Division of Nephrology, Department of Internal Medicine, Kayseri City Hospital, Kayseri, Turkey.
Ther Apher Dial. 2021 Apr;25(2):179-187. doi: 10.1111/1744-9987.13552. Epub 2020 Aug 20.
Serum albumin is a major determinant of hospitalization in patients with end-stage renal disease (ESRD). Previous reports generally use the Poisson model to evaluate the relationships between outcome and response variables. However, hospitalization data are often overdispersed, and few studies using appropriate methods exist in the literature.This retrospective cohort study included 426 patients with ESRD receiving hemodialysis treatment between 2014 and 2018. Using a negative binomial regression model with hierarchical multivariable adjustments, we investigated the relationship between serum albumin, hospital admissions, and total hospitalization days. Mean age and mean baseline serum albumin levels were 64.7 ± 11 years and 3.5 ± 0.5 g/dL, respectively. At least one hospitalization was identified in 402 (94%) patients. The incidence rate was 1.48 (95% CI, 1.41-1.56) admissions per patient-year during the follow-up period of 5 years. A negative linear association was observed between serum albumin and hospitalization frequency. Hospitalization rates (95% CI) were 1.81 (1.65-1.98), 1.44 (1.3-1.59), 1.36 (1.22-1.51), and 1.33 (1.2-1.48) per patient-year in serum albumin levels ≤3, 3.1 to ≤3.3, 3.4 to ≤3.7, and ≥3.8 g/dL, respectively. Case mix-adjusted incidence rate ratio was 0.82 (95% CI, 0.70-0.94), while it was robust to further adjustments for malnutrition and inflammation markers. Similar results were observed in hospitalization days and time to the first hospitalization. These findings, which result from the negative binomial model using overdispersed data, suggest that lower serum albumin is related to increased hospitalization rates and hospital days in incident hemodialysis patients.
血清白蛋白是终末期肾病（ESRD）患者住院的主要决定因素。以前的报告通常使用泊松模型来评估结果和反应变量之间的关系。然而，住院数据通常是过分散的，文献中很少有使用适当方法的研究。本回顾性队列研究纳入了 2014 年至 2018 年间接受血液透析治疗的 426 名 ESRD 患者。使用带有分层多变量调整的负二项回归模型，我们研究了血清白蛋白、住院和总住院天数之间的关系。平均年龄和平均基线血清白蛋白水平分别为 64.7±11 岁和 3.5±0.5 g/dL。在 402 名（94%）患者中至少有一次住院。在 5 年的随访期间，每位患者的年住院率为 1.48（95%CI，1.41-1.56）。观察到血清白蛋白与住院频率呈负线性关系。血清白蛋白水平≤3 g/dL、3.1 至≤3.3 g/dL、3.4 至≤3.7 g/dL 和≥3.8 g/dL 的患者年住院率（95%CI）分别为 1.81（1.65-1.98）、1.44（1.3-1.59）、1.36（1.22-1.51）和 1.33（1.2-1.48）。病例组合调整后的发病率比为 0.82（95%CI，0.70-0.94），而进一步调整营养不良和炎症标志物后结果仍然稳健。在住院天数和首次住院时间方面也观察到了类似的结果。这些结果来自使用过分散数据的负二项模型，表明较低的血清白蛋白与新发生的血液透析患者的住院率和住院天数增加有关。
