This work compared administrative data obtained from the Department of Veterans Affairs (VA) databases with structured chart review.
We set out to determine whether a decision tool using administrative data could discriminate acute from nonacute cases among the many patients seen for a low back pain (LBP)–related diagnosis.
Large health care systems' databases present an opportunity for conducting research and planning operations related to the management of highly burdensome conditions. An efficient method of identifying cases of acute LBP in these databases may be useful.
This was a retrospective review of all consecutive Iraq and/or Afghanistan Veterans seen in a VA primary care service during a 6-month period. Administrative data were extracted from VA databases. Patients with at least 1 encounter that was coded with at least 1 LBP-related ICD-9 code were included. Structured chart review of electronic medical record free text was the “gold standard” to determine acute LBP cases. Logistic regression models were used to assess the association of administrative data variables with chart review findings.
We obtained complete data on 354 patient encounters, of which 83 (23.4%) were designated acute upon chart review. No diagnostic code was more likely to be used in acute cases than nonacute. We identified an administrative data model of 18 variables that were significant and positively associated with an acute case (C-statistic = 0.819). A reduced model of 5 variables including a lumbar magnetic resonance imaging order, tramadol prescription, skeletal muscle relaxant prescription, physical therapy order, and addition of a new LBP-related ICD-9 code to the electronic medical record remained reasonable (C-statistic = 0.784).
Our results suggest that a decision model can identify acute from nonacute LBP cases in Veterans using readily available VA administrative data.
Level of Evidence: N/A
Supplemental Digital Content is Available in the Text.An automated method to discriminate acute from nonacute cases of low back pain in electronic medical records can be helpful. This pilot work compared administrative data with chart review and presents models that can identify acute cases with reasonable to strong likelihood in a subpopulation of Veterans.
*VA Connecticut Health Care System, West Haven, CT
†University of Bridgeport College of Chiropractic, Bridgeport, CT; and
‡Department of Internal Medicine
§Department of Emergency Medicine; and
¶Department of Psychiatry, Yale University School of Medicine, New Haven, CT.
Address correspondence and reprint requests to Anthony J. Lisi, DC, VA Connecticut Healthcare System, 950 Campbell Ave, 111D, West Haven, CT 06516; E-mail: Anthony.firstname.lastname@example.org
Acknowledgment date: November 25, 2013. Revision date: March 18, 2014. Acceptance date: March 19, 2014.
The manuscript submitted does not contain information about medical device(s)/drug(s).
Department of Veterans Affairs, Veterans Health Administration, Office of Research and Development, Health Services Research and Development Research Enhancement Award Program (REA 08-266) grant funds were received in support of this work.
Relevant financial activities outside the submitted work: grants.
The views expressed in this article are those of the authors and do not necessarily reflect the position or policy of the Department of Veterans Affairs or the United States government.