A systematic review of the effectiveness of interventions to enhance the healing of chronic ulcers of the foot in diabetes

Transcription

1 DIABETES/METABOLISM RESEARCH AND REVIEWS Published online in Wiley InterScience ( REVIEW ARTICLE A systematic review of the effectiveness of interventions to enhance the of chronic ulcers of the foot in diabetes R. J. Hinchliffe 1,2,G.D. Valk 3,J.Apelqvist 4,D.G. Armstrong 5,K.Bakker 6,F.L. Game 2, A. Hartemann- Heurtier 7,M.Löndahl 8,P.E. Price 9, W. H. van Houtum 10, W. J. Jeffcoate 2 * 1 Department of Vascular Surgery, Nottingham University Hospitals Trust, Nottingham, UK; 2 Department of Diabetes and Endocrinology, Nottingham University Hospitals Trust, Nottingham, UK; 3 Department of Internal Medicine, University Medical Center Utrecht, Utrecht, The Netherlands; 4 Department of Endocrinology, University of Malmö, Malmö, Sweden; 5 Rosalind Franklin University of Medicine and Science, Chicago, USA; 6 IWGDF, Heemstede, The Netherlands; 7 Service de Diabétologie et Métabolisme, Hôpital Pitié-Salpêtrière, Paris, France; 8 Department of Internal Medicine, Helsingborg Hospital, Helsingborg, Sweden; 9 Department of Wound Healing, School of Medicine, Cardiff University,Cardiff,UK; 10 Department of Internal Medicine, Spaarne Hospital,Hoofddorp,TheNetherlands *Correspondence to: W. J. Jeffcoate, Foot Ulcer Trials Unit, Department of Diabetes and Endocrinology, Nottingham University Hospitals Trust, City Hospital Campus, Nottingham NG5 1PB, UK. wjeffcoate@futu.co.uk Received: 9 October 2007 Accepted: 11 January 2008 Summary The outcome of management of diabetic foot ulcers is poor and there is uncertainty concerning optimal approaches to management. We have undertaken a systematic review to identify interventions for which there is evidence of effectiveness. A search was made for reports of the effectiveness of interventions assessed in terms of, ulcer area or amputation in controlled clinical studies published prior to December Methodological quality of selected studies was independently assessed by two reviewers using Scottish Intercollegiate Guidelines Network (SIGN) criteria. Selected studies fell into the following categories: sharp debridement and larvae; antiseptics and dressings; chronic wound resection; hyperbaric oxygen (HBO); reduction of tissue oedema; skin grafts; electrical and magnetic stimulation and ultrasound. Heterogeneity of studies prevented pooled analysis of results. Of the 2251 papers identified, 60 were selected for grading following full text review. Some evidence was found to support hydrogels as desloughing agents and to suggest that a systemic (HBO) therapy may be effective. Topical negative pressure (TNP) may promote of post-operative wounds, and resection of neuropathic plantar ulcers may be beneficial. More information was needed to confirm the effectiveness and cost-effectiveness of these and other interventions. No data were found to justify the use of any other topically applied product or dressing, including those with antiseptic properties. Further evidence to substantiate the effect of interventions designed to enhance the of chronic ulcers is urgently needed. Until such evidence is available from robust trials, there is limited justification for the use of more expensive treatments and dressings. Copyright 2008 John Wiley & Sons, Ltd. Keywords Introduction diabetes; diabetic foot; ulcer; wound ; dressings Ulcers of the foot in diabetes are the source of major suffering and cost [1,2]. Only two-thirds eventually heal [3 5]. Up to 28% may result in some form of amputation [6]. Management is based on the simple principles of eliminating infection [7], debridement, cleansing and the use of dressings to maintain a moist wound bed [8,9]. Many different interventions have been proposed to accelerate the process but few have been subjected to formal evaluation. In an earlier, non-systematic review by this group, we reported little evidence to substantiate the choice of any particular agent or intervention [10]. A systematic review of current evidence has now been undertaken.

