Clinical Prediction Model for Surgical Outcome in Cervical Spondylotic Myelopathy Found Valid
A clinical predictive rule can help surgeons determine whether an individual patient will benefit from spine surgery to treat cervical spondylotic myelopathy (CSM). Such rules are valuable tools that can quantify a patient’s likely outcome based on objective evidence. However, any model based on data from one geographic location, and culture should be tested internationally to ensure that it is valid in other populations.
Cervical Spondylotic Myelopathy
CSM is the leading cause of spinal cord dysfunction worldwide. It is progressive and an estimated 20% to 60% of patients with symptoms will deteriorate over time without surgical intervention. Traditionally, the role for surgery in CSM has been to arrest clinical progression and stop further disability from occurring. However, it remains a clinical challenge predict accurately which patients will benefit the most from surgical intervention.
In 2013, Tetrault et al1 published a study evaluating a clinical predictive rule to determine the surgical outcome for patients with CSM. This rule was based on data from a prospective study of 278 patients in North America. Their purpose in creating the rule was to discriminate between patients with mild myelopathy (modified Japanese Orthopaedic Association [mJOA] assessment of 16 or higher) and those with substantial residual neurological impairment (mJOA assessment of less than16) at one year after surgery.
Now, this rule has been evaluated for internal and external validity based on results from a prospective study of 479 patients at 16 international locations.2
The performance of a model or rule in another population or a different geographic area may be poorer than the original model due to potential differences in patient characteristics, healthcare access/systems, or methods of measurement, the authors noted. Although the criteria for patients were the same for both the North American and international studies, there could have been differences in factors such as the underlying etiology for myelopathy or the frequency of comorbid states.
In most respects, there were no differences between the North American group of patients on which the rule was based and the international group. However, one major difference is that in the international group, there were fewer patients with a record of co-existing psychiatric illness. Potentially this could be due to under-reporting of depression, bipolar disease, or other psychiatric issues in some countries or cultures rather than to any actual difference in prevalence.
The predictive performance (the area under the receiver operating curve [AUC]) of the model based on the North American data was 0.77. Based on the international data, the predictive performance was 0.74. This difference was considered to be minimal by the authors. The AUC for the validated model demonstrates that it can be useful in determining which patients will most benefit from surgical intervention.
“The validation of this model will allow it to be implemented in clinical practice, enabling surgeons to accurately quantify outcome, manage their patient’s expectations, and ultimately improve care,” according to the authors of the study, which was published in The Spine Journal.
“Surgeons often have different perceptions as to which individuals will have a good prognosis, and a prediction model will aid in the alignment of these perceptions with more objective evidence,” the authors stated.
The clinical prediction rule for CSM is based on one imaging predictor and six clinical predictors. The clinical predictors include age, duration of symptoms, baseline severity score according to the mJOA assessment, smoking status, absence of coexisting psychiatric issues such as depression or bipolar disease, and the presence of impaired gait.