Multicenter, Randomized, Placebo-controlled Crossover Trial Evaluating Topical Lidocaine for Mechanical Cervical Pain

Permission to conduct this randomized, placebo-controlled study was separately granted by the institutional review boards (IRBs) of Johns Hopkins Medical Institutions, which served as the IRB of record for Walter Reed National Military Medical Center, the District of Columbia Veterans Administration Medical Center, and the New England Institutional Review Board on behalf of the Pain Management Institute. The study was registered on May 4, 2020, at clinicaltrials.gov (NCT04378959) by the Principal Investigator (S.P.C.). The first subject was treated on February 25, 2021, while the remaining participants were treated between September 7, 2021, and June 1, 2023, due to a COVID-19 pandemic–related supply chain delay.

Participating institutions include an inner-city tertiary care teaching hospital (Johns Hopkins), an urban military treatment facility (Walter Reed), an urban Veterans Administration hospital, and a two-site urban private practice. Inclusion criteria were age between 18 and 90 yr, axial (localized, without radiation or neurologic deficits) neck pain more than 3 months in duration, average neck pain score of at least 4 points of 10 for the week before enrollment, a stable analgesic regimen for more than 2 weeks, and tenderness to palpation (pain elicited when an average 4-kg force in a 70-kg individual is applied) in the affected areas bounded by the upper trapezius muscles, mastoid processes, and shoulders. Exclusion criteria included radiculopathy, greater than 30 oral morphine equivalents per day of opioids, structural defect presumed to be the primary pain generator, previous cervical spine surgery, serious psychiatric or medical comorbidities (e.g., elevated liver function tests) that might interfere with participation or response, diffuse pain phenotype, secondary gain, painDETECT score greater than 18 suggesting non-nociceptive pain, allergy to lidocaine, pregnancy, and skin defect over the painful area(s).

In this randomized, placebo-controlled crossover trial, patients were randomized in a 1:1 ratio via computer-generated randomization tables in blocks ranging between 8 and 16 depending on enrollment estimates at each institution into one of two groups: placebo patch for 4 weeks followed by lidocaine patch after a 1-week washout period estimated based on the short half-life (2 h) and expected duration of action (24 h) of the medication (group 1)¹⁹ ; or lidocaine patch followed by placebo patch after the same washout (group 2). All investigators except for the research pharmacist, who released 4 weeks of medications for dispensation, were blinded to treatment allocation. Participants were enrolled by investigators, who entered blinded orders for treatment and used blinded note templates. The 1.8% lidocaine patches are 10 cm × 14 cm × 0.08 cm in size and contain 36 mg of active substance that provides 40 to 50% bioavailability versus less than 10% in most other commercial products, greater than 90% adherence in 75% of individuals versus less than 60% for other commercial products at 12 h, and increased penetrance (more than 1 cm vs. 5 to 10 mm; ZT Lido, Scilex, USA).^17,18  The placebo patches were identical in dimensions, texture, and smell.

Participants could receive one to three patches depending on the extent of their pain, applied in a 12-h on–off fashion: they were instructed to keep the patches on during the day if their pain was activity-dependent or at night if it was worse during sleep. During the study course, no therapeutic interventions in addition to structured (e.g., physical therapy) or nonstructured exercise were permitted.

Baseline data were collected within 24 h before phase 1 and 2 medication dispensation and included age, sex, race and ethnicity, pain duration, laterality and number of patches, military status, inciting events, chronic pain and psychiatric comorbidities, social variables including smoking and disability status, analgesic medications, average and worst arm and neck pain score on a 0 to 10 numerical rating scale over the past week, function measured by the Neck Disability Index, sleep quality measured by the Athens Insomnia Scale, neuropathic and nociplastic contributions to pain burden estimated by painDETECT, and pressure pain threshold (PPT) (newton/mm²) estimated by the mean threshold measured at the three most affected areas.^20–22 

At the 4-week phase 1 and 2 follow-ups (window, 24 to 42 days), a blinded, disinterested investigator not involved in treatment recorded outcomes. These outcomes included average and worst pain scores measured through daily pain diaries; function; sleep quality; mean PPT; adverse effects via set (skin reactions) and open-ended questions; compliance (less than 75% of prescribed patches)²³ ; the binary variable “medication reduction,” predefined as cessation of a nonopioid analgesic or more than 20% reduction in opioid consumption; Patient Global Impression of Change measured on a 1 to 7 Likert scale, in which 1 indicates no change or worse, 3 indicates slightly better, 5 indicates moderately better, and 7 indicates definitely better; and a binary categorical outcome of success, which was predesignated as a 2-point or greater reduction in average neck pain coupled with a score of at least 5 points on the Patient Global Impression of Change scale.^24–26  The primary outcome measure was mean reduction in average neck numerical rating scale pain score calculated over the past week.

