第一部分 高钾血症与中国住院心力衰竭患者预后关联

背景：高钾血症是心力衰竭常见的并发症，国外研究提示高钾血症与心衰患者不良预后相关。然而，目前中国心衰患者合并高钾血症的流行病学特征及高钾血症对心衰患者预后影响的大样本量研究仍相对缺乏。本研究拟探究高钾血症对中国住院心衰患者出院后1年内全因死亡及心源性死亡的影响，同时探究最佳血钾水平区间。

方法：本研究纳入2017年1月至2020年12月入组的158349名住院心衰患者，获取受试者基线信息，包括人口学特征、合并症、实验室检查结果及出院用药情况，同时对患者进行随访，获取患者出院后1年内全因死亡及心源性死亡事件信息。高钾血症定义为血清钾>5.0 mmol/L。采用Cox比例风险回归模型评估高钾血症与预后事件的关联，并进行亚组分析。使用限制性立方样条模型探讨住院期间血钾水平与住院心衰患者1年内全因死亡风险之间的非线性关系。

结果：纳入的158349例住院心衰患者中，平均年龄为69.2±13.3岁，女性占41.0%。其中，8876例（5.6%）患者在住院期间合并高钾血症，轻度、中度及重度高钾血症比例分别为4.4%、0.9% 和 0.3%。Kaplan-Meier生存曲线及Cox回归分析结果显示，高钾血症是住院心衰患者出院后1年内全因死亡（HR=1.29，95%CI: 1.23-1.35, P＜0.001）及心源性死亡（HR=1.28, 95%CI:1.21-1.36, P＜0.001）风险增加的独立相关因素。亚组分析显示该关联在不同亚组人群中具有一致性。心衰患者住院血钾水平与全因死亡风险呈“U”型关系，在血钾4.3mmol/L时心衰患者全因死亡风险最小。

结论：本研究结果提示高钾血症是住院心衰患者1年内发生全因死亡和心源性死亡风险增加的独立相关因素，强调了常规血钾检测与精准血钾管理在心衰患者治疗策略中的重要意义。

第二部分 高钾血症与中国住院HFrEF患者RAASi使用的关联

背景：射血分数下降的心力衰竭（Heart Failure with Reduced Ejection Fraction, HFrEF）患者规范使用肾素-血管紧张素-醛固酮系统抑制剂(Renin-Angiotensin-Aldosterone System inhibitors, RAASi)药物治疗可显著改善预后。本研究第一部分结果提示，合并高钾血症的心衰患者出院时RAASi使用率更低。既往研究证据提示高钾血症阻碍心衰患者RAASi使用。目前国内尚无关于高钾血症对中国HFrEF患者出院后RAASi使用情况及长期用药依从性影响的相关研究证据。本研究拟探讨住院期间合并高钾血症对HFrEF患者出院时和随访期间RAASi使用率及RAASi撤药的影响。

方法：本研究纳入2017年1月至2022年12月入组的住院HFrEF患者124257例。高钾血症定义为血清钾>5.0 mmol/L。按出院前血钾水平将患者分为高钾血症组和正常血钾组，比较两组基线特征及出院时RAASi使用差异，采用多因素Logistic回归分析相关影响因素。对完成1年随访的44139例患者，进一步分析两组患者在1个月、3个月及1年随访时RAASi使用率及撤药率。肾素-血管紧张素抑制剂（RASi）使用定义为心衰患者出院带药或随访记录中使用任意一种血管紧张素转换酶抑制剂（Angiotensin-Converting Enzyme Inhibitor, ACEI）、血管紧张素Ⅱ受体阻滞剂（Renin-Angiotensin System Inhibitors , ARB）或血管紧张素受体-脑啡肽酶抑制剂（Angiotensin Receptor–Neprilysin Inhibitor, ARNI）。RASi药物撤药定义为出院时使用RASi，但在随访时间点时未被记录使用。醛固酮受体抑制剂（Mineralocorticoid Receptor Antagonist, MRA）药物使用及撤药定义参照上述，除P值外，同时采用标准化均数差(Standardized Mean Difference, SMD)评估组间差异，SMD≥10%认为两组之间差异程度大。

