University of Chicago Neurosurgery Journal Club
B Roitberg, F Brown, S Lam
M. Stamates; J Khader-Eliyas; A Bhansali, N Mansour, S Shakur, I Takagi.
Wang, et al. Trends and variations in cervical spine surgery in the United States, Medicare beneficiaries, 1992 to 2005. Spine. 2009, 34 (9), 955-961.
Rates of surgery in the Medicare population have almost doubled over a decade (1990-2000). The purpose of this article was to examine surgical trends, particularly when they have had higher reported postoperative complications. This was a retrospective cohort study reviewing ICD-9 codes reported on Medicare claims. Patient were all 65 years of age or older and not enrolled in disability. Cervical fusion was the most common procedure, with almost 2/3 through an anterior approach. Most notable was the association between location and fusion rates, which were higher in the Northwest and South Central regions, with Idaho Medicare claims totaling 140 cervical fusions per 100,000 Medicare beneficiaries – the highest in the study.
The increase of cervical fusions in Medicare beneficiaries similarly paralleled an increase in lumbar fusions. The authors suspect that rates of surgery may be increasing disproportionately in elderly adults when compared to all adults undergoing spine surgery. The authors note the difficulty in drawing conclusions from this data set as the “right” procedure for cervical disease, the surgical indications and the severity of disease have yet to be determined. This study also calls for further research into patient outcomes, cost-effectiveness studies and long-term follow-up to delineate the history of cervical spondylosis. Defining the surgical indications should lead to better efficacy when treating degenerative spine disease.
Medicare trends in cervical spine surgery are important to note and document. The information is culled from Medicare billing so the granularity of clinical details is not available, but the power of N=156,820 is impressive.
Regional variations in almost all clinical practices have been noted for over 20 years in the Dartmouth Atlas, and has been highlighted for spine surgery (http://www.dartmouthatlas.org/downloads/reports/Spine_Surgery_2006.pdf). Regional variation can be attributed to many factors, including variations in medical training, differences in financial incentives, access to resources, differences in patient populations, and density of providers. The factors behind the variations in cervical spine surgery need to be studied further especially as surgical indications and expected benchmarked outcomes have yet to be clearly defined.
As a 206% increase in rates of cervical spine surgery is noted between 1992 and 2005, further studies should examine contributors to this increase such as number of spine surgeons, changes in technology, changes in safety and efficacy, and trends in reimbursement, demographic changes, and trends in the legal environment.
Patient outcomes, cost-effectiveness of treatment, and comparative effectiveness need to be examined in cervical spine surgery. This Medicare trends analysis is a commendable and important first step. Use of population-based databases is an invaluable tool in achieving a reflection and examination of how health care is actually being delivered in the US, and a first-step in cost-containment and development of rational. consensus-based, and evidence-based care.
This Wang et al article is important to review, because it is an early attempt by a group of neurosurgeons to look at data that everyone else is paying attention to. The government and the insurers are trying to limit spending on spine surgery. Regional variability is an indicator in the minds of some, that unnecessary operations are being done. In other words – if someone in Maine does not need the procedure, why would someone in Idaho need it? In the absence of real outcome studies, and a real determination of what treatment is optimal for whom, such assumptions are premature. This article calls for further research into patient outcomes. I agree wholeheartedly. We are keeping a prospective database here at the University of Chicago; it is blinded to the surgeon, providing an extra measure of objectivity. There is also a new database coming from Neuropoint Alliance – sponsored by organized Neurosurgery; it is called N2QOD (National Neurosurgery Quality and Outcomes Database).
Wang, et al. Complications and mortality associated with cervical spine surgery for degenerative disease in the United States. Spine. 2007, 32 (3), 342-347.
A retrospective cohort analyzing the 932,009 cervical spine surgery discharges from the Nationwide Inpatient Sample (1992-2001). Despite an increase in the reporting of comorbid conditions in degenerative c-spine patient, the reported mortality & surgical complications has remained the same. The purpose of the study was to look for any associations between age, diagnosis or procedure and related complications. This was accomplished by viewing the ICD-9 codes used for each patient discharge. The main diagnoses were herniated disc (722.0), cervical spondylosis with myelopathy (721.1 & 722.71), cervical spondylosis without myelopathy (721.0, 722.4, 723.7) and spinal stenosis (723.0). Complications had to be recorded with an ICD-9 code in the discharge in order to be recognized; the authors note the likelihood of major under-reporting of all (major or minor) complications.
