gender and socio-economic status, if these are related to the outcome. We will conduct this regression with all participants who have not opted out of the study, regardless of whether or not they returned the questionnaire. Participants may not comply with their random assignment. That is, in this case, for some reason not complete and return the questionnaire. We will use instrumental variables (IV) regression techniques to estimate the causal effect of the intervention in compliers [see [38, 39]]. Here randomization acts as the instrumental variable, compliance status as the endogenous variable. This technique is a viable alternative to traditional intention to treat (ITT) analysis. ITT treats randomization as a treatment, when in actuality it is the intervention, not the randomization that is the source of any effect. By treating assignment as an instrumental variable and intervention compliance as an endogenous variable, IV regression techniques reduce the bias in standard ITT analyses, and provide as estimate of the causal effect of the intervention in compliers [38]. These analyses will be conducted in MPlus-6 [40] and Stata [41]. We will also record the time interval between the questionnaire being sent out and date of organ donor registration to test for temporal effects.
If our intervention is successful, we will test the processes through which it occurs. We will assess whether the anticipated regret intervention promotes organ donor registration by increasing people's intentions to register and decreasing their emotions and non-cognitive affective attitudes towards organ donation. Essentially, we will test whether intentions and emotions mediate the effect of our intervention on registration. Simple tests of mediation will be implemented in ZUMASTAT [42]. More detailed tests of multiple and joint mediation as well as moderated mediation will be conducted using MPlus.
Evaluation
The effectiveness of the simple anticipated regret intervention will be assessed using the instrumental variables logistic regression analyses outlined above. This will determine whether there is a significant difference in the proportion of registered organ donors in each arm 6 months following our intervention. If we find that there is greater proportion of participants registered as organ donors in the AR arm than the other three arms, we will conclude that a simple anticipated regret intervention increases organ donor registration. We will also assess the mechanisms behind this intervention. The mediation analyses outlined above will test whether any significant effects of our intervention are due to increases in people's intentions to register and decreases in their emotions and non-cognitive affective attitudes towards organ donation.
Research ethics and timetable
This study has received ethical approval from the South East Scotland Research Ethics Committee (ref: 11/SS/0093). This is an 18-month project. Months 1-4 will be spent designing, piloting and finalising the layout of the questionnaire pamphlet, and also in working closely with Perspektiv in randomising participants to the four arms of the study. The questionnaire packs will be posted out to participants in months 5 and 6. During months 7-12 the responses will be entered into the PASW spreadsheet as they are returned, and the analytic strategy will be finalised. The protocol for linking the participants to the NHSBT Organ Donor Register will also be developed and finalised during this period. During months 13 & 14 the NHSBT Organ Donor Register will be checked for new registrations by our participants. Months 15-16 will be spent conducting the final analyses, and in months 17-18 we will draft scientific papers, conference presentations and the Chief Scientist Office final report.