This crossover study was powered to detect a minimal difference between lidocaine patch and a placebo patch using STATA 14.2 (USA). The study was conservatively powered to detect the difference between the two treatment groups based on the following assumptions^10,26,27 :

• Baseline pain score: 6.0 (SD 1.8)

• Lidocaine treatment: a 2.3-point reduction in numeric pain rating score (SD 2.2)

• Placebo treatment: 1.2-point reduction in numerical pain score (SD 2.0)

• Dropout rate: 16%

• Statistical power: 90%, α: 0.05

Based on these assumptions, we estimated that 37 patients in each treatment arm (lidocaine–placebo, placebo–lidocaine) were needed to achieve 90% power to detect a clinically meaningful difference of 1.1 points between the treatments. For 80% power, we estimated that 65 patients needed to be randomized. To account for an anticipated 15% dropout rate, we planned to enroll 84 total patients.

Because this study began during the pandemic caused by the severe acute respiratory syndrome coronavirus 2 (COVID-19), a 6-month hiatus occurred between the first and subsequent patients enrolled while we addressed issues regarding in-person visits, the supply chain, and IRB administrative requirements. The IRBs requested that telehealth visits be permitted in lieu of some in-person visits before restarting the study, which prevented collection of PPT data on those participants. The study was halted after 76 patients were randomized because of expiration of the placebo and lidocaine patches.

Outcomes were assessed by treatment type and analyzed via intention-to-treat without missing data imputation. Nonparametric tests were applied for variables as indicated based on the results from the Shapiro–Wilk test, frequency histograms, and examination of Q-Q plots. Continuous variables were analyzed with the Wilcoxon rank sum test. For the binary categorical outcome of success versus failure, Pearson’s chi-squared test was utilized. Odds ratios and 95% CIs were calculated to measure the association of age, sex, and duration of pain with the binary outcome. Backward stepwise logistic regression was performed to calculate adjusted odds ratios for the association of the aforementioned explanatory variables with the binary outcome of success using variables found to be associated (P < 0.25) with lidocaine patch treatment outcome in univariable analysis and those postulated a priori to have an effect on treatment outcome (e.g., age, sex, duration of pain, and the most predictive measure of disease burden, Neck Disability Index).

The Department of Defense paid for personnel at Walter Reed and Johns Hopkins, the D.C. Veterans Administration Hospital paid for personnel costs at their hospital, while Scilex paid for personnel at Johns Hopkins and IRB fees for the Pain Management Institute. The funding sources played no role in study design, data collection and analysis, or manuscript preparation.

A total of 76 people were enrolled and randomized between February 2021 and June 2023. A total of 60 patients completed both phases, receiving both lidocaine and placebo treatment, while 12 patients only completed a single phase (7 placebo patch and 5 lidocaine patch). Five patients were noncompliant (less than 75% utilization), all in phase 1. There were no significant differences in demographics or baseline outcome measures between the treatment groups (fig. 1; table 1).

The baseline differences between phases 1 and 2 stratified by group allocation are shown in supplemental table 1 (https://links.lww.com/ALN/D395). Phase difference in baseline average neck pain was not significantly different (P = 0.84) between group 1 (median [interquartile range], 0.0 [−2.0, 0.0]) and group 2 (0.75 [−1.0, 0.0]). Phase difference in baseline worst neck pain was also not significantly different between groups (P = 0.80). There was a significant difference in the difference of mean PPT between phase 1 and 2 when comparing group 1 (−2.0 [−5.0, 0.2]) with group 2 (4.0 [0.0, 10.0]; P = 0.015; n = 30), suggesting a small carryover effect for lidocaine (supplemental table 1, https://links.lww.com/ALN/D395; fig. 2).

