Family Perceptions of Quality of End-of-Life Care for Veterans with Advanced CKD

Abstract

Background and objectives: Little is known about the quality of end-of-life care for patients with advanced CKD. We describe the relationship between patterns of end-of-life care and dialysis treatment with family-reported quality of end-of-life care in this population.

Design, setting, participants, & measurements: We designed a retrospective observational study among a national cohort of 9993 veterans with advanced CKD who died in Department of Veterans Affairs facilities between 2009 and 2015. We used logistic regression to evaluate associations between patterns of end-of-life care and receipt of dialysis (no dialysis, acute dialysis, maintenance dialysis) with family-reported quality of end-of-life care.

Results: Overall, 52% of cohort members spent ≥2 weeks in the hospital in the last 90 days of life, 34% received an intensive procedure, and 47% were admitted to the intensive care unit, in the last 30 days, 31% died in the intensive care unit, 38% received a palliative care consultation in the last 90 days, and 36% were receiving hospice services at the time of death. Most (55%) did not receive dialysis, 12% received acute dialysis, and 34% received maintenance dialysis. Patients treated with acute or maintenance dialysis had more intensive patterns of end-of-life care than those not treated with dialysis. After adjustment for patient and facility characteristics, receipt of maintenance (but not acute) dialysis and more intensive patterns of end-of-life care were associated with lower overall family ratings of end-of-life care, whereas receipt of palliative care and hospice services were associated with higher overall ratings. The association between maintenance dialysis and overall quality of care was attenuated after additional adjustment for end-of-life treatment patterns.

Conclusions: Among patients with advanced CKD, care focused on life extension rather than comfort was associated with lower family ratings of end-of-life care regardless of whether patients had received dialysis.

Keywords: bereaved family; chronic kidney disease; chronic renal insufficiency; death; dialysis; end-of-life care; hospice care; hospices; humans; intensive care units; intensive treatment; kidney failure; life expectancy; logistic models; palliative care; patient-centered care; quality of life; referral and consultation; retrospective studies; terminal care; veterans.

Copyright © 2019 by the American Society of Nephrology.

Figures

Graphical abstract

Figure 1.

Cohort derivation. *Denominator includes patients who died in VA inpatient settings during the study period (n=10,800). **Percentages do not sum to 100% due to rounding.

Figure 2.

Association of dialysis treatment status with end-of-life treatment patterns. Parameters are predicted probabilities; reference group for acute and maintenance dialysis is no dialysis and reference group for end-of-life treatment variables is no receipt; model adjusted for race, age, sex, next of kin, region, facility complexity, year of death, and individual comorbidities, standard errors adjusted for facility-level clustering. n=9993. ICU, intensive care unit.

Figure 3.

Association of dialysis and end-of-life treatment with excellent overall care. Parameters are predicted probabilities of excellent rating versus all others; reference group for acute and maintenance dialysis is no dialysis and reference group for end-of-life treatment variables is no receipt; model adjusted for race, age, sex, next of kin, region, facility complexity, year of death, individual comorbidities and Quan quantile, weighted for survey nonresponse and missing items imputed, standard errors adjusted for facility-level clustering. n=5435. ICU, Intensive care unit.

Figure 4.

Association of dialysis treatment status with most favorable response on individual bereaved family survey items. Presented are predicted probabilities of giving most favorable response versus all others; reference group is no dialysis; model is adjusted for race, age, sex, next of kin, region, facility complexity, year of death, individual comorbidities, weighted for survey nonresponse and missing items imputed; standard errors adjusted for facility-level clustering. n=5435.