At present, in the midst of the opioid crisis in the United States, increased attention is being paid to the amount of opioid pain medications prescribed following encounters, as well as the most efficacious means of reducing prescription opioid dependence and misuse1 - 4. This follows 2 decades of exponential growth in the use of prescription opioids1 - 5. Current estimates maintain that expenditures associated with the use of opioids for spine-associated disorders have increased by a factor of 6 over the last 10 years6, with as many as 55% of spine patients exposed to opioid medications prior to surgery4.
Persistent opioid use has been shown to not only contribute to tolerance, hyperalgesia, and systemic effects1 - 10; it may also lead to dependence, misuse, and, in some instances, a transition to illegal narcotics1 - 3 , 6 , 10. It remains unclear whether patients with a history of sustained opioid use prior to spine surgery cease using these medications following the successful treatment of the spinal condition. Prior studies intending to address these issues have been confounded by limited sample size9, single-center design4 , 9 , 11, lack of clinical heterogeneity4 , 6 , 8 , 9 , 11, and limited postoperative surveillance4 , 6 , 8 , 9 , 11.
In this context, we sought to utilize claims data from TRICARE insurance to evaluate sustained prescription opioid use in a population of patients previously exposed to prescription opioid medications who also underwent 1 of 4 common spine surgical procedures. We also planned to examine the impact that the duration of preoperative prescription opioid use had on the discontinuation of these medications following a surgical procedure. We hypothesized that sustained preoperative prescription opioid use, as defined by a period of continuous use of at least 90 days, would be associated with continued opioid use at ≥90 days after the surgical procedure. TRICARE insurance data have been utilized in a number of prior studies investigating prescription opioid dependence and postoperative outcomes in a variety of clinical contexts3 , 5 , 7 , 12 , 13. Previous work has also maintained that the population insured through TRICARE is sociodemographically representative of the civilian American population who are 18 to 64 years of age3 , 5 , 7 , 12 - 16.
Materials and Methods
The data used for this investigation was derived from TRICARE claims maintained by the Military Health System Data Repository (MDR) for the time period 2006 to 2014. As described in previous investigations, TRICARE is the insurance program of the U.S. Department of Defense and currently covers approximately 9 million beneficiaries, 80% of whom are civilians3 , 5 , 7 , 12 , 13 , 15. These include retired military personnel, medically separated individuals with a disability, and dependents. The MDR captures all medical services paid through TRICARE insurance, including prescription medications, irrespective of the site of service (e.g., military or civilian health-care facility) or location3 , 5 , 7 , 12. TRICARE insurance is not used in the delivery of medical care in combat theaters or through Veterans Health Administration hospitals. These data have previously been successfully employed in the study of surgical care delivery13 , 16 and sustained opioid use3 , 5 , 7 , 12.
We queried the MDR using International Classification of Diseases, Ninth Revision (ICD-9) procedure codes to identify all adult patients (18 to 64 years of age) who underwent 1 of 4 common spine surgical procedures (lumbar interbody arthrodesis, lumbar posterolateral arthrodesis, lumbar discectomy, and lumbar decompression or laminectomy) in the period 2007 to 2013 (to allow for the 1-year observation period on either end of the procedure). Eligible patients were required to have 1 year of claims data available before and after the surgical intervention. MDR drug claims were then surveyed to identify those who received at least 1 prescription for any medication defined by the U.S. Drug Enforcement Agency as an opioid combination, opioid partial agonist, and/or opioid agonist category II (high abuse risk) or III (moderate abuse risk). Patients who had not received any prescription opioid medications prior to the surgical procedure were excluded from further review.
The remaining patients had their medical records abstracted and relevant sociodemographic, clinical, and care-related characteristics recorded. Sociodemographic data included age at the time of the surgical procedure, biologic sex, marital status (married or single), race (white, African-American, Hispanic, Asian, other, or unknown), beneficiary category, and military rank. Military rank was used as a proxy for socioeconomic status in this investigation, with junior enlisted rank considered indicative of low socioeconomic status. This approach is based on findings in a variety of prior studies that demonstrate that junior enlisted personnel and their families disproportionately derive from low socioeconomic backgrounds, receive the lowest pay in the military system, and are prone to health-related behaviors and conditions similar to those of their civilian counterparts without any military affiliation13 , 16 - 18.
