Towards precision well-being in medical education.

Medical trainee well-being is often met with generalized solutions that overlook substantial individual variations in mental health predisposition and stress reactivity. Precision medicine leverages individual environmental, genetic, and lifestyle factors to tailor preventive and therapeutic interventions. In addition, an exclusive focus on clinical mental illness tends to disregard the importance of supporting the positive aspects of medical trainee well-being. We introduce a novel precision well-being framework for medical education that is built on a comprehensive and individualized view of mental health, combining measures from mental health and positive psychology in a unified, data-driven framework. Unsupervised machine learning techniques commonly used in precision medicine were applied to uncover patterns within multidimensional mental health data of medical students. Using data from 3,632 US medical students, clusters were formulated based on recognized metrics for depression, anxiety, and flourishing. The analysis identified three distinct clusters. Membership in the 'Healthy Flourishers' well-being phenotype was associated with no signs of anxiety or depression while simultaneously reporting high levels of flourishing. Students in the 'Getting By' cluster reported mild anxiety and depression and diminished flourishing. Membership in the 'At-Risk' cluster was associated with high anxiety and depression, languishing, and increased suicidality. Nearly half (49%) of the medical students surveyed were classified as 'Healthy Flourishers', whereas 36% were grouped into the 'Getting-By' cluster and 15% were identified as 'At-Risk'. Findings show that a substantial portion of medical students report diminished well-being during their studies, with a significant number struggling with mental health challenges. This novel precision well-being framework represents an integrated empirical model that classifies individual medical students into distinct and meaningful well-being phenotypes based on their holistic mental health. This approach has direct applicability to student support and can be used to evaluate the effectiveness of personalized intervention strategies stratified by cluster membership.

Optimal cholesterol treatment plans and genetic testing strategies for cardiovascular diseases.

Atherosclerotic cardiovascular disease (ASCVD) is among the leading causes of death in the US. Although research has shown that ASCVD has genetic elements, the understanding of how genetic testing influences its prevention and treatment has been limited. To this end, we model the health trajectory of patients stochastically and determine treatment and testing decisions simultaneously. Since the cholesterol level of patients is one controllable risk factor for ASCVD events, we model cholesterol treatment plans as Markov decision processes. We determine whether and when patients should receive a genetic test using value of information analysis. By simulating the health trajectory of over 64 million adult patients, we find that 6.73 million patients undergo genetic testing. The optimal treatment plans informed with clinical and genetic information save 5,487 more quality-adjusted life-years while costing $1.18 billion less than the optimal treatment plans informed with clinical information only. As precision medicine becomes increasingly important, understanding the impact of genetic information becomes essential.

Projected increase in obesity and non-alcoholic-steatohepatitis-related liver transplantation waitlist additions in the United States.

Nonalcoholic steatohepatitis (NASH) cirrhosis is the fastest growing indication for liver transplantation (LT) in the United States. We aimed to determine the temporal trend behind the rise in obesity and NASH-related additions to the LT waitlist in the United States and make projections for future NASH burden on the LT waitlist. We used data from the Organ Procurement and Transplantation Network database from 2000 to 2014 to obtain the number of NASH-related LT waitlist additions. The obese population in the United States from 2000 to 2014 was estimated using data from the U.S. Census Bureau and the National Health and Nutrition Examination Survey. Based on obesity trends, we established a time lag between obesity prevalence and NASH-related waitlist additions. We used data from the U.S. Census Bureau on population projections from 2016 to 2030 to forecast obesity estimates and NASH-related LT waitlist additions. From 2000 to 2014, the proportion of obese individuals significantly increased 44.9% and the number of NASH-related annual waitlist additions increased from 391 to 1,605. Increase in obesity prevalence was strongly associated with LT waitlist additions 9 years later in derivation and validation cohorts (R2 = 0.9). Based on these data, annual NASH-related waitlist additions are anticipated to increase by 55.4% (1,354-2,104) between 2016 and 2030. There is significant regional variation in obesity rates and in the anticipated increase in NASH-related waitlist additions (P < 0.01). Conclusion: We project a marked increase in demand for LT for NASH given population obesity trends. Continued public health efforts to curb obesity prevalence are needed to reduce the projected future burden of NASH. (Hepatology 2017).

Identifying key risk factors for premature discontinuation of opioid use disorder treatment in the United States: A predictive modeling study.