结果：纳入的124257例住院HFrEF患者，整体平均年龄为 65.3±14.0 岁，女性占比为 30.7%。高钾血症组HFrEF患者出院时RASi和MRA使用率显著低于正常血钾组（RASi:76.6%比86.1%，MRA: 77.1%比85.4%，均P<0.001，SMD≥10%）。多因素logistic回归校正性别、年龄、合并症等因素后，高钾血症为出院时RASi(OR=0.79, 95%CI:0.63-0.98, P=0.030)和MRA(OR=0.54, 95%CI:0.43-0.68, P<0.001)使用率减低的独立相关因素。随访结果显示，高钾血症组在1个月、3个月、1年各随访时间点RASi、MRA及两者联合药物使用率显著低于正常血钾组(P<0.001，SMD≥10%)。高钾血症组在随访各时间点的撤药率与正常血钾组无统计学差异。将高钾血症组患者进一步分为中重度高钾组(基线血钾≥5.5mmol/L)，结果显示中重度高钾血症组1年RASi和MRA撤药率均高于正常血钾组（RASi:28.1%比16.3%，MRA:33.8%比23.7%，均P<0.001，SMD≥10%）。

结论：住院期间合并高钾血症是HFrEF患者出院RAASi药物使用率减低的独立相关因素；中重度高钾血症(血钾≥5.5mmol/L)与HFrEF心衰患者出院1年内RAASi撤药风险升高相关。需要优化血钾管理以改善HFrEF患者的药物治疗依从性及长期预后。

第三部分 心衰患者1月内高钾血症发生预测模型建立

背景：高钾血症使心衰患者预后不良，且阻碍心衰患者肾素-血管紧张素-醛固酮系统抑制剂（Renin-Angiotensin-Aldosterone System inhibitors, RAASi）规范使用。识别高钾血症高危心衰患者对于制定个体化随访策略与药物调整方案，改善患者临床结局具有重要意义。目前国内尚缺乏针对心衰患者的高钾血症风险预测模型。本研究拟构建心衰患者1月内高钾血症发生预测模型，有助于指导临床血钾监测的频率。

方法：本研究纳入2017-2020年入组的14,727例住院心衰患者。高钾血症定义为血清钾>5.0 mmol/L。研究人群按7:3比例随机分配为训练集与验证集。采用随机森林算法和最小绝对收缩和选择算子回归(Least Absolute Shrinkage and Selection Operator, LASSO)回归筛选预测变量，进一步建立多因素Logistic回归模型，绘制列线图以建立预测工具。采用受试者工作特征（Receiver Operating Characteristic, ROC）曲线和决策曲线分析（Decision Curve Analysis, DCA）分别对训练集与验证集评估模型的预测效能和临床实用性。

结果：纳入的14727例心衰患者中，平均年龄为67.9±13.9岁，女性占比39.8%，基线血清钾水平为4.2±0.5mmol/L，出院后1月内共700例患者（4.8%）发生高钾血症。对训练集进行LASSO和随机森林算法筛选变量，最终挑选出临床易获取且解释性良好的5个变量：基线血清钾、估算肾小球滤过率、糖尿病、RASi使用、MRA使用，并构建高钾血症发生预测评分表。模型在训练集和验证集中ROC曲线下面积（Area Under the Curve, AUC）分别为0.746（95%CI: 0.722-0.766）和0.723（95%CI: 0.703-0.743）。训练集及验证集的临床实用性表现良好。

结论：基于基线血清钾、估算肾小球滤过率、糖尿病、RASi使用、MRA使用5个临床易获得的变量，本研究构建了中国心衰患者出院后1月内高钾血症发生预测模型，在中国心衰患者中具有良好的判别性能和临床应用潜力，未来有待进一步的外部验证以拓展其适用性。

第四部分 高钾血症、衰弱与老年心血管病患者预后关联

背景：衰弱为生理储备下降导致机体易损性增加、抗应激能力减退的非特异性状态，被证实可用于预测心血管病患者临床结局。本部分研究拟探讨高钾血症与衰弱在老年心血管病患者中的关联，并探索两者与老年心血管病患者5年全因死亡的预后关联。