Additional clinical information included Charlson Comorbidity Index, based on the Deyo modification19, as well as the presence of ICD-9 codes associated with certain depressive disorders (296.20-296.26, 296.30-296.36, 296.82, 311) or generalized anxiety disorder (300.02) based on the correlation of these conditions with sustained prescription opioid use in other work5 , 12. Preoperative diagnosis, classified according to the ICD-9 diagnosis code, was recorded and defined as deformity, degenerative disc conditions, revision surgical procedure, spinal fracture, spinal stenosis, spondylolisthesis, and other spinal conditions. The development of postoperative complications, defined using a previously published algorithm13, and postoperative length of stay were also documented. Health-care-related characteristics included the environment of care in which the surgical procedure was performed (e.g., civilian or military facility) and the census region of the patient’s residence. Opioid prescription characteristics included the number of pills prescribed, the duration of the prescription, and the number of times additional opioid medications were reordered. Time of discontinuation was determined as the date on which the last pill on a final prescription would have been used given that a new prescription was not issued.
The primary outcome in this study was the duration of postoperative prescription opioid use up to 1 year following spine surgery. (We will be using the term “opioid” to refer to legally prescribed opioids and thus exclude any illegal use in the patients in this study.) The primary predictor variable was the duration of preoperative opioid use. On the basis of prior studies1 , 2 , 5 , 6, we defined sustained prescription opioid use as the continuous use of opioids for any period exceeding 90 days without an interruption in prescription refills. This takes advantage of the fact that opioid medications of the type under consideration here cannot be issued for a period of >30 days without a refill prescription. In line with prior research5 , 6, the duration of preoperative opioid use was defined as follows: acute exposure (first prescription for opioids issued within 30 days of the surgical procedure, exposed (non-sustained use; receipt of opioids within a 12-month period prior to the date of the surgical procedure, but no continuous use of the medication as defined above), intermediate sustained use (continuous use of prescription opioids for <6 months prior to the surgical intervention), and chronic sustained use (continuous use of prescription opioids for ≥6 months prior to the surgical intervention).
Initial statistical comparison between the groups was performed with a chi-square test. We then used Cox proportional-hazard regression analysis, evaluating the effect of the duration of preoperative opioid use on postoperative consumption and adjusting for the sociodemographic, clinical, and health-care-related characteristics enumerated above. These factors have been shown to influence opioid dependence in prior work5 - 8 , 12. In these models, patients with acute opioid exposure were used as the reference. Adjustment for surgical intervention was performed with discectomy or decompression compared with interventions including arthrodesis, and adjustment for preoperative diagnosis utilized patients with a diagnosis of degenerative disc disorders as the reference. Patients with missing data (<1% for region and rank) were excluded. Extended Cox models were employed to evaluate for the proportionality assumption necessary to Cox regression analysis20. In a sensitivity analysis, consideration was limited to those patients who underwent arthrodesis. Significance, set at p < 0.05, was determined to be present for those variables that demonstrated hazard ratios (HRs) and 95% confidence intervals (CIs) independent of 1.0 following adjusted testing.
In the time period of the study, there were 27,031 patients who met the full inclusion criteria. The majority of patients were male (61.4%), white (65.7%), and 45 to 64 years of age (55.4%) (Table I). Only a minority of patients (n = 9,936 [36.8%]) were active-duty service members, and the remainder were largely retirees or dependents. More than three-quarters of patients underwent the surgical procedure at a civilian hospital, with 64.4% residing in the U.S. South. Approximately 10% of the population was identified as having ≥1 of the depressive disorders surveyed and 1% carried a diagnosis of generalized anxiety disorder.
