County-Level Characteristics Associated with Variation in ESKD Mortality in the United States, 2010-2018

Early prediction of end-stage kidney disease using electronic health record data: a machine learning approach with a 2-year horizon

Objectives

In the United States, end-stage kidney disease (ESKD) is responsible for high mortality and significant healthcare costs, with the number of cases sharply increasing in the past 2 decades. In this study, we aimed to reduce these impacts by developing an ESKD model for predicting its occurrence in a 2-year period.

Materials and Methods

We developed a machine learning (ML) pipeline to test different models for the prediction of ESKD. The electronic health record was used to capture several kidney disease-related variables. Various imputation methods, feature selection, and sampling approaches were tested. We compared the performance of multiple ML models using area under the ROC curve (AUCROC), area under the Precision-Recall curve (PR-AUC), and Brier scores for discrimination, precision, and calibration, respectively. Explainability methods were applied to the final model.

Results

Our best model was a gradient-boosting machine with feature selection and imputation methods as additional components. The model exhibited an AUCROC of 0.97, a PR-AUC of 0.33, and a Brier score of 0.002 on a holdout test set. A chart review analysis by expert physicians indicated clinical utility.

Discussion and Conclusion

An ESKD prediction model can identify individuals at risk for ESKD and has been successfully deployed within our health system.

Geography, inequities, and the social determinants of health in transplantation

Among the causes of inequity in organ transplantation, geography is oft-cited but rarely defined with precision. Traditionally, geographic inequity has been characterized by variation in distance to transplant centers, availability of deceased organ donors, or the consequences of allocation systems that are inherently geographically based. Recent research has begun to explore the use of measures at various geographic levels to better understand how characteristics of a patient's geographic surroundings contribute to a broad range of transplant inequities. Within, we first explore the relationship between geography, inequities, and the social determinants of health. Next, we review methodologic considerations essential to geographic health research, and critically appraise how these techniques have been applied. Finally, we propose how to use geography to improve access to and outcomes of transplantation.

Receiving hemodialysis in Hispanic ethnic dense communities is associated with better adherence and outcomes among young patients: a retrospective analysis of the Dialysis Outcomes and Practice Patterns Study

BACKGROUND

Hispanic ethnic density (HED) is a marker of better health outcomes among Hispanic patients with chronic disease. It is unclear whether community HED is associated with mortality risk among ethnically diverse patients receiving maintenance hemodialysis.

METHODS

A retrospective analysis of patients in the United States cohort of the Dialysis Outcomes and Practice Patterns Study (DOPPS) database (2011-2015) was conducted (n = 4226). DOPPS data was linked to the American Community Survey database by dialysis facility zip code to obtain % Hispanic residents (HED). One way ANOVA and Kruskal Wallis tests were used to estimate the association between tertiles of HED with individual demographic, clinical and adherence characteristics, and facility and community attributes. Multivariable Cox proportional hazards models were used to estimate the mortality hazard ratio (HR) and 95% CIs by tertile of HED, stratified by age; a sandwich estimator was used to account for facility clustering.

RESULTS

Patients dialyzing in facilities located in the highest HED tertile communities were younger (61.4 vs. 64.4 years), more commonly non-White (62.4% vs. 22.1%), had fewer comorbidities, longer dialysis vintage, and were more adherent to dialysis treatment, but had fewer minutes of dialysis prescribed than those in the lowest tertile. Dialyzing in the highest HED tertile was associated with lower hazard of mortality (HR, 0.86; 95% CI, 0.72-1.00), but this association attenuated with the addition of individual race/ethnicity (HR, 0.92; 95% CI, 0.78-1.09). In multivariable age-stratified analyses, those younger than 64 showed a lower hazard for mortality in the highest (vs. lowest) HED tertile (HR, 0.66; 95% CI, 0.49-0.90). Null associations were observed among patients ≥ 64 years.

CONCLUSIONS

Treating in communities with greater HED and racial/ethnic integration was associated with lower mortality among younger patients which points to neighborhood context and social cohesion as potential drivers of improved survival outcomes for patients receiving hemodialysis.