Video Education and Behavior Contract to Improve Outcomes After Renal Transplantation (VECTOR): A Randomized Controlled Trial

Introduction

Transplantation, the preferred treatment for improving health outcomes for those with end-stage kidney disease,¹ is demanding. Kidney recipients must manage lifestyle modifications and maintain strict adherence to immunosuppressant medications despite frequent adverse effects. An inability to adapt to these changes may result in an increased risk of transplant rejection and graft loss.^2,3 Routine monitoring of medication adherence is recommended as a “fifth vital sign” in all transplant recipients³ as even a 5% deviation from the prescribed immunosuppressant regimen can have negative consequences.²

Adherence-enhancing interventions have been generating recent attention, with multi-dimensional behavioural strategies showing efficacy in randomized controlled trials (RCT).⁴ In the MAESTRO-Tx trial (n = 205; heart, liver, and lung recipients), a tailored intervention involving electronic monitoring feedback and motivational interviewing contributed to 16% higher dosing adherence compared to the control.⁵ Likewise, an electronic medication tray and a smartphone app with monitoring and reinforcement messages significantly reduced tacrolimus intrapatient variability (IPV) in non-adherent kidney recipients (n = 80).⁶ Multicomponent interventions such as these seem promising, but they can be expensive, time consuming and difficult to implement widely.⁷ Given that nearly a quarter of kidney transplant patients are estimated to struggle with medication adherence,² sustainable solutions should be explored.

The Video Education and behavior Contract to improve Outcomes after Renal transplantation (VECTOR) RCT evaluated whether a low-cost, simple intervention improves adherence, compared to usual care. VECTOR combined an educational component (three-part video series) with a behavioural component (goal setting exercise and adherence contract) delivered electronically to kidney recipients after their transplant.⁸ The evidence-based patient-centered videos⁹ were shown to improve knowledge and satisfaction when administered before kidney transplantation in a previous RCT.¹⁰ The behavioral component was inspired by a study showing education provided by a pharmacist combined with a behavioral contract post-transplant improved immunosuppressant adherence and decreased hospitalizations.^11,12 The intervention, however, involved intensive collaboration between patients and a study pharmacist, and to our knowledge has not been widely implemented. Therefore, we sought to determine whether a streamlined simpler approach could improve adherence.

Materials and Methods

Trial Design

A prospective, multicenter, parallel design randomized-controlled trial was conducted in de novo kidney transplant recipients, in 6 centers in North America. Methods have been described elsewhere,⁸ clinicaltrials.gov (NCT03540121). The study timeline is presented in SDC Figure 1.

Study Population and Recruitment

Study participants were 18 years or older and received their first kidney transplant at one of the participating centers. They were enrolled within 2 weeks of hospital discharge. Patients were excluded if they had previously participated in an educational study, were not fluent in English and did not have a support person to assist them with participation.

Intervention (Video Education + Behavior Contract + Standard Education)

On demand video education was delivered immediately after enrolment along with standard education. Participants were provided access to password-protected videos, available for viewing on any electronic device. The three-part video series (Introduction; Medications; Your New Life) was adapted from an existing educational program.⁹ These patient-centered videos incorporate principles from adult learning theory and were developed according to best-practices for education for patients awaiting transplantation.¹³ Participants were asked to watch the videos (in hospital or at home) and replay as desired. Participants without electronic access were provided with a study device to view the videos in hospital.

Approximately 1-month post-enrolment an email link invited participants to work through an electronic self-directed goal setting exercise and sign an adherence contract pledging to take immunosuppressant medications as directed. This autonomous activity encouraged participants to reflect on their motivation for remaining adherent, potential barriers and consequences, and set an action plan for taking medications as directed.¹¹ A signature box enabled participants to commit to the contract. It was revisited during the 3-month and 12-month post-enrolment surveys and was not reviewed by the healthcare team.

Control Condition (Standard Education)

Participants in the control condition received the standard education only, including one-on-one medication teaching from a healthcare provider, medication teaching sheets and a personalized medication schedule. These participants received the same assessments throughout the study and were sent a control message at the 1-month period, in substitution for the behavior contract.

