Article Text

Download PDFPDF
A randomised controlled trial of subcutaneous sodium salicylate therapy for osteoarthritis of the thumb
  1. A S Smith1,
  2. C J Doré2,
  3. L Dennis3,
  4. A Julius3,
  5. C G Mackworth-Young1
  1. 1Department of Rheumatology, Charing Cross Hospital, London, UK
  2. 2ARC Clinical Trials Collaboration, MRC Clinical Trials Unit, London, UK
  3. 3Department of Physiotherapy, Charing Cross Hospital, London, UK
  1. Correspondence to Dr C G Mackworth-Young, Department of Rheumatology, Charing Cross Hospital, London, UK; c.mackworth-young{at}


Background Current treatment for osteoarthritis (OA) is limited. Many patients with OA of the hand have areas of tender subcutaneous thickening in the forearm and upper scapular region. A pilot study showed an improvement in pain from OA at the first carpometacarpal joint after injection of such areas with 0.5% sodium salicylate or saline, an inexpensive treatment that can be administered by general practitioners and nurses. The study indicated that a randomised, sham-controlled trial was justified.

Methods 40 patients with OA of the first carpometacarpal joint were randomised to receive either injections of sodium salicylate into tender, thickened areas of subcutaneous tissue on the forearm (baseline) and upper scapular region (week 1) or sham injections consisting of pressure without skin penetration. Blinded assessments were made at weeks 3, 7 and 13 after baseline.

Results Pain and tenderness during follow-up were both significantly lower in the active treatment group compared with the sham group: 19% and 14% greater reduction in mean visual analogue scale (VAS) score, respectively (p=0.007 and 0.02, baseline mean 5.65 and 5.35 cm, average difference in change from baseline VAS 1.9 and 1.4 cm, 95% CI 0.6 to 3.2 and 0.2 to 2.5). Active and sham injections were painful, the former significantly more so; however, there was no significant correlation between the pain of active injections and response.

Conclusion The data show that subcutaneous sodium salicylate injections are an effective symptomatic treatment for OA of the thumb. The results provide a basis for further physiological and therapeutic research in this area.

  • Osteoarthritis
  • pain
  • hand joint
  • rheumatic patch
  • sodium salicylate
  • rheumatology

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Osteoarthritis (OA) is a common cause of pain and disability.1 It can have a major financial impact on individuals and healthcare providers.2 Currently available treatment is limited. Analgesics, anti-inflammatory agents, physical therapies and surgery are limited in applicability, efficacy and tolerability. This study examines a different form of medical treatment for OA, which is inexpensive and could be administered by physicians, general practitioners and trained nurses.

In many forms of arthritis, areas of thickening and tenderness in cutaneous and other soft tissues can be found. These have been referred to by a number of different names, including interstitial fibrositis, myofasciitis and myofascial trigger points.3–5 They are also well described by acupuncturists.6 Fox and Freed7 used the term ‘rheumatic patch’ to distinguish those areas of thickening and tenderness that are relatively superficial—that is, those that are found clinically in cutaneous and subcutaneous tissue. The term is used here, although it is not used to imply a causal link between the soft tissue lesion and joint pathology. Such patches are found at sites near to or distant from inflamed or painful joints, typically in areas of a limb proximal to an affected joint, and in paraspinal regions in the cervical and lumbar areas. The patches are generally tender, and are easily detectable by palpation: the localised area of thickened tissue can be identified by the examiner using a gentle pinch or rolling technique between finger and thumb. Unlike myofasciitis or myofascial trigger points (which often involve deep tissues such as muscular fascia), rheumatic patches do not involve structures deeper than the subcutaneous tissue.

Treatment of arthritis (generally OA) or myofascial pain by injection of tender areas of soft tissue (usually trigger points) has been described. Some clinicians have used local anaesthetic, anti-inflammatory agents, saline or water.5 8–13 Improvement in symptoms was found, although this was usually short-lived. In a large uncontrolled study, Fox14 reported a better response from injecting 0.5% sodium salicylate into superficial (‘rheumatic’) patches: using multiple injections, he recorded an improvement in symptoms in 79% of patients at 12 weeks. Most of these subjects had OA.

In a preliminary study, 33 patients with different arthritic conditions were treated with salicylate injection into rheumatic patches, following the protocol of Fox and Freed7 (CGM-Y, unpublished observations). Significant symptomatic improvement was seen in many patients. In a subsequent pilot study, 16 patients with OA of the first carpometacarpal (CMC) joint were treated with injections of 0.5% sodium salicylate or saline into rheumatic patches.15 There was a significant improvement in pain score in both groups. The improvement was better maintained in the patients that received salicylate. The results indicated that a randomised controlled trial of sodium salicylate injections was justified.

