TriNetX Database: Clinical Research & Cohort Guide

The TriNetX database is a global health research platform and federated real-world data network connecting healthcare provider organizations, academic medical centers, and biopharmaceutical companies. Founded in 2013, the platform enables investigators to explore electronic health record (EHR) data, evaluate trial feasibility, build patient cohorts, and generate real-world evidence without custom statistical programming for baseline queries.

For clinical investigators and research administrators, using TriNetX requires understanding how its federated infrastructure protects patient privacy, how medical ontologies are harmonized across hospital systems, and how to apply sound epidemiological principles when designing a retrospective cohort study. Users access the platform through institutional Single Sign-On (SSO) portals managed by their health system’s Clinical and Translational Science Award (CTSA) informatics core or via the web portal at live.trinetx.com following approved data use agreements.

What it is

The TriNetX platform is an interactive, web-based analytics appliance built on a federated real-world data network. Unlike centralized repositories that aggregate patient records into a single warehouse, TriNetX leaves electronic medical record data behind the firewalls of each participating healthcare organization (HCO). When a researcher submits a query, criteria travel to local TriNetX appliances at member institutions. These appliances execute searches locally and return only aggregate, de-identified statistical results to the central interface.

This federated architecture aligns with HIPAA in the United States and GDPR in Europe. Under HIPAA regulations (45 CFR 164.512(i)(1)(ii)), researchers may use aggregate de-identified counts for activities preparatory to research—such as assessing protocol feasibility or establishing cohort size—without upfront Institutional Review Board (IRB) review. Downstream workflows requiring patient re-identification for trial recruitment or row-level de-identified data extraction require formal institutional review and governance approvals.

The platform also integrates automated analytics, allowing researchers to explore cohort demographics, generate Kaplan-Meier survival curves, and perform propensity score matching directly within the web application.

Who runs it and what it covers

TriNetX is operated by TriNetX LLC, a commercial health research network company based in Cambridge, Massachusetts, founded in 2013. The company partners with healthcare organizations through data-sharing agreements, granting academic investigators access to network-wide data in exchange for institutional participation in sponsored trial feasibility queries.

According to platform documentation verified in November 2026, the global TriNetX network encompasses:

• 309M+ patient lives across participating healthcare organizations worldwide.
• 99B+ clinical data points mapped across longitudinal clinical encounters.
• 14,200+ clinical sites across more than 20 countries.
• 65M+ linkable patient records available for longitudinal tracking and comparative studies.
• 4,000+ peer-reviewed publications supported by TriNetX real-world data.

Member health systems extract records from EHR systems (including Epic, Oracle Health/Cerner, and MEDITECH) and transform variables into standardized clinical vocabularies: diagnoses (ICD-9-CM, ICD-10-CM), procedures (CPT, HCPCS, ICD-10-PCS), medications (RxNorm, ATC), labs and vitals (LOINC), and specialized oncology staging (TNM) or genomic biomarkers.

The platform offers three core product tiers: TriNetX LIVE™ (interactive web cohort discovery and survival analytics), TriNetX Dataworks™ (downloadable, row-level de-identified datasets for external statistical software), and TriNetX Linked™ (tokenized linkage of EHR data with closed claims, mortality registries, and Social Determinants of Health).

How to search or use it

Constructing a cohort in the TriNetX Query Builder mirrors clinical trial inclusion and exclusion criteria. The following six-step workflow outlines the core interface operations verified from platform documentation:

1. Select Network Scope: In the Query Builder, toggle between the Local/Institutional Network (home medical center records only) and the Global Collaborative Network (multicenter federated data).
2. Define Cohort Criteria in “Must Have” and “Cannot Have” Containers: Place inclusion parameters into the Must Have panel and exclusion criteria into the Cannot Have panel, applying Boolean AND or OR operators across groups.
3. Search Standardized Terminologies: Search clinical concepts using standard coding systems (ICD-10, CPT, LOINC, RxNorm) or keywords across Diagnoses, Procedures, Medications, Labs, and Genomics. Selecting a parent category automatically includes child subcodes.
4. Configure Temporal Logic: Establish chronological relationships relative to an index event (such as first diagnosis or drug start date), defining observation windows before or after that event.
5. Evaluate Attrition with “Analyze Criteria”: Run the built-in Analyze Criteria tool to review attrition diagrams and identify which specific inclusion or exclusion criteria eliminate candidate patients.
6. Execute Advanced Analytics and Balancing: Move cohorts into Advanced Analytics. In Compare Cohorts, execute 1:1 nearest-neighbor propensity score matching to balance baseline covariates, then transition to Compare Outcomes to calculate hazard ratios and Kaplan-Meier curves.

Strengths and limits

Investigators evaluating TriNetX must balance operational efficiency against methodological constraints common to secondary electronic health record research.