The inpatient group ranged in age from 20-75+, with the most frequent diagnosis (56%) being herniated disc. Associations found in the study included: an increase in the complication rate in patients with the diagnosis of cervical spondylosis with myelopathy and with the following techniques: posterior fusion, combined anterior & posterior fusion. These associations also held true when describing mortality. All of these associations proved significant (p<0.05) with the exception of mortality with a posterior or combined approach. It is important to note that cervical spondylosis with myelopathy, the second most common cervical spine diagnosis after herniated disc was also found in higher proportion in older adults; with herniated disc being more popular in the under-50 subset.
In analyzing reported complications, the authors note the shortcomings of using the National Inpatient Sample, which only recognized 15 ICD-9 codes & may not have recognized reported complications that happened not to fit one of the accepted codes.
The Nationwide Inpatient Sample in the 10 year period of 1992 to 2001 was examined. This may not entirely reflect how cervical spine surgery is currently practiced since the time period studied was more than a decade ago, but it is very important to examine where our profession has been and how we have been faring in a country-wide assessment. While drawing from ICD-9 CM codes sacrifices clinical granularity, N=932,009 of hospital discharges associated with degenerative cervical spine diagnoses is a rich and powerful sample size for analyzing a large number of cases over multiple years. Limitations such as inability to track readmissions, severity of clinical condition, information about surgical practice volume or surgeon experience are indeed pertinent and important to note. Using this large nationally-representative database, identification of trends and rates of rare events are most salient. The authors have identified associations with increased morbidity and mortality, which they have framed appropriately and insightfully in relation to the rest of the literature. While this type of study does not substitute for well-designed clinical studies to track patient outcomes, it does effectively identify trends, complications, and rare events, and as such is important to take note. Large-scale population based studies are increasingly important in this era of heightened focus on cost-efficiency and patient-centered care so that we can ask the right questions and design clinical trials better.
In general this article is useful in order to establish a lower limit to the actual incidence of complications with cervical spine surgery. In the absence of systematic prospective and independent audits of surgical complications, they remain underreported, and thus the true balance of risks and benefits for the various procedures remains unknown. The existing reports of surgical results do provide numbers, yet it is difficult to avoid reporting bias, or to generalize the results. This study provides a certain reference point.
Ghogawala Z, Martin B, Benzel EC, Dziura J, Magge SN, Abbed KM, Bisson EF, Shahid J, Coumans JV, Choudhri TF, Steinmetz MP, Krishnaney AA, King JT Jr, Butler WE, Barker FG 2nd, Heary RF. Comparative effectiveness of ventral vs dorsal surgery for cervical spondylotic myelopathy. Neurosurgery 68(3) 622-631, 2011.
Cervical Spondylotic Myelopathy (CSM) is a very common problem - about 20% of cervical spine surgery is performed for CSM. This study was designed as a pilot project comparing outcomes of different surgical approaches for treatment of CSM before the authors could perform a randomized control trial on the same subject. It is a prospective non-randomized multicenter study conducted in a total of 7 centers (3 community hospitals and 4 tertiary referral centers).
The authors, for the purpose of this study, defined CSM as presence of 2 or more clinical features of myelopathy (clumsy hands, gait disturbance, hyper-reflexia, Babinski sign and bladder dysfunction). To be eligible for this study, myelopathic patients needed to have cord compression at 2 or more levels and equipoise for both dorsal and ventral approaches. Presence of kyphosis of more than 5 degrees, segmental kyphosis, ossified posterior longitudinal ligament, congenitally narrow canal, history of previous cervical spine surgery and ASA grade III or higher excluded patients from the study.
The question of equipoise was very important for this study. All eligible patients had MRI, CT and flexion-extension views of their cervical spine reviewed by all the participating surgeons (totally 12 in number) and 2 other spine experts and voted for equipoise. Patients with equipoise were then presented with both options by their surgeon before enrolling in the study. The authors decided on anterior cervical discectomy and fusion with instrumentation as the desired anterior approach and similarly laminectomy with lateral mass screw fixation as the desired posterior approach. The modified Japanese Orthopedic Association score (mJOA) and Neck disability Index (NDI) were selected for measuring the functional outcome of the surgery; Short Form – 36 (SF-36) and EuroQOL – 5D (EQ-5D) were the two Health Related – Quality Of Life (HR-QOL) measurement tools used for determining general well-being.