Ideally such a study would be fully blinded and placebo-controlled. However, it is not possible to administer a true placebo, as subcutaneous injection of any substance that results in a wheal could have a physiological effect at a distant site as well as locally.4 Even the simple introduction of a needle beneath the skin (without the injection of any substance) could cause such an effect: this ‘mock’ acupuncture may not be physiologically inert, and could produce analgesic effects that are not specific to the points used.16 Furthermore, our local research ethics committee felt this form of control to be unethical, because it is invasive and painful, and it is not a recognised treatment.

An established control for acupuncture is the firm pressing of a needle against the skin without penetrating it.17 18 Acupuncture needles are generally inserted for ∼20 min. In subcutaneous sodium salicylate therapy (SSST), the injections last for between 60 and 90 s. It is unlikely that pressure over a rheumatic patch with a needle for this time would cause a significant lasting physiological effect, and this sham treatment was therefore chosen as a control. The pilot study and other preliminary data had shown that, in the majority of patients, two treatment sessions were sufficient to obtain a symptomatic response (injection of patches in the forearm and then in the scapular region).15

The objective of the study was to determine if SSST is superior to sham therapy in treatment of OA of the first CMC joint.

Patients and methods

Patient selection

Male and female patients were selected who had OA of the first CMC joint, as defined by the American College of Rheumatology.19 Required clinical features included local pain, hard tissue enlargement of the joint, and absence of soft tissue swelling at this joint or other hand joints. The patients were recruited from rheumatology clinics at Charing Cross and Chelsea and Westminster Hospitals, London, UK. Participants were recruited between February 2003 and February 2007. All patients invited to participate were informed about the study (including the nature of the active and sham treatments).

The inclusion criteria were as follows: persistent local symptoms for at least three months; radiological confirmation of OA; normal erythrocyte sedimentation rate, full blood count, platelet count, serum uric acid and ferritin, corrected plasma calcium, plasma phosphate and alkaline phosphatase, and a Latex test for rheumatoid factor either negative or within normal limits; the presence of rheumatic patches in the ipsilateral forearm and upper scapular region; informed consent.

The exclusion criteria were: patient age under 18 years; any other detected rheumatological disease; pregnancy or current breast feeding; any patient of child-bearing potential not taking adequate contraceptive measures; asthma; allergy to aspirin, other salicylates or other non-steroidal anti-inflammatory agents; haematological disorders leading to impaired haemostasis; widespread rashes, dermatological conditions affecting the areas overlying rheumatic patches, and those expressing the Köbner phenomenon; active peptic ulceration or history of peptic ulcer; local sepsis; any condition in which an anti-platelet action might be harmful (eg, recent stroke); known underlying conditions such as malignancy, renal failure or serious infection; the concomitant administration of anticoagulants, corticosteroids, immunosuppressive agents or non-steroidal anti-inflammatory agents; any change in arthritis therapy within 2 weeks of starting the study.


Upon entry into the study, patients were randomly allocated 1:1 to one of two groups: active treatment or sham treatment. During recruitment to the study, both patients and assessors were blinded as to this allocation. The randomisation code was generated by Charing Cross Hospital Pharmacy.

Treatment protocol

All patients underwent an initial assessment, and were then left on their current drug treatment unchanged for 3 weeks. At the end of this time (week 0), they were assessed again. At this point, they received the first of two injections: subcutaneous sodium salicylate or sham (see below for details of technique). A week later (week 1), the subjects received a second injection (sodium salicylate or sham). They were then assessed at 3, 7 and 13 weeks.

Patients in both groups were asked not to make any changes to their use of analgesics during the study, if taken. No physiotherapy was allowed during the trial period, except exercises already being performed by the patients at home. Patients were free to withdraw from the study at any time. After week 13, patients who had been in the sham-treated group were offered active sodium salicylate injections outside the study.

Injection technique

Injections and sham injections were performed by either AS or CGM-Y. Rheumatic patches were identified by palpation in the extensor region of the forearm and in the upper scapular region on the same side as the CMC joint being treated and assessed.

In the active group, up to 20 ml 0.5% sodium salicylate was injected on any one occasion, given all into one large patch, or divided between two to four smaller patches. Patches on the forearm were injected on the first occasion, and patches in the upper scapular region a week later. The solution was delivered through a 23-gauge needle subcutaneously so as to produce a firm wheal.

