Background Quality Improvement (QI) is used globally to improve care. Equity is one of the six domains of health care quality, as defined by the Institute of Medicine. If this domain is ignored, QI projects have the potential to maintain or even worsen inequalities. We currently do not know the impact of QI on inequalities. We undertook a realist review that aimed to understand why, how, for whom, and in which contexts quality improvement approaches might increase, decrease, or not change health inequalities in healthcare organisations.

Methods We first developed an initial programme theory informed by background literature and a workshop with our expert panel.We searched MEDLINE, Embase, CINAHL, PsychINFO, Web of Science, and SCOPUS to identify quality improvement studies using any study design in any clinical setting that reported health or care inequalities published from January 2000 to January 2023. We included studies that either explored the impact of the QI initiative on closing health gaps, or focusing on specific disadvantaged groups. We did not include studies that did not look at inequalities, studies undertaken in low and middle-income countries, or studies not published in English. Included studies were assessed for rigour and relevance and uploaded to NVivo v.12 for both deductive and inductive coding. Through this analysis we generated Context-Mechanism-Outcome Configurations (CMOCs) and developed an overall programme theory.

Results We screened 6259 records. 36 records met our inclusion criteria, the majority of which were from the USA. We developed 20 CMOCs covering four clusters: values and understanding; resources; data; and design. Five of these CMOCs described circumstances in which QI may increase inequalities, and fifteen described circumstances where it may reduce inequalities. We found that QI projects that are values-led and incorporate diverse, patient-led data into design may help to address health inequalities. However, when practical or technological barriers mean that staff and patients cannot engage fully with equity-focused projects, QI projects may worsen inequalities.

Conclusion The potential for QI projects to positively impact inequalities depends on embedding equity-focused values across organisations, ensuring sufficient and appropriate resources are provided to staff delivering QI, and using diverse disaggregated data alongside considered user involvement to inform and assess the success of QI projects. Policymakers and practitioners should ensure that QI projects are used to address inequalities, rather than to worsen them.