Systematic Review and Meta-Analysis of the Effect of Osteoporosis on Reoperation Rates and Complications after Surgical Management of Lumbar Degenerative Disease
Article information
Abstract
Background
There is considerable heterogeneity in findings and a lack of consensus regarding the interplay between osteoporosis and outcomes in patients with lumbar degenerative spine disease. Therefore, the purpose of this systematic review and meta-analysis was to gather and analyze existing data on the effect of osteoporosis on radiographic, surgical, and clinical outcomes following surgery for lumbar degenerative spinal disease.
Methods
A systematic review was performed to determine the effect of osteoporosis on the incidence of adverse outcomes after surgical intervention for lumbar degenerative spinal diseases. The approach focused on the radiographic outcomes, reoperation rates, and other medical and surgical complications. Subsequently, a meta-analysis was performed on the eligible studies.
Results
The results of the meta-analysis suggested that osteoporotic patients experienced increased rates of adjacent segment disease (ASD; P=0.015) and cage subsidence (P=0.001) while demonstrating lower reoperation rates than non-osteoporotic patients (7.4% vs. 13.1%; P=0.038). The systematic review also indicated that the length of stay, overall costs, rates of screw loosening, and rates of wound and other medical complications may increase in patients with a lower bone mineral density. Fusion rates, as well as patient-reported and clinical outcomes, did not differ significantly between osteoporotic and non-osteoporotic patients.
Conclusions
Osteoporosis was associated with an increased risk of ASD, cage migration, and possibly postoperative screw loosening, as well as longer hospital stays, incurring higher costs and an increased likelihood of postoperative complications. However, a link was not established between osteoporosis and poor clinical outcomes.
INTRODUCTION
Osteoporosis is a common disease characterized by compromised bone strength with a global prevalence of approximately 18%.[1] It carries a substantial cost, estimated at 18 billion dollars per annum in the United States alone.[2] It also presents a significant burden to patients, who are far more likely to suffer from a variety of debilitating fractures, including vertebral and hip fractures.[2] Osteoporosis is an age-related process of decreased bone strength and quality. Decreased bone mineral density (BMD) is a significant contributor not only to increased fracture risk but also to impaired fracture healing, failure of fixation, hardware, and increased reoperation rates.[3] Moreover, osteoporosis has also been linked to increased hospital-related morbidity, including an increase in hospital visits, longer stays, and increased costs.[4]
Lumbar degenerative spine disease (DSD) is a challenging clinical condition in elderly individuals. In a 2018 meta-analysis on the global epidemiology of lumbar DSD, Ravindra et al. [5] estimated that 266 million individuals worldwide are diagnosed yearly. Generally, lumbar DSD denotes an age-related deformation of the lumbar spine architecture and includes processes such as lumbar disc degeneration, herniation, and degenerative spondylolisthesis. Several surgical interventions are available for patients with lumbar DSD based on specific indications, including discectomy, decompression, and various types of lumbar fusion and have shown favorable improvements in terms of pain and disability, especially among osteoporotic patients.[6–8] However, there is considerable heterogeneity of findings and a lack of consensus with respect to the interplay of osteoporosis and outcomes of patients with lumbar DSD. Osteoporosis has profound effects on outcomes following lumbar spine surgery. For example, Formby et al. [9] found that following transforaminal lumbar interbody fusion surgery, patients with osteoporosis were more likely to experience radiographic complications such as cage non-union, subsidence, and postoperative fractures. Similarly, Sharma et al. [4] identified osteoporosis as a risk factor for readmission rates, longer hospital stays, and increased costs following surgical management of thoracolumbar spinal diseases. Therefore, given the prevalence of osteoporosis, it is crucial to understand its influence on patients undergoing surgical intervention for lumbar DSD.
The purpose of this systematic review and meta-analysis was to gather and analyze current data on the effect of osteoporosis on radiographic, surgical, and clinical outcomes following surgery for lumbar DSD. We hypothesize that patients with osteoporosis will tend to have greater odds of adverse surgical and functional outcomes following surgery for lumbar DSD and may experience longer hospital stays and incur greater costs.
METHODS
Following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines,[10] we performed a systematic review to determine the effect of osteoporosis on the incidence of adverse outcomes after surgical intervention for lumbar DSDs.
