# Risk prediction models for chemotherapy-induced nausea and vomiting: a systematic review and meta-analysis

## TL;DR

This study reviews and evaluates existing models for predicting chemotherapy-induced nausea and vomiting, finding them to have moderate accuracy but significant methodological flaws.

## Contribution

The study systematically reviews and meta-analyzes CINV risk prediction models, highlighting their limitations and suggesting directions for improvement.

## Key findings

- 15 studies with 16 CINV models were reviewed, with AUC values ranging from 0.629 to 0.850.
- Common predictors include age, gender, chemotherapy regimen, and history of nausea.
- Models showed a pooled AUC of 0.74, indicating moderate discrimination but high risk of bias.

## Abstract

To systematically review and critically appraise currently available risk prediction models for chemotherapy-induced nausea and vomiting (CINV).

We searched nine electronic databases from inception to April 2025. Data extraction followed the CHARMS checklist. Risk of bias and applicability were assessed using the PROBAST tool, and reporting transparency was evaluated against the TRIPOD statement.

15 studies describing 16 distinct CINV risk prediction models were included. Reported area under the curve (AUC) values ranged from 0.629 to 0.850. Frequently incorporated predictors included age, gender, history of anticipatory nausea and vomiting, chemotherapy regimen, and number of chemotherapy cycles. All studies demonstrated a high risk of bias, primarily attributable to suboptimal data sources and inadequate reporting in the analytical domain. Meta-analysis of AUC values from eight development models yielded a pooled estimate of 0.74 (95% CI: 0.68-0.81), indicating moderate discrimination.

Existing CINV risk prediction models exhibit significant methodological limitations and remain largely in the developmental phase. While common predictors emerge, controversies persist. Future research should prioritize developing novel models with larger sample sizes, rigorous methodology, multicenter external validation, enhanced clinical utility, and improved reporting transparency.

https://www.crd.york.ac.uk/prospero/, identifier CRD42023395416.