At the time of the surgical procedure, the majority of patients (n = 16,287 [60%]) had been exposed to opioid medications but were not actively using them. A further 9,183 patients (34%) were acutely using opioids at the time of the surgical procedure. Three percent (n = 907) of the cohort were classified as chronic sustained opioid users, and 2% were categorized as intermediate sustained users. Nearly half (49%) of all patients underwent a discectomy, 29% underwent a lumbar interbody arthrodesis, 13% underwent a standalone decompression or laminectomy, and 9% underwent a posterolateral lumbar arthrodesis. Three percent of patients (n = 754) were found to have sustained ≥1 postoperative complication. Following surgical intervention, 18,133 patients (67.1%) had discontinued opioid use by 30 days, and 23,350 (86.4%) had ceased opioid use by 90 days. Overall, 2,379 patients (8.8%) continued to use opioid medications at 6 months following the surgical procedure.
There were substantial baseline differences between the groups stratified by duration of preoperative opioid use (Table II), supporting our inclusion of these covariates in the adjusted Cox analyses. Following Cox proportional-hazard regression analysis (Fig. 1) that adjusted for all confounders, chronic sustained users of opioids were found to be 74% less likely to discontinue use following the surgical procedure compared with those acutely exposed to opioids before surgical intervention (HR, 0.26 [95% CI, 0.24 to 0.28]; p < 0.001) (Table III). At the same time, intermediate sustained users were 65% less likely to discontinue use following the surgical procedure (HR, 0.35 [95% CI, 0.32 to 0.38]; p < 0.001), and those exposed to opioids but not actively using them at the time of the surgical procedure were 29% less likely to discontinue use postoperatively (HR, 0.71 [95% CI, 0.69 to 0.72]; p < 0.001).
There were also a number of important clinical and sociodemographic factors associated with a reduced likelihood of opioid discontinuation following the surgical procedure in this population (Table IV). These included undergoing an arthrodesis-based procedure (HR, 0.76 [95% CI, 0.74 to 0.79]; p < 0.001), preoperative depressive disorder (HR, 0.79 [95% CI, 0.75 to 0.82]; p < 0.001) or generalized anxiety disorder (HR, 0.81 [95% CI, 0.72 to 0.91]; p < 0.001), and a preoperative diagnosis of a spinal fracture (HR, 0.80 [95% CI, 0.70 to 0.92]; p = 0.002). Increased hospital length of stay was also correlated with a reduced likelihood of discontinuing opioid use following the surgical procedure (HR, 0.96 [95% CI, 0.95 to 0.97]; p < 0.001). Junior enlisted rank, our proxy measure for socioeconomic status, demonstrated a 28% reduction in the likelihood of postoperative opioid discontinuation (HR, 0.72 [95% CI, 0.68 to 0.77]; p < 0.001). These results were relatively well preserved in the sensitivity test that limited consideration to those undergoing arthrodesis, apart from a preoperative fracture diagnosis, which lost significance (Table V).
Over the course of the first decade of the twenty-first century, an exponential increase in the use of opioid medications transpired within the American population as a whole1 - 3 , 12, particularly among individuals treated for spinal disorders4 , 6 , 8 , 11. Current estimates maintain that as many as 50% of patients undergoing spine surgery may be consuming opioids at the time of the surgical procedures4, and 20% of this population may actually be opioid-dependent11. Although the adverse impact of chronic opioid use on postoperative morbidity and functional outcome is well accepted, the potential for opioid consumption in the setting of spine surgical pathology to result in continued use more than 90 days after the surgical procedure remains underexplored. We endeavored to utilize recent claims data from the TRICARE insurance program on >27,000 patients undergoing spine surgery to comprehensively investigate these issues.
