Trajectory of Serum Albumin Prior to Death in Patients Receiving Hemodialysis

Objective Although some clinical expert guidelines recommend regular monitoring of serum albumin levels in patients undergoing maintenance hemodialysis, little is known about the serum albumin trajectory patterns over time, and it is unclear how the trajectory change before death. Methods We performed this retrospective study using data from 421 patients receiving hemodialysis in a dialysis facility. We divided patients into died and survived groups according to whether they died during the observation period. To compare the albumin trajectories during the observation period between the died and survived groups, linear mixed-effect models and a backward timescale from the year of death or study end were used. Results During the observation period (median, 5.1 years; maximum, 8.4 years), 115 patients receiving dialysis died. The serum albumin level showed steeper decline 3 years before death in the died group than in the survived group. The difference in albumin between the died and survived groups became apparent 3 years before death (difference, 0.08 g/dL; 95% confidence interval, 0.00-0.15 g/dL; P = .04), and the difference widened over time (difference at 1 year before death, 0.24 g/dL; 95% confidence interval, 0.14-0.33 g/dL; P < .001). Furthermore, in an analysis of albumin trajectories according to cause of death, the albumin level showed a downward trend regardless of the cause of death. Conclusion The serum albumin trajectory differed between patients undergoing hemodialysis who died and who survived, supporting the importance of monitoring the albumin trajectory in clinical practice. Although some clinical expert guidelines recommend regular monitoring of serum albumin levels in patients undergoing maintenance hemodialysis, little is known about the serum albumin trajectory patterns over time, and it is unclear how the trajectory change before death. We performed this retrospective study using data from 421 patients receiving hemodialysis in a dialysis facility. We divided patients into died and survived groups according to whether they died during the observation period. To compare the albumin trajectories during the observation period between the died and survived groups, linear mixed-effect models and a backward timescale from the year of death or study end were used. During the observation period (median, 5.1 years; maximum, 8.4 years), 115 patients receiving dialysis died. The serum albumin level showed steeper decline 3 years before death in the died group than in the survived group. The difference in albumin between the died and survived groups became apparent 3 years before death (difference, 0.08 g/dL; 95% confidence interval, 0.00-0.15 g/dL; P = .04), and the difference widened over time (difference at 1 year before death, 0.24 g/dL; 95% confidence interval, 0.14-0.33 g/dL; P < .001). Furthermore, in an analysis of albumin trajectories according to cause of death, the albumin level showed a downward trend regardless of the cause of death. The serum albumin trajectory differed between patients undergoing hemodialysis who died and who survived, supporting the importance of monitoring the albumin trajectory in clinical practice.