Designing an Implementable Clinical Prediction Model for Near-Term Mortality and Long-Term Survival in Patients on Maintenance Hemodialysis

Abstract

Rationale & Objective

Life expectancy of patients treated with maintenance hemodialysis (MHD) is heterogeneous. Knowledge of life-expectancy may focus care decisions on near-term vs. long-term goals. Current tools are limited and focus on near-term mortality. Here, we develop and assess potential utility for predicting near-term mortality and long-term survival on MHD.

Study Design

Predictive modelling study.

Setting & Participants

42,351 patients contributing 997,381 patient months over 11 years, abstracted from the EHR system of mid-size, non-profit dialysis providers.

New Predictors & Established Predictors

Demographics, laboratory results, vital signs, and service utilization data available within dialysis EHR.

Outcomes

For each patient month, we ascertained death within the next 6-months (i.e., near-term mortality) and survival over more than 5-years during receipt of MHD or following kidney transplantation (i.e., long-term survival).

Analytical Approach

We used LASSO logistic regression and gradient-boosting machines to predict each outcome. We compared these to time-to-event models spanning both time horizons. We explored the performance of decision rules at different cut-points.

Results

All models achieved AUROC ≥ 0.80 and optimal calibration metrics in the test set. Long-term survival models had significantly better performance than near-term mortality models. Time-to-event models performed similarly to binary models. Applying different cutpoints spanning from the 1^st to 90^th percentile of the predictions, a positive predictive value (PPV) of 54% could be achieved for near-term mortality, but with poor sensitivity of 6%. A PPV of 71% could be achieved for long-term survival with a sensitivity of 67%.

Limitations

The retrospective models would need to be prospectively validated before they could be appropriately used as clinical decision aids.

Conclusions

A model built with readily available clinical variables to support easy implementation, can predict clinically important life expectancy thresholds and shows promise as a clinical decision support tool for patients on MHD. Predicting long-term survival has better decision rule performance than predicting near-term mortality.