The need for more real-world data-driven reviews and meta-analyses in haematology: an essential call

Randomised controlled trials remain essential for establishing treatment efficacy in haematology, but they cannot fully represent the diversity of patients, care pathways, and resource constraints encountered in routine practice, particularly in low- and middle-income countries (LMICs). This commentary argues for an integrated evidence system that combines locally generated real-world data with fit-for-purpose evidence-synthesis methods to advance precision haematology. Routinely collected clinical, laboratory, treatment, toxicity, and outcome data can clarify real-world effectiveness, long-term safety, uncommon adverse events, and variations in access to care. However, credible real-world evidence requires standardised datasets, clearly defined causal questions, appropriate handling of confounding and missing data, transparent reporting, and robust governance. Different synthesis methods should be matched to specific questions: individual participant data meta-analysis can examine treatment-effect heterogeneity; network meta-analysis can compare multiple interventions; living systematic reviews can maintain evidence currency; and realist and meta-narrative reviews can explore context and implementation. Machine-learning and federated approaches may further strengthen evidence generation, but only after data quality, interoperability, validation, and clinical relevance have been established. Progress in LMICs will require sustained investment in registries, data managers, biostatisticians, secure infrastructure, and equitable international collaboration. Connecting trial evidence, real-world evidence, and appropriate evidence synthesis rather than treating them as competing approaches could make precision haematology more context-responsive, implementable, and equitable.