First, the Medline (PubMed) database was queried from 1990 through August 2022 using the following terms: “osteoporosis” AND “lumbar” AND (“outcomes” OR “revision” OR “reoperation” OR “complication”). Articles written in English, both prospective and retrospective, were initially included to assess patients diagnosed with osteoporosis and report on the incidence of radiographic, surgical, clinical, and postoperative outcomes. Patients were diagnosed with osteoporosis based on the International Classification of Diseases, Tenth Revision codes or a confirmed diagnosis of decreased bone density on dual energy X-ray absorptiometry (DXA) or computed tomography.
In our initial exclusion criteria, we eliminated the following studies: systematic reviews, case reports, studies without osteoporotic patients, studies without relevant postoperative outcomes, and studies in which vertebral surgery was performed due to specific indications, such as trauma, neoplasia, or compression fractures. Furthermore, studies with limited sample sizes, defined as fewer than 20 patients, and those without operative interventions were removed. A minimum sample size of 20 was determined for appropriate assessment of our independent variable, osteoporosis status, using a more conservative variation of the customary statistical analysis criterion for regression analyses, wherein a minimum of 10 events per independent variable (EPV) was suggested to minimize overfitting.[11,12] Therefore, we sought to use a more conservative iteration, 20 EPV, established by Austin and Steyerberg [13], to mitigate the risk of bias associated with including multiple smaller-sized studies.[13,14]
After this preliminary exclusion, full-text articles were excluded if they were unrelated to lumbar pathology, if cement augmentation was used as part of the surgical intervention, or if the study was performed before 1990.
Two independent reviewers performed the electronic database search and screening. Papers were screened based on title, abstract, and full text, and duplicates were removed. The references of the full texts were then examined to identify any relevant studies missed in our initial query. These were subjected to the aforementioned screening criteria to ensure that only articles reporting postoperative outcomes for lumbar DSD were included. Finally, data collection, systematic reviews, and meta-analyses are conducted.
A data collection sheet was created using the Cochrane Consumers and Communication Review Group’s Data Extraction Template for Included Studies.[15] The eligible articles were reviewed by three authors who collected the following information: author name, publication year, study type, patient populations, indications for surgery, and the outcomes assessed. The study outcomes were further classified into radiographic, surgical, clinical, and other complications. Finally, the risk of bias was assessed using the Grading of Recommendations, Assessment, Development, and Evaluations (GRADE) method,[16] and articles were assigned a GRADE of very low, low, moderate, or high.
A meta-analysis was subsequently performed for eligible studies that included only outcomes present in at least two studies and measured similarly. The three authors obtained the following information from the respective studies: number of osteoporotic and non-osteoporotic patients with a particular outcome and the associated statistical analyses, radiographic outcomes (i.e., screw loosening or breakage, adjacent segment disease [ASD], cage subsidence, and fusion rates), and complications (i.e., reoperation rate, wound infection, and overall complications). Patient-reported outcomes included the Oswestry Disability Index (ODI), visual analog scale (VAS), and EuroQol-5-dimensions (EQ-5D).
Statistical analysis was performed using Comprehensive Meta-Analysis (version 2), where heterogeneity among studies was considered using a random-effects model. This model was used to generalize the results to the included samples. Means were calculated between conditions and differences were compared while considering the standard deviations and sample sizes of the included studies. Categorical or means with reportable standard deviations were included, and P-values less than 0.05 were considered significant.
RESULTS
1. Study characteristics
The process of study inclusion and exclusion is shown in the PRISMA diagram (Fig. 1). A total of 1,912 records were initially identified in the database, and 1,432 remained after applying the exclusion criteria. After screening the titles for relevance, 528 records were assessed for eligibility through an abstract review based on the inclusion criteria. A total of 223 records underwent a full-text review, and the references of studies that were eligible for inclusion were also screened, yielding an additional 53 studies. A total of 38 patients were ultimately included in this systematic review, including 35 retrospective and three prospective studies (Table 1). Sixteen studies were eligible for inclusion in the meta-analysis (Table 2).
Study selection process.
