Ethical use of real-world data in health research

Health research increasingly draws on information generated during ordinary care: electronic medical records, prescription data, registries, hospital admissions, wearable devices, surveys, and linked government datasets. These sources can reveal how treatments perform outside controlled trials and show which services reach patients in practice.

The value is substantial. Real-world evidence can identify gaps in cancer care, track chronic disease outcomes, support safer trauma pathways, and improve services for mothers, children, and people living with mental illness. Yet the routine nature of these records can make ethical questions less visible. A patient may agree to treatment without expecting their information to support a future research project.

Responsible use therefore requires more than removing names from a dataset. Researchers must consider consent, privacy, data security, fairness, community expectations, cultural authority, scientific validity, and accountability throughout the data lifecycle. These responsibilities are shared by researchers, health services, universities, data custodians, ethics committees, and the communities represented in the evidence.

Why real-world evidence matters

Clinical trials remain essential for testing safety and efficacy under defined conditions, but they cannot answer every practical question. A trial may exclude older people, patients with several conditions, rural populations, or those who face barriers to care. Routine health information can show whether an intervention works across these groups and whether benefits reach patients equitably.

Data from everyday practice can also support faster responses to emerging health needs. Researchers may identify changing patterns in hospital presentations, vaccination uptake, medicine use, or mental health service demand. When connected responsibly with clinical expertise, these findings can guide better policy and care rather than remaining in academic publications.

The Brisbane Diamantina Health Partners network illustrates the importance of collaboration between research institutes, universities, and health services. Such partnerships can help ensure that data analysis is connected to clinical priorities and produces outcomes that matter to patients, families, carers, and communities.

Consent and reasonable expectations

Consent is one of the clearest ways to respect autonomy, but obtaining meaningful consent for every future use is not always practical. Health records may be collected years before a research question is developed, and repeated requests could exclude valuable populations or place an unnecessary burden on patients. In some circumstances, an ethics committee may approve a waiver or alteration of consent when the research presents low risk and strong public value.

A waiver does not remove ethical duties. Researchers should explain the intended use wherever feasible, limit the information collected, avoid unnecessary contact, and provide accessible avenues for questions or complaints. Public notices, patient advisory groups, and clear institutional policies can help people understand how secondary use of health information works.

Consent should also reflect the sensitivity of the subject. Information about reproductive health, genetic risk, mental health, disability, substance use, or childhood experiences can create harm if misused or disclosed. Ethical review should consider whether individuals would reasonably expect a particular use, not simply whether a legal permission exists.

Privacy, security, and data governance

De-identification reduces risk but does not guarantee anonymity. A person may be re-identified by combining supposedly anonymous records with location, dates, rare diagnoses, public posts, or other datasets. The possibility is especially important in small communities and research involving uncommon conditions.

Strong governance uses proportional controls. These may include data minimisation, secure research environments, role-based access, encryption, audit logs, disclosure checks, retention limits, and plans for destroying or returning data. Researchers should document who can access the information, what analyses are permitted, whether data can leave the jurisdiction, and what happens if a breach occurs.

Good governance also assigns clear responsibility. Data custodians and project leaders should monitor compliance throughout the project rather than treating approval as a one-time administrative step. Agreements between organisations need to cover publication, commercial involvement, intellectual property, incident reporting, and the handling of linked datasets.

Fairness, representation, and community authority

A dataset can appear large while still producing an incomplete picture of health. People who have limited access to services, unstable housing, low digital connectivity, or distrust of institutions may be missing from routine records. If researchers treat available data as representative, their findings may reinforce existing inequalities.

Algorithms and predictive models can reproduce these problems. A model trained on historical service use may interpret under-treatment as lower need. A risk score may perform differently across cultural, language, age, gender, or socioeconomic groups. Fair analysis requires examining missingness, measuring performance across populations, and involving affected communities in interpreting results.

For Aboriginal and Torres Strait Islander peoples, ethical practice must include cultural governance and respect for Indigenous data sovereignty. Community-controlled organisations and knowledge holders should have meaningful influence over research questions, interpretation, access, and dissemination. This is a matter of authority and self-determination, not simply better consultation.

Balancing benefits, risks, and evidence quality

Ethical approval should address both the social value of a project and the reliability of its methods. Poorly designed research can expose people to privacy risks without producing useful knowledge. Researchers should define the purpose narrowly, justify each data field, pre-specify key analyses where possible, and explain how limitations will affect conclusions.

Transparent reporting is part of the ethical bargain. Publications should describe the data source, linkage process, exclusions, uncertainty, and conflicts of interest without exposing individuals. Where appropriate, researchers can share code, protocols, aggregate findings, and plain-language summaries so that others can assess the work.

Research translation should also be evaluated after publication. If a model influences triage, resource allocation, or clinical decisions, its real-world effects need monitoring. An intervention that improves average outcomes but widens gaps between groups requires review and possible redesign.

Turning safeguards into practice

Ethical data use is strongest when patients and communities help shape the project from the beginning. Advisory groups can identify harms that technical teams may overlook, improve consent materials, and clarify which outcomes matter. Patient partners should be supported with accessible information, fair payment, and enough time to contribute meaningfully.

Research teams should also connect governance with clinical implementation. For example, evidence about maternal vaccination can be translated into service changes only when researchers understand patient concerns, workforce pressures, communication needs, and local access barriers. The guidance on maternal vaccination strategies demonstrates how research-informed approaches can be considered alongside practical delivery.

Practical safeguards for a responsible project include:

• Define a specific public-interest purpose before requesting data.
• Involve patients, carers, clinicians, and relevant communities in design and oversight.
• Use the minimum necessary information and review re-identification risks throughout the study.
• Test findings for bias, missing populations, and unequal effects across groups.
• Publish limitations, incidents, funding sources, and pathways for accountability.

Real-world information can improve health services when it is handled as a trust relationship rather than an unlimited resource. Researchers and partner organisations should build projects that are lawful, scientifically sound, culturally safe, and understandable to the people whose experiences make the evidence possible. Explore Brisbane Diamantina Health Partners’ research, governance, and translation resources to support health research that earns confidence and delivers meaningful benefit.