A Clinical Risk Prediction Tool for 6-Month Mortality After Dialysis Initiation Among Older Adults

Abstract

Background: Information on an individual's risk for death following dialysis therapy initiation may inform the decision to initiate maintenance dialysis for older adults. We derived and validated a clinical risk prediction tool for all-cause mortality among older adults during the first 6 months of maintenance dialysis treatment.

Study design: Prediction model using retrospective administrative and clinical data.

Setting & participants: We linked administrative and clinical data to define a cohort of 2,199 older adults (age ≥ 65 years) in Alberta, Canada, who initiated maintenance dialysis therapy (excluding acute kidney injury) in May 2003 to March 2012.

Candidate predictors: Demographics, laboratory data, comorbid conditions, and measures of health system use.

Outcomes: All-cause mortality within 6 months of dialysis therapy initiation.

Analytical approach: Predicted mortality by logistic regression with 10-fold cross-validation.

Results: 375 (17.1%) older adults died within 6 months. We developed a 19-point risk score for 6-month mortality that included age 80 years or older (2 points), glomerular filtration rate of 10 to 14.9mL/min/1.73m² (1 point) or ≥15mL/min/1.73m² (3 points), atrial fibrillation (2 points), lymphoma (5 points), congestive heart failure (2 points), hospitalization in the prior 6 months (2 points), and metastatic cancer (3 points). Model discrimination (C statistic = 0.72) and calibration (Hosmer-Lemeshow χ²=10.36; P=0.2) were reasonable. As examples, a score < 5 equated to <25% of individuals dying in 6 months, whereas a score > 12 predicted that more than half the individuals would die in the first 6 months.

Limitations: The tool has not been externally validated; thus, generalizability cannot be assessed.

Conclusions: We used readily available clinical information to derive and internally validate a 7-variable tool to predict early mortality among older adults after dialysis therapy initiation. Following successful external validation, the tool may be useful as a clinical decision tool to aid decision making for older adults with kidney failure.

Keywords: End-stage renal disease (ESRD); chronic kidney failure; decision tool; dialysis initiation; hemodialysis; mortality; older adults; predictive model; risk score; shared decision-making; treatment decisions.