方法：本研究纳入2018-2019年北京医院入组的524例住院老年心血管病患者。高钾血症定义为血清钾>5.0 mmol/L，衰弱由Fried衰弱表型评估。采用多因素逻辑回归评估高钾血症与衰弱的关联。采用Cox回归探索高钾血症、衰弱对住院老年心血管病患者5年内全因死亡的影响。采用C-指数评估衰弱及高钾血症对老年心血管病患者5年全因死亡的预测价值。

结果：纳入的524例心衰患者中，平均年龄为75.2±6.5岁，女性占比48.3%，33例（6.3%）患者住院期间合并高钾血症，136例（26.0%）患者住院期间合并衰弱。多因素Logistic回归结果提示，衰弱与老年心血管患者住院期间合并高钾血症独立相关(OR=22.13, 95%CI:5.38-91.09, P<0.001)。多因素Cox回归结果表明，住院期间合并高钾血症（HR=2.54, 95%CI:1.36-4.74, P=0.004）与衰弱（HR=1.89, 95%CI:1.05-3.39, P=0.033）均与住院老年心血管病患者5年全因死亡风险增加独立相关。衰弱预测老年心血管病患者5年全因死亡的C指数为0.69(95%CI:0.59-0.79)，衰弱联合高钾血症预测结局的C指数为0.72(95%CI:0.62-0.83)，C指数的变化值为0.03，p=0.003。

结论：在老年住院心血管病患者中，住院期间合并高钾血症与衰弱独立相关。住院期间合并高钾血症与衰弱均与住院老年心血管病患者5年全因死亡风险增加独立相关。衰弱联合高钾血症可更好用于预测老年心血管病患者预后。

Part Ⅰ: Association Between Hyperkalemia and Prognosis in Chinese Hospitalized Patients With Heart Failure

Background:

Hyperkalemia is a common complication in patients with heart failure (HF). Evidence from studies suggests that hyperkalemia is associated with adverse outcomes in this population. However, large-scale epidemiological data on hyperkalemia among Chinese HF patients and its prognostic implications remain limited. This study aims to investigate the association between hyperkalemia and 1-year all-cause and cardiovascular mortality in hospitalized HF patients in China, as well as to explore the optimal serum potassium range.

Methods:

A total of 158,349 hospitalized HF patients enrolled between January 2017 and December 2020 were included in this study. Baseline information, including demographics, comorbidities, laboratory results, and discharge medications, was collected. Patients were followed up for 1 year after discharge to obtain data on all-cause and cardiovascular mortality. Hyperkalemia was defined as a serum potassium level >5.0 mmol/L. Cox proportional hazards regression models were used to evaluate the association between hyperkalemia and mortality outcomes, with subgroup analyses performed. Additionally, restricted cubic spline models were applied to explore the nonlinear relationship between in-hospital serum potassium levels and the risk of all-cause mortality.

Results: 

Among the 158,349 patients, the mean age was 69.2±13.3 years, and 41.0% were female. Hyperkalemia occurred in 8,876 patients (5.6%) during hospitalization, with 4.4%, 0.9%, and 0.3% having mild, moderate, and severe hyperkalemia, respectively. Kaplan-Meier survival curves and Cox regression analyses revealed that hyperkalemia was an independent risk factor for all-cause mortality (HR=1.29, 95% CI: 1.23–1.35, P<0.001) and cardiovascular mortality (HR=1.28, 95%CI:1.21-1.36, P<0.001) within one year after discharge. Subgroup analyses demonstrated consistent associations across different patient group. A U-shaped relationship was observed between in-hospital potassium levels and all-cause mortality, with the lowest risk at 4.3 mmol/L.

Conclusion: 

This study suggests that hyperkalemia is an independent risk factor for 1-year all-cause and cardiovascular mortality among hospitalized HF patients in China. These findings underscore the importance of routine potassium monitoring and precise potassium management as part of the therapeutic strategy for HF.