There was no difference in the primary outcome measure, reduction in average neck pain score, between lidocaine (−1.0 [−2.0, 0.0]) and placebo (−0.5 [−2.0, 0.0]) patches (P = 0.17). Average neck pain score at the end of treatment was also not significant between lidocaine (4.0 [2.0, 6.0]) and placebo (5.0 [3.0, 7.0]) treatments (P = 0.26). A successful outcome was observed in 27.7% of patients during lidocaine treatment versus 14.9% during the placebo phase (P = 0.073; table 2).

The secondary outcomes are presented in table 2. The reduction in worst neck pain score with lidocaine treatment (−1.0 [−2.0, 0.0]) was again twice as high as with placebo (−0.5 [−2.0, 0.0]) treatment, but neither this difference (P = 0.38) nor the difference in median worst neck pain (7.0 [5.0, 8.0] for lidocaine vs. 7.0 [5.0, 9.0] for placebo; P = 0.43) was statistically significant. The change in the Neck Disability Index was higher after the placebo phase (−3.0 [−10.0, 2.0]) than after lidocaine treatment (0.0 [−6.0, 5.0]; P = 0.16). Individuals reported a slightly higher Patient Global Impression of Change score after the lidocaine phase than after placebo treatment (3.0 [1.0, 5.0], corresponding to “slightly better” vs. 2.0 [1.0, 3.5] corresponding to “almost the same”), but again this difference fell shy of statistical significance (P = 0.24). There were no other trends or significant differences between treatment groups for other secondary outcome measures.

Factors associated with successful treatment outcome in the lidocaine phase are shown in table 3. In subgroup analyses, those with a shorter duration of neck pain were more likely to experience a positive outcome (odds ratio, 0.91; 95% CI, 0.82 to 0.98; P = 0.033) as were those with a lower baseline Neck Disability Index (odds ratio, 0.97; 95% CI, 0.94 to 1.00; P = 0.037). However, when adjusted for multiple comparisons, shorter duration of neck pain and Neck Disability Index failed to reach significance. Individuals who were nonadherent experienced only marginally less pain relief (odds ratio, 1.36; 95% CI, 0.21 to 26.71; P = 0.78) compared to compliant patients. Neither PPT (odds ratio, 0.98; 95% CI, 0.94 to 1.02; P = 0.34), anxiety (odds ratio, 0.60; 95% CI, 0.22 to 1.50; P = 0.30) or depression history (odds ratio, 0.58; 95% CI, 0.21 to 1.43; P = 0.26), age (odds ratio, 1.02; 95% CI, 0.99 to 1.05; P = 0.11), or sex (odds ratio, 1.44; 95% CI, 0.59 to 3.76; P = 0.43) was associated with lidocaine outcome at the 4-week endpoint. In the placebo phase, older age exhibited the strongest association with positive outcome but did not reach statistical significance (odds ratio, 1.05; 95% CI, 1.00 to 1.11; P = 0.08). When analyses were conducted using difference of difference (lidocaine–placebo) scores, the results did not significantly change (table 3).

Of all patients, 27.5% reported adverse effects in the lidocaine phase compared to 20.5% of patients in the placebo treatment phase (P = 0.36). The most common complication in both phases was pruritus, which 9.0% of placebo phase patients and 9.2% of lidocaine phase patients experienced. Rash localized to the sites of patch administration occurred in three patients each with lidocaine and placebo. Mild erythema was reported in three patients during the placebo phase and one patient in the lidocaine phase. In the lidocaine phase, other complaints included trapped sweat (n = 1), worsening pain (n = 1), nausea (n = 1), and neck numbness (n = 1). In the placebo phase, one patient complained of difficulty removing the patches, and one patient complained of difficulty putting on patches. Side effects necessitated aborting the treatment in only two patients, with one each in group 1 and group 2 not proceeding to phase 2 because of a rash.

A total of 58.3% of patients correctly guessed the sequence of their treatment. Three patients (4.17%) guessed that they received the same treatment in both phases. Among the remaining patients, the James Blinding Index was 0.77 (95% CI, 0.68 to 0.86), indicating effectiveness of blinding.

