We apply the science behind game theory to model the interactions, beliefs and preferences of a transplant team and a patient with ALD. While we are not the first to note that game theory provides the proper analytical framework to study medical policies in liver transplant decisions [23], we provided a comprehensive approach by incorporating how incomplete information and relying on signaling behavior of the patient facilitates a medical decision using a paradigm of eligibility of LT for ALD. Our findings showed that the 6-month abstention criterion is intrinsically flawed in its ability to distinguish LT candidates who would develop an alcohol relapse from those who will not even if there is a variability in the observed curable rates. Our findings align with recent research that has shown similar post-LT relapse rates and survival outcomes between early-LT patients and patients who abstained for 6 months.
We offer a new, mathematical approach that incorporates the uncertainties related to signaling to show many inadequacies of the 6-month rule. Several examples in probability theory have demonstrated how intuition can result in inaccurate deductions. An example relevant to medicine is the Bayseian paradox for diagnostic screening tests [24]. For example, let’s say a particular screening test for Disease A has a false positive rate of 5%, and a false negative rate of 1%. The prevalence of Disease A is 3%. Assuming a population of 1000, 30 people would have Disease A and 970 people do not. Given the false negativity rate, we would expect 3 false negatives and 27 true positives. Given the false positivity rate, we would expect 49 false positives, and 921 true negatives. This means that if the entire population was tested, 78 would test positive, though only 30 have Disease A, meaning a probability of less than 50% in distinguishing an individual with the disease from one without. Intuition, however, often suggests that the accuracy of this test is higher. Some authors refer to this concept as statistical illiteracy [25,26]. In this study, we demonstrated that similar inference blunders may occur during transplant decision-making.
Despite the trends in recent research and updated practice guidelines by the American Association for the Study of Liver Diseases (AASLD), the 6-month rule is still being actively used [18,27]. Currently, LT policy for listing candidates on the waitlist is at the discretion of the local transplant center, leading to variability in decision-making across different programs [28]. Our study reinforces the need for a national policy for LT in ALD patients to ensure fair and equitable allocation of organ. Specifically, our findings point towards the need for a more evidence-based protocol of predicting sobriety that considers the patient’s full psychosocial profile. Even at the inception of the 6-month rule, addiction experts disagreed with the premise, especially as early studies in addiction medicine suggest a minimum of 5 years to guarantee long-term sobriety [21,29]. Early efforts to identify an improved alternative approach involved engaging an addiction specialist to conduct an evaluation encompassing different domains of the patient’s psychosocial profile including their social isolation/social integration, acceptance of a drinking problem; prior history of treatment of alcohol use disorder and presence of other psychological disorders [30]. Stemming from this, several prognostic instruments have been developed to predict post-LT drinking. These tools include the Alcohol Relapse Risk Assessment (ARRA) [31], Stanford Integrated Psychosocial Assessment for Transplantation (SIPAT) [32], and the Sustained Alcohol Use Post-LT (SALT) score [33]. These scoring systems consider other important risk factors that have been found to predict relapse including psychiatric comorbidities, social support, and tobacco use [19]. It is important to note that more prospective studies are needed to validate these tools, but initial findings are promising and enable more fair and appropriate selection of ALD patients suitable for transplant [34].
Our approach in this paper has notable limitations. Our model is mostly based on assumptions, though these assumptions reflect real-world preferences. We considered the most critical factors related to decision making in LT for ALD patients. However, we elected not to consider other confounding variables like pre-existing physician biases (i.e. belief that alcoholism with is a disease with genetic predisposition versus a consequence of self-inflicted harm), other aspects of the patient’s psychosocial and clinical profile given the wide variability from physician to physician, country to country, and so forth. However, this was also important because the objective of the study was to evaluate whether the 6-month rule alone is sufficient. Furthermore, we acknowledge that patients are often characterized by a gradient and our binary classifications might be limiting. However, we also recognize that decisions are often binary. Of importance, our goal is to apply game theory to show that the 6-month rule may be irrelevant. We are not intending to guide therapeutic decisions or predict who may or may not benefit from LT."
Our game theoretic framework facilitates an explanation on why the 6-month abstinence criterion is not an effective strategy for detecting LT candidates who would develop an alcohol relapse. We also illustrated that our results remained unaltered by modifying the behavioral strategies of patients and the transplant team.
This framework may also have applications for other medical scenarios where a healthcare provider must decide about a treatment that uses a scarce resource (i.e. organ, expensive treatment for chronic or rare medical conditions) and must rely on signaling behavior from the patient to evaluate the best candidate (e.g. disease modifying therapy for patients with multiple sclerosis, immunosuppressive therapy for patients with inflammatory or autoimmune diseases, gene therapy for children with spinal muscular atrophy). Having an equitable and efficient health care system is critical when delivering time-sensitive medical interventions.
Previous studies shed light on how physicians’ beliefs and preferences may affect the decision-making process and clinical outcomes [35–38]. LT for ALD is not the exception. Although there is no perfect selection system to identify candidates for a LT in ALD, a combination of measurable biomarkers, epidemiological, clinical, and behavioral factors may achieve lower relapse rates and better survival outcomes. Further studies are needed to identify the best evidence-based strategy under the framework of precision medicine.