AoU_CLAD_ROA-OTA-23-003 Appendix D - Data Tier Alignment.docx.pdf

APPENDIX D: ALIGNMENT OF DATA LINKAGES WITH DATA TIERS AND

INFORMED CONSENT

The All of Us Research Program can only pursue data linkages that align with the informed consent that program participants agree to. The primary consent to participate in the program is broad and should cover most data linkages. The consent informs prospective participants that we will add passive data about their health from sources other than the data they provide to the All of Us Research Program. The consent informs participants that linkages may be done based on where they live and work, or based on data that identifies them. Participants are provided with examples. The consent also informs participants that the Institutional Review Board will review and approve each data linkage.

All of Us currently makes participant data available to researchers on the RW through the following three tiers:

a) Public Tier: aggregate data and summary statistics;

b) Registered Tier: participant-level data subject to a high degree of privacy-preserving obfuscation and transformation; and

c) Controlled Tier: participant-level data with additional data types and fewer such transformations.

Use of any tier falls under a non-human subjects research (non-HSR) designation, in which participant identities may not “readily be ascertained by the investigator or associated with the information”(45 CFR 46.102, pre-2018 Common Rule).

The DRC has privacy experts that work to ensure that each tier maintains its non-HSR status and that the risks of re-identification are kept sufficiently low. The following table indicates what data types can currently be made available through the Registered and Controlled Tiers based on the assessments they have conducted to date and as approved by the All of Us Research Program Committee on Access, Privacy, and Security (CAPS). Information on the current data dictionaries can be found on the RW User Support website. More information on the RW data tiers can be found on the Research Hub.

Based on these determinations, there are data types that cannot currently be made available under any existing tiers. Consequently, the awardee will need to develop data de-identification, degradation, or obfuscation methods to ensure that linked data can be incorporated, as appropriate, into one or more tiers. Note, the All of Us Research Program is currently investigating an additional tier that may allow for inclusion of additional data types.

Data Element Registered Tier Controlled Tier

Explicit identifiers Suppress Suppress

Free text fields in survey and unstructured clinical documents

Suppress Suppress

Dates (of events) Random shift As Collected (unshifted) https://support.researchallofus.org/hc/en-us/articles/360033200232-Data-dictionaries-for-Curated-Data-Repository-CDR-https://www.researchallofus.org/data-tools/data-access/

Backward by a random number between 1 to 365

Date of Birth Random shift Backward by a random number between 1 to 365

Generalize to year of birth

Date of Death Random shift Backward by a random number between 1 to 365

As Collected (unshifted)

Data of participants age >89 Suppress As Collected

Geolocation Generalize to U.S. state Generalize the first three digits of zip code. In cases of less than 20,000 of participants in a 3-digit zip code area, those participants will be aggregated into another nearby local 3-digit zip code

Marital status As Collected As Collected

Living situation PPI (survey): Where are you currently living?

Suppress As Collected

Own or rent As Collected As Collected

Self-identified race and ethnicity (broad categories) e.g.: Asian, White, Black, MENA, etc.

Generalize As Collected

Self-identified race and ethnicity (subcategories) e.g.: Hmong, Filipino, Caribbean

Suppress Suppress

Sex assigned at birth (PPI)* Generalize As Collected

Gender identity (PPI) Generalize As Collected

Sexual orientation (PPI) Generalize As Collected

Race and Ethnicity (EHR) Suppress Value from EHR is suppressed to harmonize with PPI data

As Collected

Sex/Gender (EHR) Suppress Value from EHR is suppressed to harmonize with PPI data

As Collected

ICD codes indicative of suppressed sex/gender

Suppress As Collected

Educational attainment level Generalize As Collected

Employment status Generalize As Collected

Annual household income As Collected As Collected

Death cause i.e., Death cause noted in the EHR, including relevant diagnosis codes

Suppress As Collected

Diagnosis codes subject to public knowledge

Suppress As Collected

ICD Codes indicative of motor vehicle accidents

ICD9 E80*-E84*, ICD10 V*

Suppress Suppress

Active duty military status Suppress As Collected

Born in U.S. or not As Collected As Collected

Genomic data Includes program-generated Whole Genome Sequencing and Array data

Suppress As Collected

Note: ‘As Collected’ indicates that there will be no change to the data for the purpose of privacy protection *Free text responses will be suppressed.