Outcome predictors for sacroiliac joint (lateral branch) radiofrequency denervation.

BACKGROUND AND OBJECTIVE Sacroiliac (SI) joint pain is a challenging condition characterized by limited treatment options. Recently, numerous studies have reported excellent intermediate-term outcomes after lateral-branch radiofrequency (RF) denervation, but these studies are characterized by wide variability in technique, selection criteria, and patient characteristics. The purpose of this study was to determine whether any demographic or clinical variables can be used to predict SI joint RF denervation outcome. METHODS Seventy-seven patients with refractory, injection-confirmed SI joint pain underwent SI joint denervation at 2 academic institutions. A composite binary variable "successful" outcome was predefined as greater than 50% reduction in pain lasting at least 6 months coupled with a positive global perceived effect. Secondary outcome measures included Oswestry Disability Index scores, medication reduction, and retention on active duty for soldiers. Factors retrospectively evaluated for their association with outcome included demographic variables, duration of pain, opioid usage, pain referral pattern, physical examination signs, number of blocks and percentage of pain relief after SI joint injection, prognostic lateral-branch blocks, previous surgery, levels lesioned, RF technique, disability status, and coexisting medical conditions. RESULTS Forty patients (52%) obtained a positive outcome. In multivariate analysis, preprocedure pain intensity, age older than 65 years, and pain radiating below the knee were significant predictors of failure. A trend was noted whereby patients receiving regular opioid therapy were more likely to experience a negative outcome. The use of cooled, rather than conventional RF, was associated with a higher percentage of positive outcomes. CONCLUSIONS Whereas several factors were found to influence outcome, no single clinical variable reliably predicted treatment results. The use of more stringent selection criteria was not associated with better outcomes.