Study Notes on Expanding the Role of Justice in Secondary Research Using Digital Psychological Data

Jonathan Herington, Department of Health Humanities and Bioethics, University of Rochester
Kevin Li, Department of Family Medicine, University of Rochester
Anthony R. Pisani, Department of Psychiatry, Center for the Study and Prevention of Suicide, University of Rochester; Department of Pediatrics, University of Rochester

Definition: Secondary analysis of Digital Psychological Data (DPD) is a research methodology gaining traction in behavioral health research.
Human Subjects Review: Current guidelines allow secondary research on de-identified DPD without requiring human subjects research review.

Primary Critique: The existing ethical framework focusing on de-identification fails to address various ethical issues inherent in secondary research using DPD.
Emphasis on Justice: More importance should be given to the ethical principle of justice when conducting secondary research.

Social Risks: These arise when individuals collaborate to generate public goods, such as research knowledge, making it impractical to exempt individuals from participation.
Just Allocation: Emphasizes the need to balance benefits and burdens amidst collective social cooperation.

Create Socially Valuable Knowledge: Research conducted should aim for broader societal benefits.
Fair Distribution of Benefits and Burdens: It is paramount that the outcomes and responsibilities of research are shared equitably.
Transparency About Data Use: Researchers should clarify how data are utilized without compromising participant anonymity.
Create Mechanisms for Data Withdrawal: Allow participants to withdraw their data if desired, without extensive complications.
Stakeholder Engagement: Integrate input from relevant stakeholders in research plans to ensure diverse perspectives.
Responsible Reporting of Results: Findings should be reported with particular care to their implications and potential misuses.

The authors advocate modifying guidelines related to ethical practices in secondary research with de-identified digital data to ensure equitable fair allocation of research benefits and burdens.

Historical Burden: Mental health conditions previously accounted for approximately 7% of the global disease burden (Rehm & Shield, 2019).
Impact of COVID-19: The pandemic has exacerbated mental health conditions like anxiety and depression, especially within disadvantaged communities, complicating access to care (Safran et al., 2009).

The integration of big data and machine learning in psychological clinical practice can enhance management of behavior health disorders.
Ethical concerns surrounding secondary analysis need continual consideration.

Secondary Research Definition: Involves repurposing data collected for other primary activities for research purposes (Office for Human Research Protections, 2018).
Framework Reference: Uses data rather than biospecimens, though similar ethical principles apply.
Focus on Information: The primary emphasis in this paper revolves around psychological health evaluated through individual-level data (e.g., electronic health records, social media posts).

Meaning of De-identified Data: Data where subjects' identities are not discernible through identifiers (HHS Common Rule, 2017).
Levels of De-Identification:
- Coded Data: Direct identifiers replaced by codes.
- Brokered Data: Codes held by an external broker allowing data retrieval.
- Anonymized Data: Irreversible extraction of identifiable information.

Common Rule and Principles: Govern secondary research ethics, focusing on respect for persons (autonomy), beneficence (harm minimization), and justice (equitable treatment).
High Risk vs. Low Risk: Secondary analyses, deemed low risk under the Common Rule, are exempt from many regulations (including direct participant consent).

De-identification does not completely guard against ethical risks and introduces new concerns such as consent issues and re-identification risks.
Consent Issues: Secondary research often involves minimal participant consent, which is crucial for transparency and ethical responsibility.

De-identification Spectrum: Not binary, as broad identification measures may fail in establishing privacy safeguards in complicated datasets.
Reidentification Risks: Data merging across datasets enhances reidentification potential, complicating privacy efforts.
Long-term Data Risks: Prolonged storage of identifiable data can lead to privacy breaches.

Data Quality Concerns: Secondary research leveraging existing data can result in misleading findings due to potential data pollution (De Nadai et al., 2022).
Bias in Data: Historical biases present in collected datasets influence research conclusions and outcomes.

Particular analyses may inadvertently promote stereotypes related to behavioral health, particularly in marginalized groups leading to adverse societal implications.

Premature Deployment: The analysis and translation of data into clinical practices may overlook the limitations and possible misuses of findings.
Circumstances of Justice: The ethical landscape changes when researchers must balance ethical risks laid on the community versus individual subjects, obligating the necessity to ensure divided benefits and burdens.

Emphasizes the importance of collective justice within the broader community regarding the secondary research ethics, expanding beyond traditional concepts focused on identifiable individuals.
Broad Understanding of Justice:
- Just treatment of all impacted parties, irrespective of direct participation.
- Obligations extend to respecting interests and well-being of entire communities.

Societal Value of Knowledge: Emphasizes producing research that is beneficial on a community level.
Fair Distribution of Research Outcomes: Address health disparities and ensure equitable access to findings.
Research Transparency: The community has the right to be informed and involved in the research process.
Data Withdrawal Mechanisms: Allow participants to opt-out where risks are evident.
Collective Stakeholder Inputs: Ensure collective governance of research sharing dynamic ensures fairness.
Responsible Reporting: Findings must be honestly and ethically communicated.
Obligation to Participate: Recognizes a societal responsibility to contribute to research and knowledge-sharing in fair terms.