AVURT: aspirin versus placebo for the treatment of venous leg ulcers a Phase II pilot randomised controlled trial

Article history

The research reported in this issue of the journal was funded by the HTA programme as project number 13/87/08. The contractual start date was in January 2015. The draft report began editorial review in December 2016 and was accepted for publication in June 2017. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The HTA editors and publisher have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the draft document. However, they do not accept liability for damages or losses arising from material published in this report.

Declared competing interests of authors

Catriona McDaid is a member of the National Institute for Health Research Health Technology Assessment and Efficacy and Mechanism Evaluation Editorial Board. Christine Moffatt has received grant funding from 3M UK PLC and Smith and Nephew, two health science-based technology companies, outside the submitted work.

Copyright statement

© Queen’s Printer and Controller of HMSO 2018. This work was produced by Tilbrook et al. under the terms of a commissioning contract issued by the Secretary of State for Health and Social Care. This issue may be freely reproduced for the purposes of private research and study and extracts (or indeed, the full report) may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising. Applications for commercial reproduction should be addressed to: NIHR Journals Library, National Institute for Health Research, Evaluation, Trials and Studies Coordinating Centre, Alpha House, University of Southampton Science Park, Southampton SO16 7NS, UK.

2018 Queen’s Printer and Controller of HMSO

Chronic venous leg ulcers

Chronic venous leg ulcers (VLUs) are wounds of the lower limb caused by a diseased venous system, which results in swollen legs and damage to the tissues, usually around the ankles. VLUs are most commonly the result of severe varicose veins, a previous deep-vein thrombosis, trauma or failure of the calf muscle pump, all of which result in impaired venous return. Obesity and immobility are additional important factors contributing to venous dysfunction. 1

The VLUs may take many months to heal (with approximately 25% failing to heal completely), during which time they result in significant suffering and reduction in quality of life for patients. 2 VLUs have a tendency to become recurrent, with rates of recurrence estimated at between 18% and 28%. 3 As a result, the management of VLUs represents a substantial cost to the NHS, the majority of which is attributed to nurse time. Estimated lifetime prevalence of VLUs is between 1% and 3% of the elderly population in the USA and Europe. 4 It is estimated that 1% of the adult population will suffer from leg ulcers at some point in their life. 5 Furthermore, incidence and prevalence of ulceration is predicted to increase as a result of the increasing age and obesity of the population in the USA and Europe. A recent cohort study conducted in the UK estimated that 278,000 VLUs per year are managed by the NHS. 6 Furthermore, the annual cost to the NHS of this management was estimated to be £941M, with substantially more cost associated with unhealed wounds. 7

Current treatment strategies

Current treatment strategies for VLUs focus on efforts to reduce venous hypertension. At present, compression is the main treatment for venous ulceration and few additional therapies have robust evidence to suggest they improve healing rates.

Potential role of aspirin as a treatment for venous leg ulcers

Aspirin (also known as acetylsalicylic acid) has been widely used as a medication for > 100 years and is inexpensive, readily available and generally safe to use. Aspirin is a cyclo-oxygenase inhibitor that irreversibly reduces prostaglandin 2 and thromboxane A2. 21 At low doses, it is used very widely to reduce cardiovascular events in those at high risk. 22

The exact mechanism by which aspirin may improve time to healing of VLUs is unclear but it is potentially associated with both the inhibition of platelet activation and the reduction of inflammation. 23

Possible adverse events (AEs) associated with the use of aspirin include gastric ulceration and other gastrointestinal effects. Other effects include liver and renal toxicity, exacerbation of asthma and dermatological reactions. Antiplatelet drugs, when administered in combination with anticoagulants, are associated with a higher risk of gastrointestinal bleeding than that associated with each drug class used alone. 24

Explanation of rationale

The Aspirin for Venous leg Ulcers Randomised Trial (AVURT) was undertaken to address the primary question of whether or not the addition of 300 mg of daily aspirin to standard evidence-based therapies demonstrates evidence of a reduction in time to healing of VLUs. This pilot trial was developed to explore this question as well as assessing the feasibility (especially in terms of participant recruitment and treatment compliance) and safety (in terms of aspirin-related AEs) of conducting a larger-scale pragmatic study, powered to investigate the clinical effectiveness and cost-effectiveness of aspirin for VLU healing.

This research is important because leg ulcers are common and costly and result in significant patient suffering. 2 If aspirin, which is commonly used in many patients, was able to reduce the time to healing of VLU with limited risk of treatment-related harm, then this would result in a potentially important reduction in resource use and an improvement in patients’ health-related quality of life. Because aspirin is generally safe, cheap, well tolerated (for most patients) and widely available, the potential impact on this population is large.

Two previously conducted RCTs have been performed on the use of aspirin in the treatment of VLUs. 25,26 The findings of both trials suggested that there may be benefit in patients with VLUs taking aspirin: one reported that a greater proportion of patients healed with 300 mg of aspirin together with standard compression bandaging25 and one reported a shorter time to healing with 300 mg of aspirin in conjunction with gradual compression therapy. 26 However, both trials have been assessed as being at a risk of bias. 23 The authors of the Cochrane review23 concluded that the low-quality and insufficient evidence from the two included trials meant that they were unable to make definitive claims on the benefits and potential harm of oral aspirin, as an adjunct to compression therapy, on the recurrence and healing of VLUs. The Cochrane review23 recommended that further high-quality studies were needed.

