- Two behaviorally informed messages for organ donation registration will perform better (in terms of organ donation registration numbers) than the current (business-as-usual) organ donation message presented at Service BC offices.
- A workshop organ donation registration helps customer service representatives feel more conformable initiating organ donation conversations with customers presenting at Service BC offices.

*Setting*
The study will be conducted at Service BC offices, which assist British Columbians with permits, property taxes, and/or driver’s licenses. Offices are located across the Canadian province of British Columbia (in both rural and urban environments), and receive, on average 1,000 customers every month.

Eight Service BC offices located across the province of British Columbia will be selected to participate in the study. Offices of any size and municipality that offer full-service will be included. Offices will be excluded if they: (a) do not have full service (e.g., they process hunting licences, business permits, but not driver’s licenses); (b) do not or did report monthly number of customers at any point since 1st April of 2016; or (c) did not exist by the 1st of April 2016. The eight offices were chosen to represent, as much as possible, the different types of locations: medium size with a reception desk, small offices where customers pick a number and wait for their turn, and offices where simply queue up.

*Interventions*
The intervention has two components: (1) training of staff; (2) organ donation messaging located at Service BC offices.

Training of Customer Service Representatives (CSRs). The objective of the CSR training is to increase self-efficacy among CSRs so they feel more comfortable talking to clients about organ donation. Training consisted of a 25-minute online session covering: (1) the importance of organ donation registration, (2) psychological barriers to registration, (3) CSRs role in helping to reduce the barriers, and (3) guidelines, tips, and recommendations.

Message frames. Two message frames will be tested: “Imagine you or someone you love needed an organ transplant and couldn’t get one” (reciprocity frame)[3] and “90% of British Columbians support organ donation but only 25% are registered as organ donors” (cognitive dissonance frame)[4]. Each message will be placed on a professionally designed poster.

Each poster has corresponding 4x6 inch card, which provides additional facts about organ donation on the reverse side:
- “There are more than 500 people waiting for an organ transplant in BC.”
- “One organ donor can save up to 8 lives.”
- “There is no age limit to register for organ donation. The oldest organ donor in Canada was 92 years old!”
- “You may still be able to donate if you had a medical condition. The specialist will determine if you can.”
- “A doctor trying to save someone’s life has no way of knowing they are an organ donor.”
- “Organ donation is considered only after all life-saving efforts are made and it’s certain a person will not survive.”
- “You are more likely to need a transplant than you are to be an organ donor.”

*Study design*
Pilot. One office was selected to pilot the trial, using Poster 1. The purpose of the pilot was twofold: (1) to identify previously undetected barriers, logistic issues related to the procedures, messages, and training; and (2) to pre-test the training session. The pilot run between the 1st and the 30th of November of 2022. CSRs at the pilot office were received training on the 10th of November 2022. During the pilot CSRs were encouraged to report any problem, difficulty, or barrier resulting from the trial (e.g., poster location interferes with operations in some way) or any interference with the normal operation of the office (e.g., other customer service priorities take precedence).

Full study. We will use a multiple baseline cluster design.[5] Each cluster corresponds to a Service BC office and each data point within each cluster corresponds to a month (not a person). Four locations (including the pilot office) were assigned the reciprocity frame, four were assigned the cognitive dissonance frame, and offices that were not selected will be used as controls and will use the "business-as-usual" message ("It takes 2 minutes to save a life".

Figure 1 indicates timing of the training and installation of the posters for each office. All locations will receive training on December 8. Installation of posters at selected offices will be staggered according to the schedule in Figure 1. Staggering the intervention will enable researchers to acquire experience implementing the intervention and modify and adapt the intervention to the specific circumstances of each office as needed. All adaptations will be documented and included as covariables during analyses.

*Procedure*
Customers presenting at Service BC offices with reception desks are greeted by a CSR to help them navigate services. At that moment, customers will be given the BC Transplant organ donation registration form for them to fill out while they wait. The form takes less than two minutes to complete. The postcards will be available at the reception desk in case customers are interested in more information. When customers are at the wicket, CSRs will ask them if they are interested in registering as organ donors and/or will receive completed organ donation registration forms, as appropriate.

At locations without a reception desk, organ donation registration forms will be placed next the poster for customers to pick up if interested. At the service wicket, CSRs will ask customers if they want to register as organ donors and hand them the registration form for them to fill out and/or receive completed organ donation registration forms, as appropriate. The postcards will be available at the wicket, in case customers are interested in more information.

CSRs at reception desks will be instructed to simply hand out the form to customers. CSRs at service wickets will be instructed to use their own judgment to determine if and when it is appropriate to talk to a customer about organ donation. CSRs will be advised that they can invite customers to read the postcards or visit www.transplant.bc.ca. CSRs will also be advised that they do not need to answer organ donation questions or dispel misconceptions about organ donation.