2 S120 R. J. Hinchliffe et al. Materials and methods Prospective and retrospective controlled studies, published in any language, that evaluated interventions for the treatment of chronic foot ulcers in people aged 18 years or older with either type 1 or type 2 diabetes mellitus were considered. Studies were included if they assessed the effect of interventions on, time to, reduction in ulcer area or amputation. The following databases: Medline ( ), Embase ( ), the Cochrane database of systematic reviews, and the Cochrane Central Controlled Trials Register (2006) were searched using strategies that included the search term on study design, patient group, clinical problem and interventions of interest (see Appendix). There were no language restrictions. Randomized controlled trials (RCTs), case-control studies, prospective and retrospective cohort studies, control before-and-after (CBA) design and interrupted time series (ITS) designs were included. Bibliography of the identified articles was not tracked down. Previously performed high-quality systematic reviews and relevant Cochrane reviews were searched to determine the need for an extension to the literature search. One reviewer assessed all identified references by title and abstract on the basis of patient group, intervention and outcome. Full paper copies of identified articles were then assessed for eligibility by two independent reviewers and an agreement was reached on inclusion and exclusion following discussion. Each included paper was further assessed by two reviewers, working independently, and information was extracted on study design, patient group, intervention, outcomes, duration of and loss to followup, using standard data extraction sheets. Each study was scored for methodological quality using designspecific scoring, based on checklists developed by the Dutch Cochrane Center ( with equal weighting being applied to each validity criterion. Findings on data extraction and methodological quality were discussed between co-reviewers and a final decision was made. Quality items were rated as done, not done, or not reported and only those rated as done contributed to methodological quality score. This quality score was translated into a level of evidence according to the SIGN instrument [11]: (1) RCTs and (2) studies with case-control, cohort, CBA or ITS design. Studies were also rated as: ++ (high quality with low risk of bias), + (well conducted with low risk of bias) and (low quality with higher risk of bias). Meta-analyses, other reviews and studies reporting non-analytic case reports and case series were not included. Extracted data were summarized in evidence tables on a study-by-study narrative basis. Because of the heterogeneity of study designs, interventions, followup and outcomes, no attempt was made to pool the results. These evidence tables were compiled following collective discussion by the working party and conclusions were drawn and recommendations formulated. Results Sharp debridement and wound bed preparation with larvae The term debridement refers to the removal of surface debris, slough and infected matter with the aim of leaving clean, viable tissue that may then heal by secondary intention. Debridement may be undertaken surgically (sharp debridement), biologically (larvae), biochemically (enzymes), or chemically (antiseptics). There were 626 relevant papers identified in which effectiveness of an intervention was compared with a contemporary control group. Only 35 of these were judged potentially eligible after screening the abstract and only 3 studies were included following full text review (Table 1). Sharp debridement The evidence of benefit of sharp debridement is not strong and is based on a single study comprising a subgroup analysis of cases from an RCT of another intervention [15]. Healing at 12 weeks was more likely following a more vigorous debridement. An earlier paper of similar design was not included because of insufficient reported detail [16]. Larvae Two studies evaluated the effectiveness of debridement with larvae. The first was small and the structure complex, but reported a significant effect in 2 weeks [13]. The second was a study of elderly, non-ambulant (bed or wheelchair bound) people with peripheral arterial disease, and reported decreased time to, decreased incidence of major amputation and reduced requirement for antibiotics in the intervention group [14]. Wound bed preparation using antiseptics, applications and dressing products Dressing products may encourage debridement by either dessicating or binding to debris such that it is removed with dressing changes. Dressings may also facilitate autologous debridement by providing a warm moist environment. Some may have antimicrobial activity. The search identified 98 papers, of which 37 were considered for selection after screening the titles and abstracts and 11 were included after full text review (Table 2). Antiseptics One study demonstrated that cadexomer-iodine showed no benefit in cavity wounds [17]. One RCT suggested that zinc oxide tape reduced the area of necrotic wounds within 5 weeks when compared with a hydrocolloid [18].

3 Interventions for Diabetic Foot Ulcers S121 Table 1. Wound bed preparation by sharp debridement and the use of larvae Reference Study design Study population and characteristics Saap 2002 [15] Cohort study Study quality: 5/8 143 evaluable subjects with neuropathic superficial diabetic foot ulcers followed for 12 weeks in a parent RCT Sherman 2003 [13] Cohort study Study quality: 3/8 18 subjects with 20 chronic, non- ulcers divided into three groups: 6 conventional therapy, 6 debridement therapy with larvae, 8 conventional therapyfollowedbylarval debridement therapy Followed for 14 weeks total Armstrong 2005 [14] Case-control study Study quality: 1/7 30 people (mean age 72 years; 26 M) with diabetes and peripheral arterial disease and confined to either bed or wheelchair, who had their foot ulcers treated with larvae, compared to 30 age- and sex-matched controls who had not I, intervention; C, controls. Wound bed preparation/debridement Intervention and control conditions Outcome category Assessment of the extent of sharp debridement on Day 0, using a debridement index Closure of ulcer Debridement therapy with larvae Decrease in extent of necrotic tissue at 2 weeks History of wound debridement with larvae Healing; time to ; major amputations Antibiotic use (antibioticfree days) Results primary outcome + statistic A wound with a debridement index of 3 6 was 2.4 times more likely to heal than one with an index of 0 2 (p = 0.03) Decrease in necrotic tissue (4.1 versus 0 cm 2 )(p = 0.02) Larvae: complete debridement at 4 weeks versus 33% at 5 weeks (p = 0.001) Trendtodifferenceinulcer (p = 0.07); Shorter time to (I: 18.5 versus C: 22.4 days, p = 0.04); Fewer major amputations (I: 10% versus C: 33%, p = 0.03) andmore antibiotic-free days: (I: 127 versus C: 82, p = ) Level of evidence SIGN Comments on weaknesses 2+ This was a sub-analysis of a study of the effectiveness of another intervention, (Apligraf) Veves, et al. (2001) [12] 2 Complex study. Comparison between groups difficult because of the use of different times to outcome High percentage male. Unusual population. Cases and follow-ups selected by those in whom 6-month follow-up data were available. Not clear if controls matched for criteria other than age and sex