A RCT is required to assess the potential effectiveness and safety profile of aspirin in this population. However, it would be premature to conduct a full trial initially, not least as it is not clear how many people with VLUs currently take aspirin or other antiplatelet medications and the potential impact of this on the design and feasibility of any future study.

During the registration of AVURT, we identified two other RCTs investigating aspirin for VLUs. ASPiVLU (ASPirin in Venous Leg Ulcer Healing)28 was a trial being conducted in Australia that was planning to randomise patients with VLU to receive either 300 mg of aspirin or placebo. The primary end point of that study was time to complete healing of reference ulcer at or before 12 weeks post randomisation. Aspirin4VLU (Low Dose Aspirin for Venous Leg Ulcers) was a trial being conducted in New Zealand that was planning to randomise VLU patients to either 150 mg of aspirin or placebo in addition to standard care. The primary end point was time to complete healing of reference/largest ulcer. A secondary outcome of this study is change in estimated reference ulcer area from baseline to 24 weeks. 29

Research objectives

To assess the efficacy of aspirin on time to healing of VLUs, to examine safety issues in this cohort of patients and to assess the feasibility of proceeding from a Phase II trial to a Phase III trial of clinical effectiveness and cost-effectiveness.

Design

A multicentred, pilot, Phase II randomised double-blind, parallel-group, placebo-controlled efficacy trial.

Setting

Patients presenting at community leg ulcer clinics/hospital outpatients’ clinics, or registered with a leg ulcer clinic but receiving care at home, were recruited. Some sites could use patient identification centres (PICs) to identify patients to take part. Participants were recruited from 10 centres in England, Wales and Scotland (see Appendix 1) from leg ulcer hospital outpatient clinics (n = 5), community leg ulcer clinics or community caseloads (n = 3), a wounds clinic in a university (n = 1) and a primary care leg ulcer clinic (n = 1). At each of the nurse-led community centres (n = 3), a doctor was identified to work with the centre to review and confirm the patient’s eligibility for the trial, to prescribe the investigational medicinal product (IMP) and to review changes to concomitant medication.

Participants

Participant eligibility for the trial was assessed according to the criteria below.

Recruitment

Patients were pre-screened on the basis of three criteria (concomitant aspirin, wound size and ulcer duration or history of venous ulceration) by study research nurses to determine those potentially eligible for the study. The reason(s) for ineligibility or not approaching patients were recorded on pre-screening logs (see Appendix 2). Two pre-screening logs were issued. Completion of the first log (version 1.0) was non-mandatory as stipulated by the trial sponsor (based on the sponsor’s belief that the clinics received a heterogeneous referral pattern of mixed aetiology ulcers not thought to be truly representative of the total population of patients with chronic VLUs). Following a recommendation by the Data Monitoring Committee (DMC), the sponsor permitted a new pre-screening log that was made mandatory (version 2.0). Patients attending clinics as part of their routine care and who satisfied the pre-screening criteria were approached by study research nurses or designated health-care professionals and provided with both verbal and written information about the trial in a face-to-face meeting (see Appendix 3). Patients were given a minimum of 24 hours to consider participation in the trial. Study research nurses then obtained voluntary full written consent from those patients who wanted to enter the trial (see Appendix 4). After they gave consent, patients were screened against the study’s full eligibility criteria by the study research nurses or designated health-care professionals using the screening case report form (CRF) (see Appendix 5). The reason(s) for a patient’s ineligibility were recorded. Patients were informed that their eligibility would be subject to confirmation by a medical practitioner and in all cases a medical practitioner determined and confirmed patient eligibility following screening. If a potential participant was not known to the medical practitioner, provision was made for the participant to be contacted by telephone by the medic to check for any possible contraindications. When the medic was satisfied of patient eligibility, they would sign off the prescription for the IMP (see Appendix 6).

Randomisation

Patients were randomly allocated in a 1 : 1 ratio to either aspirin or placebo by the Research Pharmacy (St George’s University Hospitals NHS Foundation Trust, London, UK). Randomisation was stratified according to ulcer size (≤ 5 cm² or > 5 cm²) as this is the strongest known predictor of outcome. 4

Interventions

Investigational medicinal product supply

The sponsor had responsibility for the order and purchase of trial medication and for arranging labelling of medication for the trial with St George’s University Hospitals NHS Foundation Trust.

Outcomes

Baseline assessment

Following confirmation of a participant’s eligibility, and before randomisation, a baseline assessment was conducted by the study or research nurse using the baseline CRF (see Appendix 7).

Outcome assessments

Participants were followed up weekly or fortnightly, depending on their usual pattern of attendance at clinic, for a minimum of 25 weeks post randomisation. Participants were not asked to make any additional visits for the purposes of the trial. The participant weekly data collection file, made up of CRFs and forms, was completed during follow-ups by research nurses or treating nurses (see Appendix 10). In addition, recorded in the file were the weeks in which participants missed or did not have an appointment, the randomisation date, the date that the first IMP dose was taken and the time of day that the IMP was generally taken. A summary flow chart of participant follow-up is shown at Appendix 11.

Planned participant follow-up was for 25 weeks post randomisation, but participants who had a wound initially judged as healed in week 24 or 25 were followed up for 2 further weeks (26 and 27 weeks post randomisation, respectively) to confirm healing. Table 1 summarises the schedule of assessments. All anonymised completed CRFs were faxed or sent via the University of York’s secure electronic system to the YTU.

