Cervical fusion for treatment of degenerative conditions: development of appropriate use criteria

BACKGROUND CONTEXT: High quality evidence is difficult to generate, leaving substantial knowledge gaps in the treatment of spinal conditions. Appropriate use criteria (AUC) are a means of determining appropriate recommendations when high quality evidence is lacking.

PURPOSE: Define appropriate use criteria (AUC) of cervical fusion for treatment of degenerative conditions of the cervical spine.

STUDY DESIGN/SETTING: Appropriate use criteria for cervical fusion were developed using the RAND/UCLA appropriateness methodology. Following development of clinical guidelines and scenario writing, a one-day workshop was held with a multidisciplinary group of 14 raters, all considered thought leaders in their respective fields, to determine final ratings for cervical fusion appropriateness for various clinical situations.

OUTCOME MEASURES: Final rating for cervical fusion recommendation as either "Appropriate," "Uncertain" or "Rarely Appropriate" based on the median final rating among the raters.

METHODS: Inclusion criteria for scenarios included patients aged 18 to 80 with degenerative conditions of the cervical spine. Key modifiers were defined and combined to develop a matrix of clinical scenarios. The median score among the raters was used to determine the final rating for each scenario. The final rating was compared between modifier levels. Spearman's rank correlation between each modifier and the final rating was determined. A multivariable ordinal regression model was fit to determine the adjusted odds of an "Appropriate" final rating while adjusting for radiographic diagnosis, number of levels and symptom type. Three decision trees were developed using decision tree classification models and variable importance for each tree was computed.

RESULTS: Of the 263 scenarios, 47 (17.9 %) were rated as rarely appropriate, 66 (25%) as uncertain and 150 (57%) were rated as appropriate. Symptom type was the modifier most strongly correlated with the final rating (adjusted ρ2 = 0.58, p<.01). A multivariable ordinal regression adjusting for symptom type, diagnosis, and number of levels and showed high discriminative ability (C statistic = 0.90) and the adjusted odds ratio (aOR) of receiving a final rating of "Appropriate" was highest for myelopathy (aOR, 7.1) and radiculopathy (aOR, 4.8). Three decision tree models showed that symptom type and radiographic diagnosis had the highest variable importance.

CONCLUSIONS: Appropriate use criteria for cervical fusion in the setting of cervical degenerative disorders were developed. Symptom type was most strongly correlated with final rating. Myelopathy or radiculopathy were most strongly associated with an "Appropriate" rating, while axial pain without stenosis was most associated with "Rarely Appropriate."