Skip to main content
Expert Talk
Amakomaya — Aamako Maya Apps Pvt. Ltd.

Data Governance

Data governance answers a question no technology answers: who decides? Who may define an indicator, change a data element, grant access, approve a release, or retire a dataset — and how those decisions are recorded.

Security controls protect data. Governance decides what protection is for.

What it covers​

Decision rights​

Named accountability for each domain of data: who owns clinical data, registry data, financial data. "The IT department" is not an answer — IT operates systems; it does not own clinical definitions.

Stewardship​

Data stewards are the people who know what a field means and are accountable for its quality. Usually programme or clinical staff, not engineers.

Metadata management​

Definitions, code lists, indicator formulas and their change history. In a DHIS2 deployment this is the metadata layer, and it is where governance either exists or does not — see metadata design principles.

Quality accountability​

Defined dimensions, defined thresholds, defined follow-up — see health data. Quality that is everyone's responsibility is no one's.

Access management​

Who may see which data at which granularity, on what basis, reviewed on a schedule. Access granted for a project that ended three years ago is a live risk.

Lifecycle​

Retention, archival and deletion. Health records have long statutory retention periods; other data does not, and keeping everything forever is a liability, not prudence.

Making change management real​

The single highest-value governance practice in a health information system is disciplined change control over shared metadata:

  1. A proposed change is documented with its rationale and impact.
  2. Affected stewards review it.
  3. Changes are batched into a release, not applied ad hoc.
  4. Downstream users are notified before it takes effect.
  5. The change is recorded with date, author and reason.
  6. Breaks in indicator time series are documented explicitly.

Without this, a well-meaning correction mid-year silently invalidates a national time series.

Governance bodies​

Typically a small standing group with representation from clinical/programme areas, information systems, legal or privacy, and analysis. Meets on a schedule. Keeps decisions in writing. What matters is not the org chart but whether decisions are documented and findable a year later.

Signs governance is missing​

  • Two dashboards showing different values for the same indicator
  • Nobody can say who approved a definition change
  • Access lists that only grow
  • Data quality discussed but never assigned
  • Every integration renegotiating the meaning of the same fields
Last updated on