Total cholesterol and mortality in peritoneal dialysis: a retrospective cohort study

Study design and patients

We conducted a retrospective real-world cohort study that included 3565 incident Chinese CAPD patients from five PD centers between January 1, 2005, and May 31, 2020. To maximumly represent the real-world setting of the CAPD population, no patient was excluded from this study. The requirement for informed consent was waived by the Ethics Committee of (The First Affiliated Hospital of Nanchang University, Nanchang, China. The First Affiliated Hospital of Zhengzhou University, Zhengzhou, China. Jiujiang No. 1 People’s Hospital, Jiujiang, China. Zhujiang Hospital of Southern Medical University, Guangzhou, China. The Second Affiliated Hospital of Guangzhou Medical University, Guangzhou, China) because of the retrospective nature of the study. The study protocol complied with the Declaration of Helsinki and had full approval from each Clinical Research Ethics Committee.

Follow up and data collection

We respectively collected demographic data, comorbidities, medication use, and laboratory variables one week (5.3 ± 1.2 days) before the start of PD, including age at study entry, sex, body mass index, current smoker, current alcohol use, systolic blood pressure, diabetes mellitus, prior cardiovascular disease, hypertension, beta-blockers, angiotensin-converting enzyme inhibitors/angiotensin II receptor blockers (ACEI/ ARB), diuretics, statins, albumin, estimated glomerular filtration rate (eGFR), total cholesterol, high-density lipoprotein cholesterol (HDL-C), and LDL-C.

Our primary and secondary endpoints were all-cause and cardiovascular mortality, respectively. Details for the CAPD follow-up were previously described elsewhere [11]. The follow-up period was from the start of PD to the date of death, transfer to hemodialysis, receiving renal transplantation, transfer to other dialysis centers, loss of follow-up, or May 31, 2020. Patients lost to follow-up were censored at the date of the last examination.

Statistical analysis

Differences in the baseline characteristics stratified by total cholesterol were compared using the chi-square test for categorical variables and analysis of variance for continuous variables. Restricted-cubic-spline plots were used to explore the shape of the association between total cholesterol and mortality, fitting a restricted-cubic-spline function with four knots (at the 25th, 50th, 75th, and 95th percentiles) [12].

Based on our restricted-cubic-spline plots for the primary endpoint, we selected a level of 4.10 to 4.50 mmol/L as the reference category for total cholesterol. To explore the association of total cholesterol with mortality, we primarily used cause-specific hazard models. We then constructed sub-distribution hazard models to confirm the association observed in the primary analysis. All factors were included in the multivariate analysis based on their clinical significance. Transfer to hemodialysis, receiving renal transplantation, transfer to other centers or loss of follow-up was considered a competing risk. The main difference between the two hazard models is that subjects experiencing a competing risk event remain in the risk set in the sub-distribution hazard model. In contrast, they are removed in the cause-specific hazard model [13, 14]. We tested for interactions of age, sex, diabetes mellitus, prior cardiovascular disease, hypertension, and malnutrition (albumin < 36.0 g/L).

Sensitivity analysis

To minimize the potential for reverse causation, we conducted analyses that excluded patients with prior cardiovascular disease or those deaths in the first 2 years of follow-up. As for those patients with a short-term follow-up period, the interesting endpoints may not be observed, under-estimating the incidence of mortality. We further analyzed the association in patients with at least 24 months of follow-up for fully observing endpoints. In addition, to minimize the effect of beta-blockers, diuretics, or statins on total cholesterol, we also analyzed the association in patients without beta-blockers, diuretics, or statin use. Missing data for total cholesterol (n = 35) or any other explanatory variables (n = 121) at the start of PD were replaced by the most recent available values by checking patients’ medical records at the first PD procedure. All analyses were conducted with the use of Stata 15.1. statistical software (StataCorp, College Station, TX).

留言 (0)

沒有登入
gif