This was a retrospective study using information imparted to the Military Health System Data Repository (MDR) for beneficiaries receiving care under the auspices of the TRICARE insurance program between 2006 and 2014. TRICARE insurance is provided to all uniformed military employees of the DoD and their dependents as well as retirees and those separated as a result of a medical condition associated with greater than 30% disability [3, 18, 22]. TRICARE beneficiaries may receive care in the fee-for-service civilian setting (purchased care) or through a DoD medical facility (direct care) [3, 18, 22].
All major medical centers in the DoD provide spine surgical services through salaried employees (uniformed and civilian) or contractors who are orthopaedic spine surgeons or neurosurgeons . TRICARE is not involved in the care of servicemembers in combat theaters  nor does the plan provide services through the Veterans Administration [18, 22]. TRICARE beneficiaries may receive care in the purchased care setting for a variety of reasons, including personal preference and location of their residence. There are currently no categories of patients that are systematically referred from the direct care to the purchased care setting. The MDR captures all information on inpatient and outpatient insurance claims handled through TRICARE irrespective of the environment of care or the location of service. TRICARE data have previously been utilized in other initiatives evaluating the delivery of surgical services in a number of different contexts [3, 18, 22]. Moreover, the broad demographic insured through TRICARE, which covers individuals of extremely varied sociodemographic, educational, professional, and vocational backgrounds, has been considered representative of the American population aged 18 to 64 years in prior studies [3, 16, 18, 22].
A query was performed using International Classification of Diseases, 9th Revision (ICD-9) codes to identify all TRICARE beneficiaries aged 18 to 64 years surgically treated for a disc herniation (722.10, 724.4), lumbar spinal stenosis (721.42, 721.91, 724.02), and spondylolisthesis (738.4, 756.11, 756.12). Patients aged 65 years and older as well as those otherwise eligible for Medicare services were excluded from review. Revision procedures were not included. Demographic data were extracted for identified individuals including age at the time of surgery, race, biologic sex, census region, number of medical comorbidities as defined by the modified Charlson index , sponsor military rank, type of surgery performed, and whether services were delivered in the direct or purchased care setting. Race was defined as white or nonwhite with nonwhite including individuals characterized as black, Asian, Hispanic, and other race. Sponsor military rank was classified as officer, enlisted junior (lowest four ranks in any branch of service), senior enlisted (noncommissioned officers), and other (warrant officers and cadets). In line with prior research using TRICARE data, sponsor military rank was considered a proxy for socioeconomic status in this analysis with individuals in the enlisted junior category maintained to be representative of those in lower socioeconomic circumstances [6, 16, 18, 22]. The type of surgical intervention was determined using ICD-9 procedure codes and defined as discectomy, decompression, posterolateral fusion, and interbody fusion using a previously published algorithm .
Within each lumbar disorder (such as disc herniation, spinal stenosis, spondylolisthesis), initial bivariate comparisons were made between the type of surgery (discectomy, decompression, posterolateral fusion, and interbody fusion) and the environment of care (purchased versus direct care) using the chi-square test. Adjustments were then made for case-mix (age, race, biologic sex, medical comorbidities, sponsor rank, and census region) using multinomial logistic regression. In regression testing, the direct care setting was used as the referent for the environment of care and interbody fusion was compared with the other types of surgical intervention. The results of regression tests were reported using an odds ratio (OR), 95% confidence interval (CI), and p value. Statistically significant results were maintained to be those that demonstrated p values < 0.05 and 95% CIs exclusive of 1.0 after multinomial logistic regression analysis. All statistical testing was performed using SAS Version 9.3 (Cary, NC, USA). Our institutional review board determined that this investigation was exempt from full review.
A total of 28,344 patients met all inclusion criteria, 21,290 (75%) treated in the purchased care setting and 7054 (25%) in direct care. The average age of the entire study cohort was 43 years (SD 12) and 63% (n = 18,005) of the population were men (Appendix 1 [Supplemental materials are available with the online version of CORR
®.]). Sixteen percent of the population had at least one medical comorbidity. The majority of patients (n = 17,046 of 28,344 [60%]) derived from the South. The fee-for-service setting treated patients who were older on average and a higher percentage of nonwhites, females, and those with a higher comorbidity count (Table 1). As a result of the size of our sample, there were differences in case-mix between the fee-for-service and salaried healthcare settings. Most patients (n = 18,900 of 28,344 [67%]) in both settings were treated for a diagnosis of disc herniation. Spondylolisthesis was the indication for surgery in 4705 of 28,344 instances (17%) with spinal stenosis diagnosed in 4739 of 28,344 (17%).