The Relationship Between Repeated Epidural Steroid Injections and Subsequent Opioid Use and Lumbar Surgery

Methods 

We used national VA administrative data during a 2-year period from October 2001 through September 2003 to examine the use of ESIs and their relationship to lumbar surgery and opioid use. Clinical encounters in the VA's health care system are captured in this administrative dataset in a similar manner to commercial claims data. These national data are stored in a central location and made available for research purposes. We abstracted clinical data relating to all ESIs performed within the VA health care system during the study period. The dataset included the Current Procedural Technology (CPT) procedure codes, the International Classification of Diseases–9th Revision–Clinical Modifications (ICD-9-CM) diagnosis codes, dates of service, patient age, sex, ethnicity, use of fluoroscopy, lumbar surgeries, and use of opioids in the 6-month period before and after ESI administration. Patients undergoing ESIs during the study period with CPT codes 62311 (interlaminar or caudal) and 64483 (transforaminal) for specific LBP diagnoses were included in this analysis. In addition, the injection codes must have been associated with 1 of 5 diagnostic categories of LBP ICD-9 codes (herniated disk, radiculopathy, spinal stenosis, degenerative disease, other LBP). These categories were chosen to be consistent with studies on LBP and lumbar surgery rates and have been previously described and validated.30, 31 The other LBP category included ICD-9 codes for LBP, lumbago, and lumbosacral sprain. The scope of this analysis was limited to lumbosacral injections. We excluded cervical and thoracic steroid injections for 2 reasons. First, cervical spine disorders differ clinically from lumbar spine disorders and may be the result of different disease processes. Second, cervical and thoracic ESIs are much less common, representing less than 10% of all ESIs performed.1, 32 Lumbar surgeries were identified by using Healthcare Common Procedure Coding System codes for 6 months before and after ESIs, and opioid use was determined by examining VA pharmacy data for 6 months before and after an ESI. Opioid use was classified into use or nonuse such that if a patient had any prescription for opioids filled during the 6-month period, he/she was considered to use opioids. For those patients who underwent multiple ESIs during the study period, prior lumbar surgery and opioid use were calculated from the date of the first injection performed, whereas subsequent surgery and opioid use were calculated based on the date of the last injection performed.

Statistical analyses were performed by using SPSS software.a Demographics were analyzed descriptively. The Pearson chi-square test was used for dichotomous, nominal outcomes measures (ie, surgery and opioid use) that met the assumptions for using this test (ie, large enough sample size, categorical data). ESI use (exposure) was divided into 2 groups (>3, ≤3). A P value of .05 (2-tailed for all analyses) was considered statistically significant. Odds ratios (ORs) with 95% confidence intervals (CIs) were reported for the outcome measure, lumbar surgery after ESI.

Results 

Opioid Use 

The majority of all patients were using opioids in the 6 months before their ESI (64%) as were the majority after their ESI (67%). Therefore, overall opioid use did not decrease in this patient population in the 6 months after an ESI. Of the patients not on opioids before the ESI, 38% were prescribed opioids afterward, whereas only 16% of people on opioids before the ESI stopped using opioids afterward. The majority of patients on opioids before their ESI continued to use opioids after ESI (84%). Patients on opioids before ESI were more likely than patients not on opioids before ESI to take them within 6 months after ESI (84% vs 38%, P<.001). Conversely, patients not on opioids before were more likely not to be taking them after ESI (62% vs 16%, P<.001). Diagnosis, age, and sex were not correlated with prior or subsequent opioid use.

Patients who received multiple injections during the study period (ie, >3) were less likely to stop taking opioids than patients receiving fewer injections (8% vs 11%, P<.001) and were more likely to start taking opioids after treatment with ESIs (19% vs 13%, P<.001) (Table 1, Table 2). These findings persisted when patients with a history of surgery either before or after ESI were removed from the sample. A smaller percentage of patients who received greater than 3 injections were taking opioids before ESI treatment than those patients who received fewer injections (55% vs 64%, P<.001) (see table 2).

