Real-world data · Latin America
Real-world health evidence, observed at the source.
Longitudinal, de-identified data from Latin America, beginning with an integrated payer and provider in Argentina. Claims, clinical records and real costs on the same patient, delivered under governance.
- visit
- lab result
- dispensing
- admission
- 01 · members payer
400,000
Private payer, Argentina
- 02 · patients a day provider
~1,000
High-complexity hospital, Buenos Aires
>3,000 deliveries / yr - 03 · critical-care patients per year
~1,700
A year, with laboratory trajectories and discharge outcomes
The data
Four layers on the same patient
One pseudonymous person key links the claim, the clinical encounter and its real cost. Each layer maps to OMOP CDM domains.
- 01
Financial
Claims · real costClaims and real costs: authorizations, reimbursed services, copayments and the cost actually borne by the payer, not list prices.
costpayer_plan_perioddrug_exposure - 02
Clinical
Primary recordConsultations, laboratory results, imaging reports, admissions and discharges, captured in the provider’s electronic record.
condition_occurrencemeasurementprocedure_occurrencevisit_occurrence - 03
Network
DerivedThe group’s own physicians, referral paths between specialties and sites, and where care actually takes place.
providercare_sitereferral paths - 04
Behavioural
DerivedFrequency of use, continuity of care and adherence patterns in chronic conditions, derived from longitudinal activity.
adherence (PDC)persistencegaps in care
Access
How a client works with the data
Three modes of access and one rule: identified data never leave their country of origin, and clients never see an individual row.
- 01
Clean Room
SQL over de-identified OMOP tables inside our environment. Results are aggregated and reviewed before release.
- 02
Federated node
Your protocol travels to the data. Summary tables travel back. Designed for compatibility with OHDSI network studies.
- 03
Feasibility API
How many patients meet your criteria? Rounded counts and distributions, returned without a data transfer.
Feasibility API
Counts before commitment.
Submit inclusion and exclusion criteria. Receive rounded counts and distributions, nothing below the minimum cell size, without a data transfer.
- Inclusion
- Type 2 diabetes, first diagnosis in window SNOMED 44054006
- Measurement
- HbA1c ≥ 8 % within 90 days LOINC 4548-4
- Exposure
- New users of GLP-1 analogues ATC A10BJ
- Observation
- ≥ 365 days continuous enrolment before index
Eligible persons · rounded
n rounded
min cell 10 · dates shifted · aggregates only
Age distribution · illustrative shape
Regional focus
Built for Latin America, one country at a time
Built for Latin America
Designed for the region
The platform, vocabularies and governance model are built for Latin American health systems, where payer and provider data are rarely linked and regional populations are missing from most evidence.
First market
Argentina
Where the first source is live: an integrated payer and provider with 400,000 members and around 1,000 patients a day, hosted in Argentina under Law 25.326.
Next
Country by country
Each additional market gets its own source, its own data residency and its own legal basis on the same OMOP architecture. The region is not one population, and we do not model it as one.
Depth
Where the clinical layer is densest
Volumes reflect the hospital’s profile. Every variable in the catalogue carries its vocabulary, its origin and its completeness.
| Specialty | Volume | Typical content | Vocabularies | Evidence |
|---|---|---|---|---|
| Obstetrics & neonatology | >3,000 deliveries / yr | Antenatal visits, delivery, neonatal course, maternal follow-up | SNOMED · LOINC | Primary |
| Critical care | ~1,700 patients / yr | ICU stays, laboratory trajectories, procedures, discharge outcomes | SNOMED · LOINC · ATC | Primary |
| Interventional cardiology | reference service | Catheterisation, prior and subsequent utilisation in claims | SNOMED · ATC | Primary + claims |
| Diagnostic imaging | reference service | Modality, body region, report; no pixel data | DICOM · SNOMED | Primary |
Governance
Identified data never leave the country of origin.
Pseudonymisation happens before the data reach the platform. Everything a client touches sits downstream of that step.
- 01 · Source AR
FHIR R4 export
Payer claims and hospital clinical systems.
person_id: identified
- 02 · Pseudonymise AR
Hash + salt, Δt, bands
Salt held apart from the data. Dates shifted per person; intervals exact.
token: a3f9…e71c · age: 50–59 · Δt shifted
- 03 · Fabrica AR
OMOP CDM
OHDSI vocabularies, quality checks, Tabulae data dictionary.
condition_concept_id: 201826 (T2DM)
- 04 · Theatrum Client
Aggregates only
Clean Room, federated node, Feasibility API. Output review before release.
rows exported: 0 · queries logged: all
- OMOP CDM v5.4 · data model
- FHIR R4 ingestion
- SNOMED CT conditions · procedures
- LOINC laboratory
- DICOM imaging metadata
- ATC drugs · with RxNorm
Why it matters
Why an integrated source changes the evidence
The patient is complete
Most real-world datasets see either the claim or the chart. Here the same person appears in both, with the cost attached, so outcomes, utilisation and spend are measured on one population rather than stitched across three.
Costs are real, not imputed
Financial data come from the payer’s own ledger: what was authorised, delivered and paid. No reference tariffs standing in for actual expenditure.
Evidence from a region that is rarely observed
Latin American populations are under-represented in real-world evidence. We build governed sources country by country, beginning in Argentina, so a submission can include the region without moving a single identified record.
Contact
Open a conversation.
We are in pre-launch and working with a small number of pharma, CRO and health-system partners. Tell us the question you need to answer and we will return a feasibility assessment.