BACKGROUND: Treatment for opioid use disorder (OUD), particularly medication for OUD, is highly effective; however, retention in OUD treatment is a significant challenge. We aimed to identify key risk factors for premature exit from OUD treatment. METHODS: We analyzed 2,381,902 cross-sectional treatment episodes for individuals in the U.S., discharged between Jan/1/2015 and Dec/31/2019. We developed classification models (Random Forest, Classification and Regression Trees (CART), Bagged CART, and Boosted CART), and analyzed 31 potential risk factors for premature treatment exit, including treatment characteristics, substance use history, socioeconomic status, and demographic characteristics. We stratified our analysis based on length of stay in treatment and service setting. Models were compared using cross-validation and the receiver operating characteristic area under the curve (ROC-AUC). RESULTS: Random Forest outperformed other methods (ROC-AUC: 74%). The most influential risk factors included characteristics of service setting, geographic region, primary source of payment, and referral source. Race, ethnicity, and sex had far weaker predictive impacts. When stratified by treatment setting and length of stay, employment status and delay (days waited) to enter treatment were among the most influential factors. Their importance increased as treatment duration decreased. Notably, importance of referral source increased as the treatment duration increased. Finally, age and age of first use were important factors for lengths of stay of 2-7 days and in detox treatment settings. CONCLUSIONS: The key factors of OUD treatment attrition identified in this analysis should be more closely explored (e.g., in causal studies) to inform targeted policies and interventions to improve models of care.

A machine learning approach for the prediction of overall deceased donor organ yield.

BACKGROUND: Optimizing organ yield (number of organs transplanted per donor) is a potentially modifiable way to increase the number of organs available for transplant. Models to predict the expected deceased donor organ yield have been developed based on ordinary least squares regression and logistic regression. However, alternative modeling methodologies incorporating machine learning may have superior performance compared with conventional approaches. METHODS: We evaluated the predictive accuracy of 14 machine learning models for predicting overall organ yield in a cross-validation procedure. The models were parameterized using data from the Organ Procurement and Transplantation Network database from 2000 to 2018. The inclusion criteria for the study were adult deceased donors between 18 and 84 years of age that had at least 1 organ procured for transplantation. RESULTS: A total of 89,520 donors met the inclusion criteria. Their mean (standard deviation) age was 44 (15) years, and approximately 58% were male. Our cross-validation analysis showed that a tree-based gradient boosting model outperformed the remaining 13 models. Compared with the currently used prediction models, the gradient boosting model improves prediction accuracy by reducing the mean absolute error between 3 and 11 organs per 100 donors. CONCLUSION: Our analysis demonstrated that the gradient boosting methodology had the best performance in predicting overall deceased donor organ yield and can potentially serve as an aid to assess organ procurement organization performance.

The Magnitude of the Health and Economic Impact of Increased Organ Donation on Patients With End-Stage Renal Disease.

Objectives. There are several approaches such as presumed consent and compensation for deceased donor organs that could reduce the gap between supply and demand for kidneys. Our objective is to evaluate the magnitude of the economic impact of policies to increase deceased donor organ donation in the United States. Methods. We built a Markov model and simulate an open cohort of end-stage renal disease patients awaiting kidney transplantation in the United States over 20 years. Model inputs were derived from the United States Renal Data System and published literature. We evaluate the magnitude of the health and economic impact of policies to increase deceased donor kidney donation in the United States. Results. Increasing deceased kidney donation by 5% would save $4.7 billion, and gain 30,870 quality-adjusted life years over the lifetime of an open cohort of patients on dialysis on the waitlist for kidney transplantation. With an increase in donations of 25%, the cost saved was $21 billion, and 145,136 quality-adjusted life years were gained. Policies increasing deceased kidney donation by 5% could pay donor estates $8000 or incur a onetime cost of up to $4 billion and still be cost-saving. Conclusions. Increasing deceased kidney donation could significantly impact national spending and health for end-stage renal disease patients.

Assessment of National Organ Donation Rates and Organ Procurement Organization Metrics.