Table 1. The Relationship Between Repeated ESIs and Opioid Use

	Number of ESIs	Started Opioids After ESI (n=1840)	No Change in Opioid Use (n=10,465)	Stopped Opioids After ESI (n=1436)	Total No. of Patients Receiving ESI
	≤3 ESIs	1607(13)⁎	9571(76)⁎	1344(11)⁎	12,522(100)
	>3 ESIs	233(19)	894(73)	92(8)	1219(100)

NOTE. Values are n (%).

⁎ P<.001.

Table 2. VA Patients Receiving ESIs and Opioid Use and Surgery

	ESIs and Opioid Use	Total No. Receiving ESIs	No. Receiving ≤3 ESIs	No. Receiving >3 ESIs
	Total	13,741	12,522	1219
	Opioid use before	8728 (64)	8051 (64)	677 (55)
	No use after	1436	1344	92
	No change	7292	6707	585
	No opioid use before	5013 (36)	4471 (36)	542 (45)
	No change	3173	2864	309
	Use after	1840	1607	233
	Surgery before	1791 (13)	1626 (13)	165 (14)
	No surgery after	1577	1431	146
	Surgery after	214	195	19
	No surgery before	11,950 (87)	10,896 (87)	1054 (86)
	No surgery after	11,175	10,213	962
	Surgery after	775	683	92

NOTE. Values are n (%) for the main subgroups and n otherwise.

Lumbar Surgery 

Thirteen percent (n=1791) of patients had at least 1 surgery within 6 months before ESIs, and 7% (n=989) of patients had at least 1 surgery within 6 months after ESIs. Of the patients who had a history of surgery before the ESIs, 960 (54%) underwent a fusion, 92 (5%) underwent a laminectomy with diskectomy, 280 (16%) had a diskectomy, and 459 (26%) had a laminectomy alone. Patients with a history of surgery in the 6 months before ESIs were twice as likely to undergo a subsequent surgery after their ESIs (12% vs 6%; OR=1.75; 95% CI, 1.54–1.99; P<.001). Controlling for the history of prior surgery, patients who received 4 or more injections during the study period were also more likely to undergo lumbar surgery after receiving the ESIs (8.7% vs 6.3%, P=.003) (see Table 2, Table 3). Patients with diagnoses of radiculopathy were slightly more likely than patients without radiculopathy to undergo lumbar surgery within 6 months after their injection (8.3% vs 6.9%; OR=1.16; 95% CI, 1.04–1.293; P=.008). Patients with radiculopathy were less likely to undergo a fusion procedure as compared with patients without radiculopathy (50% vs 54%, P=.012). Age and sex did not correlate with prior or subsequent lumbar surgery.

Table 3. The Effect of Frequency of ESIs on Surgery Rates After ESI

	No. of ESIs	Surgery After ESI		Total
		Yes	No
	≤3	878(7.0)⁎	11,644(93.0)	12,522(100.0)
	>3	111(9.1)⁎	1108(90.9)	1219(100.0)
	Total	989(7.2)	12,752(92.8)	13,741(100.0)

NOTE. Values are n (%).

⁎ P<.001.

Discussion 

In this national study of VA patients, we found that the majority receiving ESIs were also using opioid medications. This finding is consistent with a prior study3 of opioid use among patients in a VA orthopedic spine clinic (66% use). There has been growing concern regarding chronic opioid use because of risks associated with dependence, side effects, and addiction. Opioid use is frequently considered 1 outcome measure of interventions for a variety of pain syndromes. Although in this study we did not examine the indications for opioid prescription and concomitant pain syndromes in addition to LBP, it is reasonable to hypothesize that opioid use would decrease after ESI for LBP. However, contrary to our hypothesis, opioid use did not decrease in the 6 months after an ESI in this population, and over a third of patients who were not taking opioids before the ESI received opioids afterward. In addition, we found that patients who underwent more frequent injections were more likely than patients receiving fewer injections to start opioids if they were not taking them before their first ESI and were less likely to stop taking opioids if they were taking them before their first ESI. This finding may reflect that patients with more severe pathology or pain are more likely to undergo multiple ESIs and therefore may be more likely to use opioid medications. However, patients undergoing repeated ESIs in this study were no more likely to have been taking opioids before receiving the injections than those only undergoing 1 to 2 injections. In fact, patients receiving fewer injections were more likely to be taking opioids before ESI treatment (64%) than those receiving a repeat ESI (56%). In addition, diagnoses did not vary between those receiving 4 or more injections and those receiving fewer injections (ie, each group had similar numbers of patients with documented radiculopathy and spinal stenosis).