Study Overview

Ethics approval was obtained at each site from the local institutional review boards. A site lead oversaw activities at each center. During the study period, a front-line healthcare worker identified patients who received a transplant and used a standardized script to ask recipients whether they were interested in learning about an educational study. A research assistant contacted patients who expressed interest to provide details and take informed consent. This process was performed in-person, until the COVID-19 pandemic declaration (March 2020). The study was temporarily halted until a virtual enrolment process was approved by each regional ethics board (March–September 2020). The Canadian Hub for Applied and Social Research (CHASR) at the University of Saskatchewan oversaw the experimental research activities, achieving a standardized process across the sites. They disseminated surveys, videos, adherence contracts and reminders. All data from the surveys was immediately transmitted and managed in a central repository and deidentified prior to analysis by the research team. Gift cards in the amount of $20 and $25 were issued after completion of the 3-month and 12-month post-transplant questionnaires, respectively. The research activities were consistent with the principles outlined in the “Declaration of Istanbul on Organ Trafficking and Transplant Tourism” and the “Declaration of Helsinki on Ethical Principles for Medical Research Involving Human Subjects”.

Randomization and Blinding

Randomization (1:1) was performed centrally in permuted blocks of six or eight and stratified according to site. Pre-generated codes were embedded into the participation links and numbered sequentially to conceal treatment. Groups received communications of similar content and frequency. Participants in the intervention group were given unique login credentials and passwords and were reminded not to discuss the nature of the intervention with anyone; the research team was blind to participant allocation.

Outcome Measures

The primary outcome was medication adherence measured by the differences between the two groups in percentage of patients self-reporting optimal adherence using the Basel Assessment of Adherence to Immunosuppressive Medications (BAASIS) Scale¹⁴ sent electronically to each participant at 3 and 12 months post-enrolment. A patient was considered “non-adherent” if they answered “yes” to any of the questions (pertaining to missing doses/drug holidays, timing, and dose adjustment). Other adherence measures included the visual analogue scale (VAS) on the BAASIS as a continuous measure of adherence and tacrolimus blood levels (IPV and immunosuppressant time in therapeutic range (TTR)) at the end of the study. Immunosuppressant blood levels (tacrolimus, cyclosporine, sirolimus and everolimus) were collected according to routine practice. TTR was calculated using the Rosendaal method¹⁵ where immunosuppressant blood levels were imputed for every day of the follow-up period based on linear extrapolation between dates of actual (ie, measured) results. Observed and imputed levels were then coded as either 1 (within target) or 0 (outside target) based on drug-specific targets for calculation of TTR. For IPV, tacrolimus trough variability was defined as the coefficient of variance ([CV]; calculated as [SD/Mean] × 100%). Only patients taking tacrolimus were included since validity studies indicate different immunosuppressants should not be combined within the same CV calculation.¹⁶ Secondary outcome measures were collected with a self-reported survey administered 3 and 12 months post-enrolment. The survey contained several components; transplant knowledge was measured using the Kidney Transplant Understanding Tool (KTUT),¹⁷ self-efficacy, quality of life and medication beliefs were measured by the generalized self-efficacy scale (GSE),¹⁸ short form-12 version 2 (SF-12^®),¹⁹ and Beliefs about Medications Questionnaire (BMQ),²⁰ respectively. Satisfaction was measured using Likert scale questions where patients rated their confidence, understanding and satisfaction with education. Adherence to appointments or “no shows” (percentage of times a patient missed a scheduled transplant appointment or test without calling to cancel) and the number of days in hospital were collected retrospectively. Potential predictors or covariates collected at baseline included age, sex, marital status, race/ethnicity, education level, province, distance to transplant center, and health literacy.

Statistical Analysis

A sample of 100 patients in each group was estimated to detect a 15% absolute increase (from 75% to 90%) in percentage of adherent patients at an alpha of 0.05 and a beta of 0.20.

Randomized participants with at least one immunosuppressant blood level were included in the ITT analysis. A rolling monthly average compared immunosuppressant blood level measures (TTR and CV) between the intervention and the control. To account for the dependence of repeated measures from the same patient, generalized linear mixed models were used to compare groups during the 1-year follow-up period. For CV, individual-specific random intercepts were included in both mean and standard deviation models. For TTR, we used the zero-one-inflated beta regression model to accommodate the 0’s and 1’s. A Bayesian MCMC method, implemented by the R package brms, estimated model parameters. A square-root transformation was applied to rolling average CVs to adjust for the skewness in the original scale.

For the assessment of self-reported outcomes, we analyzed our results using the last observation carried forward when final assessments (ie, 12-month questionnaire) were missing. The percentage of patients indicating optimal adherence in each group by the BAASIS was compared using χ². The K-TUT, GSE, SF-12^® and BMQ were scored according to standardized methods.^17–20 Independent t-tests compared the mean adherence calculated according to the VAS, and the secondary endpoints. A “per protocol” analysis was conducted after excluding intervention patients that did not have objective evidence (confirmed by time spent on each video webpage) of watching >80% of the videos and completing the adherence contract.

