Complication Trends and Costs of Surgical Management in 11,086 Osteoporotic Patients Receiving Lumbar Fusion

Shane Shahrestani, MS; Xiao T. Chen, BA; Alexander M. Ballatori, BA; Andy Ton, BS; Joshua Bakhsheshian, MD; Raymond J. Hah, MD; Jeffrey C. Wang, MD; Zorica Buser, PhD


Spine. 2021;46(21):1478-1484. 

In This Article


The purpose of this study was to compare different outcomes of fusion surgery in patients with osteoporosis regarding graft subtype and surgical approach. Findings on univariable analysis suggested that the use of nonautologous graft material in single-level and multilevel lumbar fusion may result in lower postoperative complications in osteoporotic patients compared to complications associated with use of autologous graft material. Furthermore, multivariable analysis of single and multilevel fusion of the anterior or posterior column found similar postoperative results in osteoporotic patients between fusion groups, although the charges associated with anterior fusion were found to be significantly higher than that of posterior fusion when controlling for covariates.

Although many studies discuss methods of perioperatively decreasing surgical risks of lumbar spine fusion, few studies focus on the optimization of surgical approach. A retrospective observational cohort study by Lee et al compared 53 elderly patients with osteoporosis who underwent multilevel anterior lumbar interbody fusion (ALIF) with either posterolateral fusion (PLF) (n = 28) or percutaneous pedicle screw fixation (PPF) (n = 25).[14] ALIF + PPF was associated with less blood loss, shorter operative times, and fewer complications at 1-year follow-up compared to the ALIF + PLF group, although fusion rates were statistically equivalent.[14] Similarly, Putzier et al conducted a trial of 40 patients with degenerative spinal disease treated with either autologous iliac crest cancellous bone or allogenic cancellous bone graft, and their results showed that after 12 months both groups had similar fusion rates upon radiological evaluation, although patients treated with autologous graft material had higher rates of hematoma formation and persistent pain.[15] Similarly, patients who received fusion with autologous graft material in our study were found to also have higher rates of complications, including lumbar vertebral fracture and hardware, compared to those treated with nonautologous graft material. Although our short-term analysis within 6 months of primary discharge suggested that autologous graft material may increase the rate of acute complications, literature on this subject remains limited and future studies evaluating complications in osteoporotic patients treated with different osteobiologic graft materials is warranted.

Additionally, a retrospective study by Cavagna et al evaluated 39 patients older than 65 years with osteoporosis who underwent lumbar decompression and fusion with instrumentation consisting of a less rigid titanium rod adapted for bone of poor mechanical quality.[16] Although there was no control group for the study, the authors concluded that clinical complications (VAS, ODI, and Japanese Orthopedic Association scores) were satisfactory at 2 to 4 years postoperatively. Furthermore, the reported fusion rate was 89.74% (35/39) on plain radiographs and although two screws and two rods failed at 2 years, those four patients had adequate fusion and were asymptomatic. The findings of this study suggested that instrumentation material may significantly improve fusion success in patients with degenerative lumbar pathologies. In the present study patients receiving single-level fusion with autologous graft material had higher rates of hardware failure at 180 days of discharge compared to those receiving single-level fusion with nonautologous graft material, however not significant. A potential explanation for significantly higher rates of hardware failure at early time points in the autograft group might be due to the hardware placement rather than fusion itself. However, due to the lack of imaging data our study was unable to determine the exact reasons. In addition, nonautograft group was smaller so sample size might be another factor.

Other database studies include those by Chen et al, Saleh et al, and Toy et al, which all query complication and readmission rates associated with lumbar fusion surgery.[17–19] Chen et al document an 11.6% 90-day readmission rate in all patients who received a lumbar fusion procedure. Our findings suggested similar 90-day readmission rates in osteoporotic patients. Furthermore, the study conducted by Saleh et al found that longer operative time and unnecessary surgical instrumentation increased postsurgical complication rates in 2320 frail and elderly patients who received lumbar fusion surgery.[19] However, neither study analyzed osteoporotic patients as a subgroup and both lack the necessary follow-up time and sample size to draw strong conclusions. Conversely, our analysis of 11,086 osteoporotic patients provides significant power to report more accurate associations regarding lumbar fusion surgical approaches.

Osteoporosis of the spine and associated risks such as vertebral fracture or accelerated disc degeneration are especially problematic in elderly patients who often have poorer muscle quality and kyphotic posture leading to decreased stability/balance and abnormal biomechanical stress to the lumbar spine. Previous studies have evaluated the efficacy of perioperative factors, such as exercise training and pharmacologic intervention, as an adjuvant therapy for osteoporotic patients undergoing lumbar fusion.[20–22] As the prevalence of osteoporosis continues to increase in the United States due to the aging population, it is of utmost importance to understand methods of optimizing lumbar spine fusion surgery in this vulnerable population.[1,23] Not only would optimization of lumbar surgery reduce the high rate of osteoporosis-related postoperative complications following lumbar fusion surgery, but it would also ease the economic burden of osteoporosis-related complications on the hospital and United States health system.[23]


There are several limitations to this study. First, the retrospective nature of this database study is limited by the quality of medical coding. Second, NRD also lacks important patient-reported complications such as pain and function, which makes clinical conclusions difficult. Third, this study only included NRD years 2016 and 2017. However, this time span was chosen due to the implementation of mandatory ICD-10 coding in late 2015, which allowed for more detailed codes. Thus, despite the limited time window and decreased potential sample size, this study should have greater accuracy and granularity than any previous studies that used ICD-9 or CPT codes. Fourth, ICD-10 coding only allows the distinction between single and multilevel fusion, so the exact number of vertebrae fused may vary within the multilevel group. Fifth, both synthetic graft materials and bone morphogenic protein were not queried for analysis because of limited ICD coding and a lack of specificity in coding. Additional some of the common risk factors were not considered due to the lack of codes for levels of severity or poor coding of medication/drug use. As such, we use CCI as an imperfect (does not include recreational tobacco use, drugs that may induce osteoporosis, among others) metric of patient comorbidity status. Finally, ICD-10 coding does not capture the degree of osteoporosis as there is no information such as DEXA scan t scores. To confront this issue, we chose to include only patients with osteoporosis codes and did not use a control group of normal or osteopenic patients with unknown bone densities. Additionally, our results showed no significant differences in age, sex, or CCI in the majority of analyzed timepoints, which improves comparability between our groups of interest.

The strength of this study lies in its sample size of >11,000 patients with vertebral osteoporosis who underwent lumbar spine fusion, which to our knowledge is the largest such cohort within the current literature. Additionally, the comparability between groups of interest (autologous vs. nonautologous biologics and anterior vs. posterior column fusion) in terms of average age, sex, and CCI allowed for analysis of complications and readmission rates between groups. Both univariable and multivariable analyses provided more detailed information on the impact of understudied variables such as biologic type and fusion location on osteoporotic lumbar spine fusion complications. Although the results in this article are limited by factors characteristic of any database study relying on ICD coding, the rigorous methodology and statistical analysis provide preliminary information on an important topic within spine surgery.