In this population, ESIs were primarily used for spinal stenosis (73%) despite the lack of evidence to support their use for this diagnosis. The best evidence regarding the effectiveness of ESIs is for symptoms of sciatica or radiculopathy. However, during this study period, nearly half of the veterans receiving injections had spinal stenosis without documented radiculopathy. We also found that patients over the age of 60 were more likely to receive repeat injections (>3 in the 2-year study period) and were more likely to be diagnosed with spinal stenosis. Patients who received more than 3 injections during the study period were also no more likely to stop using opioids and were more likely to undergo subsequent lumbar surgery. These findings suggest that ESIs in this VA population are not being substituted for lumbar surgery and are not decreasing opioid use.

There are several potential reasons for these findings. First, it must be acknowledged that this study reflects outcomes in actual clinical practice rather than in the controlled environment of a randomized clinical trial. Therefore, these findings suggest that the use of ESIs in an uncontrolled environment (ie, outside of the realm of a controlled trial with very specific inclusion and exclusion criteria and standardized techniques) is not associated with decreasing opioid use and is not substituting for lumbar surgery. These outcomes certainly may be different than those in a controlled setting with standardized patient selection criteria and injection techniques. These findings suggest that it may be important to have more standardized criteria for patient selection and/or injection technique to improve outcomes in actual clinical practice.

Second, chronic pain syndromes may be more prevalent within the VA population and therefore patients may be using opioids for reasons other than LBP or sciatica. One would not expect to see changes in opioid use after ESI for those patients who take opioids for other pain conditions. We did not have access to specific information regarding comorbidities and other pain complaints or diagnoses for which opioids may have been used. We also did not have access to data regarding the indication for the opioid prescription. Pain relief from the ESI in this study may also not have been substantial enough for patients to discontinue use of opioids or to prevent surgery. We did not include self-reported pain as an outcome measure in this study and chose to use opioid use as 1 measure of pain.

Additionally, the reasons for choosing to undergo an ESI may be different in the VA health care system than in others. For example, patients may undergo ESI as routine care preceding an anticipated surgery to better delineate the pain generator or to provide temporary pain relief in anticipation of surgery. The VA health care system may be different from the private sector in that there is less direct financial influence on procedures performed and less constraint in the use of procedures because of insurance coverage.

Currently, there are no evidence-based guidelines regarding the appropriate number of injections, the optimal frequency and timing of injections, or the appropriate use for the injections in patients considering lumbar surgery. Our data suggest that patients who receive repeat injections are more likely to use opioids and are more likely to undergo surgery, calling into question the rationale for using repeat injections, particularly in this population of patients with primarily spinal stenosis.

Further research is needed to better understand this relationship between the use of repeated ESIs and the effect on subsequent surgical rates and opioid use. In addition, more research is needed to determine if repeated ESIs are of benefit in the long-term management of spinal stenosis for those patients who are not candidates for surgery, particularly in terms of function and quality of life measures.

Conclusions 

Three quarters of all lumbar ESIs were performed for spinal stenosis, and nearly half were for spinal stenosis without radiculopathy despite the equivocal data to support their use in this clinical situation. Patients with spinal stenosis were more likely to receive repeated ESIs (≥4). Patients receiving repeated ESIs were more likely to start opioid use after ESI than patients receiving fewer ESIs and were more likely to undergo subsequent lumbar surgery than patients receiving fewer ESIs. These findings are concerning given the equivocal data to support ESI use for lumbar spinal stenosis and our data that suggest ESIs are not reducing opioid use in this population of veterans.