How do I access and use the Trinitex database to retrieve cardiovascular research data?

Accessing and Using TriNetX for Cardiovascular Research

TriNetX is a global federated health research network that provides access to de-identified electronic health record (EHR) data from healthcare organizations, enabling researchers to conduct large-scale retrospective cohort studies and real-world evidence analyses without requiring direct data transfers.

What is TriNetX?

TriNetX operates as a cloud-based platform that connects healthcare organizations and research institutions through a federated network architecture 1, 2, 3. The platform maintains patient privacy through de-identification while allowing real-time access to clinical data for research purposes 4, 3.

Key Network Characteristics

The TriNetX Dataworks-USA Network contains:

• Over 150 million de-identified patient EHRs 3
• Data from approximately 38-42 million patients across 35+ healthcare organizations in the United States 5, 2
• Regularly updated clinical information including demographics, diagnoses (ICD codes), procedures, medications, and laboratory results 1, 6, 2

The network is broadly generalizable to the US healthcare-seeking population, though it skews slightly older (approximately 5 years) and has a higher proportion of female patients compared to US Census data 2.

How to Access TriNetX

Access requires institutional affiliation with a participating healthcare organization or research institution that has joined the TriNetX network 4, 3. Individual researchers cannot directly purchase access; rather, their institution must establish a collaboration agreement with TriNetX.

Establishing Access:

1. Institutional Partnership: Your healthcare organization or academic institution must join the TriNetX network as a participating member 4
2. Data Use Agreements: Institutional agreements ensure HIPAA compliance and appropriate data governance 3
3. User Credentials: Once institutional access is established, individual researchers receive login credentials to the web-based platform 3

Using TriNetX for Cardiovascular Research

Platform Capabilities

The TriNetX interface allows researchers to:

• Query patient cohorts using ICD-10 codes, medications, procedures, and laboratory values 1, 6, 7
• Perform propensity score matching to create comparable control groups 7, 5
• Calculate hazard ratios, risk ratios, and survival analyses 8, 7
• Generate Kaplan-Meier survival curves 8
• Export de-identified aggregate statistics (individual patient-level data cannot be exported) 3

Cardiovascular Research Applications

TriNetX has been successfully used for multiple cardiovascular studies:

Post-MI lipid management: Researchers assessed real-world lipid-lowering treatment gaps after acute myocardial infarction, finding that only 7-12% of patients achieved LDL-C <55 mg/dL targets within 3 years 6

SGLT2 inhibitor cardiovascular outcomes: Analysis of 46,909 patients on SGLT2 inhibitors versus 189,120 on DPP-4 inhibitors showed a risk ratio of 0.63 (95% CI: 0.60-0.66) for stroke or MI, replicating RCT findings in real-world settings 5

LVAD long-term survival: Investigators identified CMV infection, hepatitis A exposure, atrial fibrillation, ischemic cardiomyopathy, and renal dysfunction as risk factors impacting long-term survival in LVAD recipients 8

COVID-19 cardiovascular sequelae: A cohort of 4.1 million patients demonstrated increased 12-month risks of stroke (HR 1.62), atrial fibrillation (HR 2.41), myocarditis (HR 4.41), and heart failure (HR 2.30) in COVID-19 survivors 7

Methodological Approach for Cardiovascular Studies

Step 1: Define Your Research Question

Identify specific cardiovascular outcomes using standardized coding:

• ICD-10 codes for diagnoses (e.g., I21 for MI, I63 for stroke, I50 for heart failure) 6, 7
• Medication codes for treatments (e.g., statins, SGLT2 inhibitors, anticoagulants) 6, 5
• Procedure codes for interventions (e.g., PCI, CABG, TAVR) 9

Step 2: Build Your Cohorts

• Exposure cohort: Define using diagnosis codes, medication prescriptions, or procedures 7, 5
• Control cohort: Select comparable patients without the exposure 7, 5
• Inclusion/exclusion criteria: Specify age ranges, comorbidities, and time periods 8, 6, 7

Step 3: Propensity Score Matching

Match cohorts on confounding variables including:

• Demographics (age, sex, race) 7, 2
• Comorbidities (hypertension, diabetes, CKD) 5
• Concomitant medications 5

This creates balanced groups for fair comparison 7, 5.

Step 4: Define Follow-Up Period

• Avoid reverse causality by starting follow-up 30 days after index event 7
• Specify observation windows (e.g., 1 year, 3 years) 8, 6, 7
• Account for loss to follow-up 3

Step 5: Analyze Outcomes

Calculate:

• Hazard ratios with 95% confidence intervals using Cox regression 8, 7
• Risk ratios for binary outcomes 5
• Kaplan-Meier survival curves 8
• Absolute risk reductions 7

Critical Limitations and Caveats

Laboratory Data Constraints

TriNetX only provides the most recent laboratory test result within a specified observation period 1. This severely limits analysis of:

• Serial laboratory measurements over time
• Trends in biomarkers (e.g., troponin kinetics, BNP trajectories)
• Laboratory-based treatment monitoring

Workaround: Define multiple discrete time windows and query the most recent value within each window separately 1.

Missing and Incomplete Data

• High proportion of missing race/ethnicity data in some healthcare systems 2
• Laboratory values not universally captured across all participating institutions 1
• Medication adherence cannot be assessed (only prescriptions, not actual consumption) 6

Selection Bias

• Healthcare-seeking population only: Patients without regular healthcare access are underrepresented 2
• Geographic clustering: Network composition may not represent all US regions equally 2

Coding Accuracy

• Diagnosis codes depend on clinical documentation quality and may not capture subclinical disease 3
• Procedure codes may not reflect technical success or complications 3

Advantages Over Traditional RCTs

TriNetX complements randomized controlled trials by:

• Providing larger sample sizes for rare outcomes and subgroup analyses 8, 3
• Including diverse, real-world patient populations often excluded from RCTs 5, 3
• Enabling rapid hypothesis generation and feasibility assessments for clinical trials 3
• Detecting rare adverse events that RCTs are underpowered to identify 3
• Assessing long-term outcomes beyond typical RCT follow-up periods 8, 7

Practical Example: Cardiovascular Outcomes Study

To assess cardiovascular outcomes after a specific intervention:

1. Define exposure: ICD-10 code for index procedure/diagnosis + date
2. Build control cohort: Patients without exposure, matched 1:1 on age (±5 years), sex, diabetes, hypertension, prior CVD
3. Exclude: Patients with outcome of interest in 30 days pre-index (prevalent cases)
4. Follow-up: Start 30 days post-index, continue 12 months
5. Outcomes: Composite MACE (MI [I21], stroke [I63], cardiovascular death), individual components, heart failure hospitalization [I50]
6. Analysis: Cox proportional hazards regression, report HR with 95% CI, Kaplan-Meier curves

This approach has been validated across multiple cardiovascular studies in TriNetX 8, 6, 7, 5.