Measurement and verification of primary outcome measure

Measurement of secondary outcomes

Patient safety

Each participant was regularly reviewed by their treating nurse and/or physician working closely with the AVURT research team and was continually assessed for any increased dyspepsia, other gastrointestinal symptoms, skin rashes and any other possibly linked AEs that could be attributable to the IMP.

Withdrawal

Participants were deemed to have exited the trial when they:

• withdrew consent

• were lost to follow-up

• died

• had completed follow-up (i.e. 25 weeks post randomisation or, for patients whose leg ulcer was first assessed as healed in weeks 24 and 25, weeks 26 and 27, respectively).

If a participant chose to withdraw from the trial then reasonable effort was made to establish the reason for this withdrawal. For participants leaving the trial before final follow-up, nurses completed a change to study status form (see Appendix 10, Form F), giving the main reason for the participant’s exit. No further follow-up data were collected. Participants withdrawing from the study were given the option for their data not to be used.

Participants stopped treatment for any one of the following reasons, but continued with follow-up:

• Unacceptable treatment toxicity that, in the investigator’s opinion, is attributable to the IMP or a SAE.

• Intercurrent illness that prevents further protocol treatment.

• Any change in a participant’s condition that, in the investigator’s opinion, justified the discontinuation of treatment.

• If a participant became pregnant or suspected that they were pregnant.

• Reference ulcer confirmed as healed.

• A participant chose to discontinue treatment.

Participants whose leg ulcer was initially assessed as healed were encouraged to take the IMP during the 2-week observation period. If healing was confirmed after 2 weeks, the participant stopped taking the trial medication and follow-up was suspended until a final follow-up in week 25.

Sample size

The target sample size was 100 participants. This sample size is sufficient to test the feasibility of study procedures, such as recruitment and retention, and is large enough to demonstrate whether or not there is evidence for efficacy in line with two previous trials of aspirin for leg ulcers. 25,26

The primary outcome was time to healing of the largest eligible leg ulcer (reference ulcer). Ulcer area and duration of ulcer are known prognostic factors for healing. In a previous leg ulcer study, Venous leg Ulcer Study IV (VenUS IV), after adjustment for log-area of ulcer and log-duration of ulcer, the standard error for the time to healing estimate was 0.105, with data on 448 participants. 3 Applying this to a smaller sample of 100 participants implies that the standard error of such a sample would be increased to 0.22 [obtained from 0.105 × √(448/100)]. A 95% confidence interval (CI) for the log-hazard ratio (HR) would thus be the estimate of the log (HR) ± 1.96 × 0.222 = log(HR) ± 0.435. The antilog of this is 1.54 and the 95% CI for the HR would be the observed value divided or multiplied by this. Hence, if our HR were the same as that suggested by the existing studies (i.e. about 1.5), then our CI would be 0.97 to 2.31, which just includes 1.00. It would be unlikely that, if the HR is as these two previous smaller studies suggest, we would observe an overall HR of < 1.00. Compliance and follow-up were measured as part of the study and so there is no formal inflation of the recruitment target for drop out.

A secondary outcome was change in wound area. Using data from the Venous leg Ulcer Study I (VenUS I) of compression bandaging,40 ulcer area was measured for 245 participants who were measured within 60 days of recruitment. Ulcer area has a highly skewed distribution, so we calculated a difference in log-area at follow-up, after adjustment for log-ulcer area at baseline and time elapsed until follow-up. The residual standard deviation (SD) was 1.09. Two groups of 50 participants would give us 80% power to detect a difference of 0.62 on the natural-log scale, corresponding to a reduction of 46% in ulcer area at follow-up. In the current study, we had multiple measurements of wound area and so predicted that we should be able to detect smaller differences.

Statistical methods

The statistical methods for the analysis of the trial data were prespecified and detailed in a statistical analysis plan (SAP) before the completion of data collection. The SAP was prepared by the trial statisticians and reviewed by members of the Trial Management Group (TMG) and DMC. However, given that the final number of participants randomised was much lower than the 100 planned (n = 27), many of the pre-planned analyses were infeasible or inappropriate. In this section, we describe the analyses as performed, highlighting any deviations from the SAP.

Approvals obtained and governance

Patient and public involvement

At the grant application stage, the views of six patients attending a leg ulcer clinic were elicited. Specifically, they were asked for their views on the likely willingness of patients to take aspirin on a daily basis, given its possible side effects, if it were shown to improve healing of leg ulcers. All responded that they would be willing to take medication if it meant that their ulcer was likely to heal more rapidly. They thought that the risks of aspirin were acceptable given that many patients already take it regularly for cardiovascular disease. In terms of a trial, they thought that they would be happy to receive a dummy tablet (placebo) if it meant that more information could be gleaned about the efficacy of aspirin in terms of healing ulcers – even though a further larger trial might be required to confirm the results. Some patients questioned the benefit about taking a high dose of aspirin. However, the feeling expressed by some was that the perceived increased risks would be worthwhile if it significantly decreased time to healing.

There were two patient coapplicants. Ellie Lindsay, president of the Leg Club Foundation, was a member of the TMG, and our other patient representative, Laurie Williams, was a member of the TSC. Both were involved in the development of the trial during the application stage and throughout the study. They were also involved in the development of the trial’s patient information resources.

Protocol amendments

Amendments to the protocol were required by REC prior to approval. Following approval, no substantial amendments were made to the protocol. Details of all ethics and MHRA amendments are detailed in Appendix 13.

Site recruitment

Ten sites opened to recruitment. Prior to recruitment, the sites indicated the approximate number of participants they could recruit (see Appendix 1). Recruitment was largely based in leg ulcer community clinics and hospital outpatient clinics. Many of the recruiting sites were chosen as they had been high recruiters to other leg ulcer studies.

