Eutectic mixture of local anaesthetics (EMLA®) as a primary dressing on painful chronic leg ulcers: a pilot randomised controlled trial

Study design

This feasibility study was a pilot, parallel group, non-blinded, randomised, controlled trial (RCT). The study protocol was approved by the Northern Sydney Health (AU RED Ref. HREC/09/HARBR/162) and the Griffith University Human Research (GU Ref No: NRS/16/12/HREC) Ethics Committees (HREC), registered with the Australian New Zealand Clinical Trials Register (ACTRN12609001080213) and conducted in accordance with the Declaration of Helsinki (revised 2013); written informed consent was obtained from all participants. The study is reported according to the CONSORT 2010 statement [13] (Fig. 1). In line with Good Clinical Practice (GCP) [14], complications such as adverse reactions to EMLA® or wound infection were reported to the Data Safety and Monitoring Board (DSMB) and HREC.

Open image in new window

Setting and sample

Participants considered for inclusion in this study were individuals already referred to a large health district community nursing service in New South Wales, Australia. This study was conducted across six procedure clinics within the service where approximately three-quarters of patients required wound management; 76% had one or more CLUs [15].

A sample size of 60 was selected for this study as it was thought to be a good representation of the target population, large enough to provide practical information about the feasibility aspects of the study [10]. Although reported sample sizes for pilot studies vary, the median sample size per arm is often 30 [16]. In an external pilot study such as this, there is a ‘trade-off’ between maximising the precision of estimates of important parameters and sample size which impacts resources, time and costs of a study [17]. To address feasibility, we enrolled 60 patients to accommodate possible attrition throughout the 12-week study period. Attrition rates are often high in wound studies due to comorbidities associated this patient group particularly when the study period is long, where wounds may deteriorate, and the innate difficulty to adhering to the protocol over many weeks [18]. Even so, based on the community service patient profile, we estimated it would take approximately 12 months to enrol 60 patients assuming a 10% drop out rate [19]; a 20% drop-out is considered acceptable [20].

Randomisation, blinding and allocation concealment

Following eligibility assessment and consent, a simple randomisation method (1:1) was used to randomise participants to either the intervention (EMLA®) or control group. PASS 2008 Power Analysis and Sample Size software (NCSS, Kaysville, UT) was used to generate the allocation sequence by a researcher not involved with screening patients; the investigator was blinded to the allocations. Patients were allocated to study groups by retrieving the next in a series of sequentially numbered, opaque, pre-prepared sealed envelopes. This method for allocation concealment can achieve a low risk of bias [21].

Similar to other wound care studies [18], it was not possible to blind the participants, treating nurses or the investigators to the treatment allocation as the intervention was compared to usual care and not a placebo. However, the statistician was blinded. To minimise bias, the intervention and control groups were treated as equally as possible apart from the intervention itself. Treating clinicians were required to adhere to the Australian Standards for Wound Management [22] and health service policies and procedures.

Interventions

An intervention period of 4 weeks was selected as healing rates over a 4-week period can determine intervention effectiveness [23, 24, 25]. Patients were followed-up over a 12-week period because wound margin advancement, initial healing rate and percentage of wound surface area reduction are strong indicators of complete healing at 12 weeks independent of topical dressing used [25].

Management of adverse events

Participants recorded any problems or adverse reactions to the intervention in their Pain Diary which was reviewed at each clinic visit. Treating clinicians observed for known reactions to EMLA® such as blanching, erythema, oedema, pruritus, burning, purpura and contact hypersensitivity [28] at each dressing change and documented them in the participant’s electronic medical record. Additionally, a Notification of Adverse Event form was completed followed by immediate notification and discussion with the chief investigator (CI). In the event of any adverse reaction in the intervention group, treatment was ceased immediately and the medical team notified. Notifications of adverse events were mandatory and reported promptly to the NSH HREC and the DSMB by the CI.