Importance: Organ transplant is a life-saving procedure for patients with end-stage organ failure. In the US, organ procurement organizations (OPOs) are responsible for the evaluation and procurement of organs from donors who have died; however, there is controversy regarding what measures should be used to evaluate their performance. Objective: To evaluate OPO performance metrics using combined mortality and donation data and quantify the associations of population demographics with donation metrics. Design, Setting, and Participants: This national cohort study includes data from the US organ transplantation system from January 2008 through December 2017. All individuals who died within the US, as reported by the National Death index, were included. Exposures: Death, organ donation, and donation eligibility. Main Outcomes and Measures: Evaluation of the variation in donation metrics and the use of ineligible donors by OPO and demographic subgroup. Results: This study included 17 501 742 deaths and 75 769 deceased organ donors (45 040 men [59.4%]; 51 908 White individuals [68.5%]). Of these donors, 15 857 (20.9%) were not eligible, as defined by the OPOs. The median donation metrics by OPO were 0.004 (range, 0.002-0.012) donors per death, 0.89 (range, 0.68-1.30) donors per eligible death, and 0.72 (range, 0.57-0.86) eligible donors per eligible death. The OPOs in the upper quartile of the overall eligible donors per eligible death metric were in the upper quartile of annual rankings on 90 of 140 occasions (64.3%). There was little overlap in top-performing OPOs between metrics; an OPO in the upper quartile for 1 metric was also in the upper quartile for the other metrics on 37 of 570 occasions (6.5% of the time). The median donor eligibility rate, defined as the number of eligible donors per donor, was 0.79 (range, 0.61-0.95) across OPOs. Age (eg, 65 to 84 years, coefficient, -0.55 [SE, 0.03]; P < .001; vs those aged 18 to 34 years), sex (male individuals, -0.09 [SE, 0.02]; P < .001; vs female individuals), race (eg, Black individuals, 0.35 [SE, 0.02]; P < .001; vs White individuals), cause of death (eg, central nervous system tumor, 0.48 [SE, 0.08]; P < .001; vs anoxia), year (eg, 2016-2017: -0.10 [SE, 0.03]; P < .001; vs 2008-2009), and OPO were associated with the use of ineligible donors; OPO was a significant factor associated with performance in all metrics (χ256, 500.5; P < .001; coefficient range across individual OPOs, -0.15 [SE, 0.09] to 0.75 [SE, 0.09]), even after accounting for population differences. Female and non-White individuals were significantly less likely to be used as ineligible donors. Conclusions and Relevance: We demonstrate significant variability in OPO performance rankings, depending on which donation metric is used. There were significant differences in OPO performance, even after accounting for differences in potential donor populations. Our data suggest significant variation in use of ineligible donors among OPOs, a source for increased donors. The performance of OPOs should be evaluated using a range of donation metrics.

Estimated Association Between Organ Availability and Presumed Consent in Solid Organ Transplant.

Importance: Presumed consent, or an opt-out organ transplant policy, has been adopted by many countries worldwide to increase organ donation. The implication of such a policy for transplants in the United States is uncertain, however. Objective: To simulate the potential implications of a presumed consent policy in the United States. Design, Setting, and Participants: In a decision analytical model, a simulation model was developed using cohort data from January 1, 2004, to December 31, 2014, in the Organ Procurement and Transplantation Network Standard Transplant Analysis and Research files. All US patients (n = 524 359) who were on the waiting list for at least 1 solid organ and all deceased organ donors during the study period were included in the analyses. All data and statistical analyses were performed from January 30, 2019, to July 31, 2019. Main Outcomes and Measures: Increase in the organs available for donation and life-years gained associated with a 5%, 15%, or 25% increase in deceased donors, based on the published changes from a presumed consent policy. Results: This study considered 524 359 unique candidates (aged ≥18 years; 320 908 [61.2%] male) for a solid organ transplant from January 1, 2004, to December 31, 2014. With a base case scenario of a 5% presumed consent-associated increase in donors, the removals (owing to death or illness) from the waiting list for all organs would have an associated 3.2% to 10.4% mean reduction, depending on the random or ideal allocation of new organs to patients on the waiting list. Sensitivity analyses showed that waiting list removals could be decreased up to 52%; however, this reduction was not enough to completely eliminate waiting list removals during the study period. The biggest estimated increases in annual life-years gained associated with a presumed consent policy were in kidney transplant candidates (95% CIs by deceased donor increase: 5% increase, 3440-3466 years; 15% increase, 10 321-10 399 years; 25% increase, 17 201-17 332 years) and liver transplant candidates (95% CIs by deceased donor increase: 5% increase, 898-905 years; 15% increase, 2693-2714 years; 25% increase, 4448-4523 years). Adoption of a presumed consent policy could result in a 4295-year (95% CI, 4277-4313 years) to 11 387-year (95% CI, 11 339-11 435 years) increase in life-years, accounting for the survival advantages associated with a transplant. Conclusions and Relevance: In this study, presumed consent was estimated to be associated with modest but important improvement in the number of organ transplants and increases in life-years gained for patients awaiting an organ transplant. Further consideration and even debate about the ethical and public policy implications of a presumed consent policy are warranted.