Eleven sites were initially interested in participating. Nottingham University Hospitals NHS Trust subsequently declined after undertaking a complete screening review of its patient population, which consisted of 3300 patients referred with chronic oedema of all forms with many suffering from venous leg ulceration. An analysis of their patient profile indicated that they would have little access to non-complex patients. The remaining 10 sites were submitted to REC for approval:

1. St George’s Healthcare NHS Trust London

2. Bradford Teaching Hospitals NHS Foundation Trust

3. Leeds Community Healthcare NHS Trust

4. Newcastle upon Tyne Hospitals NHS Foundation Trust

5. Cardiff & Vale University with Aneurin Bevan University Health Board (Newport)

6. Hull and East Yorkshire Hospitals NHS Trust

7. Harrogate and District NHS Foundation Trust

8. Mid Yorkshire Hospitals NHS Trust (Wakefield)

9. Lancashire Care NHS Foundation Trust

10. Sussex Community NHS Trust (Brighton).

Sites were opened throughout the participant recruitment phase and, of these sites, three (Bradford, Leeds and Wakefield) did not open but were in various stages of contracts, training, site initiation visits and approvals when the trial closed to recruitment. During the recruitment phase, we received interest from three other sites that opened after they received REC and local research and development (R&D) approvals: NHS Tayside (Dundee), NHS Lanarkshire and Kent Community Health NHS Foundation Trust. We were also in discussion with Birmingham Community Healthcare NHS Trust towards the end of the recruitment phase.

Participant recruitment

Once open to recruitment there were fewer than expected eligible patients at sites. The pre-screening data from sites indicated various and multiple reasons for patients not being approached (Figure 1). The main reasons for participant ineligibility were:

• already taking aspirin or other prohibited medication

• having a small or otherwise ineligible ulcer.

FIGURE 1.

The AVURT pre-screening study flow. a, The actual number of patients pre-screened was higher than 457 as the first version of the pre-screening log was not mandatory; and b, reasons not mutually exclusive.

In the last few months of recruitment, when all sites were open, the trial was recruiting three to five participants per month (see Appendix 14). It was generally thought by some sites that a large proportion of leg ulcer patients were receiving treatment in general practices (i.e. in primary care) or outside primary care in specialist clinics and by district nurses. It was therefore likely that the patients being seen by the secondary recruiting sites were older, more likely to have mixed disease and, therefore, more likely to be already taking aspirin.

Strategies to improve recruitment

Strategies to improve recruitment were explored. Modification of the eligibility criteria was assessed with reference to the pre-screening log data. The only acceptable modification to the exclusion and inclusion criteria was to include a smaller wound size. However, those with a wound area of < 1 cm² were excluded, as these ulcers usually heal very rapidly.

In October and November 2015, the chief investigator and a trial manager contacted the recruiting sites that had been the first to open (Kent, Hull, Brighton, Harrogate and Newcastle) to discuss possible solutions to improve participant recruitment. During meetings with sites, a number of ideas were discussed, including:

• Advertising via social media.

• Posters to inform patients about the study.

• Newsletter for sites.

• Flyers for sites to remind staff to recruit to the study.

• Radio advertising (Hull and Harrogate).

• An amendment to protocol to allow participants to visit sites for follow-up purposes (the protocol stated that patients will not be invited to attend clinics for research purposes), and to introduce per-patient payments for visits that were not part of routine care. This was a particular problem for one site (Newcastle) that did not routinely see patients in clinic after their initial visit.

• Allowing for more telephone follow-ups so that participants did not need to come in to clinic as regularly as once per week or once per fortnight.

• Remove minimum ulcer size from the eligibility criteria, in line with the study being conducted in New Zealand. 29

Apart from removing the minimum ulcer size criterion, each of the options considered would have potentially benefited just one or two of the recruiting sites and, therefore, a variety of strategies would need to be implemented across the trial. The funder and REC had required for regular follow-up to monitor patient safety and, therefore, less frequent follow-up was not viewed as a feasible option by the research team.

A flyer was produced and sent to sites to remind clinic staff to recruit to the trial (December 2015) and three electronic newsletters were sent to sites (two during recruitment, in October 2015 and December 2015) to update on participant recruitment in the trial (the third newsletter was sent in June 2016 after recruitment had closed). The flyer and newsletters were relatively cheap to produce and did not require ethics approval and so could be sent out swiftly. During training, sites were reminded to identify potential participants before opening so that the participants could be approached as soon as the sites were given the green light to recruit.

Summary

There were external factors outside the control of the research team that meant that sites were slow to open. Nine out of the 10 recruiting sites were based in secondary care and, once open, there were fewer than anticipated eligible participants. As this was a Clinical Trial of an Investigational Medicinal Product (CTIMP) study, it required more input from a doctor where nurses would otherwise often take the lead. The site make up was very different from other wound trials in which almost all sites were community based with tissue viability nurses acting as principal investigator, which was not possible here. A range of options were considered and explored to improve recruitment, including recruitment from primary care. Preliminary searches of records in primary care also indicated very few potentially eligible participants. In addition, the trial’s short recruitment window and budget constraints meant that many of the options considered to improve recruitment were not viable without a funded extension.