Measurements used to address study aims

The feasibility of conducting an RCT in a public community health service to prevent unnecessary spending or wasting of resources in a larger study was measured by assessing quantitative and qualitative data to address key study processes, resources, management and scientific feasibility. Quantitative feasibility outcome measures included validation of the recruitment rate and randomisation processes, consent rate, retention rate, and suitability of the eligibility criteria and, data collection instruments. Additionally, to measure intervention adherence, the number of protocol deviations or violations during the study by treating nurses and participants were identified, and the reasons why these occurred were documented. Further quantitative and qualitative measures included the availability and commitment of human resources, the time to perform study processes, availability and quality of study equipment, data management outcomes and challenges and the cost estimates to conduct such a study. The feasibility of the intervention was measured by evaluating the patient’s physiological responses to the intervention and the monitoring of adverse events. Qualitative measures included participant and investigator burden. Scientific feasibility was assessed to address the clinical responses to intervention. Scientific feasibility is reported elsewhere [11, 12].

Data analysis

Data were entered and checked for missing and invalid values in Microsoft Excel® then imported into Statistical Analysis for Social Scientists (SPSS Version 22, Chicago, USA) for analysis. A random sample (10%) of the data was verified against the original case report form. Quantitative data were analysed using descriptive statistics. Qualitative data from field notes were analysed using descriptive content analysis [30].

Study process assessment

Resource assessment

Human resources required to conduct this study were impacted by availability of the CI and RA who had competing full-time clinical roles. The RA was only available 1 day per week for 17 months to support the study; funding was required for the RA position.

The nurse time required for dressing change and pain data collection was the same in both groups (30 min) as was the added time taken for HRQoL and wound photography data collection at the 2, 4, 8, and 12 week time-points (1 h).

Some participants required home vists during business hours and weekends mostly due to participant frailty, lack of transport options, lack of clinic capacity, and the intervention protocol that is, daily dressings for the intervention period (weekdays: intervention group n = 9 (30%); control group: n = 6 (20%); weekends: intervention group: n = 30 (100%); control group n = 4 (33.3%): range: 1 to all visits). Seven participants required all visits in their homes (intervention group: n = 5 (16.6%), range: 24 to 77 visits); control group: n = 2 (6.6%), range: 7 to 24 visits).

A limited economic feasibility assessment was informed by a comparison of cost estimates for the intervention and usual care primary and secondary dressings. The intervention group had higher overall costs over the 12-week study period, with increased costs attributed to increased dressing frequency. Throughout the data collection period of 33 months, the intervention group required almost double the number of dressings compared to the control group (intervention group, 1232 dressings (65.4%); control group, 651 dressings (34.6%)). Daily dressings during the intervention period for the intervention group contributed to the considerable difference between groups (intervention group, 741dressings (74.6%); control group 252 dressings (25.3%). The overall cost of dressings per dressing change was less in the intervention group (intervention group, A$6.03; control group A$8.73). However, the intervention group had a 13.2% higher overall cost of primary and secondary dressing consumables over the study period (intervention group A$7441; control group A$5684), due to the increased frequency of dressings compared to the control group.

Management assessment

Participants were initially seen in the community health service clinics. However, as the study progressed, 13 participants had difficulty attending the clinics for treatment. Hire cars were provided for these participants as a short-term solution however, due to budget restrictions, this strategy was not feasible and was discontinued after 20 months. Subsequently, participants unable to attend the clinic were treated in their homes during business hours and weekends as previously reported using existing community nursing resources. This meant that some visits were attended by nurses not educated in the intervention protocols at the beginning of the study; nurse continuity was also an issue. The potential impact on the data collected during home visits was anticipated; the CI attended the visit particularly on weekends or made phone contact with the nurse to explain the protocols and procedures prior to the home visit.

Participant burden was observed in this study. The length of the eligibility interview, randomisation and recruitment processes (1 to 3 h), the length of the intervention (4 weeks) and study periods (12 weeks), the length of some data collection tools (up to 50 questions), the frequency of the data collection (at each dressing change), the average age and health status of the participants (73 years) and requirement of participants to come to the clinics, all contributed to participant burden.