Over a period of 2 years, 50 patients with equipoise for both anterior and posterior approach to the cervical spine were enrolled. Ventral approach was used in 28 patients while dorsal approach was employed in 22 patients. The various outcome tools mentioned earlier were measured once pre-op and repeated 3, 6 and 12 months after the surgery. The authors also looked at complication rates, length of hospital stay and cost incurred from respective approaches. Patients who underwent dorsal approaches were found to stay longer in the hospital and also had more levels on an average treated, both of which were statistically significant. Rostral levels (C2-3, and C3-4) were mostly treated by dorsal surgery and most of these patients with rostral disease were older than rest of the group. Similar proportion of patients in either group had complications (around 16% overall); in the ventral cohort it was mainly swallowing problems while in the dorsal group it was C5 palsy.
Around 90% of patients in both cohorts had follow up information at 1 year. At this time both groups showed increase in HR-QOL measures but there was no significant change in EQ-5D. On the contrary SF-36 scores were higher in the ventral cohort, which was statistically significant when controlled for confounders. There was more neck pain and disability in the dorsal group, which reflected in the NDI but was not statistically significant. mJOA scores showed no difference after correcting for the initial higher score in ventral group during pre-op. The authors also looked at direct cost of either treatment related to surgery and hospital stay, excluding ancillary cost such as surgeon’s fee, travel expenses and rehabilitation cost. Approaching the cervical spine dorsally was approximately 10 K costlier than doing so from the front but we need to remember that more levels were operated from the back than ventrally.
Overall, it is a well-conducted study where information was collected prospectively thus preventing bias associated with retrospective data gathering. The authors have endured to create a study group with equipoise for either treatment, which is commendable, but does equipoise really exist in CSM? Since the study was not randomized the dorsal cohort had more levels operated on an average and this influenced various outcome measures such as hospital stay, cost incurred, HR-QOL tools and NDI score. Unfortunately the authors were not able to enroll even 50% of their CSM patients during the same time period as the study. The number of subjects being small also does not power any of the results but the study’s objective, which was to gauge feasibility of a randomized control trial on surgical treatments of CSM, was achieved. Finally due to strict selection of patients, the results of the study cannot be generalized to the general population and again leads us to question the existence of equipoise in practical terms, in surgical treatments of CSM.
This study is carefully crafted and executed with the support of a lot of research resources - multiple grants, study surgeons, and clinical coordinators. Eligibility was carefully reviewed with group majority vote on equipoise. Follow-up rate was good, 92% at a year. However, enrollment was low (around 20% of cases encountered by the study sites), the dorsal group had a preexisting worse myelopathy and more operated levels. These inherent biases make internal and external validity questionable even though this study was well-designed to try to withstand the many difficulties with conducting clinical trials in surgery.
The economic analysis is limited in that the authors chose to look only at inpatient hospital charges, but given the lack of consensus in this type of analysis and given that physician behavior is a known major driver of health care costs, this is a very reasonable place to start.
HR-QOL measurements and follow-up are good here. This is a pilot/feasibility study, so the relatively short 1 year follow up is understandable. With an anticipated randomized controlled trial, it would be even better to see longer-term follow up with HR-QOL measurements to see how durable the results are and how the trends in QOL improvement play out in these cervical spine patients.
This groups’ capable and painstaking efforts at a prospective clinical pilot trial should be commended. A randomized controlled trial will be the next step, which will require a very large amount of expertise, time, money, manpower, and resources. Given the need for greater understanding of comparative effectiveness, especially for spine surgery with aging demographics in this country with increasing limitations on Medicaid/Medicare spending, a definitive RCT is urgently needed.
One of the most common problems in surgical studies is the difficulty in randomizing patients. Surgeons and patients tend to have an opinion regarding which treatment is “really” better, so compliance with randomization is low. SPORT studies we reviewed in a previous journal club suffered from crossover on the order of magnitude of 50%, essentially invalidating the study. Ghogawala et al are investing a major effort in agreeing before randomization that patients can indeed be randomized and both options are really considered equivalent in the mind of the surgeon – equipoise. Without equipoise a randomized surgical study is either not feasible (the surgeons will just do what they believe if best for the patient) or unethical (the surgeons will perform procedures they believe are not the best for their patients). The results of this particular study demonstrate how difficult it is to find equipoise. Moreover, like with any other randomized prospective trial – any attempt to make the population more uniform and exclude all kinds of outliers risks exclusion of the majority of real patients. This generates a problem with extrapolating from the small group in the study to the universe of real patients, the problem of external validity.
However, we have no choice but to keep working on improvement of our studies and analytical tools. Limited data, cautiously analyzed, is better than no data.