Friday, October 21, 2016
The reoperation rate following ACDF is relatively low, making it difficult to study risk factors for reoperation. The most common problems leading to reoperation are adjacent segment degeneration (ASD) and pseudarthrosis, with overall reoperation rate reportedly between 2% and 5% per year.1,2 In an effort to better understand risk factors for re-operation, Dr. Park and colleagues from Korea used the Korea Health Insurance Review and Assessment Service database to evaluate reoperation following ACDF. They identified approximately 8,000 patients who underwent ACDF for cervical spondylotic radiculopathy or myelopathy in 2009 and followed them through 2014. Their main goal was to compare reoperation rates for the radiculopathy and myelopathy cohorts, though they also evaluated risk factors such as age, sex, diabetes, osteoporosis, other comorbidities, number of levels fused, and hospital type. They found an overall 5 year reoperation rate of 2.45%, with the myelopathy patients undergoing reoperation at a significantly higher rate than the radiculopathy patients (3.69% vs. 2.19%). After controlling for potential confounders, the hazard ratio for reoperation rate for the myelopathy patients was 1.7 (i.e. 70% more likely to undergo reoperation). Males, diabetics, those with multiple comorbidities, and those treated at larger hospitals were also more likely to undergo reoperation.
This study provides high quality data on a very large number of patients that allows for meaningful analysis of risk factors for reoperation following ACDF. One of the most interesting findings was a very low reoperation rate—approximately 0.5% per year. This is 5 to 10 fold lower than the rate cited in most studies performed in the United States, indicating very different thresholds for reoperation in the two different countries. The other marked difference compared to practice in the United States is that only 5% of the Korean patients underwent multilevel procedures, compared to over 50% of patients in the US.2 The main finding of the paper, namely that myelopathy patients have a higher reoperation rate than radiculopathy patients after the same procedure, is not that surprising. The myelopathy patients are older, have more comorbidities, and typically have persistent symptoms after surgery, all of which can drive reoperation and are difficult to control for completely in statistical models. The authors also point out that myelopathy patients frequently have a greater degree of congenital stenosis that may predispose them to developing cord compression at adjacent levels. While this study does provide some interesting big picture insights on ACDF reoperation rates in Korea, it has the typical limitations of large, administrative database studies. The main limitation with this study is that the reason for reoperation (i.e. pseudarthrosis, adjacent segment disease, hardware problems, etc.) was not able to be determined. Additionally, smoking status was not recorded, which is a significant risk factor for pseudarthrosis. The most striking finding from this study is the very low rate of reoperation following ACDF in Korea. Multiple industry-sponsored FDA IDE trials evaluating cervical total disk arthroplasty devices have reported very high reoperation rates—up to 20% at 5 years--in the ACDF control groups.3 This paper demonstrates how it is possible to have much lower reoperation rates, and that reoperation rate is an outcome highly dependent on surgeon and patient preference.
Please read Dr. Park's article on this topic in the October 15 issue. Does this change how you consider reoperation rates following ACDF? Let us know by leaving a comment on The Spine Blog.
Adam Pearson, MD, MS
Associate Web Editor
1. Singh K, Phillips FM, Park DK, Pelton MA, An HS, Goldberg EJ. Factors affecting reoperations after anterior cervical discectomy and fusion within and outside of a Federal Drug Administration investigational device exemption cervical disc replacement trial. Spine J 2012;12:372-8.
2. Veeravagu A, Cole T, Jiang B, Ratliff JK. Revision rates and complication incidence in single- and multilevel anterior cervical discectomy and fusion procedures: an administrative database study. Spine J 2014;14:1125-31.
3. Jackson RJ, Davis RJ, Hoffman GA, et al. Subsequent surgery rates after cervical total disc replacement using a Mobi-C Cervical Disc Prosthesis versus anterior cervical discectomy and fusion: a prospective randomized clinical trial with 5-year follow-up. J Neurosurg Spine 2016;24:734-45.
