The use of the Department of Veterans Affairs (VA) centralized data sources to study the impact of post-traumatic stress disorder (PTSD) on returning veterans' health has become commonplace. Unfortunately, research is hampered by the lack of a well validated method to identify PTSD using the centralized data. Often a singular definition is used to answer different kinds of research questions.
The purpose of this study was to examine the validity of three algorithms tested against medical chart evidence for two different types of basic research questions: 1) which definitions best predict for a diagnosis of PTSD and, 2) which definitions best predict for the receipt of medication treatment for PTSD.
VA centralized data sources were used to identify a random sample (n=136) of veterans nationwide enrolled in VA care in 2006. Using the ICD-9-cm code of 309.81, three administrative data algorithms commonly used in research were used to define half of the study sample with PTSD and the other half without PTSD: 1) one or two outpatient encounters with PTSD, 2) three or more outpatient encounters with PTSD, and 3) one or more inpatient encounter with PTSD. Next, medical records over a three year period (FY 2006 - 2008) were abstracted for each subject by trained psychology research assistants focusing on two relevant clinical judgments: 1) was there evidence of medical documentation of a diagnosis of PTSD, and 2) was there evidence of prescription treatment for PTSD. The two independent raters used a 5 point Likert scale to determine whether or not the providers of record were documenting each of these relevant clinical situations. Sensitivity, specificity, positive and negative predictive (PPV and NPV) values were then calculated across each algorithm for each situation.
Medical records of n=573 Veterans were reviewed for 1) a diagnosis of PTSD; and 2) treatment for PTSD. The records were compared against the Veterans Health Administration (VHA) administrative data in order to determine the positive predictive value (PPV) and negative predictive value (NPV) of three commonly used administrative coding data (ACD) approaches. The PPV and NPV varied according to the ACD approach. Relative to a record review, the ACD approach of one or two PTSD coded outpatient encounters had a PPV was 78% and an NPV of 91%; whereas the PPV was 97% and the NPV was 98% for three or more PTSD codes. A similar review compared ACD approaches for PTSD pharmacotherapy. The ACD approach with one or two codes for PTSD had a PPV of 33% (NPV=93%), whereas three or more PTSD coded encounters improved the PPV to 85% (NPV=100%).
When using VHA administrative data, we recommend tailoring the identification strategy according to the research aims. An ACD approach identifying one or more PTSD outpatient encounters is likely to be sufficient for a diagnosis of PTSD. Assessments for PTSD associated pharmacotherapy require using an ACD approach that identifies veterans with the presence of three or more outpatient PTSD encounters.
- Abrams TE, Vaughan-Sarrazin M, Keane TM, Richardson K. Validating administrative records in post-traumatic stress disorder. International Journal of Methods in Psychiatric Research. 2016 Mar 1; 25(1):22-32.
- Abrams TE, Blevins A, Weg MW. Chronic obstructive lung disease and posttraumatic stress disorder: current perspectives. International journal of chronic obstructive pulmonary disease. 2015 Oct 15; 10:2219-33.
- Abrams TE, Lund BC, Alexander B, Bernardy NC, Friedman MJ. Geographical diffusion of prazosin across Veterans Health Administration: Examination of regional variation in daily dosing and quality indicators among veterans with posttraumatic stress disorder. Journal of rehabilitation research and development. 2015 Sep 1; 52(5):619-27.
- Abrams TE, Vaughan-Sarrazin M, Richardson K, Cram P, Rosenthal GE. Patterns of illness explaining the associations between posttraumatic stress disorder and the use of CT. Radiology. 2013 May 1; 267(2):470-8.
- Abrams T, Mengeling M, Torner J, Booth B, Sadler AG. Lifetime Sexual Assault and Emergency Room Use: A study examining the temporal relationship of single and repeat assaults on ER use in women Veterans. Poster session presented at: VA HSR&D / QUERI National Meeting; 2015 Jul 9; Philadelphia, PA.
- Mawanda F, McCoy K, Wallace R, Abrams T. Systemic Bacterial infections and the Risk of Dementia among U.S. Veterans: A Retrospective Cohort Study. Paper presented at: VA HSR&D / QUERI National Meeting; 2015 Jul 9; Philadelphia, PA.
- Abrams T, Friedman M. Initial Treatment Variation in Previously Untreated Veterans with PTSD. Poster session presented at: International Society for Traumatic Stress Studies Annual Symposium; 2014 Nov 5; Miami, FL.
- Abrams T, Lund BC, Alexander B, Bernardy N, Friedman M. Dissemination of Prazosin use for PTSD in the Veterans Health System. Poster session presented at: International Society for Traumatic Stress Studies Annual Symposium; 2014 Nov 4; Miami, FL.
- Abrams T, Richardson KK, Lund BC. Variations in Pharmacotherapy Associated with a Diagnosis of PTSD. Poster session presented at: VA HSR&D / QUERI National Meeting; 2012 Jul 16; National Harbor, MD.
Military and Environmental Exposures, Mental, Cognitive and Behavioral Disorders, Health Systems
Diagnosis, Prevention, Treatment - Comparative Effectiveness