We retrospectively analyzed the medical records of 91 patients (30 males and 61 females; age, 17-79 years) with lumbar radicular pain who received a CESI and were hospitalized in the Department of Rehabilitation Medicine between December 2010 and July 2013. We investigated patients with HLD for whom physical therapy in the outpatient department or at other hospitals for >2 weeks was ineffective and who had an MRI <3 months prior to their CESI treatment. All patients had undergone MRI scans confirming the existence of lumbar disc-related pain (herniation) that was expected to improve following a CESI. A positive result on the Straight Leg Raise Test was not mandatory for a patient to be enrolled in the study, as it is well accepted that degenerative changes do not necessarily correlate with pain generation [6]. The exclusion criteria were 1) motor weakness due to radiculopathy or cauda equina injury, 2) no prior surgery, and 3) a medical history of steroid injection treatment within 6 months.

All CESIs and ultrasonography (US) examinations were performed by a physician with more than 10 years experience. US was provided by an Accuvix V10 (Samsung Medison, Seoul, Korea). After placing a pillow on the lower abdomen, patients were placed in a prone position for the CESI procedure. After positioning the US transducer on the transverse midline of the sacral hiatus, a US view of the sacral hiatus was identified. The diameter between two sacral cornua was measured. Next, the depth of the hiatus from the skin was measured through a longitudinal US view. A 23-gauge spinal needle was advanced while facing the bevel down at a 45° angle toward the sacral canal and positioned between the two cornua. Needle position was reconfirmed using the 'pop' or 'give' feeling when the needle penetrated a sacrococcygeal ligament. The needle was advanced and redirected cephalad, horizontal, and parallel for further advancement. A sufficient volume of fluid was injected consisting of interspersed 8 mL normal saline, 2 mL 10 mg dexamethasone, and 20 mL lidocaine. An inhalation test was used to verify that there was no blood.

CESI effectiveness was evaluated using the visual analogue scale (VAS) and the Roland Morris Disability Questionnaire (RMDQ). The VAS is a measurement instrument that assesses a characteristic or attitude that ranges across a continuum of values and cannot easily be measured directly [7]. A VAS is usually a horizontal line, 100 mm in length, anchored by word descriptors at each end [7]. The patient marks the point on the line that they feel represents their perception of their current state [7]. The VAS score is determined by measuring in mm from the left side of the line to the point marked by the patient [7]. The RMDQ is a self-administered outcome measure of disability where greater levels of disability are reflected by higher numbers on a 24-point scale [8,9]. Patients were asked to read a list of 24 sentences and place a tick against appropriate questions based on how they feel each sentence described them today [8,9]. If the sentence does not describe their symptoms, the patients were asked to leave the space next to the sentence blank [8,9]. 'Responder' was defined as a patient with a VAS score that decreased by >50% or an RMDQ that improved by >50%. 'Non-responder' referred to a patient whose VAS score did not decrease by >50% or the RMDQ did not improve by >50%.

One radiologist analyzed the MRI images without previous knowledge of patient symptoms. The MRI images were classified by HLD type, HLD zone, HLD volume, relationship of the HLD to nerve roots, disc height loss, and disc degeneration grade. Classification of HLD type, HLD zone, and HLD volume followed the standards used by the American Society of Spine Radiology, the American Society of Neuroradiology, and the North American Spine Society [10]. HLD type was divided into 'protrusion' and 'exclusion', and HLD zone consisted of 'central', 'subarticular', 'foraminal', and 'extraforaminal' (Figs. 1, 2). HLD volume was sorted into 'mild', 'moderate', and 'severe'. When the volume of a herniated disc accounted for onethird of the canal, it was categorized as 'mild'; 'moderate' was defined as a case when a herniated disc covered onethird to two-thirds of the canal volume; and 'severe' was used when a herniated disc's volume accounted for more than two-thirds of the canal. To investigate the relationship between the HLD and nerve roots, we examined whether the prominent HLD contacted nerve roots, and whether the nerve roots were displaced and compressed by the HLD [11]. Degeneration grade was categorized into five stages by looking at T2 MRI images [12]. This was divided into two categories: grades 1, 2, and 3 vs. grades 4 and 5. The level of disc height loss was defined as 'none', 'decrease by <50%', and 'decrease by >50%' [10].

The characteristics of age, sex, duration of symptom attack, and MRI findings between responders and non-responders were evaluated by the Mann-Whitney U-test and the chi-square test. Logistic regression analysis was performed to assess the relationship between the factors. Patient age, sex, duration of symptoms, HLD type, HLD zone, HLD volume, relationship between HLD and nerve roots, disc height loss, and disc degeneration grade were included in the analysis. The effect of the CESI was obtained through an adjusted odds ratio. The statistical analysis was performed with the SPSS Korean ver. 20.0 statistical package for Windows (IBM SPSS Inc., Armonk, NY, USA). A p-value <0.05 was considered significant.