The main findings in this study were that no significant differences were observed between the lidocaine and placebo patches, with the increased pain relief associated with lidocaine patches translating to higher, albeit still low, success rates. Our findings were comparable to the placebo-controlled trials of Lin et al.,¹⁰  who found a small difference (mean, 0.44) after 2 weeks of application but not earlier or at 3 weeks (mean difference, 0.22), but less propitious than those of Affaitati et al.,¹¹  who found an enormous effect at 9 days (mean difference, 61.85 mm on a 100-mm visual analog scale), 4 days after cessation of a 4-day treatment protocol. Differences between our selection criteria and these studies were that Lin et al.¹⁰  required a taut band of muscle that elicited a local twitch response (trigger point) and used only clinical findings to rule out neurologic disease, while Affaitati et al.¹¹  enrolled only those with acute myofascial pain.

Although our study was negative and indicates that this topical lidocaine preparation is unlikely to produce clinically important benefits in a general population, the small, nonsignificant effects favoring lidocaine suggest the possibility that an enriched population, refined by measurable characteristics, might experience clinically meaningful benefit. Skin and muscles are the two largest organ systems in the body, with both capable of being pain generators. One reason why topical local anesthetic patches have previously failed to elicit benefit is poor penetrance and adherence, which is why newer formulations with greater adherence and penetrance may theoretically improve outcomes.²⁶  The distance from skin to the deep intrinsic lumbar spinal muscles ranges from 4 to 6 cm, which may help explain the negative studies for back pain.^9,28  The musculature in the neck generally extends between 6 and 12 mm under the skin, which is slightly greater than the penetrance of traditional topical patches but comparable to those of the product we studied.^17,18,29  However, myofascial pain frequently co-occurs with spine pathology that does not respond to topical therapies,⁶  and one study that examined neck pain patients with normal imaging failed to find correlations between myofascial pathology and pain, suggesting the possibility of nociplastic pain (i.e., nonspecific neck pain).^30,31  Although we attempted to rule out non-nociceptive causes of pain by screening with painDETECT, this instrument is not designed to identify central sensitization as a mechanism. Individuals with central sensitization have been shown in numerous studies to be more likely to fail interventional and pharmacologic treatments, including for neck pain.^32,33 

In multivariable analysis, greater duration of pain and higher Neck Disability Index scores were negatively correlated to treatment outcome. This is consistent with other studies demonstrating an inverse relationship between disease burden (including duration of pain and disability) and treatment effect for neck and back pain.^34,35  The recent emphasis on personalized medicine and phenotyping has led to attempts to identify likely responders, which may improve the cost-effectiveness and risk:benefit ratios. Studies performed in diabetic neuropathy suggest an “irritable nociceptor” phenotype may be more likely to respond to both topical and intravenous lidocaine, but whether higher baseline pain thresholds predispose to failure for mechanical pain is unknown.^36,37  An irritable nociceptor phenotype that might respond to topical lidocaine could present with lower mechanical pain thresholds in the neck but not diffusely, with preserved small fiber function. These individuals would lack evidence of pronounced central sensitization in the form of temporal summation or impaired conditioned pain modulation on quantitative sensory testing or abnormal scores on instruments such as the Central Sensitization Inventory.³⁸  In contrast, those with central sensitization may have lower pain thresholds in both painful and nonpainful areas. We found no significant effect for baseline PPT, but reasons other than sensitized nociceptors can cause low PPTs, including nonorganic pathology, nociplastic pain, and secondary gain, all of which are associated with treatment failure.^31,34,35  The trend toward younger age being associated with a positive lidocaine treatment outcome may be attributed to the higher contribution of myofascial pain (greater muscle mass) in younger individuals compared to the degenerative processes such as facetogenic and discogenic neck pain that predominate in the elderly.⁶ 

The IMMPACT guidelines assert that either greater than 2 points or greater than 30% reduction in pain represents the minimal clinically important difference for an individual patient, and in our protocol, we predesignated the former.²⁵  When reanalyzed using the latter, 15% of placebo period (N = 10 of 67) and 23% of lidocaine period (N = 15 of 65) attained a positive outcome (P = 0.22), which did not alter the analysis. Studies in chronic pain have yielded mixed results on the linearity of pain scales, which would suggest that in some scenarios, utilizing a percentage cutoff would be more sensitive to identifying responders.^39,40 

Our results indicate that when applied to a general population of individuals with mechanical neck pain, the benefits afforded by a high bioavailability lidocaine patch are small and questionably meaningful. However, the risk:benefit and cost:effectiveness ratios may be more favorable in a clinically refined or enriched population consisting perhaps of younger, thinner, or more muscular individuals; those with evidence of irritable nociceptors (e.g., low thermal pain thresholds without evidence of central sensitization)^38,41 ; and those at high risk of pharmacologic side effects. Given the strong signal for clinically meaningful efficacy, determining whether formulations that contain higher doses or greater penetrance increase the effect size is an avenue for future investigation.