Population-Based Analysis and Projections of Liver Supply Under Redistricting.

BACKGROUND: To reduce the geographic heterogeneity in liver transplant allocation, the United Network of Organ Sharing has proposed redistricting, which is impacted by both donor supply and liver transplantation demand. We aimed to determine the impact of demographic changes on the redistricting proposal and characterize causes behind geographic heterogeneity in donor supply. METHODS: We analyzed adult donors from 2002 to 2014 from the United Network of Organ Sharing database and calculated regional liver donation and utilization stratified by age, race, and body mass index. We used US population data to make regional projections of available donors from 2016 to 2025, incorporating the proposed 8-region redistricting plan. We used donors/100 000 population age 18 to 84 years (D/100K) as a measure of equity. We calculated a coefficient of variation (standard deviation/mean) for each regional model. We performed an exploratory analysis where we used national rates of donation, utilization and both for each regional model. RESULTS: The overall projected D/100K will decrease from 2.53 to 2.49 from 2016 to 2025. The coefficient of variation in 2016 is expected to be 20.3% in the 11-region model and 13.2% in the 8-region model. We found that standardizing regional donation and utilization rates would reduce geographic heterogeneity to 4.9% in the 8-region model and 4.6% in the 11-region model. CONCLUSIONS: The 8-region allocation model will reduce geographic variation in donor supply to a significant extent; however, we project that geographic disparity will marginally increase over time. Though challenging, interventions to better standardize donation and utilization rates would be impactful in reducing geographic heterogeneity in organ supply.

Predictors of Deceased Donor Kidney Discard in the United States.

BACKGROUND: Renal transplantation is a lifesaving intervention for end-stage renal disease. The demand for renal transplantation outweighs the availability of organs; however, up to 20% of recovered kidneys are discarded before transplantation. We aimed to better characterize the risk factors for deceased donor kidney discard. METHODS: We performed a secondary analysis of the Organ Procurement and Transplantation Network database from 2000 to 2012 of all solid organ donors. The cohort was split into training (80%) and validation (20%) subsets. We performed a stepwise logistic regression to develop a multivariate risk prediction model for kidney graft discard and validated the model. The performance of the models was evaluated with respect to calibration, and area under the curve (AUC) of receiver operating characteristic curves. RESULTS: There were no significant baseline differences between the training (n = 57 474) and validation (n = 14 368) cohorts. The multivariate model validation showed very good discriminant function in predicting kidney discard (AUC = 0.84). Predictors of increased discard included age older than 50 years, performance of a kidney biopsy, cytomegalovirus seropositive status, donation after cardiac death, hepatitis B and C seropositive status, cigarette use, diabetes, hypertension, terminal creatinine greater than 1.5 mg/dL and AB blood type. The model outperformed the Kidney Donor Risk Index in predicting discard (P < 0.001). Subgroup analysis of expanded criteria donor kidneys demonstrated good discrimination with an AUC of 0.70. CONCLUSIONS: We have characterized several important predictors of deceased donor kidney discard. Better understanding of factors that lead to increased deceased donor kidney discard can allow for targeted interventions to reduce discard.

Data-Driven Markov Decision Process Approximations for Personalized Hypertension Treatment Planning.

Background: Markov decision process (MDP) models are powerful tools. They enable the derivation of optimal treatment policies but may incur long computational times and generate decision rules that are challenging to interpret by physicians. Methods: In an effort to improve usability and interpretability, we examined whether Poisson regression can approximate optimal hypertension treatment policies derived by an MDP for maximizing a patient's expected discounted quality-adjusted life years. Results: We found that our Poisson approximation to the optimal treatment policy matched the optimal policy in 99% of cases. This high accuracy translates to nearly identical health outcomes for patients. Furthermore, the Poisson approximation results in 104 additional quality-adjusted life years per 1000 patients compared to the Seventh Joint National Committee's treatment guidelines for hypertension. The comparative health performance of the Poisson approximation was robust to the cardiovascular disease risk calculator used and calculator calibration error. Limitations: Our results are based on Markov chain modeling. Conclusions: Poisson model approximation for blood pressure treatment planning has high fidelity to optimal MDP treatment policies, which can improve usability and enhance transparency of more personalized treatment policies.