4 S122 Alginate and collagen-alginate products No significant effect on either wound area or rate of was found with a collagen-alginate dressing product, compared to a saline-moistened gauze in a nonblinded RCT [19]. An alginate appeared no better than vaseline gauze in a second RCT [20]. Hydrogels Three controlled studies of hydrogels were identified: one RCT [21] and two cohort studies [22,23]. The open label RCT reported a significant benefit in terms of of non-ischaemic foot ulcers when a hydrogel was compared with saline-moistened gauze. The evidence for a beneficial effect from the two cohort studies was not strong: neither reported any hard data on wound and one used no statistical analysis. Carboxymethylcellulose hydrofibre dressing A single, small RCT of subjects with deep foot ulcers reported a highly significant reduction in days to when a carboxymethylcellulose hydrofibre dressing was used, compared with a saline-moistened gauze [24]. Polymeric semi-permeable membrane The use of a polymeric semi-permeable membrane dressing was reported to reduce the ulcer area better when compared with a saline-moistened gauze, over a 6-week period [26]. Resection of the chronic wound The rationale for resecting a chronic ulcer and its bases is that a chronic wound will be replaced by one that resembles an acute one more closely and will proceed to heal more rapidly. Healing may be promoted if underlying bony prominences are also resected. The search strategy identified 879 papers and 152 of these were considered for inclusion, but only 4 were selected after full text review (Table 3). Excision of plantar ulcers with/without removal of underlying bone Wide excision of chronic plantar ulcers combined when indicated with removal of underlying bone reduced time to but had no effect on eventual rate [29]. A retrospective cohort study of the effect of excising the 5th metatarsal head underlying a chronic ulcer revealed a weak effect when compared to nonsurgical management [30]. A similar study involving excision of wounds under the interphalangeal joint of the hallux or first metatarsophalangeal joint, combined with arthroplasty of the metatarsophalangeal joint, reported of more ulcers by 6 months as well as reduced recurrence [31]. R. J. Hinchliffe et al. Early excision of infected soft tissue A cohort study compared outcomes in patients admitted to hospital with extensive infection and who either underwent surgical excision of infected tissue within 3 days of admission, or not [32]. The incidence of major amputation was significantly lower in the group treated surgically (13% versus 41%). Hyperbaric oxygen The search strategy identified 114 studies, of which 6 were selected on the basis of title and abstract. All the six fulfilled the inclusion criteria and were selected (Table 4). Topical treatment involves inclusion of the affected foot into a sealed chamber containing hyperbaric oxygen (HBO) Systemic HBO involves the patient spending prolonged periods of time in a large HBO chamber. Topical HBO Two studies evaluated topical HBO. One was randomized and reported no apparent reduction in the cross-sectional area of ulcers at either 7 or 14 days [33]. The other was only partially randomized but reported an apparent benefit at 4 weeks [34]. Systemic HBO The four RCTs [35 38] provided some evidence to suggest that systemic HBO may reduce the rate of major amputation, although all were limited by aspects of trial design and/or small numbers. A fifth study [39] reported a significant reduction in the ulcer area, but was not randomized. Reduction of tissue oedema Reduction of tissue oedema may promote. In addition, topical negative pressure (TNP) may also reduce accumulation of surface debris. The search identified 66 studies and 40 were considered for inclusion on the basis of title and abstract. Four (three on TNP; one on compression) were selected after full text review (Table 5). All studies were undertaken in populations who had previously undergone surgery. TNP Two of the three RCTs of TNP were very small but reported significant benefits in both rate and time [40, 41]. A third, much larger study reported a significant benefit of TNP in both time to and proportion of persons in those who had recently undergone pedal surgery [42], even though the definition of used included those who healed after repeat surgery and this weakens the conclusions to be drawn from the results.

5 Interventions for Diabetic Foot Ulcers S123 Table 2. Wound bed preparation using antiseptics, applications and dressing products Reference Study design Study population and characteristics Intervention and control conditions Apelqvist 1996 [17] RCT Study quality: 3/9 41 patients with diabetes, >40 years old, with toe/ankle pressure >30/80 mmhg, respectively, and with exudating, cavity wounds with an area of 1 25 cm 2 Intervention group 22, control group 19 Lodosorb daily initially and then less often for 12 weeks or until the wound was less exudative versus saline-moistened gauze Lost to follow-up 5 Apelqvist 1990 [18] RCT Study quality: 3/9 44 patients with necrotic ulcers. Intervention group 22, control group 22 Adhesive zinc oxide tape versus hydrocolloid Followed for 5 weeks and lost to follow-up: 2 Donaghue 1998 [19] RCT Study quality: 5/9 Patients with non-ischaemic foot ulcers, area >1 cm 2 Intervention group 50, control group 25 Followed for 8 weeks Lost to follow-up: 14 Collagen-alginate wound dressing versus saline-moistened gauze Lalau 2002 [20] RCT Study quality: 4/9 77 with both chronic and acute wounds, area >1 cm 2 Intervention group 39, control group 38 Calcium alginate versus vaseline gauze Outcome category Healing and decrease in area >50% Necrotic ulcer area reduction greater than 50% Ulcer, reduction in ulcer area >75% wound granulation plus decrease in ulcer area by >40% Results primary outcome + statistic Healing in intervention group 5/17 versus 2/18 (NS) Outcome achieved in 14/21 in the intervention group versus 6/21 controls (P < 0.025) 48% of the intervention group healed versus 36% controls (NS). Mean reductioninulcerarea:81% versus 61% in controls (NS) Combined endpoint achieved in 42.8% intervention group versus 28.5% in controls (NS) Level of evidence SIGN Comments on weaknesses 1 Primarily a health economic analysis, with limited results presented on clinical outcomes Per protocol analysis; 5 said to be lost to follow-up but results given on only 35 1 Uncertain numbers of withdrawals 1+ Open label study 1 Included acute wounds Study duration reduced from 6 weeks to 4 weeks because of high dropout rate Mean ulcer area at recruitment was very high at 8 cm 2 High percentage with type I diabetes suggests selected population (continued overleaf)