There are several limitations in this study that warrant consideration. First, although the sample size is extremely large for a placebo-controlled crossover study,⁴²  we enrolled slightly fewer patients than our maximum allotment owing to an IRB-mandated pandemic hiatus, during which we had to implement risk-mitigation strategies and owing to the patches expiring; hence, the study may have been underpowered to detect a small difference. One of the risk-mitigation measures applied was permitting telehealth appointments, which resulted in many visits being conducted over the phone or via video, which decreased the number of patients in whom PPTs were recorded. This may have also resulted in lower overall success rates, as most telephone visits were performed by a nurse rather than physician, with some studies reporting lower pain scores when visits are conducted by a doctor.⁴³  Second, although we tried to eliminate individuals with significant central sensitization, this was accomplished via history and physical exam and painDETECT rather than a validated instrument such as the Central Sensitization Inventory.⁴⁴  Third, it is possible that measuring pain at different time points (e.g., at multiple points, or at a distal point to exploit the unforeseen possible carryover effect) rather than only at the end of treatment may have increased the effect size. Whereas a 1-week washout period is consistent with other crossover trials evaluating topical lidocaine, evidence for this includes the carryover effect in the lidocaine group observed in our study, as well as the finding by Lin et al.¹⁰  finding no difference at treatment end between lidocaine and placebo patches for neck pain but a significant treatment effect 1-week after treatment cessation.^10,17,45  Last, accounting for a possible small carryover effect in a crossover trial creates statistical implications with no consensus on analysis, with some statisticians believing that only the first period should be analyzed when a clear carryover effect exists.⁴⁶  In our study, analyzing data only for the first period would have changed our primary outcome from a trend to statistical significance.

In summary, although most important measures showed trends for greater improvement with lidocaine, the differences were small and not significant. This suggests a need for additional studies evaluating products with greater dosing and penetrance, applied with more stringent selection criteria based on phenotyping, to determine whether topical lidocaine may provide meaningful benefit in a subset of patients with chronic neck pain.

The authors acknowledge Daniel Scharfstein, Sc.D., and Vwaire Orhurhu, M.D., for their assistance with statistical analyses. The views expressed in this article are those of the authors and do not reflect the official policy of the Uniformed Services University, the Departments of the Army and Navy, the Department of Defense, the Department of Veterans Affairs, or the U.S. Government. The identification of specific products or scientific instrumentation is considered an integral part of the scientific endeavor and does not constitute endorsement or implied endorsement on the part of the authors, the Department of Defense, or any component agency. The views expressed in this article are those of the authors and do not reflect the official policy of the Departments of the Army and Navy, the Department of Defense, or the U.S. Government.

Supported in part by grants from MIRROR, Uniformed Services University of the Health Sciences (Bethesda, Maryland); U.S. Department of Defense (Bethesda, Maryland) grant No. HU00011920011; and Scilex Pharmaceuticals (Palo Alto, California). The District of Columbia Department of Veterans Affairs Institute for Clinical Research (Washington, D.C.) funded the Veterans Affairs Medical Center. The sponsors did not play a role in study design or performance, or analysis or interpretation of data or manuscript preparation.

Dr. Cohen is a consultant for Avanos (Alpharetta, Georgia), SPR Therapeutics (Cleveland, Ohio), Scilex (Palo Alto, California), Persica Pharmaceuticals (Canterbury, Kent, UK), and Sword Health (Draper, Utah; active) and previously was a consultant for Releviate Therapeutics (San Diego, California) and Clearing (New York, New York; ended within the past 3 yr). Dr. Cohen also conducted institutionally funded research by Avanos and Scilex. The other authors declare no competing interests.

Full protocol available at: steven.cohen@northwestern.edu. Raw data available at: steven.cohen@northwestern.edu.

Supplemental Table 1. Difference in Phase Baseline Stratified by Treatment, https://links.lww.com/ALN/D395