Our study was able to leverage the advantages of TRICARE claims and its covered population, including a large patient sample derived from centers across the United States with longitudinal coverage over the course of multiple years without interruption3 , 5 , 7 , 12 , 13 , 15 , 16. We were able to track prescription drug use among beneficiaries, irrespective of the environment of care (e.g., civilian or military) in which treatment is administered or the location where prescriptions are filled, inclusive of centers outside the continental United States3 , 5 , 7 , 12. This allowed for accurate surveillance of opioid use in the time period leading up to spine surgery and for 1 year following the procedure. Moreover, the covered population has been reported to be representative of the American demographic age of 18 to 64 years and possesses diverse ethnocultural, sociodemographic, vocational, occupational, and functional characteristics as a whole3 , 5 , 7 , 12 - 18. In the current study, the demographic and medical characteristics of the population, as well as the postoperative complication rate and types of procedures performed, lie within the range of prior large series studying patients undergoing lumbar spine surgery4 - 6 , 11 , 21 , 22. In the time period leading up to the surgical procedure, 60% of the cohort had been exposed to opioids but were not actively using the medications, and 5% were considered sustained users of some degree. Following surgical intervention, close to 1 in 10 individuals continued use of opioids at the 6-month postoperative time point. This can be contrasted with a previous investigation with regard to opioid-naive patients, in whom only 0.2% of the population were found to continue use at 6 months after spine surgery5.
A dose-response effect was also noted for preoperative opioid use on postoperative cessation. Patients defined as chronic users preoperatively (continuous use for a period of ≥6 months prior to the surgical procedure) had the lowest likelihood of postoperative discontinuation, followed by individuals who were sustained users for a period of 3 to 6 months. Interestingly, even those patients exposed to opioid medications in the preoperative period, but not actively using them at the time of the surgical procedure, had a significantly greater risk of continuing use postoperatively. Additional factors identified as predictors of prolonged postoperative opioid use included our proxy measure for socioeconomic status, the use of spinal arthrodesis, and the presence of certain mental health disorders. Many of these other factors have previously been maintained to influence prolonged opioid use following spine surgery4 - 6 , 8 , 9 , 11, a finding that reinforces the external validity of our results. For example, in a series of more than 550 patients undergoing various spine procedures, Armaghani et al. reported that depression and anxiety were associated with an increased risk of persistent opioid use at 1 year following the intervention4. Similar findings were documented by Connolly et al. in a private insurance database of patients treated with lumbar spine fusion6. Sustained preoperative use might be indicative of underdiagnosed and undertreated psychosocial issues, such as unrecognized psychological disorders, pain catastrophizing, and ineffective coping strategies4 - 6 , 11. Taken as a whole, these results highlight the importance of surgeons providing more comprehensive care, including recognition of risk factors in the biopsychosocial paradigm preoperatively and in the perioperative period. Consensus strategies on the part of providers, medical centers, and societies are needed to address the optimal means of managing preoperative and postoperative opioid use in patients undergoing spine surgery.
We acknowledge the presence of important limitations associated with this work, including the fact that the data were derived from insurance claims. As a result, we possess limited clinical granularity in certain respects, such as factors considered in the decision for the surgical procedure, the extent of some surgical interventions, the type of instrumentation, and provider experience. In addition, we cannot be certain of the rationale behind the administration of opioids or decision-making associated with their use, which imparts the potential for selection and indication bias. In this context, we must assume that patients used the pain medications as prescribed by their surgeon and cannot account for inappropriate diversion, abuse, or attainment of illegal narcotics3 , 5 , 7 , 12. Likewise, given the design of this investigation, we were unable to comprehensively account for the use of non-opioid pain medications, particularly if over-the-counter formulations were utilized3 , 7. Patients with missing data had to be excluded from regression testing, but, given the large size of the sample and the small number of patients with missing data, this was unlikely to adversely impact our determinations. Although the population under study has been shown to be representative of the broad American demographic age of 18 to 64 years, this also meant that the findings may have not been translatable to select populations undergoing spine surgery such as the elderly, patients receiving Workers’ Compensation, and individuals with a history of substance abuse. Lastly, there may be concerns with regard to the military affiliation of the population insured through TRICARE. We do not believe that this was a substantial limitation based on the findings of prior efforts that have used those insured through TRICARE as a model for the civilian population3 , 5 , 7 , 12 - 18. In addition, the majority of patients in this study were civilian dependents or retirees, and 77% of care was received through the civilian sector.
