Predictors of the efficacy of epidural steroid injections for structural lumbar degenerative pathology

Background: Lumbar epidural steroid injection (LESI) is a valuable therapeutic option when administered to the appropriate patient, for the appropriate disease process, at the appropriate time. There is considerable variability in patient-reported outcomes (PROs) after LESI, creating uncertainty as to who will benefit from the therapy and who will not.

Purpose: We set out to identify patient attributes, which are important predictors for the achievement of a minimum clinically important difference (MCID) in the Oswestry Disability Index (ODI) after LESI.

Study design: A prospective cohort study was carried out.

Patient sample: A total of 239 consecutive patients undergoing LESI for back-related disability, back pain (BP), and leg pain (LP) associated with degenerative pathology comprised the patient sample.

Outcome measures: Baseline and 3-month patient self-reported ODI, numeric rating scale-BP and LP, Euro-Qol-5D, and Short Form (SF)-12 scores were recorded.

Methods: A total of 239 consecutive patients undergoing LESI for degenerative pathology over a period of 2 years who were enrolled into a prospective web-based registry were included in the study. Using the previously reported anchor-based approach, an MCID threshold of 7.1% was established for ODI after LESI. Each enrolled patient was then dichotomized as a "responder" (achieving MCID) or a "non-responder." Multiple logistic regression analysis was then performed, with the achievement of MCID serving as the outcome of interest. Candidate variables included in the regression analyses were age, gender, employment, insurance type, smoking status, preoperative ambulation, preinjection narcotic use, comorbidities, predominant LP or BP symptoms, symptom duration, diagnosis, number of levels, prior surgery, baseline PROs, type of stenosis (central, lateral recesses, or foraminal), injection route (transforaminal, interlaminar, or caudal), and number of injections. Subsequently, we also randomly selected 80% of the patients to serve as the training data for a multiple logistic regression model. Once this predictive model was built, it was validated using the remaining 20% of patients.

Results: There were 124 (62%) patients who achieved MCID for ODI. The existence of central stenosis (p=.006), TF or IL injection route (p=.02) compared with caudal epidural steroid injection, higher baseline ODI (p=.00001), and a diagnosis of disc herniation (p=.02) increase the odds of achieving MCID for ODI at 3 months. Symptom duration for over a year (p=.006), prior surgery (p=.08), and preinjection anxiety (p=.001) decrease the odds of achieving MCID. The area under the curve (AUC) for our predictive model's receiver-operator characteristic was 0.81 when using the 80% training data set, and the AUC was 0.72 when using the 20% validation data.

Conclusion: We have identified patient attributes that are important predictors for the achievement of MCID in ODI 3 months after LESI. The use of these attributes, in the form of a predictive model for LESI efficacy, has the potential to improve decision making around LESI. Spine care providers can use the information to gain insight into the likelihood that a particular patient will experience a meaningful benefit from LESI.