Virtual Data Warehouse – Most MHRN research relies on the Virtual Data Warehouse (VDW) developed by the Health Care Systems Research Network. Data are extracted from a range of health system databases, including electronic health records, insurance claims, laboratory results, and pharmacy dispensings. Each participating research center maintains a series of data tables following a common data model. This distributed or federated data structure supports population-based research while protecting privacy of patients’ health information.
Key Data Areas – VDW data tables most often used in MHRN research include:
- Demographics – includes age, sex, race, ethnicity, sexual orientation, and gender identity
- Enrollment – records periods of enrollment in health system insurance plan OR alternative utilization-based measures of health system affiliation
- Encounters – records inpatient and outpatient encounters, including video visits and other virtual encounters
- Procedures – records outpatient and inpatient procedures using standard CPT and/or ICD codes
- Diagnoses – records diagnoses assigned at all encounters using standard ICD codes
- Outpatient medication dispensings – records ingredients, strength, quantity, duration, and instructions for all outpatient medication dispensings
- Patient-reported outcomes – records selected patient-reported outcome measures (e.g. PHQ-9 depression scale, GAD anxiety scale, Columbia Suicide Severity Rating Scale, AUDIT-C alcohol questionnaire) completed at visits or via the electronic health record patient portal
- Vital signs – height, weight, blood pressure
- Mortality – records data and cause of death via linkages to hospital records, Social Security Administration databases, and state or national mortality records
- Census – describes neighborhood characteristics (income, educational attainment, housing) according to residential address in health system records
Useful Tools – MHRN has developed several sets of useful tools (lists of diagnosis codes, lists of drug codes, etc.) to facilitate use of health system records in mental health research. More here.
Routine Descriptive Analyses – All MHRN members participate in routine descriptive analyses of VDW data to both assess data quality and identify trends in mental health diagnosis or treatment. More detailed annual analyses examine counts and rates for a wide range of mental health diagnoses and treatments, including stratification by various sociodemographic characteristics. More here.
Custom “Add-On” Queries – Ad hoc custom queries of VDW data tables can address more specific or detailed research questions. More here.
Distributed vs. Central Data Analysis – Analyses of VDW data for MHRN research typically follow one of two models:
- Distributed data analysis with pooling of results – A common program is executed in each health system so that counts, rates, and frequency distribution data can be shared and combined. No individual-level data are shared.
- Centralized analysis of individual-level data – A common program is executed to extract and process data into parallel analytic datasets. Those individual-level data can be shared and combined for more detailed analyses.