FactFinders for patient safety: Do epidural steroid injections increase the risk of fracture or lumbar surgical site infection?

FactFinder for patient safety: Increased fracture risk associated with epidural steroid injections (Updated from 2015)

Ben Marshall, DO; Christin Tiegs-Heiden, MD; Kunj Amin, MD; Alison Stout, DO; Clark C. Smith, MD, MPH; Benoy Benny, MD; David O'Brien, MD; Wade M. King, MMed; Andrew J. Engel, MD; and David Levi, MD

On behalf of the International Pain and Spine Intervention Society's Patient Safety Committee.

Acknowledgment:

We wish to acknowledge Nikolai Bogduk, MD; Jeffrey Laseter, MD; and Alan Swearingen, MD, who were not involved in authoring this update but contributed significantly to the original FactFinder (2015).

Myth: Epidural steroid injections increase a patient's risk of fracture.

Fact: There is conflicting and somewhat limited evidence in the literature. While studies suggest some transient relative increased risk of osteoporotic fractures, the absolute risk is likely highly dependent on baseline risk and frequency of steroid administration.

Epidural steroid injection (ESI) is a common non-operative treatment option to help patients with radiculopathy who have failed to improve with conservative care [1]. As such, it is important to be aware of any associated adverse systemic effects of steroid use, including the increased risk of fractures. This is particularly important because a fracture in at-risk populations can nearly triple an individual's mortality risk in the subsequent year, with a 1.2–3.5x estimated mortality risk in patients over 80 [2].

Exogenous oral corticosteroids are well-established as a common cause of secondary osteoporosis. A large study of oral corticosteroid users showed a dose-dependent increased relative risk of fracture in the hip of 1.61 [95 % confidence interval (CI): 1.47–1.76] and vertebral body of 2.60 (95 % CI: 2.31–2.92). Notably, this elevated fracture risk tends to decline rapidly following the cessation of oral corticosteroids and generally returns to a baseline level within 1 year after discontinuation [3].

The epidural administration of corticosteroids has also been shown to result in systemic absorption and can elicit similar metabolic and hormonal responses when compared to oral or intravenous corticosteroids [4]. However, data also suggest that these findings could be steroid-specific. In a randomized trial of 400 patients over 50 years of age who received either epidural steroid or lidocaine, neither dexamethasone nor betamethasone showed a significant difference in 3-week cortisol suppression compared to lidocaine, whereas suppression was significant with triamcinolone and methylprednisolone (MP) [5].

Studies directly evaluating the changes in bone turnover markers following a single ESI are scarce but have consistently demonstrated up to 6 months of increased markers of bone osteolysis, such as C-telopeptide (CTX), and either a reduction or no significant change in markers of bone formation, such as procollagen type 1 N-terminal propeptide (P1NP). These hormonal and paracrine abnormalities would suggest that ESIs may result in reduced bone formation and, therefore, have potentially deleterious effects on systemic bone density. However, these results should be interpreted cautiously as the few studies available in the published literature were uncontrolled and typically included at-risk populations, such as postmenopausal women [[6], [7], [8], [9]]. Additionally, the steroid administered and steroid dose both varied, with dexamethasone and betamethasone used in only one study each [9].

Literature focused directly on the potential effect of ESIs on bone mineral density (BMD) primarily utilizes serial dual-energy x-ray absorptiometry (DEXA) scanning, the most clinically relevant tool to evaluate such changes in bone quality. Low BMD is an established and significant risk factor for fragility fractures, with a ∼2 % decrease in density equating to a 20–30 % increased fracture risk [10]. Therefore, any significant detriment to BMD from ESI would increase one's risk for a fragility fracture and associated morbidity. Unfortunately, retrospective analyses of the potential correlation between diminished BMD and ESI exposure are somewhat conflicted and wrought with confounding issues, including population matching and heterogeneity of DEXA timing [[11], [12], [13], [14]]. The most informative of these may be a quantitative CT study of 121 patients who received at least three ESIs (mean cumulative dosage of 340 mg triamcinolone or equivalent) over 5 years. Their analysis showed a dose-dependent ∼20 % lower average spine BMD compared to the matched control group [12]. Interestingly, this difference was uniform throughout the lumbar spine and not concentrated in the injection region, suggesting the potential effect is from systemic, and not local, effects of the corticosteroid [12]. Another relevant retrospective study compared a population of postmenopausal patients receiving anti-osteoporotic medication who underwent ESIs (mean 2.2 injections or ∼80 mg triamcinolone total) to a control group who did not. Although both groups actually had an increase in spine BMD, ESIs seemed to modestly truncate this effect at the first but not the second year [14].