Participant flow

The CONSORT diagrams in Figures 1 and 2 show the flow of participants pre-screened and the flow of eligible participants during the trial. Figure 1 illustrates the number of patients pre-screened (n = 457) and the number of those who consented (n = 20). Pre-screening data were unavailable for the nine remaining patients who consented. Pre-screening was under-reported because some sites did not complete the first pre-screening log. The flow of participants in the trial is illustrated in Figure 2, which shows that 29 patients consented. The number of patients randomised by treatment group, receiving the intended treatment, completing the study protocol, and analysed for the primary outcome is presented. Two patients were excluded after they gave consent: one patient developed potential gastric problems and the other patient, affected by multiple comorbidities, was admitted to hospital.

FIGURE 2.

The AVURT CONSORT diagram.

Recruitment

Recruitment took place over 8 months, 2 months longer than originally scheduled. The trial opened on 23 June 2015 and the first participant was recruited on 13 July 2015. Recruitment closed on 29 February 2016. In total, 10 sites were opened and eight randomised a total of 27 patients (Table 2). The sites in Cardiff and Dundee did not recruit any participants. For accumulative recruitment over time see Appendix 14.

Baseline data

The baseline participant and ulcer-related characteristics, as well as the baseline treatments, are shown in Tables 3–5. The average age of the 27 randomised participants was 62 years (SD 13 years) and two-thirds were male (n = 18). Participants had had their reference ulcer for a median of 15 months and the median size of ulcer was 17.1 cm². All participants were receiving compression therapy at baseline.

Withdrawals and losses to follow-up

Withdrawals and losses to follow-up were recorded on a change of status form (see Appendix 10, Form F) for five patients (placebo, n = 3; aspirin, n = 2). However, there was only one withdrawal during the course of the study (who was in the placebo group) for whom data on primary outcome was not possible to obtain. This was reported at week 2 and so no follow-up data are available for this participant beyond week 1. The other four participants (placebo, n = 2; aspirin, n = 2) either agreed to withdraw from treatment but provided full follow-up until week 25 (i.e. the end of planned follow-up) or healed at a point before withdrawal and, thus, all four provided primary outcome data.

Overall, this means that, in terms of analyses related to primary outcome, only one patient was not included (see Figure 2).

Primary outcomes

Overall, 13 out of the 26 (50.0%) participants followed up were recorded as healing during the course of the study. All the reference ulcers reported to be healed were later confirmed healed approximately 2 weeks later. The first date of reported healing (as reported on the weekly CRF data collection) was used in the statistical analysis (as per the SAP).

Over the course of the trial, 7 out of 12 participants (58.3%) followed in the placebo group were observed to have a healed reference ulcer, and in the aspirin group the corresponding figure was 6 out of 14 (42.9%). It was not possible to estimate median time to healing and/or corresponding 95% CIs as < 50% of participants healed during the 25- to 27-week maximum follow-up period of the study (Table 6). Therefore, the 25th percentile of time to healing was also estimated. Figure 3 shows the unadjusted Kaplan–Meier plot of proportion of reference ulcers healed over time. The unadjusted log-rank test investigating the difference between the survival curves showed no statistically significant difference (test statistic = 1.02; p = 0.30).

TABLE 6 - Healing of the reference ulcer (unadjusted analysis)

NE, not possible to estimate.

One participant was lost to follow-up immediately after randomisation and provided no outcome data.

Outcome	Placebo (N = 12)a	Aspirin (N = 14)	Overall (N = 26)
Number healing, n (%)	7/12 (58.3)	6/14 (42.9)	13/26 (50.0)
Kaplan–Meier estimate of median time to healing (days) (95% CI)	98 (21 to NE)	NE (84 to NE)	147 (97 to NE)
Kaplan–Meier estimate of 25th percentile time to healing (days) (95% CI)	36 (20 to 97)	111 (69 to NE)	84 (21 to 111)

FIGURE 3.

Kaplan–Meier plot of time to ulcer healing by trial arm (unadjusted).

The primary analysis as written in the SAP investigates difference in time to healing using a Cox model adjusting for baseline ulcer area, baseline ulcer duration and centre. The covariates in this model were to be baseline area and duration of the reference ulcer (planned log transformation), randomised allocation and centre as a shared frailty effect. The model was originally chosen with a view to recruiting the target of 100 participants (if successful to follow with a larger definitive study) and was based on analyses performed for the VenUS IV study, which enrolled 457 participants. Given the final distribution of participants across centres (26 participants across eight centres, with six centres contributing three or fewer participants), the shared frailty model is an impractical choice for this analysis. Adjustment with centre as a covariate alongside ulcer area and duration would yield a very low number of events per variable (at approximately four events) and lower than the 10 events that has been previously recommended from simulation studies using Cox regression. 43 Therefore, we present the Kaplan–Meier curves (see Figure 3), log-rank test and unadjusted Cox regression as planned but caution against over-interpretation; we also present the results from the Cox regression adjusted for log-area and log-ulcer duration but without adjustment for centre, giving HRs and corresponding 95% CIs (Table 7).

TABLE 7 - Healing of the reference ulcer (log-rank test, unadjusted and adjusted analysis)

df, degrees of freedom.

Parameter	HR (95% CI)	p-value
Unadjusted Cox regression
Aspirin vs. placebo (allocation)	0.58 (0.19 to 1.72)	0.322
Adjusted Cox regression
Aspirin vs. placebo (allocation)	0.58 (0.18 to 1.85)	0.357
Area (log-transformed)	0.42 (0.22 to 0.81)	0.009
Duration (log-transformed)	0.61 (0.34 to 1.08)	0.089

Overall, these data do not provide evidence of a difference in time to healing with the addition of aspirin to usual care. The placebo group tended to heal more rapidly but this difference is not statistically significant. The numbers within this feasibility study are small and the results are inconclusive in terms of the primary outcome.