Changes to the Study Protocol

Saskatoon, Calgary, Chicago, and Halifax were the planned study sites. The site lead in Halifax was unable to initiate the study. Enrolment rates were lower than anticipated necessitating the addition of Edmonton and two Vancouver transplant centres, but we could not obtain missed appointments and hospitalizations from Vancouver. During the pandemic, the study was amended to allow a central research coordinator in Saskatchewan to obtain consent remotely (phone or virtual meeting) rather than face-to-face by a research assistant. This study did not have a Data Safety Monitoring Board due to the low-risk nature of the intervention; however, the research team (clinicians, statisticians, patients) decided to halt enrollment due to futility after an interim analysis of the first 126 patients to complete the trial. The original protocol specified that self-reported adherence would be collected at baseline (in addition to 3 and 12 months) but since sites varied on their in-hospital medication administration protocols (eg, nursing administration versus self-administration) this baseline timepoint was removed. A cost-utility analysis was not completed due to the lack of difference between the groups.

Results

Of 203 patients approached by a research assistant, 175 met inclusion criteria and agreed to randomization. Two patients passed away immediately post-transplant leaving 173 (91 intervention, 82 control) for the ITT analysis. One hundred and twenty participants (69%) completed a post-intervention questionnaire and were analyzed as the complete case population. Video viewing statistics revealed that 43 participants randomized to the intervention group (67% of the complete case group) had objective evidence of watching at least 80% of the videos and 37 participants (58%) also completed the electronic goal setting exercise and adherence contract (per protocol analysis). Fifty-seven participants were enrolled after the onset of the pandemic using the virtual process. Comments from those who viewed the videos were generally positive (SDC Table 1). The study participation flow chart is presented in Figure 1.

Figure 1 Study participant flow chart.

Participant Characteristics

Participants had a mean age of 51.0 (12.3) and 86% spoke English as a first language. Baseline characteristics were similar between both groups (Table 1). Significantly more participants who completed a post-baseline questionnaire were white and from a Canadian center (p < 0.001). Over three-quarters (85%, 45/53) of the participants who did not finish the study were enrolled in Illinois, and nearly half (42%, 22/53) were Black or African American (Table 2).

Table 1 Characteristics of the Intention to Treat (ITT) Study Population Stratified According to Arm

Table 2 Characteristics for Those Who Did and Did Not Complete the Study

Medication Adherence

No significant differences were found between the groups in any measure of adherence for the ITT, complete case or per protocol analysis. According to the self-reported scales, the percentage of participants reporting optimal adherence was 55% in the intervention group and 64% in the control, while the mean score on the VAS was 98.4% (3.1) and 98.5% (4.5) for the intervention and control groups, respectively (Table 3). There were no significant differences for patient’s self-reported adherence between the 3-month and 12-month data collection points.

Table 3 Adherence Measured the BAASIS Self-Report Questionnaire Stratified by Condition

No significant differences were found between the groups for IPV or TTR (Table 4 and Figure 2). For the ITT CV (transformed), the 95% credible interval for the slope difference between the two groups was (−0.018, 0.071), while for the TTR it was (−0.005, 0.044). Since 0 is covered within these intervals, there is no statistically significant difference in the groups (instead of Bayesian p-values, we reported interval estimates because of their transparent calculation and interpretation).

Table 4 The Rolling Monthly Average IPV and TTR for the Intervention and Control Groups

Figure 2 The mean tacrolimus intrapatient variability and time in therapeutic range for the immunosuppressants from baseline to 12 months post-randomization in the treatment and control arms.

Secondary Outcomes

Knowledge, self-efficacy, beliefs of medicine, quality of life and educational satisfaction assessments are reported in Tables 5 and 6. The mean number of days spent in the hospital was 9.9 (16.5) and 9.7 (15.2) and missed appointment rates were 0.03% and 0.02% for the intervention and control, respectively. No significant differences were found in any measure.