Most important to evaluating the “myth” are studies that have aimed to directly assess the incidence of fractures in patients exposed to ESI to determine if there is an increased relative risk. The most robust and clinically relevant of these is likely a 2024 cohort study of over 7,000 patients that summated all of the participants' corticosteroid injections (e.g., spine, joint, tendon) and subsequently found no association between cumulative injection dosing and subsequent fragility fracture risk, including in patients with osteoporosis [15]. Previously compiled data have been assessed in two systematic reviews in 2018 and 2019; both concluded that BMD is likely diminished transiently by the administration of ESIs. However, the subsequent increased risk of fragility fracture is more nebulous, and neither review asserts that an association has been well established, citing a paucity of underpowered studies [9,13]. Kerezoudis et al. also postulate that decrements in BMD were almost entirely associated with a cumulative MP dose of 200 mg over 1 year or 400 mg over 3 years, but not in doses of less than 200 mg of MP equivalents for postmenopausal women and 3g for healthy men [9]. Focusing on patients with the highest absolute baseline risk of fragility fracture (those with diagnosed osteoporosis, the elderly, or postmenopausal women) appears to demonstrate more of a potential association [14,[16], [17], [18]]. A large retrospective analysis of Medicare data suggests that greater than three ESIs per year carry an increased hazard ratio of ∼1.3–1.6 for fragility fracture, primarily concentrated in the 6 months following injection [17]. Interestingly, at least one study has demonstrated a significant attenuation of this association in patients concurrently receiving osteoporosis treatment [18]. The overall heterogeneity of these data makes it difficult to draw firm conclusions, and there are potentially other confounding factors. For example, any improvement in radicular symptoms is likely to increase a patient's weight-bearing tasks and compliance with exercise programs, which have also been shown to have similar BMD benefits to such medications [19].

Conclusion/Recommendation

Based on available evidence, even a single ESI is expected to temporarily decrease BMD in the 6–12 months following ESI. This effect may be more pronounced in specific at-risk populations, such as postmenopausal women. Although data linking this to fracture risk are conflicting and limited, it is reasonable to exercise caution in these patients due to the potential for a transient increase in fracture risk. Concurrent osteoporosis treatment or an increased volume of weight-bearing exercise may truncate this risk. Cumulative corticosteroid exposure from all sources and premorbid fragility fracture risk are likely more significant factors and should be considered when assessing the appropriateness of ESIs for an individual patient.

FactFinders for patient safety: Epidural steroid injections and risk of lumbar surgical site infection (updated from 2020)

Eric K. Holder, MD; Mathew Saffarian, DO; Zheyan Chen, MD, PhD; Jaymin Patel, MD; Adrian Popescu, MD; Byron J. Schneider, MD; Zachary L. McCormick, MD; Clark C. Smith, MD, MPH; and David Levi, MD on behalf of the International Pain and Spine Intervention Society's Patient Safety Committee.

Myth: Epidural steroid injections (ESIs) increase the risk of surgical site infection in patients undergoing any subsequent lumbar spine surgery.