For sham injections, a blunt 23-gauge probe was pressed on to the skin over each patch as if the patch were being injected. The skin was not penetrated, but sufficient pressure was applied to cause local discomfort. The probe was kept in this position for the duration of the sham injection, this being determined by the number of individual patches suitable for injection. The total time taken for a sham injection on each visit was 75 s, divided equally between the individual patches.

All injections (genuine and sham) were performed out of view of the patient. For those to the forearm, an opaque shield was used. Sticking plaster (Bandaid; Smith and Nephew) was put over all of the injection sites and adjacent areas in order to mask puncture points and bruising. Patients were asked not to take this off for 3 days.


Assessments were performed on enrolment and at weeks 0, 3, 7 and 13. They were made in a blinded fashion by a clinician (LD or AJ) who had not administered the injections.

At all assessments, the patient was asked to use a 10 cm visual analogue scale (VAS) from 0 (none) to 10 (greatest) to record: (i) the degree of pain suffered in the affected joint during the preceding week; (ii) the degree of overall disability experienced in that joint over the same period. The clinician also assessed the degree of tenderness in the joint. A standard pressure was applied to the joint,20 and the amount of induced pain was recorded on a VAS (score 0–10) by the patient.

Patients in both groups were also asked to use a VAS (score 0–10) at weeks 0 and 1 to record the degree of pain experienced during and immediately after each of the two active or sham injections.

Statistical methods

A prospective sample size calculation was performed to test the null hypothesis that there was no difference in effectiveness of the two randomised treatments A sample size of 17 patients per group would provide 80% power to detect a difference in means of one standard deviation as significant at the 5% significance level. The sample size was increased to 20 per group to allow for patients who failed to attend for all scheduled assessments. An intention-to-treat approach was used for the analysis using data from all randomised patients who attended for at least one follow-up assessment.

For each variable, a repeated-measures analysis of variance was performed. The average of the screening and baseline assessments was used as a covariate. The effect of treatment was assessed relative to the variation between subjects, while the effect of time (weeks 3, 7 and 13) and the interaction between treatment and time were assessed relative to the variation within subjects. p Values for the within-subject terms were adjusted using the Huynh–Feldt procedure for repeated-measures analysis of variance.


Recruitment and randomisation

Fifty-four patients with OA of the first CMC joint were assessed for eligibility to participate in the study. Fourteen were excluded because they did not meet the inclusion/exclusion criteria, and three declined to participate. The remaining 40 patients were randomised to receive either two subcutaneous injections of sodium salicylate—at baseline and one week—or two sham injections. Twenty patients were randomised to active treatment, and 20 to sham treatment.

Table 1 summarises the baseline characteristics of the two randomised groups. The two groups were similar at baseline. The mean ages and age ranges of the treatment and control groups were 65.2 years (50–86) and 68.6 years (49–80), respectively. Eighteen patients in the treatment group and 17 patients in the control group were female. All patients had radiographic evidence of joint space narrowing and marginal sclerosis at the relevant first CMC joint. No patients withdrew or were lost to follow-up, and none were excluded from the analysis.

Table 1

Baseline values of pain, disability and tenderness visual analogue scale (VAS) scores at the first carpometacarpal joint in the active and sham-treated groups

Assessments of efficacy

Patients were assessed by VAS for pain, tenderness and disability. This was performed 3 weeks before baseline, at baseline, and at weeks 3, 7 and 13 thereafter (table 2). Pain and tenderness during follow-up were both significantly lower in the active treatment group: 1.9 cm (19%) greater reduction in mean VAS score for pain (baseline mean 5.65 cm, p=0.007) and 1.4 cm (14%) greater reduction in mean VAS score for tenderness (baseline mean 5.35 cm, p=0.02). There were no significant effects of time—that is, the mean VAS score did not change significantly over weeks 3, 7 and 13. There were also no significant interactions between treatment and time—that is, the difference between the two groups did not change significantly over weeks 3, 7 and 13.

Table 2

Average effect of treatment on pain, disability and tenderness visual analogue scale scores

Figure 1 shows the changes in the variables over time. The changes seen between baseline and week 3 in pain and disability in the active group and the stability of the variables between weeks 3 and 13 in both groups are shown.

Figure 1

Mean visual analogue scale (VAS) score and 95% CI for (A) pain, (B) disability and (C) tenderness over time in the active and sham-treated groups.

Figure 2 plots the individual changes in pain score from mean pretreatment values to the value at 3 weeks. It suggests a continuum of response in the active group. As in previous studies, patients usually reported an improvement in symptoms within a few minutes of salicylate injection.

Figure 2

Individual changes in pain from mean pretreatment value (horizontal axis) to value at 3 weeks (vertical axis).

