PDF(Pubmed)
Abstract：
BACKGROUND: Most studies had shown a linear relationship between serum albumin (sALB) and the prevalence of diabetic retinopathy (DR). Thus, the purpose of this study is to investigate whether their relationship is non-linear.
METHODS: We included 426 patients with type 2 diabetes who were hospitalized in Guangdong Provincial People\'s Hospital from December 2017 to November 2018. The outcome was the prevalence of DR. A two-piecewise logistics regression model was performed to identify the non-linear relationship between sALB and the prevalence of DR. The inflection point was calculated to determine the saturation effect through the maximum likelihood ratio and a recursive algorithm.
RESULTS: DR was diagnosed in 167 of 426 type 2 diabetic patients. The relationship between sALB and DR was nonlinear. When sALB was less than 38.10 g/L, a significant negative association was observed (OR = 0.82; 95% CI, 0.72-0.94; P = 0.0037), while no significant association was observed when sALB was greater than 38.10 g/L (OR = 1.12; 95% CI, 0.92-1.35; P = 0.2637).
CONCLUSIONS: The relationship between sALB and the prevalence of DR is non-linear. sALB is negatively associated with the prevalence of DR when sALB is less than 38.10 g/L. Our findings need to be confirmed by further prospective research.
METHODS: We included 426 patients with type 2 diabetes who were hospitalized in Guangdong Provincial People\'s Hospital from December 2017 to November 2018. The outcome was the prevalence of DR. A two-piecewise logistics regression model was performed to identify the non-linear relationship between sALB and the prevalence of DR. The inflection point was calculated to determine the saturation effect through the maximum likelihood ratio and a recursive algorithm.
RESULTS: DR was diagnosed in 167 of 426 type 2 diabetic patients. The relationship between sALB and DR was nonlinear. When sALB was less than 38.10 g/L, a significant negative association was observed (OR = 0.82; 95% CI, 0.72-0.94; P = 0.0037), while no significant association was observed when sALB was greater than 38.10 g/L (OR = 1.12; 95% CI, 0.92-1.35; P = 0.2637).
CONCLUSIONS: The relationship between sALB and the prevalence of DR is non-linear. sALB is negatively associated with the prevalence of DR when sALB is less than 38.10 g/L. Our findings need to be confirmed by further prospective research.
摘要：
背景：大多数研究表明血清白蛋白（sALB）与糖尿病视网膜病变（DR）的患病率之间存在线性关系。因此，这项研究的目的是调查它们之间的关系是否是非线性的。
方法：纳入2017年12月至2018年11月广东省人民医院住院的426例2型糖尿病患者。结果是DR的患病率。使用两分段物流回归模型来确定sALB与DR患病率之间的非线性关系。通过最大似然比和递归算法计算拐点以确定饱和效应。
结果：在426例2型糖尿病患者中有167例被诊断为DR。sALB与DR之间呈非线性关系。当sALB小于38.10g/L时，观察到显着的负相关（OR=0.82;95％CI，0.72-0.94;P=0.0037），而当sALB大于38.10g/L时没有观察到显著关联(OR=1.12;95%CI,0.92-1.35;P=0.2637)。
结论：sALB与DR患病率之间的关系是非线性的。当sALB小于38.10g/L时，sALB与DR的患病率呈负相关。我们的发现需要进一步的前瞻性研究来证实。
方法：纳入2017年12月至2018年11月广东省人民医院住院的426例2型糖尿病患者。结果是DR的患病率。使用两分段物流回归模型来确定sALB与DR患病率之间的非线性关系。通过最大似然比和递归算法计算拐点以确定饱和效应。
结果：在426例2型糖尿病患者中有167例被诊断为DR。sALB与DR之间呈非线性关系。当sALB小于38.10g/L时，观察到显着的负相关（OR=0.82;95％CI，0.72-0.94;P=0.0037），而当sALB大于38.10g/L时没有观察到显著关联(OR=1.12;95%CI,0.92-1.35;P=0.2637)。
结论：sALB与DR患病率之间的关系是非线性的。当sALB小于38.10g/L时，sALB与DR的患病率呈负相关。我们的发现需要进一步的前瞻性研究来证实。