Secondary outcomes

Ulcer-related pain

The mean baseline VAS score for ulcer-related pain was 37.7 (95% CI 22.0 to 53.4) in the placebo group and was slightly higher at 45.4 (95% CI 24.6 to 66.2) in the aspirin group. At week 5, VAS scores had reduced in both groups with the mean VAS score at week 5 at 13.3 (95% CI 0.3 to 26.3) in the placebo group and 28.5 (95% CI 10.6 to 46.3) in the aspirin group (Table 9 and Figure 4).

FIGURE 4.

Plot of the mean VAS pain score at baseline and at week 5 by trial arm.

Ulcer area

Seven out of the 27 participants (26%; placebo, n = 3; aspirin, n = 4) had missing data for baseline ulcer area, as measured by the analysis of a photograph, and so the measure that was calculated manually by the research nurses was used instead for these participants. Photographs were not available for three measurements during follow-up and so the area calculated from tracings was also used in these cases. Participants who healed during the course of the study contributed to the computation of the ulcer area until healing or 2 weeks after healing, according to the availability of an analysable photograph. Table 12 reports, by trial arm, the number of participants with a valid measure of ulcer area, and the mean and the SD of the ulcer area for each week. The mean ulcer area fluctuates more widely in the placebo group than in the aspirin group (Figure 5) with a distinct fortnightly pattern. These fluctuations are caused by two participants who attended their clinic appointments fortnightly and, coincidentally, in the same calendar weeks, and whose ulcer sizes were particularly large compared with those of the other participants in the placebo group. For example, at baseline, the area of their ulcers was 168 cm² and 129 cm², while the mean of the ulcer area of the other placebo participants at baseline was 19.1 cm² (minimum of 2.14 cm² and maximum of 78.0 cm²). One of these two participants had an ulcer that was extended around the back of the leg, causing difficulties in the estimation of the area. Figure 6 presents the mean ulcer area over time for the two groups, but with the measurements from these two participants removed.

TABLE 12 - Mean of ulcer area by visit week and allocation group

No estimate of SD, as data were only available for one participant.

Visit week	Ulcer area (cm2)
	Placebo (N = 13)		Aspirin (N = 14)		Overall (N = 27)
	n	Mean (SD)	n	Mean (SD)	n	Mean (SD)
0	13	39.0 (53.1)	14	33.8 (37.1)	27	36.3 (44.7)
1	9	22.9 (31.2)	11	29.7 (25.4)	20	26.6 (27.6)
2	10	29.8 (43.5)	12	23.8 (19.2)	22	26.5 (31.9)
3	8	15.7 (12.8)	11	39.2 (42.6)	19	29.3 (34.9)
4	10	32.7 (48.4)	10	38.7 (38.9)	20	35.7 (42.8)
5	6	5.9 (7.0)	12	24.9 (24.2)	18	18.6 (21.9)
6	8	32.8 (52.3)	7	28.1 (25.9)	15	30.6 (40.8)
7	5	10.0 (6.9)	10	16.9 (18.6)	15	14.6 (15.7)
8	8	35.1 (53.4)	10	23.3 (14.5)	18	28.5 (36.4)
9	5	7.2 (7.5)	12	22.3 (16.7)	17	17.9 (16.0)
10	9	27.1 (43.8)	12	26.0 (24.9)	21	26.5 (33.3)
11	5	7.3 (7.9)	12	23.6 (24.6)	17	18.8 (22.1)
12	6	37.9 (51.3)	11	26.3 (22.3)	17	30.4 (34.1)
13	3	7.7 (9.5)	11	31.4 (31.0)	14	26.3 (29.3)
14	7	31.7 (42.8)	9	25.2 (22.5)	16	28.1 (31.8)
15	5	7.0 (8.1)	8	25.0 (18.9)	13	18.1 (17.7)
16	5	33.0 (37.1)	9	26.4 (18.6)	14	28.7 (25.5)
17	1	89.2 (.)a	8	37.5 (28.8)	9	43.2 (32.0)
18	4	24.2 (22.6)	8	25.8 (19.7)	12	25.2 (19.7)
19	3	14.2 (8.5)	5	27.6 (17.2)	8	22.6 (15.4)
20	4	41.0 (33.7)	6	26.7 (12.2)	10	32.4 (22.7)
21	2	11.5 (10.2)	6	44.8 (31.2)	8	36.5 (30.8)
22	4	39.4 (33.8)	6	22.5 (16.6)	10	29.2 (24.7)
23	2	10.9 (7.9)	7	29.1 (19.3)	9	25.0 (18.7)
24	3	34.9 (43.7)	7	27.3 (20.8)	10	29.6 (27.0)
25	2	12.2 (12.7)	7	33.0 (20.5)	9	28.4 (20.5)
26	0	–	0	–	0	–
27	0	–	1	4.8 (.)a	1	4.8 (.)a

FIGURE 5.

Mean of ulcer area and 95% CI for each week of follow-up stratified by allocation group (lower confidence limits truncated at zero).

FIGURE 6.

Mean of ulcer area and 95% CI for each week of follow-up stratified by allocation group without the two placebo group participants with rather extended ulcers (lower confidence limits truncated at zero).

