Mapping out a Medication Adherence Promotion System (MAPS): Can it reduce graft rejection in pediatric liver transplantation?

Nonadherence to lifesaving immunosuppressant medications is the leading cause of preventable graft failure, which can occur in up to 62% of recipients of liver transplant (LT),1–3 with even higher estimates among adolescent recipients of LT.4,5 The work by Bonn et al studies a multicomponent intervention to combat barriers to medication-taking.6 The 2 most novel aspects of this study are a risk prediction tool developed by the authors to stratify those at risk for medication nonadherence that is embedded in the electronic health record and the use of the Medication Adherence Promotion System (MAPS) intervention—a multicomponent quality improvement initiative developed and implemented at their transplant center. The MAPS program includes (1) psychologist training of transplant providers on how to discuss medication adherence with patients and families; (2) screening with an adherence risk score using objective drug levels and clinical history; (3) medication adherence barrier assessments; and (4) goal-directed action plans to address identified barriers. Most importantly, the medication barrier assessments and action plans were integrated into the electronic health record. In a group of predominantly adolescent LTRs, the authors found that the intervention was properly delivered in 78% of encounters. About 29% of those who completed a medication barriers assessment identified a specific barrier to medication-taking. Forgetfulness was found to be the most common (78%) reason for nonadherence by patients and families, which is well-established in the literature.5 The authors conclude that there was a 45% decrease in the incidence rate of late T-cell–mediated rejection (TCMR) at the clinic population level, which was sustained for 18 months from September 2020 to March 2022. Before the intervention, the monthly rejection rate was 0.84 per 100 patient-months. After initial implementation, rates increased to 1.25 rejections and then decreased to 0.46 rejections per 100 patient-months by September 2020. However, using an objective measure of drug-level variability—the Medication Level Variability Index (MLVI)7,8—there was essentially no difference in high-risk medication-taking behaviors, with 39% of the population having an MLVI >2.0 before intervention compared with 36% by July 2021 after intervention. We commend the authors on developing a multidisciplinary program aimed at targeting psychosocial and medical barriers to adherence among pediatric recipients of LT and families. A particular strength of the study was the use of a validated objective measure of nonadherence that has been shown to predict graft failure—the MLVI.7,8 The MLVI is calculated using the SD of trough immunosuppressant tacrolimus levels and is more reliable than self-reported or clinician-reported measures of nonadherence.7 MLVI levels >2.0 (ie, SD >2) have been associated with nonadherence, with MLVI levels >2.5 associated with graft rejection.1,8 This study highlights the need for and potential benefit of dedicated medication adherence programs among adolescent recipients of LT. Goal-directed action plans among targeted high-risk patients may be helpful in identifying and overcoming barriers to medication-taking. A key message is that a dedicated MAPS coach to administer the intervention may improve its delivery and fidelity. Results from this study may help transplant centers advocate for additional resources to implement this type of intervention (eg, by avoiding TCMR admissions, hospitals can focus on high-value, high-acuity care for other conditions). As with all interventions without a control group, there is concern for other factors influencing changes in rejection rates over time, including changes in the transplant population, provider practice, immunosuppression strategies, and other uncontrolled influences on medication adherence. Given that the study was conducted at the height of the COVID-19 pandemic, it is possible that changes in surveillance strategies through telehealth or lab monitoring, apart from the MAPS program components, could have contributed to temporal differences in TCMR. It would be useful to know if lab monitoring and telehealth visit frequency changed during the study period, which could have affected overall adherence. A salient question is whether the program actually influenced individual rates of rejection, and if so, what about the MAPS program led to meaningful change in medication-taking behavior? While population rates of late TCMR decreased, clinically significant medication nonadherence remained essentially the same. Questions related to individual changes in adherence remain unanswered. As with any clinical program, sustainability can be impacted by culture, workforce, and workflow issues. The authors discuss barriers to implementation, such as staff turnover and buy-in, as well as the COVID-19 pandemic. These challenges speak to the need for a transformational leader or champion to spearhead quality improvement initiatives and consider dedicated MAPS personnel to carry out the intervention. The champions may change over time, and a lack of resources can threaten intervention implementation. The authors acknowledge the inability to assess fidelity to the MAPS intervention, and there is a very high likelihood of delivery variation. Regarding the population, the authors use an MLVI >2.0 in their risk score to identify at-risk individuals; however, other variables in the risk (eg, missed lab appointments) have not been validated to identify nonadherent patients. The population for the cohort included patients later in their posttransplant recovery (median time since transplant was 9.8 y [IQR: 4.8–15.3]). Future work should aim to understand how to implement MAPS at earlier time points after transplant, when medication-taking may be more challenging to recipients of LT. It is possible that MAPS is more useful after transplant center and parental oversight wane over time; however, this should be further evaluated in studies. Robust intervention development and implementation require scrupulous investigation of different components of the intervention, how these are expected to influence mediators of behavior change, and assessment of modifiers of this relationship. There is a complex interaction of different environmental and behavioral influences that leads to the outcome of reducing graft rejection. If the MAPS intervention did affect graft outcomes through improved medication adherence, it is unclear what components of the intervention contributed to behavior change. Future studies should further understand which components are most effective (eg, patient education, enhanced monitoring of high-risk patients, and action plans). Implementation science methods in future studies should investigate intervention fidelity, barriers, facilitators, and uptake in diverse settings.9 Moreover, future studies should consider large-scale risk prediction tools, including the use of artificial intelligence and electronic health record data to identify high-risk individuals in need of targeted intervention. A limitation of this study was missing data regarding downstream effects, including referrals for additional psychosocial support. These care linkages can also be evaluated by measuring engagement with social work, psychiatry, and patient/family completion of the action plan. Future studies could also assess patient and family satisfaction with the program, what components they thought were most helpful in their care, as well as feedback from transplant team members including nurse coordinators who were the main facilitators of the intervention. The MAPS intervention may look quite different from one transplant center to another, which need not undermine its success. Hybrid study designs that evaluate effectiveness and implementation alongside clinical deployment are here to help guide interventions such as MAPS on a path to optimal implementation.