Fact: The evidence on this subject matter is heterogeneous and primarily based on retrospective studies, often lacking considerable detail and with potential confounders. Low-quality evidence indicates that there may be a modest, temporal association between epidural steroid injections and surgical site infection after instrumented lumbar spinal fusion surgery. The association appears non-existent, or marginal at best, with non-instrumented lumbar spine surgery.

When performed according to guidelines, ESIs are a low-risk procedure used to treat radicular pain [[20], [21], [22]]. Given that steroids modulate the immune system and may impair wound healing [23,24], there is a theoretical risk that an ESI in close temporal proximity to lumbar spinal surgery could increase the risk of surgical site infection.

The reported incidence of infection following lumbar microdiscectomy or decompression without instrumentation ranges from 0.6 to 3.0 % [25,26]. The incidence of infection is greater with instrumented fusions, ranging from 4.15 to 18 % [[25], [26], [27]]. The different types of surgery and technical approaches to each procedure carry variable risks. Specific surgical factors that influence the risk of infection include operative time, retractor placement, and interruption of sterile technique [[28], [29], [30]]. Non-procedural factors known to contribute to the risk of infection associated with spinal surgery include advanced age, obesity, smoking, urgent surgery, poor control of diabetes mellitus, hypoalbuminemia, and immunosuppression [[28], [29], [30], [31], [32]]. It is challenging to determine whether a pre-operative ESI increases the risk of infection without a large enough cohort to support a multivariate regression analysis or other means of accounting for potentially confounding factors. Large prospective cohort studies of this type have not yet been published.

Lumbar microdiscectomy

There is limited investigation of the infection risk associated with lumbar microdiscectomy after an ESI. One retrospective case-control study (n=64 total; 32 in the ESI group) without risk-factor matching examined the correlation between preoperative transforaminal ESI and postoperative infection rate [33]. The average number of ESIs performed was 2.6 (range: 1–4), and the mean interval between the last ESI performed and the lumbar microdiscectomy was 5.2 months (range: 1–26 months). This study identified no significant difference between groups, but no infections were reported in either group in this small study [33].

Similarly, another retrospective study (n=315; 129 in the ESI group and 186 in the non-ESI group) determined that a history of ESIs within 3 months of minimally invasive lumbar discectomy is a risk factor for intraoperative dural tears; however, other complications, including infections, were not increased in the patient cohort that received preoperative ESIs [34].

Given the paucity of literature on lumbar microdiscectomy, it is impossible to draw evidence-based conclusions regarding the risk of lumbar post-surgical site infection associated with a preceding lumbar ESI.

Lumbar decompression without fusion

A study of Medicare patients utilizing the PearlDiver patient records database assessed the association between infection risk and ESIs before lumbar decompression without fusion [35]. Patients were categorized into four cohorts based on the temporal relationship of an ESI to decompression. Postoperative infection was evaluated within 90 days of the index decompression. Confounding variables (e.g., diabetes, smoking status) and the reports of infection were identified only through diagnosis and procedure codes. Notably, there was no mention of surgical times, sterile practices, frequency of injections, control of diabetes, smoking status, or pre-operative nutritional status [32]. Patients who underwent an ESI within one month or 1–3 months of surgery had a significantly increased risk of infection [35]. Incidentally, the overall infection risks for all ESI cohorts ranged from 0.8 to 1.7 %, which falls within the generally accepted risk for lumbar surgery without fusion and well below the national average for surgical site infection rates. Another study reached similar conclusions using similar study methods and data from the same source [36]. The postoperative infection rate was again low (1.40–1.98 %) across the ESI cohorts. However, when compared with controls (no ESI within 6 months), a significantly higher odds ratio was noted in those that received an ESI within 1–3 months (OR 4.69; P < 0.001) and within 3–6 months (OR 5.33; P < 0.001) before surgery [36]. The paradoxical increase at 3–6 months is not explained. This finding may be related to confounding factors not accounted for in the study sample.