Scientific assessment

There was no difference in WRP intensity scores between groups before dressing change over the 4-week intervention period (intervention group: mean 4.10 [95% CI 3.55, 4.63] vs control group: mean 4.21 [95% CI 3.66, 4.76]). Nevertheless, there was a statistically significant reduction in WRP for the intervention group compared to the control group during dressing change (intervention group: mean 3.39 [95% CI 2.59, 4.19] vs control group: mean 4.82 [95% CI 3.98, 5.66]) and after dressing change: (intervention group: mean 2.71 [95% CI 1.99, 3.43] vs control group: mean 3.92 [95% CI 3.16, 4.68]) [11]. EMLA® was tolerated well for 4 weeks by 83.3% (n = 25) of the intervention group. The remaining 16.6% (n = 5) of the intervention group had adverse effects from the application of EMLA® to their leg ulcers. Erythema, pallor, itching, oedema, purpuric or petechial lesions, or allergic reaction were not reported by the attending clinicians; however, five participants required EMLA® to be ceased due to increased or unchanged WRP and increased wound size. Usual care was recommenced on all participants. There were no serious adverse events to the intervention during this study.

Interestingly, two participants required recommencement of EMLA® following the 4-week intervention period at their request for the remainder of the 12 week study period due to significant exacerbation of their WRP after cessation of the EMLA® and commencement of usual care. Once EMLA® was recommenced, WRP was reduced.

Recruitment

Participant and research process factors influenced recruitment rates in this study. This experience is not uncommon in RCTs [33] where up to two-thirds of trials are unable to successfully recruit their original target [34, 35]. Protracted or ineffective recruitment can have undesirable scientific, ethical and economic consequences [36, 37]. Although random assignment may result in refusal to participate in a study [36, 38], this was not the case in this study. We overestimated the pool of patients with CLUs in the community nursing service that would meet the eligibility criteria. Known as ‘Lasagna’s Law’, this phenomenon is a common problem in clinical research with the evidence indicating that the incidence of the disease investigated reduces to 10% of the original estimate once the study starts [38]. This is a common threat to the success of clinical research resulting in increased direct costs and challenges the commitment and morale of research staff and participants [38]. Study processes and under resourcing of research personnel contributed to slower than expected recruitment, and it is likely that eligible patients were missed.

Lack of interest, inability to commit, physical and time limitations, change to daily activities and inability to travel to the community nursing clinics were patient-related factors affecting slow recruitment in this study; all of which are frequently cited in the literature [38]. The biggest obstacle to recruitment was patient eligibility at initial screening; we excluded more than we had predicted (34%), thus potential participants with painful CLUs were omitted from the study which may have negatively affected the generalisability of our results. A literature review found that an average of 30% of patients attending eligibility screening in RCTs are ineligible [34]. The majority of potentially eligible participants in this study were older; this was expected since the older person is more likely to succumb to CLUs [39, 40, 41].

Considering the prevalence of chronic leg ulcers in society, most chronic wound trials have small sample sizes reflecting the difficulty in recruiting patients if the eligibility criteria are too restrictive [42]. It became apparent that two of the exclusion criteria were considered unnecessary hence amendments were made to the exclusion criteria to increase the recruitment rate previously described. Based on these findings, we acknowledge the importance of understanding the needs and abilities of the prospective study population prior to developing eligibility criteria for a larger study. Prior to recruitment, we over-estimated how many of the potentially eligible patients would qualify for the study. Although recruitment was initially slow, the rate improved when adequate resources for screening and recruitment were in place.

The difficulty in recruiting older individuals has been well identified; however, age itself does not determine an individual’s ability to give consent to research [43]. There are factors however that are associated with age that could impact on an individual’s ability to consent effectively such as frailty, fatigue, cognitive impairment, chronic disease and/or feelings of vulnerability [43]. Nevertheless, in this study, we had a good consent rate (86%) when compared to other RCTs relating to CLUs [44, 45, 46].

Community nurses

Most wound care is attended to in the community setting and community nurses are essential for the identification of potential participants in wound care research. They have been described as “effectively the gatekeepers to trial participation” [47]. They are an important link between the investigators and participants and can influence recruitment and retention rates [38]. In this study, community nurses were enthusiastic about being involved in an RCT and could see the benefits to themselves and to the patient. They were provided with criteria to assist them to identify potentially eligible patients however, subsequently, the demands of their clinical workload impacted their ability to undertake patient screening and contributed to delayed or missed participant identification [47].