*Primary*
We will use administrative data on the monthly number of paper-based organ donation registrations that are traceable to each Service BC office. The number of registrations coming from each office are approximated based on the postal codes belonging to the catchment area of each office. This method of assigning registration to offices is subject to two limitations: (1) a small percentage of registrations (typically less than 5%) have postal codes that cannot be assigned to an office; and (2) a small number of registrations (typically less than 5%) may be misassigned to an office in the rare cases where people register while away from their registered address.

*Secondary outcome measures*
- To address the aforementioned limitations, a secondary outcome measure will be used: registrations coming from each of the eight offices will be manually tallied, before being assigned through the postal code method. The discrepancy between the primary and secondary outcome measures, if there is any, will be used to adjust the confidence intervals and/or their interpretation.
- CSRs will be asked to rate, on a 7-point Likert scale, their level of confidence in having organ donation conversations with clients before and two months after training (see Figure 1).

Discontinuous generalized mixed-effects models for count variables will be used to analyze data. The family and model specifications will be determined depending on the characteristics of the data (e.g., overdispersion, zero-inflation). Historical data from all offices starting on April of 2016 will be included in the analysis. Offices that were not selected will be included in all models as controls. Discontinuous generalized mixed-effects models will estimate the magnitude of changes in elevation and/or slope of the time series of each office before and after interventions and will also compare the magnitude of the change (elevation change, slope change, or both) across trial conditions (Message A, Message B, and business-as-usual message). These models will also allow us to control for staff comfort and confidence having organ donation conversations, seasonality, number of customers, and autocorrelation, as well as modifications and adaptations to the intervention.

The following covariables will be included in modelling
- Adaptations and modifications to the intervention, as appropriate (categorical).
- Offset variable: number of monthly customers attending each service BC Office.
- Seasonality (e.g., “Organ Donation Month” a campaign that takes place every year in April).
- Time discontinuity (interruption).
- A measure of how busy the office was, if available (e.g., median length of stay, average time per transaction).

CSR comfort levels before and after training will be analyzed using tests for two partially overlapping samples. This type of analysis is necessary because, privacy restrictions impede matching CSRs’ ids or email to their responses.

Statistical power was estimated using the cpa.sw.count function of the clusterPower Package for R, which is used of clustered stepped wedge designs where clusters are sequentially exposed to the intervention over a period of time. For the calculations, we used the following parameters: (1) an Intraclass Correlation Coefficient (ICC) of 0.4846652, which was calculated by fitting a Zero-Inflated Negative Binomial Mixed-effects regression on historical data of organ donation registrations from April of 2016 to February of 2022; (2) an overall baseline registration rate of 0.01107575 (# of registrations/# of customers) for the current year (January and February 2022 ), the most representative post-pandemic period of Service BC offices since the beginning of the COVID-19 pandemic; (3) a relative risk of 10.35803 = 0.114723/0.011, where the numerators is the overall rate of organ donation registration across all offices from April 2016 to February 2022, and denominator is the current registration rate; (4) number of data points per clusters equal to 73, which correspond to 70 months of historical data available at the time of calculation plus three more months of the current trial. With these parameters, three clusters with three steps provide 0.9540899 statistical power to detect a 10.35803 relative risk between one control group and one intervention group with at a significance level of 0.05. This effect size is expected in so far as it reflects pre-pandemic organ donation rates.

Since we are also comparing two intervention groups against one control, while controlling for seasonality (which was detected in preliminary analyses), and adaptations of the intervention to the characteristics of the offices, we increased the number of clusters to eight and the number of steps to four. These correspond to a rule of thumb of adding 50 data points per additional variable in a multivariable analysis. Since data are clustered, 50 data points correspond to 103.164 = 50/0.4846652 (50/ICC) additional unique data points per additional covariable in analyses. Since each cluster will have 73 data points, this translates into 1.413205 = 103.164/73 additional clusters. This, in turn, translates into 1.413205 * 5 additional covariables = 7 additional clusters. Accordingly, the final power estimation is eight clusters and four steps.

We will follow the principle of "using all available data". Data-recording errors, if any, will be corrected at the source (administrative records). Sensitivity analyses will be used to consider the effect of outliers on results.

We will follow the principle of "using all available data". Sensitivity analyses will be used to consider the effect of missing observation on results.

- Given privacy limitations regarding CSR comfort level (no collection of email, CSR id, or even office), this variable will not be included as a covariable in main analyses. For the same reason, only descriptive statistics will be used to analyze these data.
- One selected office decline participation, so it was replaced with another of similar characteristics.
- After pilot: CSRs at reception desks are no longer handing out postcards at locations