Alternatively, other studies have reported no association between lumbar ESIs and an increased risk of postoperative infection in patients undergoing lumbar decompression without fusion [33,[37], [38], [39]]. A retrospective study of the military population included 847 patients who received an ESI before single-level decompression surgery and 5,688 control patients [38]. The findings revealed no statistically significant difference in infection rates between groups within 90 days of surgery. The results of this study may not be generalizable to a civilian population as the study population included predominantly younger males and the frequency of comorbidities within the military population may differ from the civilian population. The failure to address age and comorbidity variables was acknowledged as a study limitation by the authors [38].

Another retrospective study identified 945 patients who received at least one ESI and 4,366 control patients who underwent decompressive laminectomy and/or arthrodesis procedures [37]. Of the 945 patients who received at least one ESI, 683 patients (72 %) underwent decompressive laminectomy without arthrodesis. Variables such as duration of surgery, pre-operative hemoglobin level, and duration of hospital stay were recorded. The risk of postsurgical infection within 90 days was not increased in the ESI cohort compared to the control cohort. However, longer hospital stays, greater intraoperative blood loss, a posterior approach, and drain placement during surgery were associated with an increased risk of postoperative infection.

A 2021 retrospective study evaluated patients with or without preoperative lumbar ESIs who underwent lumbar decompression alone or with fusion. The subset of patients that underwent lumbar decompression without fusion included 9,903 patients. The authors found no association between infection risk and preoperative lumbar ESIs at any time from within 30 days to greater than 90 days before surgery. Regression analysis of all included surgical groups determined that fusion (P < 0.001), BMI (P < 0.001), and Charlson Comorbidity Index (CCI) (P=0.019) were independent predictors of postoperative infection [39].

Similarly, a 2024 retrospective all-payer database study was performed, with a 2:1 propensity score to account for comorbidities, level of surgery, and demographics between 43,674 patients who received an ESI within 90 days of laminectomy and 87,348 who did not receive an ESI [40]. The primary outcome was the rate of complications between groups at 30 days postoperatively. It was determined that surgical site infection rates were not significantly different between groups for all time points ranging from 1 - 30 days to 76–90 days (1.31 % vs. 1.42 % P = 0.11) [40]. The authors advised that ESIs should be avoided at least 30 days before surgery, given the higher rates of cerebrospinal fluid (CSF) leaks (0.28 % vs. 0.1 %; P < 0.001) and other medical complications (4.83 % vs. 3.9 %; P < 0.001) in the ESI group versus control [40]. However, the authors acknowledge that a causal relationship could not be confirmed because, despite attempts at matching, the patients who received ESIs had more comorbid conditions at baseline [40].

Lumbar decompression with fusion

Several studies identify a correlation between ESIs and the risk of postoperative infection after lumbar fusion.

A 2015 retrospective cohort study evaluated 280 patients (117 patients received and 163 patients did not receive preoperative ESIs). The study demonstrated an increased incidence of deep wound infections within 30 days of thoracic or lumbar fusion surgery if an ESI had been performed preoperatively (n=3 deep wound infections in preoperative ESI group) compared with no ESI (n=1 deep wound infection in the control group), but this difference was not statistically significant (p = 0.163) [41]. The temporal relationship between the ESI and surgery was not reported.

A 2017 study used a commercial insurance database to examine the association between preoperative ESIs and infection rates after lumbar fusion. The overall infection rate was 1.6 %, and there was an increased risk of infection in patients who received an ESI within one month [OR 2.6 (2.0–3.3), P < 0.0001] and 1–3 months [OR 1.4 (1.2–1.7), P < 0.002] of fusion surgery [42]. There was no difference in postoperative infection risk in patients who received an ESI more than 3 months before surgery [42].

