Validation of Appropriateness Criteria for the Surgical Treatment of Lumbar Degenerative Spondylolisthesis

Spinal stenosis is the most common reason for lumbar surgery in patients over age 65, and a substantial proportion of stenosis patients present with concomitant lumbar degenerative spondylolisthesis (LDS). This condition is characterized by the anterior translation of a vertebra due to degenerative changes in the facet joints and disk, most commonly at L4-L5, and is most prevalent in women over age 50. The translation commonly results in spinal stenosis, and patients tend to present with various combinations of low back pain, radiculopathy, and neurogenic claudication. The Spine Patient Outcomes Research Trial (SPORT) demonstrated better outcomes for patients treated with surgery compared to non-operative treatment. However, all patients met strict inclusion criteria including the presence of radicular pain or neurogenic claudication, cross-sectional imaging demonstrating spinal stenosis consistent with their symptoms, and duration of symptoms for over three months. Additionally, specific patient and disease characteristics were shown to significantly affect surgical and non-operative outcomes. While the literature makes it clear that surgery results in better outcomes than non-operative treatment for the "average" LDS patient, surgeons have to help make treatment decisions for patients with a diverse array of characteristics that can have positive and negative effects on surgical outcomes. Additionally, many LDS patients do not meet the inclusion criteria of SPORT, and extrapolating the results to these patients is inappropriate. The currently available evidence provides surgeons with some general principles to help guide decision-making but does not inform decision-making at the individual patient level.

Many studies have evaluated the effect of specific patient and disease characteristics on surgical outcomes for spinal stenosis. These have generally shown worse surgical outcomes for patients with medical and psychological comorbidities, lower socioeconomic status and educational attainment, smokers, those with predominant back pain, those receiving workers' compensation, and patients involved in litigation. However, subgroup analysis of the SPORT degenerative spondylolisthesis cohort, the first study that evaluated predictors of both surgical and non-operative LDS outcomes, demonstrated that these factors tended to predict both worse surgical and non-operative outcomes. As a result, most of these subgroups tended to have a similar treatment effect of surgery (i.e. relative advantage of surgical compared to non-operative outcomes) even if their absolute degree of improvement with surgery was less. While these data suggest that most LDS patients meeting the SPORT inclusion criteria will improve more with surgery than with non-operative treatment, many LDS patients do not meet the SPORT inclusion criteria, and evidence is needed to provide treatment guidelines for these patients.

In an ideal world, randomized-controlled trials (RCT) would be performed for multiple subgroups in order to compare outcomes among all available treatment options. The SPORT LDS RCT demonstrated the difficulty in performing a seemingly straightforward RCT comparing surgery to non-operative treatment, with only 64% of patients assigned to surgery having undergone an operation within 2 years, and 49% of patients randomized to non-operative treatment having undergone surgery by 2 years. Given that it is not possible to perform sufficiently powered RCTs to compare surgical and non-operative outcomes for multiple LDS subgroups, other methods must be considered to generate guidelines to help surgeons and LDS patients select the most appropriate treatment based on their individual characteristics. One such alternative is the RAND appropriateness method (RAM), which combines a detailed review of the literature with a modified Delphi panel approach to create appropriateness criteria based on collective expert opinion. A major strength of the RAM is that is involves a high level of clinical detail, allowing for treatment recommendations to be made based on a number of relevant patient and disease characteristics that can be applied to individual patients. The RAM has been widely used and studied by several groups in various countries in relation to different spinal problems. The RAM is considered most useful for procedures that are used frequently, associated with a substantial amount of morbidity and/or mortality, consume significant resources, have wide variation among geographic areas in rates of use, and whose use is controversial. All these criteria apply to the procedures commonly used in the surgical treatment of LDS. Depending on the definition of success, success rates at 1 year after surgery range from 50% to 94%. The rate of lumbar fusion in the United States Medicare population has been shown to vary nearly 20-fold across hospital referral regions. In the SPORT LDS cohort, significant differences were found across different study centers in patients' baseline neurological deficit, stenosis location and severity, and number of stenotic levels as well as in their functional outcome out to 4 years after surgery. High levels of geographic variation are frequently attributed to uncertainty about the most appropriate treatment for a given condition, usually reflecting a lack of strong evidence. Given the lack of treatment recommendations that can be applied at the individual LDS patient level, an international group of experts were convened to use the RAM to create appropriateness criteria for surgery for LDS.3

The RAM process included 14 experts from the USA, UK, Belgium, Sweden, Norway, Spain, Switzerland, and Hungary with backgrounds in orthopaedic surgery, neurosurgery, rheumatology, and physical medicine. The experts considered the appropriateness of decompression alone, uninstrumented fusion with or without decompression, and instrumented fusion with or without decompression for 744 patient scenarios. The variables defining the scenarios included symptomatology (back pain, radicular pain, and/or neurogenic claudication), severity of neurological deficit, presence of significant foraminal and/or central stenosis, level of disability, presence of significant medical comorbidities, clinically relevant "instability", and the presence of significant psychosocial "yellow flags". The appropriateness of each type of surgery was rated on a 1-9 scale (1-3 inappropriate, 4-6 uncertain, and 7-9 appropriate) in two separate rounds by the expert panel. Surgery of one type or another was considered appropriate in 27% of scenarios, uncertain in 41%, and inappropriate in 31%. In general, scenarios in which symptoms were limited to back pain in the absence of instability were considered inappropriate for surgery, whereas the presence of radiculopathy or claudication, neurologic deficit, and instability increased the likelihood of a scenario being considered appropriate for surgery. The presence of medical comorbidities and "yellow flags" decreased the likelihood of a scenario being considered appropriate for surgery. There was widespread disagreement about the appropriate use of instrumentation, but, in general, fusion of some variety was more likely to be considered appropriate in the presence of back pain or instability.

The RAM was successfully employed to develop appropriateness criteria for LDS surgery. Prior to widespread clinical implementation, the results need to be validated in the LDS population. If validated successfully, a computer-based algorithm could be developed to help surgeons and patients select treatment using the appropriateness criteria. This would represent a significant advancement in shared decision making for LDS, effectively moving from population-based recommendations to those that can be applied at the individual patient level.