6 S124 R. J. Hinchliffe et al. Table 2. (Continued) Reference Study design Jensen 1998 [21] RCT Study quality: 3/9 Cangialosi 1982 [22] Prospective cohort series Study quality: 1/8 Capasso 2003 [23] Cohort retrospective Study quality: 2/8 Piaggesi 2001 [24] RCT Study quality: 3/9 Study population and characteristics Patients with non ischaemic foot ulcers; area >1 cm 2 Intervention group 14, Control group17 Followed for 20 weeks Lost to follow-up: 0 28 diabetics with 37 lower extremity ulcers Intervention group 14, Control group 14 Dropout: unknown Follow-up: unknown 50 patients (28 with diabetes) with arterial disease and foot ulcers Intervention group 25, Control group 25 Follow-up 7 weeks 20 patients with foot ulcers >1 cm deep Intervention group10, Control group10 Followed for 8 weeks Intervention and control conditions Hydrogel dressing versus saline-moistened gauze Hydrogel and sterile gauze Amorphous hydrogel versus wet or dry sterile gauze Hydrofibre carboxymethyl cellulose dressing versus saline-moistened gauze Outcome category Results primary outcome + statistic Ulcer 85% in the intervention group versus 46% in controls (p < 0.05) Ulcer Healing said to be about 33% more rapid in hydrogel group Cost; Wound ; Time to No differences observed in wound Time to heal: p = 0.02 in favour of hydrogel Days to 127 (46 SD) days in the 234 [25] controls (p < 0.001) Level of evidence SIGN Comments on weaknesses 1 Open label study 2 No statistical analysis Duration of follow-up and number lost to follow-up not stated Stated results vague 2 Complex series of primarily health economic studies No raw data presented on either wound or time to 1

7 Interventions for Diabetic Foot Ulcers S125 Table 2. (Continued) Blackman 1994 [26] RCT Study quality: 4/9 Muthukumarasamy 1991 [27] Cohort Study quality: 4/8 Pai 2001 [28] RCT Study quality: 5/9 RCT, randomized controlled trial. 18 patients with diabetes and Wagner grade 1 or 2 ulcers Intervention group 7 (mean age 51 years; 6 M) Control group 11 (59 years; 11 M) 100 patients with type 2 diabetes and Wagner grade 1 or 2 foot ulcers Intervention group 50 (27 M) Control group 50 (27 M) 70 patients with type 2 diabetes and Wagner grade 1 or 2 ulcers Intervention group: 36 (mean age 56 years, ulcer area 11.9 cm 2 ; 25 M) Control group: 34 (60 years, 11.9 cm 2 ;22M) Dropouts: 13 Semi-permeable membrane dressing applied for 2 months versus wet-to-dry saline gauze; late cross-over for 5/7 control group Topical phenytoin versus saline 35 days versus an occlusive dry dressing Topical phenytoin powder for 6 weeks versus talc/silicone dioxide Healing by 2months Change in ulcer area over 2months (intervention versus control); Decrease in ulcer area, and complete %decreasein cross-sectional area Intervention group 3/11 healed versus 0/7 (no statistical analysis) Intervention: reduction in area 35 ± 16% baseline at 2 months versus 105 ± 28%, p = 0.03 Intervention group % decrease in area was 88% of baseline versus 50% (p < 0.005) 20/50 healed in the Intervention group versus 12/50 Intervention group 78.3% reductioninareaversus 73.5% (NS) 1 Further reduction in area in the cross-over group 2 No statistical analysis given for the numbers which healed 1+