Summary of findings

Owing to under-recruitment, we were unable to confirm the efficacy of aspirin for VLU healing. Recruitment was more difficult than anticipated owing to the large number of patients already prescribed aspirin medication, predominantly for cardiovascular risk factor management, or who were on concomitant antiplatelet therapies. Others were excluded because they had contraindications to aspirin therapy, had small ulcers of ≤ 1 cm² (that were anticipated to heal rapidly) or did not want to be enrolled (frequently reported to be associated with the need for regular clinic attendance).

Participants included in the trial had ulcers of a significantly long duration and may have been considered more difficult to heal. The ulcers that most participants had were large, with 80% of them having a surface area of > 5 cm².

The relatively small number of participants recruited to the trial means that any data should be interpreted with caution. There were a large number of AEs during the trial (n = 89) with most participants (77%) suffering at least one. The majority of these AEs were non-serious (n = 88) and among these 51 (58%) were not related to aspirin. However, there was one SAE of gastrointestinal bleeding requiring blood transfusion. Aspirin was generally well tolerated and there was no evidence of a difference in the number of experienced AEs in the two trial arms.

Compliance with the medication was good, with nearly three-quarters of participants being fully compliant and one-quarter partially compliant. Similarly, the compliance to compression therapy was very good overall and was similar between the treatment groups.

Before study commencement, a decision was made not to collect any data on health-related quality of life and collect only limited resource use health economic data. This was for reasons of brevity and the focus of the trial being feasibility. However, it was hypothesised that participants may receive some benefit in terms of pain relief if they received 300 mg of aspirin daily. The self-reported pain scores at 5 weeks showed no evidence that participants in the aspirin group suffered less pain. The number of clinic visits and dressing/compression changes were also similar among the groups.

The patients taking part in the study tended to have a high BMI, some were experiencing a high degree of ulcer-related pain and the majority receiving high-compression bandaging. A few months into recruitment, and finding that many of the patients were not meeting the trial criteria, we explored recruitment from primary care, which involved a limited database search of records held in primary care. However, very few patients were identified as there were limitations associated with conducting the search on the trial’s eligibility criteria. There may have been potentially eligible patients being treated by nurses in general practices. Further investigation of recruitment from primary care was not undertaken owing to time and budget constraints.

Limitations

There are a number of limitations of the trial. Because the trial under recruited (by 73%), some of the pre-planned analyses were infeasible or inappropriate. Because of the low number of participants, for example, the shared centre frailty effects could not be tested and the repeated measures mixed model for ulcer area could not be estimated. In these cases, the most appropriate analyses were performed. Information concerning compliance with both medication and compression therapy was obtained by participant self-report and pill count.

At the outset of the trial the sponsor did not enforce recruiting sites to record pre-screening log data owing to the disparate nature of the clinics seeing and recruiting patients. It was felt by the sponsor that the heterogeneous nature of the clinics and, therefore, patients would render a pre-screening log meaningless. For example, one site, at St George’s, offered a complex wound service led by a vascular surgeon (high prevalence of PAD and very chronic wounds). In contrast, other sites, such as the one in Brighton, were effectively based in the community with a large number of patients with very small and less complex ulcers that were managed solely by nurses. This omission was rectified later but, consequently, the pre-screening data reported is an under-representation of the number of patients screened. It was not possible to conduct second checks of the pre-screening data and there is the possibility of duplicate and incomplete entries. In addition, the data should be interpreted with caution owing to major differences among demographics between some of the recruiting sites.

It is worth reflecting that the relative rates of healing in this study are very far removed from those seen in the earlier UK and Spanish studies,25,26 which prompted this call for research. One explanation for this may have been the much larger ulcers in this trial or, perhaps, their more chronic nature.

Strengths

Retention and follow-up rates were high and there were few missing data. Both the trial coapplicants and the REC were concerned that participants may suffer a significant number of aspirin-related AEs and SAEs, but this did not appear to be the case. In fact, the number of related events was quite low and the medication appeared to be well tolerated. The frequency of follow-up, once weekly or fortnightly, was to ensure, in part, that information on AEs was identified early (perhaps before progression to more serious AEs). It is possible that some participants suffered other AEs that were related to aspirin, but this seems unlikely given the frequency of assessment.

The aspirin and placebo were manufactured and over-encapsulated effectively in a large capsule. There were some concerns that this may affect participant compliance. However, the sponsor’s previous experience with IMP manufacturer and capsule size was reassuring. Indeed, in this trial participant compliance with medication was generally very good and non-compliance was not associated with the size of the capsule.

Data were captured on ulcer area using two formats: paper tracings and digital image analysis. Both of these techniques are supported by published data3,44 on their reliability and have been used extensively in clinical trials previously. We also decided to use digital image analysis as we thought it may prove easier for the nursing staff. However, many nurses felt more comfortable with wound tracing, especially with the larger ulcers, which may be more difficult to capture in a two-dimensional digital photograph. At the most important time points (baseline and completion of the study), wound tracings were taken to avoid any problems associated with two-dimensional image analysis.

Interpretations

Owing to under-recruitment, we were unable to confirm the effect sizes found in two other published studies, and reliably and confidently establish the safety of aspirin in this population.

It may have been possible to recruit more patients to the trial if other centres could have been rapidly involved. However, the very short nature of the trial meant that it was impossible to involve other centres in a timely fashion.

Generalisability/contribution of this study to the evidence

AVURT was based on plausible results from two other studies investigating aspirin for VLUs. 25,26 This trial has demonstrated that it was possible to randomise those participants who could be recruited, but the study clearly demonstrated that it was not feasible to recruit the necessary participants in the context of a RCT in the UK.

