Nomogram to Predict the Risk of Protein Energy Wasting in Patients undergoing hemodialysis

SPSS software (IBM, Chicago, IL) and R software (version 3.6.1; http://www.Rproject.org) were used for statistical analyses. Missing data were handled by single imputation. Means ± SDs or median with interquartile range were used for normally and non-normally descriptive statistics of continuous variables, respectively. Chi-square test for comparison of constituent ratios and independent-Sample Test was used to compare between PEW risk and without PEW risk among the primary group and validation group. Skewed data among the groups were analyzed using Kruskal - Wallis test. Multivariable logistic regression analysis of risk factors began with the following clinical candidate predictors: age, gender, waist circumference, hipline, Kt/V, fat mass, BMI, albumin, prealbumin, Scr, total cholesterol, upper arm muscle circumference, FGF23 and Klotho. Backward step-wise selection was applied by using the likelihood ratio test with Akaike’s information criterion as the stopping rule.

To provide the clinician with a quantitative tool to predict individual probability of PEW, we built the nomogram on the basis of multivariable logistic analysis in the primary group and validation populations to determine the nomogram predicted probability of PEW. Calibration curves were plotted to assess the calibration of the nomogram, accompanied with the Hosmer-Lemeshow test. To quantify the discrimination performance of the nomogram, Harrell s C-index was measured. The PEW nomogram was subjected to bootstrapping validation (1,000 bootstrap re-samples) to calculate a relatively corrected C-index. Throughout the study, P<0.05 was taken as the minimum level of statistical significance.

本研究采用SPSS统计软件（IBM公司，美国伊利诺伊州芝加哥市）与R软件（版本3.6.1；http://www.Rproject.org）开展统计学分析。缺失数据采用单一插补法进行处理。针对连续变量的描述性统计，正态分布资料以均数±标准差（SD）表示，非正态分布资料以中位数及四分位数间距表示。采用卡方检验对各组构成比进行比较，采用独立样本t检验分别比较训练组与验证组内蛋白质能量消耗（Protein Energy Wasting，PEW）风险人群与非PEW风险人群的差异。组间偏态分布数据则采用Kruskal-Wallis秩和检验进行分析。本研究针对危险因素构建多因素logistic回归模型，初始纳入的临床候选预测变量包括：年龄、性别、腰围、臀围、Kt/V、脂肪量、体重指数（BMI）、白蛋白、前白蛋白、血肌酐（Scr）、总胆固醇、上臂肌围、成纤维细胞生长因子23（FGF23）及Klotho蛋白。采用似然比检验进行向后逐步回归筛选，以赤池信息准则（Akaike’s Information Criterion, AIC）作为回归终止的判定标准。为向临床医师提供可定量预测个体PEW发生概率的工具，本研究基于训练组与验证人群的多因素logistic回归分析结果构建列线图（nomogram），以确定PEW的列线图预测概率。通过绘制校准曲线结合Hosmer-Lemeshow检验对列线图的校准度进行评估。为量化列线图的区分能力，本研究计算了Harrell C指数（Harrell’s C-index）。本研究采用自举验证法（1000次自举重采样）对PEW列线图进行验证，以计算校正后的C指数。本研究全程以P<0.05作为统计学显著性的最低判定标准。