Characteristics of studies focusing on outcomes after surgery for lumbar degenerative disease
Results of the meta-analysis comparing outcomes after surgery for lumbar degenerative disease between osteoporotic and non-osteoporotic patients
2. Risk of bias and quality of evidence
The risk of bias was determined using the GRADE guidelines.[16] The Journal of Bone and Joint Surgery schema was used to determine the level of evidence for each article.[17] In terms of GRADE, 25 were low and 13 were very low. Three studies had Level II evidence, 32 had Level III evidence, and three had Level IV evidence.
3. Radiographic outcomes
1) Screw loosening
A total of seven studies assessed the relationship between a diagnosis of osteoporosis and the risk of postoperative pedicle screw loosening with six finding a statistically significant association.[18–23] Cho et al. [19] demonstrated that screw loosening rates were significantly higher amongst osteoporotic patients (32.3% vs. 12.7%; P=0.029). Löffler et al. [18] and Lee et al. [23] found that BMD or T-score was significantly lower in patients who developed screw loosening (P=0.001 for both), and Lee et al. [23], Zou et al. [20], Bokov et al. [21], and Sakai et al. [22] reported that lower Hounsfield units (HU) values were significantly associated with an increased risk of screw loosening (P<0.01 for all). Only Formby et al. [9] found a non-significant association with osteoporosis when exploring rates of screw loosening.
2) Adjacent segment disease/degeneration
Of the six studies assessing whether patients with osteoporosis were at an increased risk of developing postoperative ASD or degeneration, only Ankrah et al. [24] identified osteoporosis as an independent risk factor (odds ratio [OR], 14; P=0.03). The remaining five studies did not find an association.[25–29]
3) Cage subsidence, migration, and retropulsion
Twelve studies explored the relationship between osteoporosis and the incidences of cage migration, retropulsion, and subsidence, with nine reporting significant associations. Zhao et al. [30] and Jones et al. [31] found that osteoporosis was a risk factor for cage subsidence (OR, 5.976; P<0.001 and OR, 0.996; P=0.032, respectively). Cho et al. [19], Formby et al. [9], and Jones et al. [31] reported higher rates of cage subsidence in osteoporotic patients (P<0.001, P=0.05, and P=0.032, respectively). Yao et al. [32] and Mi et al. [33] found that patients who experienced postoperative subsidence had lower mean T-scores and HU values (P=0.007 and P=0.0015). Oh et al. [34] showed that BMD was significantly lower in segments that demonstrated a higher degree of subsidence (>3 mm) compared to a more moderate degree (1–3 mm) (P=0.049). Peng et al. [35] demonstrated a significant association between BMD and cage retropulsion (P=0.001). Finally, Park et al. [36] found that osteoporosis was a significant risk factor for cage migration with (P<0.001) and without (P<0.001) subsidence, as well as cage retropulsion (P<0.001). The last three studies could not establish a significant link between osteoporosis and implant subsidence or migration.[37–39]
4) Fusion and pseudarthrosis rates
Five studies assessed the relationship between osteoporosis and fusion rates or the development of pseudarthrosis.[9,19,40–42] Bjerke et al. [40] reported that non-union rates were significantly different between osteoporotic, osteopenic, and normal BMD cohorts (46.2%, 18.4%, and 18.6%, respectively; P=0.016). The remaining four studies identified no significant relationship between osteoporosis and rates of fusion.[9,19,41,42]
4. Complications
1) Reoperation rates
Of 12 studies assessing if osteoporosis is related to rates of reoperation after lumbar DSD surgery, five studies found a significant association.[4,24,42–44] Khalid et al. [42] found that both osteoporotic and osteopenic patients had a higher risk of needing revision surgery within 2 years of lumbar fusion compared to patients of normal bone density (OR, 3.25 and OR, 2.73, respectively). Sharma et al. [4], Wolfert et al. [44], and Ankrah et al. [24] found that osteoporotic patients had significantly higher risks of revision surgery at several different time points. Interestingly, Ehresman et al. [43] demonstrated that the vertebral bone quality score obtained via magnetic resonance imaging (MRI) was associated with revision surgery (P=0.010), but the DXA T-score was not (P=0.233). Seven additional articles were unable to establish significance for osteoporosis as a predictor of the need for revision surgery.[9,25,26,36,45–47]
2) Postoperative complications
Five articles assessing the risk of “overall complications” reported that osteoporosis was significantly associated with the development of any complication.[4,9,44,48,49] Sharma et al. [4] and Wolfert et al. [44] found that osteoporotic patients were more likely to develop any complication postoperatively (P<0.0001 for both) St Jeor et al. [49] and Aimar et al. [48] found that osteoporosis was associated with the development of various complications (P=0.001 and P=0.049, respectively). Formby et al. [9] reported that overall radiographic complications occurred more frequently in patients with osteoporosis (77.8% vs. 48.6%; P=0.03). Koutsoumbelis et al. [50] and Wolfert et al. [44] found that osteoporosis was associated with postoperative wound complications (P=0.010 and P<0.001, respectively). Formby et al. [9] identified greater rates of fracture amongst osteoporotic patients (16.7% vs. 1.4%, P=0.03), but Kim et al. [51] found no difference in T-scores between the fracture and no fracture cohorts (0.2±1.7 vs. 0.8±1.7; P=0.389).