8 S126 R. J. Hinchliffe et al. Table 3. Resection of the chronic wound Reference Study design Study population and characteristics Piaggesi 1998 [29] RCT Study quality: 5/9 Patients with plantar diabetic forefoot ulcers: intervention group 21, control group 20 Followed for at least 6 months None lost to follow-up Armstrong 2005 [30] Retrospective cohort study Study quality: 3/8 40 patients with a chronic ulcer under 5th metatarsal head Intervention group 22, control group 18 Followed for 6 months Armstrong 2003 [31] Cohort study Study quality: 2/8 Uninfected, non-ischaemic ulcers under the interphalangeal joint of the hallux or the 1st metatarsophalangeal joint Intervention group 21, control group 20 Followed for 6 months Tan 1996 [32] Cohort study Study quality: 3/8 112 patients hospitalized with 164 diabetic foot infections 77 patients had surgery within 3days 87 had no surgery within 3 days RCT, randomized controlled trial; MT, medical treatment. Intervention and control conditions Ulcer excision with removal of bone and closure of wound versus conservative treatment 5th MT head resection versus medical treatment only 1st MTP joint arthroplasty, and resection head of 1st metatarsal versus non-surgical management Surgery within 3 days of hospital admission versus no surgery within 3 days Outcome category Healing and time to Time to ulcer Time to ulcer and ulcer recurrence Amputation and resolution of infection Results primary outcome + statistic 21/22 ulcers treated with surgery healed compared with 19/24 controls (NS) Time to (days) shorter in the intervention group (46 versus 128 in controls (p < 0.001) 5.8 (2.9) weeks in cases versus 8.7 (4.3) in controls (p < 0.05) 24.2 days in the intervention group versus 67.1 in controls (p = ) Ulcer recurrence in intervention group 4.8% versus 35% controls (p = 0.02) Those operated early had 77 episodes of infection and 10 major amputations versus 87 infection episodes and 35 major amputations in the non-surgical group (p < 0.01) Level of evidence SIGN Comments on weaknesses 1+ Also recorded incidence of secondary infection per ulcer (not per patient): 3/24 1/22 (p = 0.72) Description of outcomes and lesion types is incomplete. The incidence of amputation in the control group was high.

9 Interventions for Diabetic Foot Ulcers S127 Table 4. Hyperbaric oxygen therapy both topical and systemic Reference Study design Study population and characteristics Leslie 1998 [33] RCT Study quality: 6/9 28 with diabetic foot ulcers (16 Hispanic, 7 black, 7 white) Intervention group 12, Control group 16 Heng 2000 [34] RCT Study quality: 3/9 Intervention group 13, Controls 13 (plus an additional 14 controls who were not randomized) Follow for 4 weeks Lost to follow-up: not clear Faglia 1996 [35] RCT Study quality: 5/9 68 diabetic patients with ulcers Wagner grade 2 4 Intervention group 35, Control group 33 Intervention and control conditions Topical HBO versus standard care Topical HBO versus standard care Systemic HBO (2.5 ATA, 90 min daily) continued until or amputation versus standard care Outcome category Results primary outcome + statistic Change in cross-sectional area at day 7 and 14 Day 7: Area reduced to 67.1% in the intervention group versus 69.6% controls (NS) Day 14: 45.6% versus 35.6% (NS) Ulcer 90% in the 28% controls Amputation 30% fewer major amputations in Wagner grade 4 patients (p < 0.016) Level of evidence SIGN Comments on weaknesses 1+ 1 Partial randomization: the control group was larger becauseoflackoftreatment spaces Complicated data presentation. No statistical analysis presented. Not all patients had diabetes 1+ Randomization process unclear. Not blinded. Time to not reported. High frequency of vascular surgery after randomization. Mean age in the Intervention group 61.7 years versus 65.6 years in the control group. (continued overleaf)

10 S128 R. J. Hinchliffe et al. Table 4. (Continued) Reference Study design Study population and characteristics Kessler 2003 [36] RCT Study quality: 6/9 28 patients with neuropathic ulcers Wagner grade 1 3 and Duration >3 months Intervention group 15, Control group 13 Followed for 4 weeks Lost to follow-up: 1 Doctor 1992 [37] RCT Study quality: 3/9 30 patients: 23 with gangrene and 5 neuropathic ulcers Intervention group 15, Control group 15 Abidia [38] 2003 RCT Study quality: 9/9 18 patients with diabetic ulcers area 1 10 cm 2 and duration >6 weeks Intervention group 9, Control group 9 Lost to follow-up: 2 RCT, randomized controlled trial; HBO, hyperbaric oxygen. Intervention and control conditions HBOT (2,5 ATA, 90 min bd 5 days a week for 2 weeks) versus standard care Systemic HBO (3 ATA, 45 min, 4 sessions mean 34 treatments) versus standard care Systemic HBO (2,4 ATA, 90 min, 30 sessions) versus hyperbaric air (2,4 ATA, 90 min, 30 sessions) Outcome category Results primary outcome + statistic Reduction in ulcer area at 2 weeks and at 4 weeks Wound area reduction at 2 weeks: 42% in the 21% (p = 0.037) andat 4 weeks: 62% versus 55% (NS) Amputation Major amputation: 2 in the 7 controls (p < 0.05) Healing; Reduction in ulcer area Number healed at 12 months 5/8 healed in the intervention group versus 1/8 controls Median area reduction 100% in the intervention group versus 52% controls (p = 0.02) Healed at 12 months I5/8 in the 0/8 controls (p = 0.026) Level of evidence SIGN Comments on weaknesses 1+ One patient excluded from evaluation due to barotraumatic otitis 1 Wound size and depth are not reported No differences in number of healed ulcers Less positive bacterial cultures in HBOT group 1++