Part Ⅱ: Association Between Hyperkalemia and the Use of RAASi in Chinese Hospitalized HFrEF Patients

Background:

Guideline-directed use of renin-angiotensin-aldosterone system inhibitors (RAASi) significantly improves outcomes in patients with heart failure with reduced ejection fraction (HFrEF). In Part I of this study, we observed that hospitalized heart failure patients with hyperkalemia were less likely to receive RAASi therapy at discharge. Prior research has suggested that hyperkalemia may be a barrier to RAASi use in heart failure patients. However, there is currently a lack of evidence in China regarding the impact of hyperkalemia on RAASi utilization and long-term adherence in HFrEF patients. This study aims to examine the influence of in-hospital hyperkalemia on RAASi use at discharge and during follow-up, as well as the rate of therapy discontinuation in HFrEF patients.

Methods:

A total of 124,257 hospitalized HFrEF patients enrolled between January 2017 and December 2022 were included. Hyperkalemia was defined as serum potassium >5.0 mmol/L. Patients were categorized into hyperkalemia and normal serum potassium groups based on pre-discharge potassium levels. Baseline characteristics and discharge RAASi prescription rates were compared between two groups, and multivariate logistic regression was used to identify associated factors. Among 44,139 patients with complete 1-year follow-up data, we further analyzed RAASi use and discontinuation rates at 1, 3 and 12 months. RAASi use was defined as prescription of an ACE inhibitor, ARB, or ARNI at discharge or during follow-up. Discontinuation was defined as RAASi use at discharge but absence of use at follow-up. MRA use and discontinuation were defined analogously. In addition to P value, standardized mean differences (SMD) were used to assess between-group differences, with SMD ≥10% considered clinically significant.

Results:

Among the 124,257 hospitalized patients with HFrEF included in the study, the mean age was 65.3 ± 14.0 years, and 30.7% were female. At discharge, patients in the hyperkalemia group had significantly lower rates of RASi (76.6% vs. 86.1%) and MRA (77.1% vs. 85.4%) use compared to those in the normokalemia group (both P < 0.001; SMD ≥10%). After adjusting for age, sex, and comorbidities using multivariable logistic regression, hyperkalemia remained an independent factor associated with lower RASi (OR = 0.79, 95% CI: 0.63–0.98, P = 0.030) and MRA (OR = 0.54, 95% CI: 0.43–0.68, P < 0.001) use at discharge. During follow-up, RASi, MRA, and combination therapy use at 1, 3, and 12 months remained consistently lower in the hyperkalemia group compared with the normokalemia group (all P < 0.001; SMD >10%). However, there was no significant difference in discontinuation rates between the two groups at any follow-up time point. Further stratification of the hyperkalemia group identified a subset of patients with moderate to severe hyperkalemia (baseline serum potassium ≥5.5 mmol/L). In this subgroup, 1-year discontinuation rates for both RASi (28.1% vs. 16.3%) and MRA (33.8% vs. 23.7%) were significantly higher than those observed in the normokalemia group (both P < 0.001; SMD ≥10%).

Conclusion:

In-hospital hyperkalemia is an independent factor associated with reduced RAASi prescription at discharge in HFrEF patients. Moreover, moderate to severe hyperkalemia (serum potassium ≥5.5 mmol/L) is strongly associated with higher 1-year RAASi and MRA discontinuation rates. Optimized potassium management may help improve medication adherence and long-term outcomes in HFrEF patients.

Part Ⅲ: Development of a Prediction Model for Hyperkalemia Within One Month in Patients With Heart Failure

Background:

Hyperkalemia is associated with poor prognosis in heart failure (HF) patients and is a major barrier to optimal use of renin-angiotensin-aldosterone system inhibitors (RAASi). Early identification of patients at high risk of hyperkalemia is essential for tailoring follow-up strategies and medication adjustments to improve clinical outcomes. However, there is currently no predictive model specifically designed for assessing hyperkalemia risk in Chinese HF patients. This study aimed to develop a model to predict the occurrence of hyperkalemia within one month in hospitalized HF patients, which may help guide the optimal frequency of serum potassium monitoring in clinical practice.