5. Clinical outcomes
1) Surgical outcomes
Sharma et al. [4] and Wolfert et al. [44] identified an association between the presence of osteoporosis, longer hospital stays, and greater total costs incurred by patients. However, Sharma et al. [4] found that osteoporotic patients incurred lower overall costs during index hospitalization than non-osteoporotic patients ($46,217 vs. $50,126; relative risk, 1.04; P=0.0001). Lastly, osteoporosis was found to be a significant independent risk factor associated with lower rates of discharge to home (OR, 0.845; P<0.0001).[4]
2) Functional outcomes
Lee et al. [52] and Cho et al. [19] demonstrated that postoperative ODI scores were not associated with BMD, Cho et al. [19] also failed to find differences in VAS and EQ-5D scores. Lastly, Formby et al. [9] found that neither persistent nor recurrent postoperative symptoms differed between patients with osteoporosis and those with normal BMD.
6. Meta-analysis results
Thirteen studies provided appropriate data that were eligible for inclusion in the meta-analysis with a focus on nine outcomes, as outlined in Table 2. Fusion rates were not different between osteoporotic and non-osteoporotic patients when patients were pooled from four studies (88.9% vs. 88.6%; OR, 0.997; P=0.954), demonstrated in Figure 2.[9,19,40,42] Three studies focused on ASD,[24–26] and the meta-analysis identified greater rates of ASD in osteoporotic patients than in non-osteoporotic patients (62.5% vs. 17.8%; OR, 1.51; P=0.015) demonstrated in Figure 3A. Seven papers assessed reoperation rates, and the need for revision surgery was greater amongst patients without osteoporosis (7.4% vs. 13.1%; OR, 0.897; P=0.038), demonstrated in Figure 3B.[4,9,24–26,42,45] Cage subsidence occurred much more frequently in patients with osteoporosis than in those without (63.8% vs. 30.7%; OR, 1.422; P=0.001) amongst the three studies, demonstrated in Figure 3C.[9,19,30] Three papers reported on rates of pedicle screw loosening (Fig. 3D),[9,18,19] two on rates of overall complications (Fig. 3E),[4,40] two on wound complications (Fig. 3F),[44,50] two on length of stay (Fig. 3G),[4,44] and two on overall costs (Fig. 4) incurred to patients,[4,44] but the meta-analysis did not identify differential rates between osteoporotic and non-osteoporotic patients. Heterogeneity existed among all outcome analyses, and the P-values are provided in Supplementary Table 1.
Forest plot showing meta-analysis findings for fusion rate. CI, confidence interval.
Forest plot showing meta-analysis findings complications and outcomes. (A) Adjacent segment disease. (B) Overall revision surgery. (C) Subsidence. (D) Pedicle screw loosening. (E) Overall complications. (F) Surgical site infection. (G) Length of stay. CI, confidence interval.
Forest plot showing meta-analysis findings for payment. CI, confidence interval.