11 Interventions for Diabetic Foot Ulcers S129 Table 5. Reduction of tissue oedema Reference Study design McCallon 2000 [40] RCT Study quality: 4/9 Eginton 2003 [41] RCT Study quality: 4/9 Armstrong 2005 [42] RCT Study quality: 5/9 Armstrong 2000 [43] RCT Study quality: 6/9 RCT, randomized controlled trial. Study population and characteristics Non- ulcers of duration >1 month Intervention group 5, Control group 5 Followed until Lost to follow-up: 0 10 patients with non-ischaemic foot ulcers Followed for 4 weeks Lost to follow-up: patients with residual wounds of mean duration 1.5 months after foot surgery Intervention group 77, Control group 85 Followed for 16 weeks and lost to follow-up: patients with post-operative infected diabetic neuropathic foot ulcers Intervention group 52, control group 45 Followed for 12 weeks Lost to follow-up: 18 Intervention and control conditions Negative pressure therapy (NPT) versus saline-moistened gauze Cross-over design Randomly allocated to start with either NPT therapy for 2 weeks or with saline-moistened gauze for 2 weeks NPT therapy versus standard dressings Pneumatic foot compression device versus placebo non-functioning device Outcome category Time to ulcer Reduction in ulcer volume Healing (but including those unhealed and rendered suitable for surgical closure Wound Results primary outcome + statistics 22.8 days in the intervention group versus 42.8 days controls (NS) 59% reduction with NPT therapy compared with 0.1% for saline-moistened gauze (p < 0.05) 56% in the intervention group versus 39% controls (p = 0.04) 39/52 healed in the 23/45 (p < 0.02) Odds ratio 2.9 ( ) Level of evidence SIGN Comments on weaknesses 1 Small numbers 1 Small numbers and with 40% dropout rate 1+ This study was of wounds after diabetic foot amputation, rather than chronic foot ulcers. It was also marred by a high rate of dropouts The strength of the observation is weakened by the definition of used 1+ In addition there was a difference in the intervention group between those who were and were not compliant

12 S130 Compression The single RCT on post-operative wounds suggested a significant benefit of compression therapy [43]. Application of products designed to correct aspects of wound biochemistry and cell biology associated with impaired wound Knowledge of the factors which inhibit the of chronic wounds is rudimentary, although the importance of different growth factors is accepted [44]. Moreover, abnormalities in growth factor expression [45] and other aspects of cell biology have been defined in chronic wounds in diabetes, including changes in the relative concentrations of matrix metalloproteinases (MMPs) and their tissue inhibitors (TIMPs) [46]. Neutrophils and macrophages are also known to have impaired function in hyperglycaemia and this could have an adverse impact on [47,48]. A number of attempts have therefore been made to evaluate interventions that might influence such alterations in wound biology. Of the 107 studies identified, 15 were suitable for inclusion (Table 6). Lyophilized collagen A small RCT reported a highly significant reduction in time to for lyophilized collagen when compared with hyaluronic acid-medicated gauze [49]. Platelets and platelet-derived products A small RCT (of both leg and foot ulcers, in patients both with and without diabetes mellitus) revealed no apparent benefit and, indeed, the intervention group appeared to do rather worse than the control group [50]. A very small study (in which the definition of was not clear) reported an apparent benefit of the product in neuropathic ulcers [51]. A large cohort study of neuropathic ulcers reported significant benefit but was weakened by being a retrospective analysis of a database of the outcome of patients treated with different doses and for different periods with a platelet-derived product, and compared with those who did not receive it but were otherwise managed but in no specified way [52]. A significant reduction in ulcer area was reported in a non-randomized study, undertaken in diabetics with chronic ulcers which were mainly, but not exclusively, on the feet [53]. A more recent study [54] was a well-designed RCT but was marred by a very high exclusion rate which necessitated a per protocol analysis. An acellular wound care product An acellular product derived from the pig intestine was evaluated in a single study. When compared with plateletderived growth factor (PGDF), no benefit was observed [55]. R. J. Hinchliffe et al. Platelet-derived growth factor (PDGF, becaplermin) The initial RCT of becaplermin in non-infected neuropathic ulcers [56] indicated a significant effect, and this was confirmed in the later definitive phase III study [57]. A further study failed to recruit sufficient numbers and no differences were observed [58]. Basic fibroblast growth factor (bfgf) One small RCT of bfgf suggested an increase in by 12 weeks but the difference from controls was insignificant [59]. Epidermal growth factor (EGF) EGF has been shown to accelerate epidermal growth in experimental wounds. Two studies reported a significantly higher rate of of ulcers (mainly of the foot) when compared to placebo [58,60]. Another study [61] was less robust and included patients with leg ulcers. Although ulcer area and depth were reduced in the intervention group, there was no difference in the numbers by 16 weeks. Hydrofibre dressing A hydrofibre dressing which may modulate expression of MMPsandTIMPsinwoundswasfoundtohave no demonstrable effect on by 12 weeks in a single large RCT [62]. Trans-retinoic acid One RCT has reported an increase in of neuropathic ulcers by 16 weeks, as well as a reduction in ulcer area [25]. Stem cell therapy (including G-CSF) The search identified 51 papers and 22 of these were considered for inclusion on the basis of the title and abstract. Although there have been small uncontrolled observational studies of administering autologous bone marrow, the only studies suitable for inclusion in this section were of granulocyte-colony stimulating factor (G-CSF). Published studies on G-CSF were designed to determine its effect of infection, but five RCTs also assessed wound and reduction of amputation as secondary endpoints [63 67]. The sixth [68] was not considered. Only one of the five [67] was associated with any apparent benefit (Table 7). Bioengineered skin and skin grafts Of 72 papers identified in the search, 51 were considered but only six met the inclusion criteria (Table 8). Three concerned the use of a dermal fibroblast culture, and one each was of a fibroblast/keratinocyte co-culture,

