Skip to main content

ADR 0003: JSON mode writes directly to calculated tables

Status​

Accepted

Context​

The JSON input currently represents patient-level demographics/diagnosis summary data, not the full source-table shape used by CSV and MySQL ingestion.

Decision​

SOURCE_TYPE=json writes directly to CALCULATED_PATIENT_DATA and CALCULATED_DX_DATA instead of attempting to reconstruct the full source-table pipeline.

Consequences​

Positive:

  • simple and fast JSON test runs
  • clear contract for patient-level ingestion
  • fewer assumptions about missing source-table detail

Tradeoffs:

  • JSON mode is not feature-equivalent to csv/mysql
  • source-table-dependent translations and ICD extraction are skipped
  • documentation must be explicit so users do not assume full parity