Pharmacological Treatments of Neuropathic Pain

Real-Life Comparisons Using Propensity Score Matching

Xavier Moisset; M. Gabrielle Pagé; Bruno Pereira; Manon Choinière


Pain. 2022;163(5):964-974. 

In This Article

Abstract and Introduction


Studies comparing different drug treatments for chronic neuropathic pain (NP) are very limited. We, therefore, examined 4 recommended treatments, namely, antidepressants (duloxetine, venlafaxine, and tricyclic antidepressants), antiepileptics (gabapentine and pregabalin), weak opioids, and strong opioids, among patients with NP evaluated before first visit in a tertiary pain treatment centre and 6 months later. Patients with both a clinical diagnosis of NP and a DN4 score ≥3/7 were selected from patients enrolled in the Quebec Pain Registry. Each participant was assigned an inverse weighting of the probability of receiving any NP treatment, taking into account their age, sex, baseline pain intensity, pain duration, pain catastrophizing tendency, education level, employment, and comedications at 6-month follow-up (M6). Patients were considered as improved if they presented at least a 30% reduction on average pain intensity at M6 compared with baseline. A total of 944 patients completed both baseline and M6 evaluations. Overall, 23.0% of patients were significantly improved for pain intensity at M6. There was no significant difference in proportions patients taking or not antidepressants, gabapentinoids, or weak opioids. Among patients taking strong opioids (N = 288), 13.9% (N = 40/288) were improved vs 27.0% (177/656) of those who were not on opioids (P < 0.004). Inverse probability of treatment weighting confirmed that the proportion of patients who improved was significantly lower among those taking strong opioids compared with those who did not (P < 0.001). In conclusion, long-term use of strong opioids is a treatment suited for a limited proportion of patients with chronic NP.


Neuropathic pain (NP) is present in 7% to 10% of the general population.[30] It is often difficult to treat, and it has a major impact on patients' quality of life along with important direct and indirect healthcare costs.[1,13,16] Several epidemiological studies have shown that many patients with NP do not receive recommended treatments.[1,17,45] Nonetheless, pharmacological and nonpharmacological therapies, although imperfect, are available.[15,23,37] Current international guidelines for pharmacological management of NP recommend tricyclic antidepressants, serotonin–noradrenaline reuptake inhibitors (SNRIs), and gabapentinoids (pregabalin and gabapentin) as first-line treatment; tramadol as second-line treatment; and strong opioids as third-line treatment.[23] To date, the number of studies comparing different recommended drug regimens for NP is very limited,[3,22,23,26,31,37,48] although comparative efficacy of different drugs could be informed, to some degree, by meta-analyses of placebo-controlled drug trials that allow for the estimation of number needed to treat for each agent allowing, with some recognized limitations, for a quantitative comparison across different drugs.[23,40] However, knowing the comparative clinical effectiveness of these treatments in real-world settings would be of major importance.

The Quebec Pain Registry (QPR) has been implemented in 2008 in 3 university-affiliated multidisciplinary pain treatment centres in Quebec (Canada) and 2 other centres joined in 2012. Close to 9,000 patients have been included and have provided consent for their data to be used for research purposes. Among these patients, around 20% were presenting chronic NP. As pain was systematically and comprehensively assessed both at baseline (before first appointment at the pain clinic) and 6 months later, it is possible to use such data to compare treatment effectiveness in this selected population of patients. Indication bias is frequent in observational studies because the choice of treatment is generally influenced by the patients' characteristics (eg, age, sex, and presence of depression or sleep problems)[2] and contraindications (eg, cardiac conduction block or postural hypotension for tricyclic antidepressants and substance use disorder for opioids), cost or healthcare provider coverage, and patient preference. However, a propensity score (PS) can be determined to adjust for several of these differences.[2]

The aim of this study was to examine the clinical evolution of patients with chronic NP treated in tertiary care centres and compare in real-life clinical settings the effectiveness of recommended medication for NP using a PS analysis.