Background and objective

This study described the relationship of HbA1c and continuous glucose monitoring (CGM) derived glucose management indicator (CGM-derived GMI) and developed a model to estimate GMI based on clinical parameters (clinical-parameter GMI) for hospitalized DKD treated with insulin pump.

Methods

This observational study collected clinical data of hospitalized DKD treated with insulin pump between February 2022 to February 2023. According to estimated glomerular filtration rate (eGFR), 156 participants were divided into G1–2 (n = 64), G3 (n = 56) and G4–5 (n = 36). Correlation between HbA1c and CGM-derived GMI was tested. Study population was divided into training and validation set based on ratio of 6:4. In training set, a linear model was established to calculate clinical-parameter GMI. In validation set, paired t-test and residual analysis was used to examine the difference between CGM-derived GMI and clinical-parameter GMI.

Results

With renal function reduced, the correlation of HbA1c and CGM-derived GMI decreased. Meanwhile, as renal function reduced, the discordance between HbA1c and CGM-derived GMI decreased as well (P = 0.009). In training set, based on eGFR stages, anemia, albumin, FBG and HbA1c, formula to clinical-parameter GMI was established. In validation set, the differences between clinical-parameter GMI and CGM-derived GMI was around 0, with 95 % confidence interval of −1.8 % to 1.5 %.

Conclusions

HbA1c may be less accurate to reflect glycemic condition for DKD with impaired renal function. A easily accessiable model based on clinical parameter to estimate GMI may help assess glycemia for hospitalized DKD treated with insulin pump.

中文翻译：

---

连续血糖监测衍生的血糖管理指标可用于评估接受胰岛素泵治疗的糖尿病肾病住院患者的血糖状况

背景和目标

本研究描述了 HbA1c 与连续血糖监测 (CGM) 衍生的血糖管理指标 (CGM 衍生的 GMI) 的关系，并开发了一种基于临床参数 (临床参数 GMI) 的模型，用于用胰岛素泵治疗住院 DKD 来估计 GMI。

方法

该观察性研究收集了2022年2月至2023年2月期间接受胰岛素泵治疗的住院DKD的临床数据。根据估计肾小球滤过率（eGFR），将156名参与者分为G1-2（n  = 64）、G3（n  = 56）和 G4-5（n  = 36）。测试了 HbA1c 和 CGM 衍生的 GMI 之间的相关性。研究人群按照6:4的比例分为训练集和验证集。在训练集中，建立线性模型来计算临床参数GMI。在验证集中，使用配对t检验和残差分析来检查 CGM 衍生的 GMI 和临床参数 GMI 之间的差异。

结果

随着肾功能的降低，HbA1c 与 CGM 衍生的 GMI 的相关性降低。同时，随着肾功能的降低，HbA1c 与 CGM 衍生的 GMI 之间的不一致也降低（P  = 0.009）。在训练集中，根据eGFR阶段、贫血、白蛋白、FBG和HbA1c，建立了临床参数GMI的公式。在验证集中，临床参数 GMI 和 CGM 衍生的 GMI 之间的差异约为 0，95% 置信区间为 -1.8% 至 1.5%。

结论

对于肾功能受损的 DKD，HbA1c 反映血糖状况的准确性可能较低。一个基于临床参数的易于访问的模型可用于估计 GMI，可能有助于评估接受胰岛素泵治疗的住院 DKD 的血糖。