Risk Prediction Models as an Emerging Trend for Managing Cancer-Related Fatigue: A Systematic Review

To systematically identify, describe and evaluate the existing risk prediction models for cancer-related fatigue.

Systematic review.

Seven databases (EMBASE, Cochrane Database, MEDLINE, CINAHL, CNKI, SinoMed and Wanfang) were conducted from inception to August 14, 2023 and updated in September 15, 2024.

A systematic search was conducted to identify studies that reported the development of risk prediction models for cancer-related fatigue. Two researchers independently performed a comprehensive assessment of the included studies. The Prediction Model Risk of Bias Assessment Tool was used to assess the risk of bias and applicability.

Eighteen studies were included in this review. These models predicted cancer-related fatigue in various cancers, including breast cancer, prostate cancer, gynaecological tumours and lung cancer. The most commonly included predictors were anxiety and depression, age, chemotherapy status, sleep quality and pain. Thirteen studies assessed the model performance by using the receiver operating characteristic curve. Although most models exhibited good predictive performance, a higher risk of bias was observed because of inappropriate handling of missing data methods and an imbalance in events per variable.

Prediction models show promise for cancer-related fatigue management and precision care, but few are ready for clinical application due to methodological limitations.

Future research should focus on improving the clinical utility of cancer-related fatigue models while balancing predictive accuracy with cost-effectiveness to promote equitable care.

This study critically systematically evaluated the prediction models of cancer-related fatigue. The existing prediction models have a weak methodological foundation, with only a few having the potential to be implemented in clinical practice.

The review is reported using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines and the Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis in Systematic Reviews and Meta-Analyses checklist (TRIPOD-SRMA).

No patient or public contribution.