Key Strengths

• Rapid Cohort Discovery: Multicenter feasibility counts that previously required months of institutional data requests return in seconds.
• Statistical Automation: Automated 1:1 propensity score matching and survival analysis allow researchers to explore comparative effectiveness without custom programming.
• Broad International Reach: Over 309 million patients across 20+ countries provide statistical power for investigating rare diseases and uncommon outcomes.
• Longitudinal Linkage: Over 65 million linkable patient records support multi-year outcome assessments within participating health systems.

Methodological Limitations

• Care Fragmentation: In baseline EHR networks without claims linkage, clinical care delivered outside member hospitals remains unobserved.
• Selection Bias: Participating institutions are largely academic, tertiary medical centers, resulting in underrepresentation of rural populations.
• Billing Code Variability: Diagnostic codes are assigned for clinical billing rather than research adjudication, leading to potential misclassification.
• Unmeasured Confounding: Propensity score matching balances only recorded variables; it cannot correct for unmeasured confounders like over-the-counter drug use.
• Literature Quality Risks: Automated analyses have contributed to a surge of low-effort retrospective studies. Investigators should consult the analysis of the TriNetX study surge and research integrity and adhere to STROBE and RECORD-PE guidelines.

Alternatives

The table below summarizes how TriNetX compares to major alternative clinical research networks:

Platform	Governing Entity	Architecture & Model	Network Scale	Access Model	Primary Research Strength
TriNetX	TriNetX LLC (Commercial)	Federated network; proprietary schema	309M+ patients; 20+ countries	CTSA or commercial subscription	Rapid no-code cohort discovery, trial feasibility, automated propensity matching
Epic Cosmos	Epic Systems (Commercial)	Centralized pooled repository	260M+ patients; US Epic health systems	Restricted to participating Epic health system researchers	Direct longitudinal EHR integration, deep clinical detail, provider consistency
N3C Data Enclave	NIH / NCATS (Public/Academic)	Centralized enclave; OMOP CDM	20M+ patients; 80+ contributing institutions	Open to authorized researchers with DUA and IRB approval	Public health research, reproducible analytical environments (R/Python/SQL)
PCORnet	PCORI (Non-profit/Federal)	Distributed network; PCORnet CDM	30M+ active patient records; US health systems	Proposal review via Clinical Research Networks	Patient-centered outcomes, multi-site pragmatic clinical trials
OHDSI / OMOP Network	Open Science Community	Decentralized open-source; OMOP CDM	Hundreds of millions globally across partners	Open-source software; data access negotiated per node	Transparent open-source analytics (ATLAS), standardized phenotyping

FAQ

Does using TriNetX require upfront Institutional Review Board (IRB) approval?

For initial aggregate cohort discovery and trial feasibility assessments, upfront IRB approval is generally not required at participating institutions. Under the HIPAA Privacy Rule (45 CFR 164.512(i)(1)(ii)), querying aggregate, de-identified counts is permitted as activity preparatory to research. Downstream activities involving row-level data analysis or patient re-identification for clinical trials require formal IRB review.

What is the difference between the Local Network and the Global Collaborative Network?

The Local Network restricts query results exclusively to electronic medical records from the investigator’s home institution, facilitating local audits and feasibility checks. The Global Collaborative Network federates queries across participating health systems worldwide, returning aggregated de-identified counts across hundreds of millions of patient lives.

How does TriNetX protect patient confidentiality in aggregate queries?

TriNetX uses a statistical de-identification process compliant with HIPAA Expert Determination standards. To prevent deductive re-identification of patients with rare conditions, small-cell suppression masks any cohort count fewer than 10 patients, displaying it as ≤ 10 rather than an exact count.

Can TriNetX track medications administered outside the hospital?

In standard EHR-based network views, TriNetX records outpatient electronic prescriptions and inpatient administrations documented within the health system. It cannot verify whether prescriptions were filled at outside pharmacies unless researchers access the TriNetX Linked™ product, which incorporates closed pharmacy claims data.

Sources

• TriNetX Global Real-World Data Network Overview. TriNetX LLC. Available at: https://trinetx.com/data/. Accessed November 8, 2026.
• TriNetX Platform and Governance. TriNetX LLC. Available at: https://trinetx.com/about/. Accessed November 8, 2026.
• TriNetX LIVE™ Platform Solutions. TriNetX LLC. Available at: https://trinetx.com/solutions/trinetx-live/. Accessed November 8, 2026.
• TriNetX Query Builder Documentation. TriNetX LLC. Available at: https://trinetx.com/solutions/trinetx-live/query-builder/. Accessed November 8, 2026.
• TriNetX Basic and Advanced Analytics. TriNetX LLC. Available at: https://trinetx.com/solutions/trinetx-live/basic-analytics/ and https://trinetx.com/solutions/trinetx-live/advanced-analytics/. Accessed November 8, 2026.
• Clinical and Translational Science Institute: TriNetX Research Resources. University of Rochester Medical Center. Available at: https://www.urmc.rochester.edu/clinical-translational-science-institute/resources/trinetx.aspx. Accessed November 8, 2026.