Dataset dissemination SOP

BNR Operations Manual | Dataset dissemination

11 May 2026

Document Type: Standard Operating Procedure (SOP)
Document Version: v1.0
Effective Date: [Insert Date]
Author: BNR Data Governance Committee
Approved by: BNR Professional Advisory Board

1. Purpose

This SOP outlines the policies and procedures governing dissemination of datasets produced by the BNR. It defines the types of datasets eligible for dissemination, the required approvals and safeguards, and how these align with national legislation including the Barbados Data Protection Act (2019). It also establishes principles for public access, academic use, and internal control.

2. Scope

This SOP applies to all stakeholders involved in the sharing, access, and reuse of BNR data. It applies to all data types derived from CVD registry operations including acute myocardial infarction and stroke datasets.

3. Definitions

Full Dataset: Complete case-level data with identifying fields (names, national IDs, exact dates, contact info).

De-identified Dataset: Individual-level data with direct identifiers removed and some quasi-identifiers retained (e.g. age, month, parish).

Anonymised Dataset: Data irreversibly stripped of identifiers and generalized (e.g. no dates, age binned).

Aggregated Dataset: Group-level summaries only. No individual data; all counts ≥ 5.

DUA (Data Use Agreement): Legal agreement for use of de-identified data.

4. Responsibilities

Role Responsibility
Data Abstractor Accurate and complete case entry
Statistician Risk analysis, documentation, and tier classification
QC Coordinator De-identification QA and suppression validation
Data Privacy Officer Disclosure risk sign-off and compliance with law
Technical Lead Final approval for dataset release
Data Access Manager Manage access requests and maintain audit logs

5. Risk Assessment for Disclosure

5.1 Re-identification Risk Assessment

All datasets except aggregated undergo disclosure risk analysis each month prior to release. Risk assessment includes: - k-anonymity and l-diversity checks on sensitive variable combinations. - Frequency tables on combinations like: - Age × Sex × Geography - Age × Outcome × Date - Event type × Length of stay - Small cell suppression for n < 5. - Scoring of risk using a 3-level matrix: - Low: Approved - Medium: Suppression/aggregation required - High: Not released

Monthly Log: Risk assessment results logged per release (BNR_DISCLOSURE_LOG_.txt) and signed by Data Privacy Officer.

5.2 Residual Risk Controls

Even after de-identification or anonymisation, residual disclosure risks exist. Mitigation includes: - Secondary suppression where back-calculation is possible. - No release of cross-tabulated data for rare combinations. - Use of binned/rounded variables where needed. - Data Privacy Officer must review each release and complete a signed residual risk clearance form.

6. Dataset Types and Access Conditions

Dataset Tier Description Access Approval
Aggregated Counts, percentages, and rates. No individual records. Open (website) None
Anonymised Fully stripped of identifiers. Cannot be re-identified. Public download BNR internal sign-off
De-identified Quasi-identifiers retained. IRB + DUA + Output review needed. Restricted IRB + Technical Lead
Full Identifiable health data. Internal QA/audits only. Internal only Technical Lead only

Data Dissemination Flowchart

flowchart TD
    A[Data Cleaning & QC Complete] --> B[Classify Dataset Tier]
    
    B --> C1[Aggregated Tier]
    B --> C2[Anonymised Tier]
    B --> C3[De-identified Tier]
    B --> C4[Full Dataset - Internal Only]

    %% Aggregated
    C1 --> D1[Suppression Rules Applied]
    D1 --> E1[Compliance Checklist Completed]
    E1 --> F1[Metadata Prepared]
    F1 --> G1[Published on Open Portal]

    %% Anonymised
    C2 --> D2[Run Re-identification Risk Assessment]
    D2 --> E2[Privacy Officer Review]
    E2 --> F2[Internal Sign-off]
    F2 --> G2[Prepare ZIP Package]
    G2 --> H2[Public Release]

    %% De-identified
    C3 --> D3[Access Request + IRB Approval]
    D3 --> E3[Review by Data Access Committee]
    E3 --> F3[DUA Signed]
    F3 --> G3[Risk Assessment + Suppression]
    G3 --> H3[Final Approval]
    H3 --> I3[Secure Transfer]

    %% Full Dataset
    C4 --> Z1[Internal QA or Audit Use Only]

    %% Logging
    G1 --> L[Monthly Log]
    H2 --> L
    I3 --> L
    Z1 --> L

8. Review and Publication

  • An internal checklist is used for every dissemination to validate:
    • Tier classification
    • Suppression rules applied
    • Risk clearance
    • Metadata completeness
  • Disseminated datasets include:
    • Data file (CSV, DTA, or JSON)
    • Metadata (.txt)
    • Suppression note
    • Recommended citation (for public use)

9. Version Control and Audit

  • Each dissemination assigned a unique name: BNR-<CVD>-<TIER>-<YYYYMM>.dta
  • Audit logs stored in BNR-ACCESS-LOG-<YYYYMM>.txt
  • Monthly sign-off required from Technical Lead and Data Privacy Officer

10. Contact and Queries

For dataset access requests or questions, contact:
Data Access Manager
datarequests@bnr.org.bb

11. Review Cycle

  • SOP reviewed every 12 months or sooner if legislation or BNR policy changes.
  • Revisions must be approved by the BNR Professional Advisory Board.
Back to top