In conclusion, we believe that our findings have immediate clinical impact of interest to patients, spine surgeons, and other health-care practitioners. Foremost, the results point to the fact that the overwhelming majority of patients who are using prescription opioids prior to a surgical procedure discontinue these medications subsequent to spine surgical intervention. Among those who continue use ≥90 days postoperatively, the duration of preoperative use appears to be the most important risk factor. In addition, patients may be best served by limiting prescription opioid use, if necessary, to the immediate preoperative period. In this context, we appreciate the need for further research as well as consensus statements that define the appropriate dosage and duration of pain medications prescribed by surgeons and other practitioners treating spinal disorders both before and after the surgical procedure. Given the near ubiquity of opioid exposure among patients who request spine surgery, those who are already sustained prescription opioid users at the time of the surgical procedure and who also possess any of the other risk factors for sustained postoperative use identified here should be counseled with regard to these facts prior to any intervention. Patients with these characteristics may benefit from enhanced surveillance in the postoperative period, and consideration can be given to multimodal pain control interventions to accelerate discontinuation of opioid medications as expeditiously as possible. Alternative strategies in the setting of elective spine surgery could consist of preoperative weaning of opioid medications to cessation before proceeding with any discretionary intervention as well as ensuring appropriate screening and treatment of psychosocial factors that might influence the disease process and recovery.
Investigation performed at the Center for Surgery and Public Health, Brigham and Women’s Hospital and Harvard Medical School, Boston, Massachusetts
Disclosure: Four authors of this study (A.J.S., W.J., T.K., and A.H.H.) received, through their institution, a grant from the Henry M. Jackson Foundation of the U.S. Department of Defense. On the Disclosure of Potential Conflicts of Interest forms, which are provided with the online version of the article, one or more of the authors checked “yes” to indicate that the author had a relevant financial relationship in the biomedical arena outside the submitted work and “yes” to indicate that the author had other relationships or activities that could be perceived to influence, or have the potential to influence, what was written in this work (http://links.lww.com/JBJS/E651).
1. Florence CS, Zhou C, Luo F, Xu L. The economic burden of prescription opioid overdose, abuse, and dependence in the United States, 2013. Med Care. 2016 ;54(10):901–6.
2. Brummett CM, Waljee JF, Goesling J, Moser S, Lin P, Englesbe MJ, Bohnert ASB, Kheterpal S, Nallamothu BK. New persistent opioid use after minor and major surgical procedures in US adults. JAMA Surg. 2017 ;152(6):e170504. Epub 2017 Jun 21.
3. Chaudhary MA, Schoenfeld AJ, Harlow AF, Ranjit A, Scully R, Chowdhury R, Sharma M, Nitzschke S, Koehlmoos T, Haider AH. Incidence and predictors of opioid prescription at discharge after traumatic injury. JAMA Surg. 2017 ;152(10):930–6.
4. Armaghani SJ, Lee DS, Bible JE, Archer KR, Shau DN, Kay H, Zhang C, McGirt MJ, Devin CJ. Preoperative opioid use and its association with perioperative opioid demand and postoperative opioid independence in patients undergoing spine surgery. Spine (Phila Pa 1976). 2014 ;39(25):E1524–30.
5. Schoenfeld AJ, Nwosu K, Jiang W, Yau AL, Chaudhary MA, Scully RE, Koehlmoos T, Kang JD, Haider AH. Risk factors for prolonged opioid use following spine surgery, and the association with surgical intensity, among opioid-naive patients. J Bone Joint Surg Am. 2017 ;99(15):1247–52.
6. Connolly J 3rd, Javed Z, Raji MA, Chan W, Kuo YF, Baillargeon J. Predictors of long-term opioid use following lumbar fusion surgery. Spine (Phila Pa 1976). 2017 ;42(18):1405–11. Epub ahead of print.
7. Chaudhary MA, Scully R, Jiang W, Chowdhury R, Zogg CK, Sharma M, Ranjit A, Koehlmoos T, Haider AH, Schoenfeld AJ. Patterns of use and factors associated with early discontinuation of opioids following major trauma. Am J Surg. 2017 ;214(5):792–7. Epub 2017 Jun 7.