13 Interventions for Diabetic Foot Ulcers S131 Table 6. Application of products designed to correct aspects of wound biochemistry and cell biology associated with impaired wound Reference Study design Study population and characteristics Intervention and control conditions Outcome category Results: primary outcome + statistic Di Mauro 1991 [49] RCT Study quality: 3/9 20 patients (6 with ischaemic, 4 with neuropathic and 9 with neuro-ischaemic ulcers) Lyophilized collagen versus hyaluronic acid-medicated gauze Time to 32 days in the intervention group versus 49 days controls (p < 0.001) Followed until Lost to follow-up: 0 Krupski 1991 [50] RCT Study quality: 8/9 18 non- ulcers of both leg and foot (14 had diabetes) Followed for 12 weeks Lost to follow-up: Nil Autologous platelet factor versus saline Healing and reduction in area 24% healed in the 33% controls; 4.3 cm 2 reduction in area per week in 1.9 cm 2 controls (NS) Steed 1992 [51] RCT Study quality: 6/9 13 subjects with neuropathic diabetic foot ulcers Intervention group 7, control group 6 Followed for 20 weeks PDWHF (CT-102) versus normal saline Proportion of and area reduction 5/7 healed in the intervention group versus 1/6 control (p < 0.05) Reduction in ulcer area 6.2 mm 2 /day in the 1.8 mm 2 /day controls (p < 0.05) Margolis 2001 [52] Retrospective Cohort Study quality: 5/ patients with neuropathic ulcers identified from the database of the Citizen Health System Followed for 20 weeks Platelet factor given to 6252 patients within 12 weeks Proportion healed 50% healed in intervention group versus 41% in controls Relative risk (RR):1.38 ( ) Feng 1999 [53] Cohort Study quality: 2/8 78 cases with diabetes and ulcers of the leg, foot (and elsewhere); 62 on the foot Mean ulcer area 10.7 cm 2 ;mean ulcer duration 8.9 days EGF or PDWHF or saline control administered daily Wound closure index at 6 weeks % healed at 2, 4, 6 and 8 weeks Closure index higher in both the EGF and PDWHF groups when compared with placebo (p < 0.01) %healed higher in EGF and PDWHF groups (p < 0.01) Level of evidence SIGN Comments on weaknesses 1 One ulcer was a wrist ulcer 1++ Both diabetic and non-diabetic patients Outcomes were for wounds and per patient 1+ Definition of unclear (3 subjects still needed dressings in one treatment arm) 2+ Retrospective analysis of treatment given in practice: Inconsistent dose and duration of treatment Selected population 2 Incomplete reporting of results Mean duration of the ulcers was short at 8.9 days (continued overleaf)

14 S132 R. J. Hinchliffe et al. Table 6. (Continued) Reference Study design Driver 2006 [54] RCT Study quality: 7/9 Niezgoda 2005 [55] RCT Study quality: 3/9 Steed 1995 [56] RCT Study quality: 2/9 Wieman 1998 [57] RCT Study quality: 6/9 Robson 2005 [58] RCT Study quality: 4/9 Study population and characteristics 72 (out of 129 screened) people with diabetes (type 1 or 2) and uninfected ulcers (UT 1A) of more than 4 weeks duration Intervention: mean age 56 years; 32 M; mean ulcer area 3.2 cm 2 Control: 58 years; 27 M; mean ulcer area 4.0 cm 2 98 with diabetic foot ulcers Intervention group 37 Control group 36 Followed for 12 weeks Lost to follow-up: 25 patients (25%) 118 subjects with diabetic foot ulcers Intervention group 61, control group 57 Followed for 20 weeks Lost to follow-up: 3 Uninfected non-ischaemic ulcers present for 8 weeks or more Intervention groups: (30 mcg/g) 132 (100 mcg/g) 123 Placebo gel 127 Followed up to 20 weeks Lost to follow-up: 73/ neuropathic plantar foot ulcers, duration >4 weeks Intervention group 74, control group 72 Lost to follow-up: 3 Intervention and control conditions Platelet autogel for 12 weeks versus placebo gel, with 11 weeks follow-up Acellular wound care product versus becaplermin (PDGF) Recombinant PDGF versus placebo gel Dose-ranging becaplermin gel applied daily versus placebo gel 0.01% becaplermin (PDGF) versus an adaptive dressing Outcome category Proportion healed (confirmed at 1 week) and time to Healing at 12 weeks, time to Proportion of patients healed at 20 weeks Proportion healed at 20 weeks, time to, reduction in ulcer area Healing at 20 weeks, time to Results: primary outcome + statistic Healing in 13/16 in the 8/19 in controls Time to significantly shorter in the Intervention group (p = 0.018) 49% healed in the 28% controls (NS) Time to 67 days in the 73 days controls (NS) 29 (48%) of 61 PDGF versus 14 (25%) of 57 patients randomized to the placebo group (p = 0.01) 100 mcg/g associated with 50% versus 35% placebo (p = 0.007) Time to 100 mcg/g 86 days versus 127 placebo (p = 0.013) No differences between 30 mcg/g and placebo Healing in 42% in the 35% in controls (NS) Time to NS (no data reported) Level of evidence SIGN Comments on weaknesses 1+ Very high exclusion rate necessitated per protocol analysis High percentage of heel ulcers 1 Unexplained high dropout rate 1 Details of treatment in the two arms unclear Although only 3 were lost to follow-up, total withdrawals were high, with only 86/118 completing the study 1+ Details of randomization not specified, nor the blinding of the assessor 1 Only 146 enrolled of target of 340

