The purpose of the project is to perform an RCT comparing patient satisfaction and outcome with or without the use of an expert panel. The purpose is also to create a registry to compare the effectiveness of decompression alone versus decompression with fusion for patients with degenerative grade I spondylolisthesis and symptomatic lumbar spinal stenosis. Primary analysis will focus on the patients' improvement from baseline patient-reported outcome questionnaires.

In addition, the SLIP II registry aims to (i) develop an algorithm which could identify cases in which surgical experts are likely to recommend one treatment (i.e. >80% of experts recommend one form of treatment) and (ii) develop a radiology-based machine learning algorithm that would prospectively classify patients as either 'stable' or 'unstable.'

In addition to patient reported outcomes, step counts will be collected in order to determine the correlation of step count with patient-reported outcomes (ODI and EQ-5D) and the need for re-operation.

This registry portion of the study aims to prospectively collect comparative data for these patients treated with either decompression alone or decompression with fusion.

Aim: To conduct a randomized control trial comparing patient outcomes and satisfaction with or without expert panel review before making a final decision about surgery for grade I degenerative lumbar spondylolisthesis. A prospective, multi-center registry aimed at addressing important issues pertaining to outcomes from treatment for degenerative spondylolisthesis and spinal stenosis will also be generated.

Background: Surgery may be offered to patients with symptomatic lumbar stenosis with degenerative lumbar spondylolisthesis who fail nonoperative treatment measures including physical therapy and epidural steroid injections. For patients with lumbar stenosis without spondylolisthesis, a decompression alone is typical, while those patients who do have degenerative spondylolisthesis and who also have significant mechanical back pain may be offered lumbar decompression with or without fusion. These guidelines were written based upon the SPORT study, which provided the highest quality of evidence available at the time. Additional studies have show that costly interventions such as lumbar fusion may ultimately be cost-effective if they provide durable clinical benefit. Two recent publications in The New England Journal of Medicine present new evidence with conflicting results on superficial review. The Spinal Laminectomy versus Instrumented Pedicle Screw (SLIP) trial provides level I evidence for the efficacy of fusion to improve clinical outcomes and lower reoperation rates compared to a standard laminectomy and medial facetectomy over a four year time frame in patients with neurogenic claudication associated with stable single level spondylolisthesis. Conversely, the Swedish study provides level II evidence that the addition of a variety of fusion techniques does not have significant benefit in the first two years following operation compared to a variety of decompression techniques in a heterogeneous population of patients with stenosis associated with spondylolisthesis. The patient populations treated, surgical techniques used, and outcome measures assessed differed between the two studies and when taken together, underline the need to new comparative effectiveness data for patients with this problem.

Additionally, one key challenge surgeons face is whether or not to recommend a spinal fusion. Spinal fusion is expensive, assoicated with greater costs and complications, but it appears to be necessary in at least 30% of patients.3 Preliminary data suggests that when when greater than 80% of an expert panel votes for one treatment opition, either a fusion or decompression alone, and when a patient's actual treatment aligns with the expert panel recommention, the patient-reported outcomes are greater than when the surgical approach is not aligned with the expert panel.10 This data highlights an interest in developing articifical intelligence (AI) that may be able to aid in both identification and predictive tasks. Any progress in this realm would be enormously powerful from a clinical standpoint and would likely result in more efficient use of surgical appraoches and in turn, healthcare spending.

All images that are captured in the registry will be used to train convolutional neural networks (CNN). These are mathematical operations which extract patterns from image data and generalize it across many images fed into the dataset. They primarily use calculations to extract patterns which are stored as a model which will be a collection of numbers. The images stored in this registry will be used to develop algorithms to assess cases in which an expert panel is more likely to suggest one treatment over the other as well as develop an algorithm that would prospectively classify patients as either 'stable' or 'unstable.'

Plan: Before making a decision regarding which specific operation should be performed in each case, each patient will be randomized to receive an expert panel review or to not receive an expert panel review. For patients who receive an expert panel review, the patients' de-identified lumbar MRI (sagittal and key axial images), 36-inch standing plain radiographs (if available), and flexion and extension radiographs will be uploaded into a web-based platform and reviewed with plans to share the reviews with patients and their treating physicians in real time. For patients who are randomized to no expert panel review, they will discuss with their surgeon the best surgical option for them and proceed as they would in standard of care. Patients with symptomatic lumbar spinal stenosis and single level degenerative grade I spondylolisthesis will be treated either with decompression or decompression with fusion. Symptomatic spinal stenosis will be defined as radicular and/or back pain either induced by or aggravated by activity and relieved by rest in a patient with either moderately severe or severe lumbar spinal stenosis. Patient-reported outcomes will be captured at baseline, at 3 and 6 months, and annually out to five years.

The imaging data will be used to create artificial intelligence (AI) algorithms that will help assess when an expert panel is more likely to suggest one treatment over the other as well as develop an algorithm that would prospectively classify patients as either 'stable' or 'unstable.' Ultimately long term follow-up will help confirm whether a case was correctly assessed as stable or unstable. A patient would be confirmed as unstable, if they underwent decompression alone and then required a re-operation to stabilize the spine at the level of spondylolisthesis within 5 years of the initial operation. In a similar way, a patient would be confirmed as stable, if no re-operation were necessary over the 5-year study follow-up period.

Select sites will participate in an assessment of the utilization of step count as an outcome. Average step count will be captured pre-operatively as well as at 3- and 6-months and annually out to 5 years. The mean step count at each time point will be compared to the mean change scores for ODI and EQ-5D. Additionally, average step counts overtime will be analyzed for those patients who undergo a re-operation.

Interim Analysis: An interim analysis is planned when 150 patients have reached eligibility at 6 month follow-up. Patients' change from baseline patient-reported outcome questionnaires will be assessed.

Inclusion Criteria:

• Symptomatic lumbar spinal stenosis: defined as radicular and/or back pain either induced or aggravated by activity and relieved by rest.
• Single-level grade I degenerative spondylolisthesis (3-14mm)

Exclusion Criteria:

• Serious medical illness (ASA Class IV or higher)
• Spondylolysis
• Multilevel spondylolisthesis or high grade spondylolisthesis (grade II or greater than 14mm)

Deidentified data would be shared with other researchers after primary analysis has been published.

After primary analysis has been published for up to 5 years.