DISCUSSION
This systematic review and meta-analysis focused on differences in outcomes between osteoporotic and non-osteoporotic patients after surgery for degenerative lumbar disorders. Overall, we identified that patients with osteoporosis have significantly higher rates of adjacent segment degeneration and cage migration, and may also experience greater rates of screw loosening and postoperative complications, such as wound problems and other medical complications (cardiac, renal, etc.). Generally, our findings are concordant with the current literature, suggesting that osteoporosis is linked to the acceleration of degenerative changes in the spine that ultimately diminish biomechanical stability. When cage migration, specifically subsidence, develops postoperatively, there is a greater theoretical risk of compromise to the construct via a decrease in intervertebral disc height.[31,34,36] An important consideration is the degree to which particular radiographic findings predict clinical outcomes, and while cage subsidence may present clinical issues due to differences in axial load distribution,[53–55] recent investigations have reported similar clinical outcomes.[9,19,32,34] In fact, we were unable to establish osteoporosis as a risk factor for inferior clinical outcomes as measured by ODI,[19,52] VAS, EQ-5D,[19] or recurrence/persistence of symptoms.[9] Furthermore, several investigations found that lumbar DSD patients with postoperative cage migration had lower fusion rates, which has been shown to subsequently negatively impact patients’ postoperative courses.[30,36] However, when considering cage migration as a partial measure of treatment success—an important limitation to note—is that although 2 mm of vertical migration is commonly referenced as a threshold for defining cage subsidence, the value remains unvalidated, and the measurement methods to quantify degree of subsidence have shown poor interobserver reliability.[30,39,56] Although our meta-analysis identified osteoporosis as a significant risk factor for cage subsidence, the clinical significance of this complication remains unclear.
The meta-analysis identified a much higher rate of ASD in patients with osteoporosis than in those with a normal BMD (62.5% vs. 17.8%; P=0.015). Across the studies included in this review, the incidence of ASD ranged from 6.3% to 62.5%, with wide variability likely attributed to discrepant follow-up durations between studies. Adjacent segment degeneration is a radiographic entity, while ASD necessitates the presence of clinical symptoms. The presumption is that when vertebral segments are fused, more force is distributed to the adjacent segments, resulting in increased mobility and accelerated degeneration.[25,27,57–59] Despite being distinct conditions, the interaction between both pathologies may drive further negative impact on biomechanical stability as a consequence of reduced bone density and atrophied paraspinal musculature; such characteristic features of osteoporotic patients can potentially worsen degeneration of adjacent segments particularly without adequate dynamic stabilizers.[60] Postoperatively, tissue trauma and ensuing postoperative muscle weakness further contribute to these proposed, degenerative processes as a result of increased axial loading through the spinal column.[27] ASD is also a notable indication for revision surgery, as the increased stress placed on these segments can cause a recurrence of the symptoms that prompted index surgery.[24–26] Maruenda et al. [61] detected rising rates of adjacent segment degeneration and revision surgery after circumferential lumbar fusion, with 79.8% of patients developing ASD 15 years after surgery and 37.5% requiring reoperation. Because osteoporosis seems to be a significant risk factor for the development of ASD, identification of at-risk patients is critical for sufficient surgical planning to optimize technique and hardware choices to prevent the need for future revision surgery.
Although rates of screw loosening were not found to differ based on BMD in the meta-analysis, cadaveric studies of the biomechanics of osteoporotic bone have demonstrated that the decreased density of the trabecular bone contributes to decreased screw fixation strength, pullout strength, and insertional torque.[62–65] In Zou et al.’s large retrospective cohort study [20] of patients who underwent lumbar pedicle screw fixation, lower HU values of the affected vertebral levels were a significant risk factor for pedicle screw loosening at the one-year follow-up. A frequently employed augmentation technique is polymethylmethacrylate cement application at the screw-bone interface, which has been shown to improve hardware strength and decrease rates of screw pullout and loosening.[66–68] This discrepancy between the systematic review and meta-analysis may be partially explained by the fact that a smaller proportion of papers reporting on screw loosening were eligible for inclusion in the meta-analysis.
