# Systematic Review and Meta-analysis of the Predictive Performance of Stroke and Bleeding Prediction Models in Atrial Fibrillation Patients With Kidney Disease

## TL;DR

This study evaluates how well existing prediction models work for stroke and bleeding risks in patients with atrial fibrillation and kidney disease, finding they perform similarly to patients without kidney issues despite some limitations.

## Contribution

The study provides the first comprehensive meta-analysis comparing prediction models for stroke and bleeding in atrial fibrillation patients with chronic kidney disease or undergoing dialysis.

## Key findings

- The CHADS2 model showed slightly better discrimination for stroke prediction in atrial fibrillation patients with chronic kidney disease compared to CHA2DS2-VASc.
- In dialysis patients, CHA2DS2-VASc and CHADS2 models performed similarly for stroke prediction, while HAS-BLED and HEMORR2HAGES were comparable for bleeding prediction.
- Calibration was good in high-risk groups but poor in lower-risk groups, and all studies had high risk of bias.

## Abstract

Patients with atrial fibrillation (AF) and chronic kidney disease (CKD) are at high risk for ischemic stroke (IS) and bleeding. The applicability of prediction models in this population remains debated. This study aimed to (1) identify external validations of CHA2DS2-VASc, CHADS2, HAS-BLED, and HEMORR2HAGES model scores in patients with AF undergoing dialysis or with CKD, (2) provide pooled estimates, and (3) assess their risk of bias (ROB).

Systematic review and meta-analysis.

We searched Web of Science, PubMed, MEDLINE, Embase, Emcare, PMC, Cochrane Library, and Academic Search Premier for studies externally validating IS and bleeding prediction models in patients with AF undergoing dialysis or with CKD.

AF and CKD or dialysis.

IS and bleeding.

Eligible studies were reviewed, discrimination was pooled using random-effects meta-analysis, calibration was calculated and plotted, and the ROB score was assessed using the prediction model ROB assessment tool.

The CHA2DS2-VASc score was validated in 35 studies, the CHADS2 and HAS-BLED scores in 19 each, and the HEMORR2HAGES score in 1. Among 627,199 patients, 28,493 (4.5%) experienced IS and 25,695 (4.1%) bleeding. Only 12 studies presented c-statistic scores. In patients with AF and CKD, the CHADS2 model score showed nominally better discrimination predicting IS (pooled c-statistic score of 0.70) than the CHA2DS2-VASc model score (0.64). In patients with AF undergoing dialysis, the CHA2DS2-VASc and CHADS2 model scores showed similar discrimination predicting IS (both 0.70), and the HAS-BLED and HEMORR2HAGES model scores showed similar c-statistic scores predicting bleeding (0.55 and 0.56, respectively). Calibration was good in the most relevant high-risk group.

All studies were at high ROB scores, contained within- and between-study heterogeneity, and often merged scoring categories or populations, limiting comparability.

Although modest, the discrimination of prediction models in patients with AF undergoing dialysis or with CKD is similar to patients with AF without CKD. Despite the described limitations, these models can be used in clinical practice for patients with CKD and patients undergoing dialysis.

Patients with both atrial fibrillation (AF) and chronic kidney disease (CKD) or patients undergoing dialysis face a higher risk of stroke and therapy-related bleeding. To estimate these risks, prediction models can be used, but their predictive value in this population is unclear. This study reviewed and analyzed existing studies on 4 commonly used stroke and bleeding prediction models in patients with AF undergoing dialysis or with CKD. We found that, although these models showed modest discrimination, their performance was similar to that in patients with AF without CKD or undergoing dialysis. Therefore, despite the weaknesses of the included studies, we believe these tools can be used for patients with CKD and patients undergoing dialysis to help guide treatment decisions.

Key findings:•This study identified 35 studies validating the CHA2DS2-VASc score, 21 the CHADS2 score, 22 the HAS-BLED score, and 1 the HEMORR2HAGES score, reflecting guideline endorsement.•In atrial fibrillation (AF) patients with chronic kidney disease (CKD), the CHADS2 score shows nominally better discriminative abilities than the CHA2DS2-VASc score.•In patients undergoing dialysis, the CHA2DS2-VASc, CHADS2, HAS-BLED, and HEMORR2HAGES scores perform similarly.•The CHA2DS2-VASc and HAS-BLED scores showed good calibration in the most relevant high-risk group.•High risk of bias (ROB) was found across studies, especially in the Outcome and Analysis domains.

This study identified 35 studies validating the CHA2DS2-VASc score, 21 the CHADS2 score, 22 the HAS-BLED score, and 1 the HEMORR2HAGES score, reflecting guideline endorsement.

In atrial fibrillation (AF) patients with chronic kidney disease (CKD), the CHADS2 score shows nominally better discriminative abilities than the CHA2DS2-VASc score.

In patients undergoing dialysis, the CHA2DS2-VASc, CHADS2, HAS-BLED, and HEMORR2HAGES scores perform similarly.

The CHA2DS2-VASc and HAS-BLED scores showed good calibration in the most relevant high-risk group.

High risk of bias (ROB) was found across studies, especially in the Outcome and Analysis domains.

What this adds to what was known:•Though modest, the discriminative abilities of these models in patients with AF undergoing dialysis or with CKD are comparable with those in patients with AF with normal kidney function.•Calibration was poor in lower-risk groups but showed better agreement in the more relevant higher-risk groups.•The ROB in these prediction model studies is high.

Though modest, the discriminative abilities of these models in patients with AF undergoing dialysis or with CKD are comparable with those in patients with AF with normal kidney function.

Calibration was poor in lower-risk groups but showed better agreement in the more relevant higher-risk groups.

The ROB in these prediction model studies is high.

What is the implication, and what should change now?•Aligning with broader AF guidelines, conventional stroke and bleeding prediction models can also be applied to patients with CKD and patients undergoing dialysis.•The high ROB scores of the included studies and heterogeneity of the included populations highlight the need for more rigorous validations of prediction models in this population.

Aligning with broader AF guidelines, conventional stroke and bleeding prediction models can also be applied to patients with CKD and patients undergoing dialysis.

The high ROB scores of the included studies and heterogeneity of the included populations highlight the need for more rigorous validations of prediction models in this population.