Friday, October 14, 2016
Learning curves in spine surgery have been investigated recently, especially those for the adoption of minimally invasive techniques. However, most of these studies have included experienced surgeons familiar with analogous open techniques who then adopt a less invasive approach. Possibly even more important is the learning curve of newly trained surgeons for common, widely-accepted procedures. All surgeons who have gone through the transition from training to independent practice are acutely aware of this process, though it has received scant attention in the literature. In an effort to fill this void, Dr. Singh and colleagues from Chicago reviewed his learning curve for one and two level ACDF over a nine year period. He completed 374 ACDFs over this time frame, which began with his first case out of fellowship. The cases were divided into early, middle, and late cohorts. They evaluated operative time, estimated blood loss (EBL), arthrodesis rate (defined by CT scans at 1 year post-op), change in visual analog neck pain scores (VAS), length of stay (LOS), and complications. The only outcomes that changed significantly across the cohorts were operative time (decreased from 86 to 60 minutes from early to late cohorts) and EBL (decreased from 100 to 40 cc). Analysis of the operative time learning curve demonstrated that 50% of the improvement had occurred by case 17, with 90% by case 56. Fusion rate increased modestly (94%-100%), though this was not statistically significant and could have been related to change in graft type or patient characteristics (fewer smokers and lower comorbidity burden in the late cohort). There were no significant changes in VAS scores (if anything a trend towards less improvement), complications (ranging from 2.4% early to 0.8% late), or LOS. One of the most notable changes was in payor mix, with Medicare patients decreasing from 27% to 8% and worker's compensation patients increasing from 25% to 44% from the early to late cohort.
This is an interesting paper as it looks at the learning curve for a common procedure for a surgeon straight out of training. Given that 100% of surgeons go through such a learning curve, the lack of literature on the topic is surprising. While surgeons believe that their skills improve over the first few years of their practice, it is unknown if patients treated by well-trained but inexperienced surgeons are at increased risk for a poor outcome or complications. This paper is very reassuring in that the improvements in operative time and EBL are not likely to be clinically meaningful, and longer-term outcomes such as change in neck pain or fusion rate were not worse in patients treated early in the surgeon's career. The complication rate was very low, and this may reflect underreporting of some relatively benign complications. Overall, ACDF is generally a safe procedure with a very low rate of major complications, so it is hard to measure complication rates without including a huge number of patients. It would have been interesting to study changes in dysphagia scores over time to see if improved tissue handling and retraction techniques helped reduce this bothersome if short-lived problem. The major limitation of this study was that it included just one surgeon's experience, and it is unclear if the results are generalizable to surgeons with different training and skills. Additionally, the patient reported outcomes were limited to VAS neck pain, and other measures (i.e. VAS arm pain, neck disability index) may have been more sensitive to changes in the surgeon's skill level over time. The change in payor mix was interesting, and it likely reflects a surgeon's ability to capture more worker's compensation patients—with the increased reimbursement that accompanies this group—as his experience and notoriety grows. This paper provides a fascinating look into the first decade of a surgeon's practice and should reassure patients that well-trained but inexperienced surgeons can provide them with safe care and good outcomes for ACDF. Whether or not this holds true for more complex procedures remains to be seen.
Please read Dr. Singh's paper on this topic in the October 15 issue. Does this change how you consider the learning curve for ACDF? Let us know by leaving a comment on The Spine Blog.
Adam Pearson, MD, MS
Associate Web Editor
Friday, October 7, 2016
The pros and cons of anterior and posterior cervical surgery have been well-studied, yet debate persists surrounding the best surgical approach for degenerative cervical problems. Anterior procedures are associated with a lower infection rate and less post-operative pain yet have a higher rate of dysphagia and pseudarthrosis compared to posterior procedures. Additionally, posterior procedures are contra-indicated in kyphosis, and anterior cases are generally limited to three levels due to high pseudarthrosis rates when performed at more than three levels. Dr. Park and colleagues from Korea wanted to obtain more reliable data on reoperation rates following anterior and posterior procedures. Given that reoperation is a relatively rare event, a very large cohort is needed to study this topic, so they elected to use the Korean Health Insurance Review and Assessment Service national database that captures almost all spine surgery performed in Korea. They identified over 9,000 cervical spine procedures performed for degenerative cervical pathology in 2009 and followed all patients through 2014, providing all patients with at least 4.5 years of follow-up. They categorized patients as having anterior surgery (ACDF or corpectomy), laminectomy with posterior fusion (LF), or laminoplasty (LP). Patients undergoing laminectomy without fusion were apparently excluded, though this is not discussed. Approximtely 90% (n=8143) underwent an anterior procedure, with about 6% (n=537) undergoing LF and 4% (n=391) LP. The patients undergoing posterior procedures were older, included a higher proportion of men, had a greater comorbidity burden, and were more likely to undergo multilevel surgery. The overall reoperation rate was 3.3%, yet there were significant differences for the three groups: 2.5% anterior, 12.5% LF, and 7.9% LP. After controlling for potential confounders (age, sex, diabetes, osteoporosis, number of levels, hospital type, and Charlson comorbidity index), the results were similar, with the LF group having a hazard ratio for reoperation of 4.7 and the LP patients 2.4 as compared to the anterior group. The authors also did separate analyses for reoperations within and beyond 90 days from the index surgery and found similar results, though the increase in reoperation rate in the early time frame for the LP group compared to the anterior group was not significant. The authors also published the survival curves, which demonstrated the LF group had a 2.5% reoperation rate on the day of admission, then a steady reoperation rate similar to the LP group out to 3 years, followed by the LP group having a lower reoperation rate from 3-5 years. The anterior group had the lowest reoperation rate across all time periods.
