Scholarlink

Background <p>Peritoneal dialysis-associated peritonitis (PDAP) remains a major complication in patients undergoing peritoneal dialysis (PD), often leading to technique failure, hospitalization, and increased mortality. This study aims to systematically evaluate and compare existing risk prediction models for PDAP and quantify their overall performance and key predictors.</p> Methods <p>Following PRISMA guidelines, eight databases were systematically searched including PubMed, Web of Science, Cochrane Library, Embase, China National Knowledge Infrastructure (CNKI), Chinese Science and Technology Journal Database (VIP), Wanfang, and SinoMed, covering literature from their inception to March 19, 2025. Data extraction focused on study design, participant characteristics, sample size, outcome definitions, predictors, and model performance. The bias risk assessment tool for prediction model studies (PROBAST) was employed to assess bias risk and applicability, while Stata 17.0 facilitated meta-analysis.</p> Results <p>Eleven studies involving 11 logistic regression models were included. The reported area under the curve (AUC) values ranged from 0.659 to 0.997, with a combined AUC of 0.88 (95% CI 0.83–0.93), indicating robust predictive performance.&#xa0;Four predictors associated with peritonitis were albumin (Alb) (OR = 0.672; 95% CI 0.475–0.868; <i>P</i> &lt; 0.001), C-reactive protein (CRP) (OR = 2.568; 95% CI 1.081–4.055; <i>P</i> &lt; 0.001), neutrophil-to-lymphocyte ratio (NLR) (OR = 1.377; 95% CI 1.065–1.689; <i>P</i> &lt; 0.001), and diabetes mellitus (DM) (OR = 3.549; 95% CI 1.272–5.825; <i>P</i> &lt; 0.002).</p> Conclusions <p>Despite demonstrating strong predictive capabilities, the models exhibited high bias risks, primarily due to data sources and statistical methods. Future research should prioritize large-scale, multicenter studies with rigorous designs and external validation to enhance reliability and clinical applicability.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Risk prediction models for peritoneal dialysis-associated peritonitis: a systematic review and meta-analysis

  • 期刊论文

  • 发表日期：

    2025年9月21日

查看全文
  • Zhongyi Zhao,
  • Yang Li,
  • Qin Quan,
  • Hongyan Wang,
  • Wen Zhang,
  • Xiangeng Zhang

摘要

Background

Peritoneal dialysis-associated peritonitis (PDAP) remains a major complication in patients undergoing peritoneal dialysis (PD), often leading to technique failure, hospitalization, and increased mortality. This study aims to systematically evaluate and compare existing risk prediction models for PDAP and quantify their overall performance and key predictors.

Methods

Following PRISMA guidelines, eight databases were systematically searched including PubMed, Web of Science, Cochrane Library, Embase, China National Knowledge Infrastructure (CNKI), Chinese Science and Technology Journal Database (VIP), Wanfang, and SinoMed, covering literature from their inception to March 19, 2025. Data extraction focused on study design, participant characteristics, sample size, outcome definitions, predictors, and model performance. The bias risk assessment tool for prediction model studies (PROBAST) was employed to assess bias risk and applicability, while Stata 17.0 facilitated meta-analysis.

Results

Eleven studies involving 11 logistic regression models were included. The reported area under the curve (AUC) values ranged from 0.659 to 0.997, with a combined AUC of 0.88 (95% CI 0.83–0.93), indicating robust predictive performance. Four predictors associated with peritonitis were albumin (Alb) (OR = 0.672; 95% CI 0.475–0.868; P < 0.001), C-reactive protein (CRP) (OR = 2.568; 95% CI 1.081–4.055; P < 0.001), neutrophil-to-lymphocyte ratio (NLR) (OR = 1.377; 95% CI 1.065–1.689; P < 0.001), and diabetes mellitus (DM) (OR = 3.549; 95% CI 1.272–5.825; P < 0.002).

Conclusions

Despite demonstrating strong predictive capabilities, the models exhibited high bias risks, primarily due to data sources and statistical methods. Future research should prioritize large-scale, multicenter studies with rigorous designs and external validation to enhance reliability and clinical applicability.