Related Experiment Video
Updated: Jun 19, 2026

Inverse Probability of Treatment Weighting (Propensity Score) using the Military Health System Data Repository and National Death Index
Published on: January 8, 2020
Using administrative data to identify mental illness: what approach is best?
Susan M Frayne1, Donald R Miller, Erica J Sharkansky
1Center for Health Care Evaluation, VA Palo Alto Health Care System, 795 Willow Road, Menlo Park, CA 94025, USA. sfrayne@stanford.edu
Identifying mental health conditions (MHCs) in administrative data requires careful algorithm selection. Optimizing positive predictive value (PPV) and negative predictive value (NPV) improves accuracy for quality improvement programs.
Area of Science:
- Health Services Research
- Psychiatry
- Health Informatics
Background:
- Administrative data are increasingly used for health research, but their accuracy in identifying mental health conditions (MHCs) requires validation.
- The Veterans Health Administration (VHA) administrative data are a valuable resource for studying MHCs in veteran populations.
- Accurate identification of MHCs is crucial for quality improvement initiatives and understanding healthcare disparities.
Purpose of the Study:
- To estimate the validity of algorithms for identifying MHCs in VHA administrative data.
- To compare the performance of different algorithms using International Classification of Diseases, 9th Revision (ICD-9) codes against self-reported survey data.
- To provide guidance on selecting appropriate algorithms for MHC identification in administrative datasets.
Main Methods:
- Utilized data from 133,068 diabetic patients in the VHA national cohort (1998) who responded to the 1999 Large Health Survey of Veteran Enrollees.
- Compared various algorithms using ICD-9 codes for MHCs against self-reported depression, posttraumatic stress disorder, and schizophrenia.
- Calculated positive predictive value (PPV) and negative predictive value (NPV) for each algorithm.
Main Results:
- The PPV and NPV for identifying MHCs varied significantly across different algorithms, ranging from 0.65-0.86 and 0.68-0.77, respectively.
- Optimizing PPV was achieved by requiring two or more instances of MHC ICD-9 codes or by exclusively using codes from mental health visits.
- Supplementing VHA data with Medicare data enhanced the NPV.
Conclusions:
- Algorithm choice significantly impacts the accuracy of identifying MHCs in administrative data.
- Requiring multiple diagnostic codes or restricting to mental health visits can improve PPV.
- Integrating data sources, such as Medicare, can enhance NPV for MHC identification.
- Findings support the use of validated algorithms for MHC identification in quality improvement and health disparity research within the VHA.
- Researchers must consider their specific research question when selecting algorithms for MHC identification in administrative data.
Related Concept Videos
Theoretical Approaches to Psychological Disorder
Biological approach
The biological approach posits that internal, organic factors are the primary causes of such disorders. This perspective emphasizes brain structure and function, genetic predispositions, and neurotransmitter imbalances. For example, schizophrenia has been associated with both genetic...
Archival Research
Community Based Intervention
Foundations of Community Mental Health Programs
Central to the success of community-based interventions is the...
Data Collection I
Treatment Strategies for Psychological Disorders
Psychological therapies focus on modifying emotions, thoughts, and behaviors through talking, interpreting, listening, rewarding, challenging, and modeling. Clinical psychologists, counselors, and social workers commonly practice psychotherapy. Clinical...
Diagnostic and Statistical Manual of Mental Disorders (DSM)