The authors have done a nice job using a national database that captures essentially all spine surgery in Korea, which allowed them to have sufficient numbers to make comparisons of reoperation rates among anterior surgery, LP, and LF. Even when using all cases in Korea for one year, the number of LF patients (n=537) and LP patients (n=391) was relatively low for the purposes of studying an uncommon event like reoperation. There are likely a variety of factors contributing to the markedly higher reoperation rates for the posterior procedures. In the short term, the infection rate is known to be substantially higher for posterior cervical surgery. Additionally, the LF and LP groups were likely at greater risk for infection given their increased comorbidity burden (though the difference persisted for the LF group even after controlling for these factors). The longer term persistence of the higher reoperation rate is somewhat harder to explain, as one would imagine that most of the reoperations occurring beyond a year out from surgery would be for adjacent segment degeneration, which should be similar for the two cohorts. Some of the later reoperations could be for nonunion, which, if anything, would likely be higher for the anterior group. Reoperations in the LP group in the later period were likely for persistent neck pain or kyphosis, and it is interesting that this rate is so similar to the reoperation rate for the LF group from 1-3 years given the likely different indications for reoperation in the two groups. Determining the indication for reoperation would have made this a much more informative paper, though such data were apparently not available in the database. Assuming similar clinical outcomes for the anterior fusion, LF, and LP groups (which earlier literature suggests), this paper provides relatively strong data supporting the use of anterior fusion techniques when possible. It seems likely that an anterior fusion is favored for up to three level disease, with posterior procedures indicated for pathology affecting four or more levels.
Please read Dr. Park's article on this topic in the October 1 issue and the accompanying commentary on The Spine Blog. Does this change how you consider anterior vs. posterior cervical surgery? Let us know by leaving a comment on The Spine Blog.
Adam Pearson, MD, MS
Associate Web Editor
Friday, September 30, 2016
Steroids are immunosuppressant medications that could theoretically increase the risk of surgical site infection (SSI). As such, there is concern that patients treated with a lumbar epidural steroid injection (LESI) could be at higher risk for SSI following lumbar surgery. Prior studies on the topic have mixed results, but some have suggested that steroid injections could be a risk factor for infection in both total joint arthroplasty and lumbar spine surgery. 1,2 As such, Hartveldt and colleagues from Boston elected to retrospectively review over 5,000 elective lumbar spine surgery cases (laminectomy and/or fusion) to determine if there was an association between LESI within 90 days of surgery and SSI. They found an overall 90-day infection rate of 2.5%, with a rate of 2.0% for the LESI group (19/945) and 2.6% rate for the non-LESI group (n=115/4,366). The LESI group was somewhat younger, had a slightly higher comorbidity index, and underwent smaller magnitude surgery, so a multivariate analysis was performed to control for potential confounders. This yielded an odds ratio of 0.85, indicating that LESI did not increase the rate of SSI in this cohort. Stratifying the analysis by duration between LESI and surgery and by number of injections yielded similar results.
This is a well-done retrospective study that found no association between LESI and SSI, regardless of the timing of injection. While the Medicare database study by Yang et al. did suggest an association between LESI and SSI, this study was subject to the limitations inherent in database studies, namely that controlling for confounders can be difficult.2 The current study can possibly be criticized for being underpowered, but it includes over 5,000 patients and did not suggest even a trend towards LESI patients having an elevated rate of SSI. The authors also noted that they may not have captured the LESIs performed out of their hospital system, which may have resulted in the misclassification of some LESI patients as non-LESI. This would tend to bias the results towards the null, though they did audit 100 charts and found an 11% rate of LESI that was not captured by the billing data. This study suggests that LESI does not markedly increase the SSI rate following lumbar surgery. Given the contradictory findings in the Yang Medicare database study, the jury remains out on this topic. That study showed the highest risk of infection was for surgery performed within 30 days of LESI (OR=3.2), so it might be prudent to try to avoid surgery in that time frame when possible. However, the current study does not show an increased risk, even in the short LESI to surgery time frame, which should reduce some surgeon anxiety about operating shortly after LESI when necessary. This is a difficult topic to study given the low rate of SSI and the fact that it is not possible to randomize patients to LESI or no LESI in the days leading up to surgery.
