The relative efficacy of topical non-steroidal anti-inflammatory drugs and capsaicin in osteoarthritis

Persson, Monica S.M. (2019) The relative efficacy of topical non-steroidal anti-inflammatory drugs and capsaicin in osteoarthritis. PhD thesis, University of Nottingham.

[img] PDF (Thesis - as examined) - Repository staff only - Requires a PDF viewer such as GSview, Xpdf or Adobe Acrobat Reader
Download (7MB)



Pain is often the most troubling issue for people with osteoarthritis (OA) and is typically the focus of pharmacological management. A multitude of treatment options are available for OA, but these produce, on average, only small to moderate reductions in pain. Often the modest reductions in pain do not outweigh the safety concerns of traditional analgesics. Topical non-steroidal anti-inflammatory drugs (NSAIDs) and capsaicin are safe alternatives, with different mechanisms of action, that are widely used for knee and hand OA. However, there is a paucity of evidence to guide the choice between the treatments. Their relative efficacies are unknown and it is unclear if certain people would benefit more from one treatment than the other. The identification of predictors of response to the treatments is essential in order to maximise treatment effects and implement a precision medicine approach in OA care.


[1] To determine the relative efficacy of topical NSAIDs and capsaicin in OA, [2] to identify predictors of response that may allow topical therapy to be tailored to the individual, and [3] to implement and appraise methods that guide precision medicine in OA.


Systematic literature searches were conducted for randomised controlled trials (RCTs) of topical NSAIDs or capsaicin in OA and neuropathic pain (NP). NP was chosen as a surrogate for phenotypes with predominantly neuropathic-like pain. Average treatment effects were examined using aggregate data, extracted from eligible publications. Placebo-controlled RCTs were pooled in random-effects conventional meta-analyses (CMA) for the specific (difference between treatment and placebo) and overall (improvement from baseline in the treatment group) treatment effects. Network meta-analysis (NMA) compared the treatments using placebo as a common comparator. Aggregate data effect sizes (ES) were presented as Hedges' g.

Predictors of response were examined using individual patient data meta-analysis (IPD MA) and a pilot n-of-1 trial series. Raw IPD from eligible RCTs, acquired through author contact and direct enquiry to pharmaceutical companies, were pooled in IPD MA. Predictors of the specific (treatment-by-covariate interactions) and overall (prognostic factors) treatment effect were investigated. Participants with radiographic, painful knee OA were recruited to an open label pilot n-of-1 trial series. Participants were randomly allocated to three treatment cycles consisting of alternating periods of 5% ibuprofen gel and 0.025% capsaicin (Zacin) cream. Treatment periods were four weeks. Treatment preferences were determined at an individual level and responses were pooled across participants to examine the variation in responses and potential predictors of response.


Of 63 topical NSAID and 10 capsaicin RCTs identified in the systematic literature search, CMA and NMA were based on 21 (n=6,191) and five (n=415) RCTs for topical NSAIDs and capsaicin, respectively. Moderate level evidence suggested that topical NSAIDs were effective for pain relief in OA (21 RCTs, ES: 0.31, 95% confidence interval [CI] 0.20 to 0.41). Very low level evidence suggested that capsaicin was only effective when used at the 0.025% concentration (4 RCTs, ES 0.41, 95%CI 0.17 to 0.64; ES of all capsaicin RCTs: 5 RCTs, ES 0.28, 95%CI -0.04 to 0.60). Absolute levels of pain relief from the treatments, compared to baseline, were substantial (overall effect: topical NSAIDs ES 1.23, 95%CI 1.06 to 1.41; capsaicin ES 1.05, 95%CI 0.52 to 1.57). Very low level evidence in NMA indicated that, on average, topical NSAIDs and capsaicin were equally effective (ES 0.04, 95% credible interval [CrI] -0.29 to 0.37).

IPD were provided for 15 (n=3,889) topical NSAID RCTs, including 11 (n=3,140) placebo-controlled trials. No IPD were obtained for capsaicin. Nine potential predictors of response to topical NSAIDs were examined. A significant but small interaction was observed between sex and topical NSAID use, with women reporting a greater effect (11 RCTs, 2,939 participants, p=0.023). Women also reported larger overall levels of pain relief on topical NSAIDs (15 RCTs, 1,857 participants, p=0.008), but this could be explained by baseline pain severity.

Twenty-two participants enrolled in the pilot n-of-1 trial series and completed 104 treatment periods. Clinically important pain relief (≥1 point on 0-10 numeric rating scale) was achieved in 64% of the completed periods. Women reported greater pain reduction from baseline irrespective of treatment (p=0.031), but this could be explained by baseline pain severity. No difference in pain relief was observed, on average, between topical ibuprofen and capsaicin (p=0.271). Individual responses to treatment varied, with approximately 60% of people preferring one treatment over the other, but without clear predictors for the preference. Twenty-five potential predictors of response were examined, of which fibromyalgia severity, distal pressure pain thresholds, temporal summation, synovial hypertrophy, and quadriceps strength inconsistently associated with a preference for one treatment over the other.

Very low level evidence suggested that low-dose capsaicin cream was not effective for NP (7 RCTs, ES -0.18, 95%CI -0.90 to 0.53). No data were available for the CMA of topical NSAIDs in NP. In addition, insufficient IPD were received for NP participants and it was not possible to examine for predictors of response to topical NSAIDs or capsaicin in NP.


Topical NSAIDs and capsaicin may both effectively relieve OA pain, but treatment effects from group comparisons do not directly translate to treatment outcomes at the individual level. Patients with different pain mechanisms may respond differently to these two agents, but definitive trial evidence is still needed. Presently, patients may benefit from trying both treatments to determine which is better for them.

To better guide the care of an individual, evidence synthesis should encompass a spectrum of methods, from study-level to participant-level evidence. Further development of IPD MA and n-of-1 trial series methodology is required to successfully guide precision medicine.

Item Type: Thesis (University of Nottingham only) (PhD)
Supervisors: Zhang, Weiya
Doherty, Michael
Walsh, David A.
Keywords: Topical non-steroidal anti-inflammatory drugs; Capsaicin; Osteoarthritis; Convention meta-analysis; Network meta-analysis (NMA; Individual patient data meta-analysis (IPD MA); Pilot n-of-1 trial series
Subjects: W Medicine and related subjects (NLM Classification) > WE Muscoskeletal system
Faculties/Schools: UK Campuses > Faculty of Medicine and Health Sciences > School of Medicine
Item ID: 56530
Depositing User: Persson, Monica
Date Deposited: 11 Sep 2019 08:45
Last Modified: 06 May 2020 08:03

Actions (Archive Staff Only)

Edit View Edit View