A 2021 study included 5,108 patients who underwent lumbar fusion. The authors found an overall 2.68 % versus 1.69 % infection rate (P=0.025) for patients who received an ESI before fusion compared to those who did not receive an ESI preoperatively [39]. There was an increased infection risk of 5.74 % (P=0.005) if given within 30 days of surgery and 2.9 % [P=0.022] if given >90 days before surgery [39]. There was no increased infection risk for injections within 30–90 days of surgery (1.53 %, P=0.78) [39]. The authors postulate that the higher infection rate in patients who developed an infection after ESIs performed >90 days before surgery was possibly related to higher comorbidity profiles or other confounders [39]. The authors determined that lumbar fusion, BMI, and CCI were independent predictors of postoperative infection, while ESI exposure was not.

Similarly, a comparative prospective study (469 in the ESI group, 2,312 in the non-ESI control group) found a postoperative infection rate of 6.98 % in patients that received an ESI within 1 month of lumbar fusion surgery as compared to a 3.51 % infection rate in the control group (odds ratio 1.99, range 1.21–3.22, p=0.0101) [43]. The infection rate did not increase in patients who received an ESI greater than 1 month before lumbar fusion compared to controls [43]. No increased risk of infection was observed in patients who received lumbar epidural injection without steroids (lidocaine only) as compared to the control group [43].

A small retrospective study conducted in 2023 compared 23 patients who received an intraoperative ESI during instrumented lumbar fusion with 23 control patients who underwent lumbar fusion alone. The authors reported an infection rate of 17.4 % (n=4) for superficial infections in the ESI group versus 4.3 % (n=1) in the control group, P 0.043. Additionally, the authors reported an infection rate of 8.6 % (n=2) for deep infections in the ESI group and no deep infections in the control group [44].

Alternatively, other studies have demonstrated no association between preoperative ESI and surgical site infection. One study found no significant increase in postoperative infection rates in patients receiving lumbar facet joint or epidural corticosteroid injections before lumbar decompression with fusion [45]. The authors performed a retrospective review of the Military Health System Data Repository, which included 612 patients who received lumbar corticosteroid injections (348 ESIs, 264 facet injections) compared to a control group of 279. The postoperative infection rate was 1.47 % overall. The infection rate in the injection group was 1.14 % versus 1.54 % in the control group, with no differences noted in subgroup analysis at all time points evaluated, including injections performed within 0–30 days and 31–90 days before surgery [45]. The results of this study may not be generalizable to the general population since the study population was predominantly male and younger, and the authors were not able to capture data related to comorbidities within the military population studied, which may differ from the general population.

A retrospective, longitudinal study using propensity-matched cohorts found no increased infection risk in patients undergoing lumbar spine surgery with or without fusion after receiving an ESI at any time point [46]. This study utilized the MarketScan database and identified 260,482 patients who underwent lumbar spine surgery, 29.9 % of which involved spinal fusion. Over 46 % of the patients had a preoperative ESI, with a median interval between ESI and surgery of 10 weeks (interquartile range: 5–19 weeks; overall range from <30 days to 365 days). Furthermore, 21.9 % of patients had surgery within 30 days, and 7.58 % had surgery within 2 weeks of the preoperative ESI. Patients who received an ESI within 30 days of surgery did have more reoperations and readmissions, but this did not correspond with increased infection rates. The authors proposed that patients who underwent surgery shortly after ESIs are likely to have more complicated pathology and disability, increasing their risk for reoperation [46].

Summary of the systematic review and meta-analysis literature

Several recent systematic reviews and meta-analyses have attempted to synthesize the literature on this subject. Some concluded that an association exists; however, the recommendations among these studies are inconsistent. One study determined that an ESI within 1 month of lumbar spine surgery is a significant risk factor for infection, while ESIs beyond that point showed no association. The authors recommend that physicians consider avoiding any spine surgery within 1 month of an ESI [47]. Another meta-analysis found that the overall risk of surgical site infection was slightly higher if lumbar decompression with fusion was performed within 30 days after a lumbar spine steroid injection. The risk was not higher for lumbar decompression without fusion procedures [48]. Other authors determined that preoperative ESIs within 1 month of lumbar decompression or fusion surgery were associated with a 0.6 % and 2.31 % greater risk of postoperative infections, respectively [49]. In adults ≥65 years of age, ESIs within 1 or 1–3 months of lumbar spine decompression or fusion surgery were associated with a 1.3 % and 0.6 % greater risk of a postoperative infection, respectively [49]. A 2024 review concluded that a small, time-dependent, statistically significant association between preoperative ESIs and postoperative lumbar infection may exist, with the statistical association found to be strongest for ESIs performed 0–30 days from surgery, less so at 31–90 days, and non-existent if > 90 days before surgery. However, the authors concluded that the effect size is “small” and that the potential surgery-sparing effect of ESIs outweighs the potential risk of infection [50].