Chronic leg ulcer types

Patients with venous, arterial, mixed (arterial/venous) and diabetic foot ulcerations were recruited so outcomes reflected real-world clinical practice. These ulcer types differ in their underlying aetiology and wound characteristics. To increase the ability to meet our required sample size, patients with any of the above CLU types were included. We acknowledge that such heterogeneity can threaten study validity and usefulness of a clinical trial [42]. Subgroup analysis would be a solution however, a much larger sample size would be required [42] which was not realistic for this single-centre pilot study. Additionally, simple randomisation may not be sufficient to provide well-balanced treatment groups regarding confounders in this broadly defined study; a large, multicentre RCT using stratified randomisation may be appropriate [42].

Attrition

To maintain statistical power, we aimed for less than 20% loss to follow-up. Attrition rates of 20% or more introduces bias and is a serious threat to the internal and external validity by altering the structure of the intervention and control groups [38, 48]. Common predictors for study withdrawal are older age and functional impairment [49]. The attrition rate for this study was only 10% which was encouraging considering the majority of participants were older, frail, in significant pain, had committed to a long study period, were subject to a large amount of data collection and were required to travel to community nursing clinics. To aid retention, some participants required home visits; however, in a larger study, this will require more time allocation plus additional costs [49].

Participant burden

Participant burden is a subjective, multidimensional construct relating to the perception of the participants physiological, physical and/or economic adversity with involvement in the research process [50, 51]. Investigators have traditionally addressed participant burden in clinical trials by focusing on direct risks associated with the intervention or data collection procedures. Nevertheless, it is the indirect burden that can vary due to factors such as study duration, intensity and invasiveness of study procedures [50] that needs to be considered. This pragmatic pilot study has identified direct and indirect factors that contributed to participant burden and will be able to inform a larger study to use a more pragmatic approach to reduce participant burden to maximise research participation and response rates (Table 5).

Missing data

In this study, we had a mean percentage of 19% missing data. The literature does not identify an established cut-off regarding an acceptable percentage of missing data; 5% or less is considered inconsequential, and more than 10% can result in a biased statistical analysis [52]. Additionally, missing data mechanisms and patterns can have a bigger influence on results than the proportion of missing data [52]. To manage missing data in this study, we attempted to follow up all participants, included all available data in the analysis making a plausible assumption about missing data and did a sensitivity analyses that weakened the assumptions about missing data [48]. Missing data was missing completely at random; therefore, systemic attrition did not occur and an unbiased treatment effect estimate was derived from the obtained data [48, 53]. For a definitive effectiveness trial, missing data would need to be minimised to reduce the threat to study validity. Evidence shows that 95% of RCTs report some missing data which can threaten the validity of an RCT, make a true intention to treat analysis difficult to achieve, reduce the power and efficiency of the study and lead to bias [48].

Strengths of the study

The strengths of the study include the recruitment of 100% of the target population, retention of 90% of our sample, assessment of the fidelity of the intervention, inclusion of objective outcome measures, and the ability to refine protocols and procedures. Generalizability of the results of this pilot study may be limited due to participant enrolment from a single health service. Consequently, context-specific issues that may be influenced by local, regional or country specific practices are unknown. Furthermore, there were fewer eligible participants than initially anticipated. The exclusion criteria may have resulted in some patients with painful CLUs being overlooked for inclusion in this study.

Bias could have been introduced since the participants, treating nurses and researchers could not be blinded to the intervention. Additionally, process evaluation was not attended by a neutral party but by the researchers themselves; therefore, there is the potential that further biases may have been introduced. The study protocols and procedures placed considerable demands on the mostly frail aged participants, the treating nurses and health service resources all contributing to missing data. We recognise the need to minimise the difficulties identified in this study that participants and investigators may encounter when designing a protocol for a larger multisite study. Furthermore, regulations for the use of EMLA® on open wounds such as CLUs and its drug schedule status would have to be ascertained prior to an international study as these factors may differ between countries.