Table 5 Comparisons for the Intervention and Control Group for Mean Scores for Knowledge Self-Efficacy, Beliefs of Medicine, Quality of Life and Adherence

Table 6 The Mean Scores on the Satisfaction Questions for the Intervention and Control Groups

Discussion

We evaluated whether a low-cost, simple intervention can improve adherence compared to usual care in kidney transplant recipients and found no impact on any outcome. Chisholm-Burns et al conducted an RCT using an educational intervention and behaviour contract delivered from a clinical pharmacist.¹² Renal transplant recipients were enrolled into the study at least one year post-transplant. Participants in the intervention group met with the study pharmacist to negotiate and sign the behaviour contract, which was revisited three times during the study period, using in person or phone interviews. Participants in the adherence group experienced significantly higher adherence rates along with decreased hospitalizations at 12 months. Our autonomous goal setting exercise and behavior contract, in contrast, was not successful at improving adherence, indicating that a simplified approach does not work. These findings are in line with others that suggest that in a post-transplant population, a one-size fits all approach is likely insufficient⁴ and support a case for investing in tailored adherence-enhancing interventions with evidence to support their efficacy.^5,6

The ability to justify the expenses of more complex adherence solutions and to implement them into real-world settings remains a challenge. Specifically focusing these efforts on patients who are identified as struggling with nonadherence may be one way to offset the potential costs. Since our intent was to investigate an inexpensive intervention that could be easily put into practice, we employed convenience sampling enrolling all patients who met the inclusion criteria. Future studies should aim to enroll nonadherent patients²¹ and specifically target this population. Interventions, however, must be engaging enough to encourage participation by nonadherent patients,²¹ and investigators should strive to show how study-related interventions are incorporated in real-world settings.²² Currently, even high-quality adherence RCTs lack implementation-relevant information.²²

Although our findings are not entirely new, well-designed, robust trials conducted with educational interventions are exceedingly rare and are difficult to draw conclusions. In a systematic review of educational interventions for renal transplant recipients, most interventions and assessments were inadequately described and had a high risk of bias.²³ In the present trial, all efforts were made to minimize bias and the research team was blind to participant allocation. The development of the intervention has been meticulously documented using a theory-informed and evidence-based process which was guided by patient stakeholders. Nevertheless, the ability to recruit and retain all participants remains a major challenge with educational and adherence studies. For example, in an RCT of 245 kidney recipients, only 58% uptake occurred when patients worked through self-directed computer-based medication modules.²⁴ In the present study, of 203 patients screened, 173 patients participated, with only 69% following through to completion. It is unlikely that participants forgot about the study because the central research coordinator provided up to 3 email and phone reminders. The videos were co-developed with patients and feedback indicated that the videos were appealing and understandable^9,10 and attrition was similar in both groups suggesting the acceptability of the video intervention was not a factor. Notably, more participants who completed the study were white and from a Canadian center (p < 0.001). Over three-quarters of the participants who did not finish the study (85%, 45/53) were enrolled in Illinois, and nearly half (42%, 22/53) were Black or African American. In a pre-transplant RCT in Missouri, transplant education presented to Black and White low-income patients receiving dialysis was more effective when presented directly compared to video and print modules.²⁵ Black patients are more likely to experience inequalities in access to technology²⁶ and exhibit mistrust of the medical system.²⁷ Familiarity with central study site and location of data storage may have also facilitated trust with the patients at Canadian centers and contributed to higher completion rates.

Slow recruitment was a limitation with this study. Research assistants relied on ward staff to identify and triage patients. Efforts were made to improve engagement, but this remained a challenge that worsened during the pandemic. After reviewing the evidence from an interim analysis, we decided to halt VECTOR enrolment at the end of 2021. Premature discontinuation is acceptable when the study hypothesis is unexpectedly unprovable based on an interim analysis²⁸ and a signal towards improvement in the outcome measures would have been expected at the interim analysis enrolment of 126 participants. We offered an honorarium for participation to maximize recruitment and the study’s potential for success. However, in a real-world setting, an honorarium would not be offered and recruitment for this type of intervention could be potentially lower without. A combination of methods is recommended for measuring adherence³ and we used self-report and immunosuppressant blood levels. While there is an increasing amount of psychometric data on the BAASIS,²⁹ electronic monitoring may have provided a more sensitive method for detecting the specific phases of adherence. Unfortunately, the cost of this type of monitoring could not be supported under the current funding allotment. This study was limited to one year and declines in medication adherence may not be apparent until later. Behavioral studies by nature have some intrinsic limitations, including the inability to truly blind participants. Efforts were made to minimize bias and the research team was blind to participant allocation. Development of the intervention has been meticulously documented.^9,10

Conclusion

We tested whether a feasible adherence intervention could be of benefit in de novo kidney transplant recipients. An autonomous goal setting exercise and electronic behavioural contract added to standard of care did not improve any outcome. The negative results and recruitment challenges highlight the challenge of fidelity with adherence interventions in the real-world setting and reiterate that nonadherence in transplantation is a difficult multifactorial problem.