Lastly, a review published in 2022 determined that the body of evidence “does not implicate preoperative ESIs in postoperative infection rates following lumbar fusion or decompression. Operative treatment should not be delayed due to preoperative steroid injections based on current evidence.” [51] Nine studies with mixed findings met the inclusion criteria, and the authors determined that there is a lack of high-quality data to implicate ESIs [51]. The authors of this review indicated that they conducted a sensitivity analysis to exclude studies at high risk for bias and stated that this was not adequately implemented in prior meta-analysis studies performed on this subject matter, leading to different results [51].

Conclusions and recommendations

• •There is a growing body of literature attempting to evaluate the association between preoperative ESIs and postoperative lumbar infection, and presently, there is no consensus agreement. Drawing definitive conclusions is challenging, as evidenced by the varying findings and recommendations of multiple systematic reviews and meta-analyses. The scope of literature on this subject matter is limited presently as it requires both granular detail and large patient cohorts to account for the multiple variables that may impact the risk of lumbar surgical site infection.
• •Most of the literature on this subject is retrospective and heterogeneous. It lacks detail regarding potential patient confounding variables, preoperative ESIs, and operative procedural factors.
• •The evidence regarding preoperative ESIs and infection risk in lumbar decompression without fusion is particularly equivocal. Most studies have found no temporal relationship. However, two large studies using the same Medicare PearlDiver database, exclusive to patients aged 65 or older, found a relationship exists. There are inherent limitations in accuracy due to coding, billing, and other methodological factors for which these large database studies have been criticized and may limit the applicability of these findings.
• •Low-quality evidence suggests an association between ESIs and surgical site infection in instrumented spinal fusion cases. Multiple studies indicate an increased risk of infection in patients receiving an ESI, mostly within 30 days of the instrumented fusion. However, other studies demonstrate no association. Further investigation is necessary to determine if ESIs are a true risk factor for infection in this scenario or if the correlation is spurious due to confounders and inadvertent bias.
• •With the recognition of the limitations of current evidence, a small temporal, statistically significant association between pre-surgical ESIs and postoperative lumbar surgical site infections may exist, particularly with lumbar fusion. In the scenario of a patient who has exhausted non-operative care and definitively plans to undergo lumbar surgery regardless but requests an ESI as a temporary pain-relieving bridge before surgery, it is reasonable to avoid an ESI within at least 30 days of lumbar spine surgery (particularly instrumented fusion).
• •In the appropriately selected patient who is scheduled to undergo lumbar spine surgery but expresses interest in trying a preoperative ESI in close temporal proximity to the planned surgery to avoid or delay surgery if relief is achieved, the argument can be made that the potential surgery-sparing effect of performing the ESI likely outweighs the small risk of lumbar surgical site infection. The decision to pursue a preoperative ESI in this scenario should be made on a case-by-case basis considering the type of surgery and patient co-morbidity profile and with consensus to proceed from both the interventionalist and surgeon after informed discussion with the patient.
• •Increased BMI and Charleston Comorbidity Index (CCI) are independent predictors of postoperative infection. These factors should be considered when assessing ESI timing and postoperative infection risk, particularly for instrumented spinal fusion.
• •Additional prospective studies that account for risk factors associated with surgical site infections would be helpful.

Funding statement

No funding was utilized in the preparation of this manuscript.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.