We thank Dr. Bhatia for taking an interest in our article,1 and are pleased to address the comments in detail.
We deliberately refrained from using the term “radiculopathy” in our title, as this would in fact have implied that dorsal nerve root compression or mechanical lesion was confirmed in our patient sample by magnetic resonance imaging or electrophysiology, as suggested in the Letter to the Editor. We decided not to require a technical confirmation for patient inclusion because the presence or absence of a visually identified mechanical compression would not have influenced the interpretation of study results. This makes a difference to trials on invasive, localized therapeutic approaches like epidural corticoid injections, e.g., as reported in the referenced article by Cohen et al.,2 where an identified lesion is the target. The primary aim of our trial, however, was to investigate whether or not systemic duloxetine is efficacious in reducing chronic low back pain (CLBP) in patients with a clinically diagnosed radicular neuropathic component. These patients—although frequently seen—were excluded from former studies on the efficacy of duloxetine in CLBP.3–6 We think that these patients may currently be undertreated in terms of addressing their neuropathic pain symptoms.
It is recognized that the International Association for the Study of Pain definition of neuropathic pain may be problematic in CLBP, as there is sometimes a lack of correlation between neurologic lesion and symptoms in these patients.7,8 Recent guidelines on neuropathic pain assessment9 clearly state that there is still no accepted standard for diagnosis of neuropathic pain, and that the relevance of clinical examination for distinguishing neuropathic from nonneuropathic pain is well documented. Therefore, we combined a clinical examination and diagnosis of radicular leg pain by two independent, experienced pain specialists, who were not involved in the study, with the additional assessment using the painDETECT questionnaire. Combining the validated painDETECT screening tool with a clinical diagnosis of radicular leg pain increases the probability of identifying the target group of CLBP patients with a clear neuropathic pain component.
We referenced a study by Baron et al.,10 which also uses the wording “Chronic Low Back Pain with a Neuropathic Component” in its title. In the Letter to the Editor, the quoted study has been misinterpreted as focusing only on patients with a neuropathic pain component localized in their lower back without neuropathic leg pain. Baron et al. enrolled patients who scored “positive or unclear” in the painDETECT questionnaire and additionally clinically tested them for the presence of radicular leg pain at baseline. Patients were clinically diagnosed with lumbar radicular pain in 70% of cases. Therefore, this study investigated a mixed population suffering from low back pain with or without a radicular neuropathic component, although a two-thirds majority actually suffered from CLBP combined with radicular pain.
We agree with Dr. Bhatia in that differentiation between duloxetine’s relative therapeutic effect on either leg or back pain would have added further information.
The painDETECT questionnaire was specifically developed for the diagnosis of a neuropathic component in CLBP. As it also scores the presence or absence of radiating pain, it is not surprising that patients who score positive are often clinically diagnosed with radicular pain. We acknowledge that the painDETECT is not validated for follow-up assessments of neuropathic pain over time, and the additional use of the Neuropathic Pain Symptom Inventory would have improved our study design.
Finally, this was a monocentric study in patients who had to fulfill very strict and ambitious inclusion and exclusion criteria: no previous use of World Health Organization III opioids, no use of antidepressants or benzodiazepines 6 months before entry, and not even mild depression present for over 12 months. In a developed public healthcare system, such individuals are difficult to find for a 12-week crossover trial. Therefore, we do not consider the number of participants unusually low, and we much appreciate their kind willingness and support.
Concerning statistics, in the initial sample size calculation, the SD was assumed to be 3.4 Visual Analogue Scale (VAS) units. The sample size was reviewed after enrollment of 14 patients. An SD of the treatment differences of 1.794 on the VAS was calculated at this point. To adjust for uncertainty in the estimation, an SD of 2 was added. Therefore, the revised total sample size of 21 patients was based on an assumption of an SD of 2.0 VAS units. In order to assign equal numbers of patients to each sequence group, the final sample size was determined as 22 patients. All sample size calculations are based on paired Student’s t tests, assuming a power of 90%, a two-sided significance level of 5%, and a relevant mean difference of 1.5 points.
The authors declare no competing interests.