Methods:

A total of 14,727 hospitalized HF patients enrolled between 2017 and 2020 were included. Hyperkalemia was defined as serum potassium >5.0 mmol/L. Patients were randomly assigned into training (70%) and validation (30%) cohorts. Variable selection was performed using both random forest and least absolute shrinkage and selection operator (LASSO) regression. A multivariable logistic regression model was then constructed based on selected predictors, and a nomogram was developed to facilitate clinical application. Model performance was evaluated using receiver operating characteristic (ROC) curves and decision curve analysis (DCA) in both the training and validation cohorts.

Results:

Among the 14,727 hospitalized HF patients included, the mean age was 67.9 ± 13.9 years, and 39.8% were female. The mean baseline serum potassium level was 4.2 ± 0.5 mmol/L. Within one month after discharge, a total of 700 patients (4.8%) developed hyperkalemia. Variable selection using both LASSO regression and random forest methods in the training cohort identified five clinically accessible and interpretable predictors: baseline serum potassium, estimated glomerular filtration rate (eGFR), diabetes mellitus, RAASi use, and MRA use. These variables were incorporated into a predictive scoring model for post-discharge hyperkalemia. The model demonstrated good discrimination, with areas under the ROC curve (AUC) of 0.746 (95% CI: 0.722–0.766) in the training cohort and 0.723 (95% CI: 0.703–0.743) in the validation cohort. DCA indicated favorable clinical utility in both datasets.

Conclusion:

This study developed a clinically practical predictive model for hyperkalemia occurring within one month after discharge in Chinese HF patients, based on five easily obtainable variables: baseline serum potassium, eGFR, diabetes status, RAASi use, and MRA use. The model showed good discrimination and practical applicability and may assist in guiding clinical decision-making. External validation is warranted to expand its generalizability.

Part Ⅳ: The Association of Hyperkalemia, Frailty, and Prognosis in Older Patients with cardiovascular disease

Background:

Frailty is a nonspecific state characterized by reduced physiological reserves, increased vulnerability, and diminished stress resistance, which has been confirmed as a predictor of clinical outcomes in cardiovascular disease(CVD) patients. This part of the study aims to explore the relationship between hyperkalemia and frailty in older patients with CVD and to investigate the prognostic association of both with all-cause mortality within five years in this population.

Methods:

This study included 524 hospitalized older cardiovascular patients enrolled at Beijing Hospital between 2018 and 2019. Hyperkalemia was defined as a serum potassium level > 5.0 mmol/L, and frailty was assessed using the Fried frailty phenotype. Multivariate logistic regression was used to evaluate the association between hyperkalemia and frailty. Cox regression analysis was employed to explore the impact of hyperkalemia and frailty on all-cause mortality within five years in elderly cardiovascular patients. The C-index was used to assess the predictive value of frailty and hyperkalemia for five-year all-cause mortality in these patients.

Results:

Among the 524 CVD patients, the average age was 75.2 ± 6.5 years, with 48.3% being female. 33 patients (6.3%) had hyperkalemia during hospitalization, and 136 patients (26.0%) had frailty. Multivariate logistic regression showed that frailty was independently associated with hyperkalemia during hospitalization in older CVD patients (OR=22.13, 95% CI: 5.38-91.09, P<0.001). Multivariate Cox regression results indicated that both hyperkalemia during hospitalization (HR=2.54, 95% CI: 1.36-4.74, P=0.004) and frailty (HR=1.89, 95% CI: 1.05-3.39, P=0.033) were independently associated with an increased risk of all-cause mortality within five years in hospitalized older CVD patients. The C-index for frailty in predicting five-year all-cause mortality was 0.69 (95% CI: 0.59-0.79), and the C-index for the combination of frailty and hyperkalemia in predicting the outcome was 0.72 (95% CI: 0.62-0.83), with a change of 0.03 (P=0.003).

Conclusion:

Among older hospitalized CVD patients, hyperkalemia during hospitalization is independently associated with frailty. Both hyperkalemia and frailty are independently associated with an increased risk of all-cause mortality within five years in older CVD patients. The combination of frailty and hyperkalemia can better predict the prognosis of older patients with CVD.