This retrospective cohort study was conducted at a single medical center. Overall, 182 participants were collected among individuals who were undergoing MHD at the Hemodialysis Center of Guangzhou Red Cross Hospital in March 2021. Inclusion criteria encompassed an age ≥ 18 years, undergoing hemodialysis thrice weekly for at least 4 h per session, and maintaining regular hemodialysis for over 3 months (n = 174). Exclusion criteria included the absence of PS and HGS data, missing follow-up data, patients with pacemakers, concurrent malignancies, recent heart failure, acute myocardial infarction, cerebrovascular accidents, severe infections within the past 3 months, physical disabilities, or inability to cooperate with the study procedures. Ultimately, a total of 140 patients were included in the primary outcome study. Additionally, when studying secondary outcomes (CVE), we excluded patients who already had CVD at baseline, and ultimately 90 patients were included in the study. After 24 months of follow-up, 36 patients developed CVE.

Demographic data encompassed age, sex, body mass index (BMI), and dialysis vintage. Comorbidity and lifestyle factors considered were hypertension, diabetes, CVD, smoking status, alcohol consumption status, and physical activity level. Before hemodialysis, venous blood samples were collected and promptly delivered to our clinical laboratory within a 2-h timeframe. The laboratory parameters evaluated included hemoglobin, serum calcium, serum phosphorus, serum parathyroid hormone, serum albumin, triglyceride, total cholesterol, low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, interleukin-6, high-sensitivity C-reactive protein, serum creatinine, blood urea nitrogen levels, estimated glomerular filtration rate (eGFR), and urea clearance index (Kt/V). Nutritional risk assessment was conducted using the Nutritional Risk Screening 2002 (NRS-2002) tool [14, 15].

BMI was calculated as weight (kg) divided by the square of height (m2). CVE encompassed heart failure, coronary heart disease, unstable angina, myocardial infarction, malignant arrhythmias, and stroke. Activity levels were categorized into low (average non-dialysis daily step count of < 4,000 steps) and moderate-vigorous activities [16]. The eGFR was determined using the isotope dilution mass spectrometry four-variable modification of diet in renal disease study equation [17]: GFR = 175 × standardized Scr ^-1.154 × age^-0.203 × 1.212 (if black) × 0.742 (if female). The urea clearance index of the single-chamber model (spKt/V) was computed using the following formula: spKt/V = -Ln(R-0.008t) + (4-3.5R) × UF/W, in which R represents the post-permeation BUN/pre-permeation BUN ratio, t stands for dialysis time, UF signifies ultrafiltration volume, and W denotes post-permeation weight. An NRS-2002 score ≥ 3 indicated individuals at nutritional risk.

Hemodialysis treatment for the patients was conducted using a Braun Dialog + device (B. Braun Co., Ltd., Melsungen, Germany) in conjunction with a REXEED-15 L high-throughput polysulfone membrane dialyzer (Asahi Kasei Corp., Tokyo, Japan). The dialyzer utilized had a membrane area of 1.5 m2, dialysis blood flow rate ranging from 200 to 300 mL/min, dialysis fluid flow rate of 500 mL/min, and dialysis duration of 4 h.

Baseline digital pinch and grip force meters (Model 12–0091, Fabrication Enterprises Inc., USA) were used to quantify the patient’s pinch and grip forces. The HGS measurement was conducted in adherence to the recommended standards for seated measurements established by the American Society of Hand Therapists. Each participant’s PS and HGS were gauged thrice on the non-arteriovenous fistula arm, and the resultant average was recorded [18]. During the HGS measurement, the grip distance was appropriately adjusted based on the patient’s hand size. The patient was positioned in a seated stance, with the elbow flexed at 90°, wrist extended within a 0–30° range, and ruler deviation maintained between 0° and 15°. The pinch force meter was positioned between the thumb pad and the radial side of the middle phalanx of the index finger. As this only requires the wrist to be in a neutral position, the measurements can be performed with or without dialysis.

The study encompassed an observation period extending from March 2021 to March 2023. Throughout this period, occurrences, such as death, CVE, regression events, and transfer out of our hemodialysis center, were recorded. Additionally, the duration until regression events (in months) was documented. The primary outcome was all-cause mortality, whereas the secondary outcome included recorded CVE during the follow-up period. Patients who transferred out of our hemodialysis center and underwent telephone follow-up and who remained alive upon the conclusion of the observation period marked the endpoints; the total observation duration was 24 months. Data of patients who were lost to follow-up were excluded from the analysis.

Statistical analysis was conducted using SPSS 26.0, R (http://www.R-project.org, Version 4.3.1) with packages “survival,” “survminer,” and “RMS,” as well as Empower Stats software (http://www.empowerstats.com). Continuous variables were presented as mean ± standard deviation or median (P25, P75), whereas categorical variables were expressed as frequency (percentage). Differences in PS and HGS groups were evaluated using χ² tests for categorical variables, analysis of variance tests for normally distributed data, and Kruskal–Wallis H tests for skewed distributions. Kaplan–Meier curves were employed to illustrate mortality and CVE trends across different levels of PS and HGS.Univariate Cox regression and multivariate proportional hazards regression analyses were carried out to identify independent prognostic factors. In the multivariate analysis, we included the variables that were found to be significantly associated with all-cause mortality and CVE in the univariate analysis (Supplementary Table S1). However, we found that there were no confounding variables that met the above criteria, possibly due to the small sample size in this study. Therefore, we explored potential confounding factors from clinical applications and related studies. Subsequently, multivariate Cox proportional hazard regression models were employed to investigate the independent association of PS and HGS with all-cause mortality and CVE. Four models were used: Model 1 was unadjusted; Model 2 was adjusted for age, sex, dialysis age, BMI, hypertension, diabetes, and CVD; Model 3 was adjusted for age, sex, dialysis age, BMI, serum creatinine, and physical activity; and Model 4 was adjusted for age, sex, dialysis age, BMI, low-density lipoprotein cholesterol, serum albumin, and high-sensitivity C-reactive protein. A Cox model with restricted cubic spline functions was utilized to analyze dose-response relationships between PS, HGS, and mortality to enhance robustness and test for trends based on the variable containing a median value for each quartile to validate continuous variable results. A p-value of < 0.05 was considered statistically significant.