Pain of injection

The pain of the injection (active or sham) was measured by VAS immediately after the procedure at baseline and at week 1 (table 3). The pain of the injection was significantly higher in the active treatment group (p<0.001). Patients in both groups experienced superficial discomfort at the sites of injection, with or without bruising, lasting for up to 3 days. There was a significant interaction between treatment and time (p=0.05)—that is, the difference between the two groups was slightly greater at week 1 than at week 0. No other adverse effects were detected or reported.

Table 3

Visual analogue scale (VAS) scores for pain of injection at baseline and week 1 in the active and sham-treated groups, and p values from repeated-measures analysis of variance for time (weeks 0 and 1)

In the active group, there was no significant correlation (r=−0.003, p=0.99) between the pain of the injection (mean of the two VAS scores for each patient) and response (difference between mean pretreatment and mean post-treatment pain VAS scores).


This study confirmed that SSST is superior to sham therapy in treatment of OA of the first CMC joint. It showed a significant improvement in mean scores for pain and joint tenderness in the active treatment group compared with the sham-treated group. This difference was maintained for up to 13 weeks.

The mechanism of action of this treatment is uncertain. The injected patches were distant from the affected CMC joint. It is possible that the treatment causes a change in the control of pain, perhaps through a modification of central sensitisation. This would be consistent with reports from patients of improvement in symptoms within minutes. The treatment may modify the neurogenic control of inflammation, which may be disturbed in musculoskeletal disease.21 22 This may occur through changes in the expression or transport of neurogenic peptides.23 This could be produced by an irritant effect of the salicylate, similar to that of topical capsaicin.24 25 Up to 100 mg salicylate was given per injection. However, a systemic anti-inflammatory effect is unlikely, since similar injections of salicylate at other sites (ie, away from rheumatic patches or in another limb) fail to produce any effect on the symptoms of OA (CGM-Y, unpublished observations).

Acupuncture has been shown in some studies to reduce the symptoms of OA,26 although the evidence is conflicting.27 It could be that SSST acts in a manner similar to acupuncture, but achieves a more prolonged effect because a substance is injected into the tissues that results in a sustained stimulus. Many of the rheumatic patches occur at standard acupuncture sites.

The chief limitation of this study is the imperfect nature of the control treatment. The sham injections were not as painful as the injections of salicylate, and it is possible that this influenced the symptomatic response of the patients. We did not ask patients if they suspected that they were in a particular treatment group, since the symptomatic outcome and the response to such a question could have influenced each other. The fact that the improvement in the active treatment group was sustained for as long as 13 weeks suggests that an effect greater than placebo was being observed. Furthermore, there was no significant correlation in the active group between the pain of the injections and response in terms of pain in the joint.

The sham treatment that we gave is a recognised control for acupuncture. We felt (and were advised) that administering a more invasive or painful control, such as injection of saline at an irrelevant site would have been unethical. Furthermore, we had already shown a modification of symptoms by injection of saline into rheumatic patches in our earlier pilot study: since the present trial was intended to be sham-controlled, such injections might have compromised its scientific validity.

OA is a common, painful and often debilitating condition, which can have a major financial impact on individuals and healthcare providers. Current medical treatments are limited. SSST is inexpensive, and could be administered by general practitioners and trained nurses. This study indicates that the therapy could be a useful adjunct to existing treatments. It also suggests areas for further scientific enquiry in the search for the regulation of pain and inflammation in OA. Further clinical trials of SSST in OA and other rheumatological conditions are justified. Investigating the pathophysiology of rheumatic patches and the effect of SSST on the neurophysiological control of pain and inflammation (including small fibre function) may help explain the mechanism of SSST, and point towards additional treatments in OA and other painful musculoskeletal disorders.

Main messages

  • Subcutaneous sodium salicylate therapy may be effective in the management of osteoarthritis (OA) of the first carpometacarpal joint.

  • Further clinical studies of the treatment are appropriate.

  • The trial indicates potential areas of research in the control of inflammation and pain in OA.

Current research questions

  • Is subcutaneous sodium salicylate therapy (SSST) effective in other forms of osteoarthritis, and in other painful musculoskeletal conditions?

  • What is the pathophysiology of rheumatic patches.

  • Does SSST modify the neurophysiological control of pain and/or inflammation? If so, how?


This work was supported by a grant from the Peacock Trust.


View Abstract


  • Funding Peacock Trust, 4 Waterstone Close, Itchenor, West Sussex PO20 7BP, UK.

  • Competing interests None.

  • Patient consent Obtained.

  • Ethics approval This study was conducted with the approval of the Hammersmith Hospital Research Ethics Committee.

  • Provenance and peer review Not commissioned; not externally peer reviewed.