Factors influencing withdrawal from dialysis: a national registry study.

Background: Dialysis withdrawal is the third most common cause of death in patients receiving dialysis for established renal failure (ERF) in Scotland. We describe incidence, risk factors and themes influencing decision-making in a national renal registry.

Methods: Details of deaths in those receiving renal replacement therapy (RRT) for ERF in Scotland are reported to the Scottish Renal Registry via a unique mortality report. We extracted patient demographics and comorbidity, cause and location of death, duration of RRT and pertinent free text comments from 1 January 2008 to 31 December 2014. Withdrawal incidence was calculated and logistic regression used to identify significantly influential variables. Themes emerging from clinician comments were tabulated for descriptive purposes.

Results: There were 2596 deaths; median age at death was 68 [interquartile range (IQR) 58, 76] years, 41.5% were female. Median duration on RRT was 1110 (IQR 417, 2151) days. Dialysis withdrawal was the primary cause of death in 497 (19.1%) patients and withdrawal contributed to death in a further 442 cases (17.0%). The incidence was 41 episodes per 1000 patient-years. Regression analysis revealed increasing age, female sex and prior cerebrovascular disease were associated with dialysis withdrawal as a primary cause of death. Conversely, interstitial renal disease, angiographically proven ischaemic heart disease, valvular heart disease and malignancy were negatively associated. Analysis of free text comments revealed common themes, portraying an image of physical and psychological decline accelerated by acute illnesses.

Conclusions: Death following dialysis withdrawal is common. Factors important to physical independence-prior cerebrovascular disease and increasing age-are associated with withdrawal. When combined with clinician comments this study provides an insight into the clinical decline affecting patients and the complexity of this decision. Early recognition of those likely to withdraw may improve end of life care.