8. Faour M, Anderson JT, Haas AR, Percy R, Woods ST, Ahn UM, Ahn NU. Preoperative opioid use: a risk factor for poor return to work status after single-level cervical fusion for radiculopathy in a Workers’ Compensation setting. Clin Spine Surg. 2017 . [Epub ahead of print].
9. Lawrence JTR, London N, Bohlman HH, Chin KR. Preoperative narcotic use as a predictor of clinical outcome: results following anterior cervical arthrodesis. Spine (Phila Pa 1976). 2008 ;33(19):2074–8.
10. Menendez ME, Ring D, Bateman BT. Preoperative opioid misuse is associated with increased morbidity and mortality after elective orthopaedic surgery. Clin Orthop Relat Res. 2015 ;473(7):2402–12. Epub 2015 Feb 19.
11. Armaghani SJ, Lee DS, Bible JE, Archer KR, Shau DN, Kay H, Zhang C, McGirt MJ, Devin CJ. Preoperative narcotic use and its relation to depression and anxiety in patients undergoing spine surgery. Spine (Phila Pa 1976). 2013 ;38(25):2196–200.
12. Schoenfeld AJ, Jiang W, Chaudhary MA, Scully RE, Koehlmoos T, Haider AH. Sustained prescription opioid use among previously opioid-naive patients insured through TRICARE (2006-2014). JAMA Surg. 2017 ;152(12):1175–6.
13. Schoenfeld AJ, Jiang W, Harris MB, Cooper Z, Koehlmoos T, Learn PA, Weissman JS, Haider AH. Association between race and postoperative outcomes in a universally insured population versus patients in the state of California. Ann Surg. 2017 ;266(2):267–73.
14. Schoenfeld AJ, Goodman GP, Burks R, Black MA, Nelson JH, Belmont PJ Jr. The influence of musculoskeletal conditions, behavioral health diagnoses, and demographic factors on injury-related outcome in a high-demand population. J Bone Joint Surg Am. 2014 ;96(13):e106. Epub 2014 Jul 2.
15. Gimbel RW, Pangaro L, Barbour G. America’s “undiscovered” laboratory for health services research. Med Care. 2010 ;48(8):751–6.
16. Pierre-Louis BJ, Moore AD, Hamilton JB. The military health care system may have the potential to prevent health care disparities. J Racial Ethn Health Disparities. 2015 ;2(3):280–9. Epub 2014 Dec 6.
17. Clark JB, Holt VL, Miser F. Unintended pregnancy among female soldiers presenting for prenatal care at Madigan Army Medical Center. Mil Med. 1998 ;163(7):444–8.
18. Masel J, Deiss RG, Wang X, Sanchez JL, Ganesan A, Macalino GE, Gaydos JC, Kortepeter MG, Agan BK. Seroprevalence and seroincidence of herpes simplex virus (2006-2010), syphilis (2006-2010), and vaccine-preventable human papillomavirus subtypes (2000-2010) among US military personnel. Sex Transm Dis. 2015 ;42(5):253–8.
19. Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol. 1992 ;45(6):613–9.
20. Valsecchi MG, Silvestri D, Sasieni P. Evaluation of long-term survival: use of diagnostics and robust estimators with Cox’s proportional hazards model. Stat Med. 1996 ;15(24):2763–80.
21. Schoenfeld AJ, Ochoa LM, Bader JO, Belmont PJ Jr. Risk factors for immediate postoperative complications and mortality following spine surgery: a study of 3,475 patients from the National Surgical Quality Improvement Program. J Bone Joint Surg Am. 2011 ;93(17):1577–82.
22. Martin CT, Gao Y, Duchman KR, Pugely AJ. The impact of current smoking and smoking cessation on short-term morbidity risk after lumbar spine surgery. Spine (Phila Pa 1976). 2016 ;41(7):577–84.
Supplemental Digital Content
Copyright 2018 by The Journal of Bone and Joint Surgery, Incorporated