Please read the article on this topic in the October 1 issue. Does this change how you consider the timing of lumbar surgery following LESI? Let us know by leaving a comment on The Spine Blog.
Adam Pearson, MD, MS
Associate Web Editor
1. Marsland D, Mumith A, Barlow IW. Systematic review: the safety of intra-articular corticosteroid injection prior to total knee arthroplasty. Knee 2014;21:6-11.
2. Yang S, Werner BC, Cancienne JM, et al. Preoperative epidural injections are associated with increased risk of infection after single-level lumbar decompression. Spine J 2016;16:191-6.
Friday, September 23, 2016
The degree of improvement following surgery for cervical spondylotic myelopathy (CSM) can vary substantially, with most patients experiencing partial recovery of function following surgery. Many studies have evaluated surgical outcome predictors, and a recent meta-analysis concluded that a longer duration of symptoms and more severe baseline symptoms predicted worse post-operative function.1 Diabetes has been evaluated as a risk factor for poor outcomes following surgery for CSM, but it has generally not had a strong association with outcomes.2 Given the unclear association between diabetes and CSM outcomes, Kusin et al. retrospectively analyzed the improvement in Nurick scores in 113 CSM patients treated surgically, 33 of whom had diabetes. Additionally, they stratified the diabetic patients based on their average perioperative glucose levels. They found that diabetic and non-diabetic patients had similar baseline Nurick scores of approximately 2.5, a score indicating the presence of a gait abnormality but no need for an assistive device in the average patient in the study. The diabetic patients improved significantly less on the Nurick score (1.1 vs. 1.6 points), and there was a strong inverse relationship between perioperative glucose level and Nurick score improvement. Stratifying the diabetes patients into groups with average perioperative glucose levels of above and below 150 mg/dL revealed that patients with poor perioperative glucose control had minimal improvement (0.5 points), while those with good control improved to a similar degree as non-diabetics (1.9 points). Multivariate analysis revealed that diabetics and smokers had less improvement, while worse baseline symptoms and a greater number of levels fused predicted greater improvement.
This study adds some important information to the CSM literature, namely that the level of glucose control in diabetics may affect outcomes. Prior studies that classified diabetes as a dichotomous variable may have missed an association between diabetes and outcomes if many of the patients had well-controlled diabetes and responded like non-diabetics. While the level of glucose control may affect CSM outcomes, the limitations of this study need to be considered before drawing strong conclusions. The authors used average perioperative glucose level as a surrogate for diabetes control, and hemoglobin A1c would have provided a much better picture of glucose control over a longer period of time. Additionally, the only outcome was the Nurick score, a relatively blunt instrument that may not detect subtle improvements in function. The JOA score is a more responsive item that would have been helpful to include. The study also included a relatively small number of diabetic patients and included the experience of a single surgeon. Its retrospective nature also limits the strength of the conclusions that can be drawn. Despite these limitations, the paper certainly raises the possibility that poorly controlled diabetics may have less improvement following surgery for CSM. Future studies will need to determine if better perioperative glucose control leads to better outcomes. A key question in CSM research at this point is determining if there are subgroups that do not improve more with surgery than with non-operative care. Given that CSM has a generally poor prognosis if treated without surgery, there is little data available on non-operative outcomes. However, there may be subgroups—i.e. poorly controlled diabetics, patients with a very long duration of symptoms, and patients with mild myelopathy—who don't benefit much from surgery. A long-term observational study that follows patients treated both with surgery and non-operative care would be required to answer these questions, and it is not clear if such a study will ever be performed.
Please read this article in the September 15 issue. Does this change how you consider treating diabetics with CSM? Let us know by leaving a comment on The Spine Blog.
Adam Pearson, MD, MS
Associate Web Editor
1. Tetreault LA, Karpova A, Fehlings MG. Predictors of outcome in patients with degenerative cervical spondylotic myelopathy undergoing surgical treatment: results of a systematic review. Eur Spine J 2015;24 Suppl 2:236-51.
2. Tetreault LA, Kopjar B, Vaccaro A, et al. A clinical prediction model to determine outcomes in patients with cervical spondylotic myelopathy undergoing surgical treatment: data from the prospective, multi-center AOSpine North America study. J Bone Joint Surg Am 2013;95:1659-66.