Although a meta-analysis identified higher rates of ASD and subsidence in patients with osteoporosis, only one systematic review hinted at associations between osteoporosis and longer hospital stays, development of postoperative complications, and greater costs. If greater costs are incurred by patients with osteoporosis, as suggested by Wolfert et al. [44], this may be explained by greater healthcare utilization during both the perioperative and follow-up periods.[4]
Given that several investigations have found a greater risk of revision surgery among patients with osteoporosis, we posit that this may contribute to the increased costs incurred by this population. However, a meta-analysis suggested that patients without osteoporosis underwent reoperation at a higher rate than their osteoporotic counterparts. Articles that included a sufficient number of patients with osteoporosis and those who required reoperation were more likely to demonstrate a higher rate among osteoporotic patients.[4,42,44] When both osteoporosis and reoperation rarely occur in a cohort, it is not surprising that a greater percentage of pooled patients undergoing reoperation are non-osteoporotic. For example, although Park et al. [69] studied a large cohort of 16,927 patients who underwent primary single-level posterior fusion for degenerative lumbar disease, only 0.78% of the patients had osteoporosis and 3.18% underwent reoperation. Osteoporosis was not determined to be a risk factor for needing reoperation within the follow-up period of 4.5 years (OR, 1.45; 95% confidence interval, 0.649–3.238; P=0.365).[69] Lastly, a possible explanation for these unexpected findings is the overall heterogeneity of indications for revision surgery, which may significantly differ across studies and between groups based on a diagnosis of osteoporosis. Future investigations should prioritize sufficiently powered cohorts that differentiate the need for reoperation based on various indications to assess their relationship with BMD.
This study has some limitations. One of the most pervasive concerns is the BMD measurement, which is commonly performed using DXA scanning. Recent investigations suggest that DXA values can differ significantly based on both the anatomic region and the measurement technique.[31,34,43,70] Other methods of assessing bone quality, such as HU values and MRI vertebral bone quality scores, are becoming more frequently utilized; however, studies have demonstrated discordance. For example, Sakai et al. [22] reported that low HU values were predictive of screw loosening within three months after single-level posterior lumbar interbody fusion, but BMD measured by DXA scan of the L1–L4 vertebra was not. The lack of concordance between different methods of determining BMD highlights the need for consistent systematically validated and standardized screening of at-risk patients, especially those who may prior to instrumented fusion to mitigate the development of preventable complications. Other limitations of this review are related to the retrospective nature of most studies, which limits the quality of evidence and our ability to draw conclusions. There was also substantial heterogeneity across the studies included in the meta-analysis. Although several studies with large cohorts were included in this review and meta-analysis, the low proportion of patients with osteoporosis across most studies resulted in small effect sizes and ultimately indicated insufficient power to establish statistical significance. Future investigations should prioritize the enrollment of a sufficient number of osteoporotic patients in prospective studies to further our understanding of the implications of poor bone quality on outcomes after lumbar spine surgery.
CONCLUSION
In summary, our systematic review and meta-analysis demonstrated that osteoporosis is strongly associated with an increased risk of ASD and cage migration. Osteoporosis may also be associated with increased rates of pedicle screw loosening as well as suboptimal outcomes related to hospitalization costs and long-term complications. Conversely, we were unable to establish a link between osteoporosis, decreased BMD, decreased clinical outcome scores, and symptomatic persistence. Our findings suggest that BMD can be a powerful tool for guiding surgical decision-making in patients with lumbar degenerative diseases. Future research should aim to prospectively assess the association between osteoporosis and postoperative outcomes, especially reoperation rates. Efforts should be made to standardize the screening of patients with osteoporosis before surgical intervention. Overall, these investigations can aid in guiding operative considerations that optimize surgical care for patients with osteoporosis.
Supplementary Information
Notes
Funding
The authors received no financial support for this article.
Ethics approval and consent to participate
Not applicable.
Conflict of interest
Lechtholz-Zey EA, Ayad M, Gettleman BS, Mills ES, Shelby H, Ton A, and Shah I have nothing to disclose. Wang JC has received intellectual property royalties from Zimmer Biomet, NovApproach, SeaSpine, and DePuy Synthes. Hah RJ has received grant funding from SI bone, consulting fees from NuVasive, and support from the North American Spine Society to attend meetings. Alluri RK has received grant funding from NIH, consulting fees from HIA Technologies, and payment from Eccentrial Robotics for lectures and presentations. No potential conflict of interest relevant to this article was reported.