Trial Steering Committee members

We would like to thank external members of the Trial Steering Group: Professor Julie Brittenden, Professor Andrea Nelson, Mrs Angela Oswald (leg ulcer nurse practitioner), Professor Janet Powell and Mr Laurie Williams (coapplicant and patient representative).

Data Monitoring Committee

We would like to thank external members of the DMC: Professor Peter Franks, Dr Sarah Brown, Mr Jonathan Earnshaw and Mr Toby Richards.

Contributions of authors

Helen Tilbrook (Research Fellow) was the lead study trial co-ordinator. She contributed to the development of the trial protocol and the first draft of the report.

Laura Clark (Research Fellow) contributed to the co-ordination of the study.

Liz Cook (Trial Co-ordinator) contributed to the co-ordination of the study and contributed to the first draft of the report.

Martin Bland (Professor of Health Statistics Emeritus) contributed to the development of the grant application and trial protocol and provided statistical expertise.

Hannah Buckley (Medical Statistician) contributed to the design of the study and wrote the SAP.

Ian Chetter (Associate Dean for Research Hull York Medical School/University of Hull) contributed to the development of the grant application and trial protocol and clinical expertise. Professor Chetter was also a principal investigator.

Jo Dumville (Senior Lecturer in Health Sciences) contributed to the development of the grant application and trial protocol and had project oversight.

Chris Fenner (Core Surgical Trainee) reviewed ulcer photographs and calculated ulcer area.

Rachael Forsythe (Clinical Research Fellow) reviewed ulcer photographs and calculated ulcer area.

Rhian Gabe (Reader in Clinical Trials and Senior Trial Statistician) contributed to the development of the grant application, reviewed the SAP and provided statistical oversight.

Keith Harding (Clinical Lead for Wound Healing) contributed to the development of the grant application and trial protocol and provided clinical expertise.

Alison Layton (Consultant Dermatologist) contributed to the development of the grant application and trial protocol, provided clinical expertise and was also a principal investigator.

Ellie Lindsay (President, Leg Club Foundation) contributed to the development of the grant application, trial protocol and participant information, and had project oversight as a member of the TMG.

Catriona McDaid (Senior Research Fellow) contributed to the development of the grant application and trial protocol and had project oversight.

Christine Moffatt (Professor of Clinical Nursing Research) contributed to the development of the grant application and trial protocol.

Debbie Rolfe (Acting Head of Research Governance and Regulatory Assurance Manager) was sponsor representative for the study, contributed to the design of the study protocol and was also responsible for project oversight and monitoring of the study.

Illary Sbizzera (Trainee Statistician) undertook the statistical analysis and contributed to the first draft of the report.

Gerard Stansby (Professor of Vascular Surgery and Honorary Consultant Surgeon) contributed to the development of the grant application and trial protocol and was a principal investigator.

David Torgerson (Director of YTU) provided advice on trial conduct and contributed to the design of the grant application and trial protocol.

Peter Vowden (Honorary Clinical Director NIHR WoundTec HTC) contributed to the development of the grant application and trial protocol and was a principal investigator.

Laurie Williams (patient representative and member of AVURT TSC) contributed to the development of the grant application, trial protocol and participant information, and had project oversight as a member of the TSC.

Robert Hinchliffe (Honorary Consultant in Vascular Surgery) lead applicant and chief investigator for AVURT. He had overall responsibility for the design and implementation of the study and the writing of the first draft of the report with final approval of report submission.

All authors contributed to the final manuscript.

Data-sharing statement

All data requests should be submitted to the corresponding author for consideration. Access to available anonymised data may be granted following review.

Patient data

This work uses data provided by patients and collected by the NHS as part of their care and support. Using patient data is vital to improve health and care for everyone. There is huge potential to make better use of information from people’s patient records, to understand more about disease, develop new treatments, monitor safety, and plan NHS services. Patient data should be kept safe and secure, to protect everyone’s privacy, and it’s important that there are safeguards to make sure that it is stored and used responsibly. Everyone should be able to find out about how patient data are used. #datasaveslives You can find out more about the background to this citation here: https://understandingpatientdata.org.uk/data-citation.

Disclaimers

This report presents independent research funded by the National Institute for Health Research (NIHR). The views and opinions expressed by authors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, NETSCC, the HTA programme or the Department of Health and Social Care. If there are verbatim quotations included in this publication the views and opinions expressed by the interviewees are those of the interviewees and do not necessarily reflect those of the authors, those of the NHS, the NIHR, NETSCC, the HTA programme or the Department of Health and Social Care.

Ethics submissions

Amendment 1. Substantial (1). Approval not required.

Amendment 2. Substantial (2) and non-substantial (1). Approved 7 May 2016.

Amendment 3. Non-substantial (2). Approved 4 August 2015.

Amendment 4. Substantial (3). Approval not required.

Amendment 5. Substantial (4). Withdrawn.

Amendment 6. Non-substantial (3). Approved 25 July 2016.

Medicines and Healthcare products Regulatory Agency submissions

Amendment 1. Substantial (1). Approved 8 May 2016.

Amendment 2. Substantial (2) and non-substantial (1). Approval not required.

Amendment 3. Non-substantial (2). Approval not required.

Amendment 4. Substantial (3). Approved 14 December 2015.

Amendment 5. Substantial (4). Approval not required.

Amendment 6. Non-substantial (3). Approval not required.

Non-substantial amendments

Substantial amendments