Circadian Variation of Death in Hemodialysis Patients

Background There is a circadian variation of death in nondialysis populations, with more cardiovascular events occurring in the morning. Whether this holds true in hemodialysis patients was never investigated. Study Design Case series. Setting & Participants All prevalent (>3 months on hemodialysis therapy) and incident (≤3 months on hemodialysis therapy) patients of a dialysis network followed up prospectively for 18 months. Predictors Patient characteristics and circumstances of death. Outcomes & Measurements Time of death. Data for time of death were collected within 72 hours of the event. The frequency of deaths occurring in the morning hours (4:01 am to 12:00 noon) was compared with that expected by chance alone. Results Time of death could be defined in 873 of 927 deaths (94.2%). In 459 prevalent hemodialysis patients, morning deaths occurred 24.8% more frequently than expected ( P < 0.001). No similar excess was observed in the 414 incident hemodialysis patients ( P = 0.9). In logistic regression, significant predictors of death occurring from 4:01 am to 12:00 noon in all subjects were being an outpatient (odds ratio [OR], 1.89; 95% confidence interval [CI], 1.24 to 2.88) and time since the end of the last dialysis treatment (postdialysis 49- to 72-hour period: OR, 1.73 [95% CI, 1.13 to 2.64] compared with 1 to 24 hours postdialysis). Considering prevalent hemodialysis patients only, being an outpatient (OR, 1.93; 95% CI, 1.17 to 3.18), postdialysis 25- to 48- and 49- to 72-hour periods (OR, 1.68; 95% CI, 1.05 to 2.68 and OR, 1.80; 95% CI, 1.03 to 3.12), diabetes (OR, 1.73; 95% CI, 1.14 to 2.63]), and β-blocker use (OR, 1.62; 95% CI, 1.08 to 2.43) were directly related and the presence of medical symptoms during the last dialysis treatment (OR, 0.53; 95% CI, 0.34 to 0.83) was inversely related to the risk of morning death. Limitations No information for causes of deaths was gathered. Conclusions Prevalent hemodialysis patients have an excess of morning deaths, and its predictors suggest potential avenues for intervention studies.