15 Interventions for Diabetic Foot Ulcers S133 Table 6. (Continued) Richard 1995 [59] RCT Study quality: 6/9 17 patients with diabetic foot ulcers Intervention group 9, control group 8 Followed for 12 weeks Fibroblast growth factor (bfgf) versus placebo vehicle Ulcer and reduction in ulcer area 5 healed in the intervention group versus 3 controls (NS) 47.2% had reduction in area in intervention group 35.8% controls (NS) Tsang 2003 [60] RCT Study quality: 7/9 61 patients with neuropathic diabetic foot ulcers Intervention groups 0.02% % 21 Control group 19 Dose-ranging study of EGF 0.02% versus EGF 0.04% versus placebo Proportion of 12/21 receiving 0.02% EGF healed, compared with 20/ % EGF, and 8/19 controls (p = ) at 12 weeks for 0.04% gel Followed for 12 weeks Afshari 2005 [61] RCT Study quality: 4/9 50 patients, including 25% with a leg ulcer Intervention group 30, control group 20 Followed for 4 weeks Lost to follow-up: 0 Topical epidermal growth factor versus placebo Proportion healed by 4 weeks; >70% reduction in ulcer area No difference in proportion of ulcers healed. 70% reductioninareain50%of the 15% in controls (p = 0.05) Veves 2002 [62] RCT Study quality: 2/9 276 diabetic foot ulcers Intervention group 138 Control group 138 Followed for 12 weeks Loss to follow-up: 27% Hydrofibre (cellulose/ collagen dressing) versus saline-moistened gauze Healing by 12 weeks No significant difference in (37.0% versus 28.3% p > 0.05) Tom 2005 [25] RCT Study quality: 7/9 24 subjects with neuropathic diabetic foot ulcers Intervention group13 Control group 11 Followed for 16 weeks. Lost to follow-up: 2 Solution of topical Tretinoin (retinoin A-) versus placebo saline solution applied for 4 weeks Proportion healed by 16 weeks Reduction in ulcer area and depth 6/13 healed in the 1/11 in controls (p = 0.03) Reductioninarea(p < 0.02), and depth (p < 0.01) greater in intervention group EGF, epidermal growth factor; bfgf, fibroblast growth factor; PDGF, derived growth factor; PDWHF, platelet-derived wound fluid; RCT, randomized controlled study. 1+ Small sample size 1+ Small sample size 1 Reduction in ulcer area adopted as an endpoint retrospectively after no difference found in primary end point 1 High dropout rate Suboptimal off-loading strategy 1+ Details of the analysis are not clear

16 S134 keratinocytes and the effectiveness of different split-skin grafts. Dermal fibroblast culture One dose-ranging study [69] reported that weekly applications of dermal fibroblast culture improved of plantar non-ischaemic ulcers by 12 weeks, when the highest dose was compared with saline-moistened gauze. It should be noted that the proportion in the control was only 8% in this study. Another study [70] found no difference between intervention and placebo. Although the definitive RCT of fibroblast culture [71] reported that by 12 weeks was significantly greater in the intervention arm than in controls, the percentage in the intervention group was only 30%, and in controls was only 18%. A report by Hanft et al. [74] was a single centre subgroup analysis of the parent study [71]. Fibroblast/keratinocyte co-culture The 12 week rate in both intervention and control groups were higher in the study of fibroblast/keratinocyte co-culture (56% and 38%, respectively) [12] than those reported for dermal fibroblast culture [69 71]. The difference between intervention and control groups was significant. Keratinocytes The single published RCT did not report full results [72], but a reduction was claimed in wound area, as well as improvement in the wound bed in the intervention group. Skin grafts The study by Puttirutvong et al. [73] reported no difference between meshed autologous and conventional split-skin grafts. The study included leg ulcers and there were some discrepancies between data in the text and in the abstract. Electrical, electromagnetic, lasers and ultrasound Of 142 identified studies, 14 were considered for selection on the basis of title and abstract and 6 were included after full text review (Table 9). Electrical stimulation Two RCTs examined electrical stimulation of the feet. The first was methodologically weaker and no benefit was observed [75]. In contrast, Peters et al. [76] reported a trend towards a greater proportion at 12 weeks. R. J. Hinchliffe et al. Ultrasound The single selected study was flawed methodologically but reported no significant difference between groups when analysed on an intention to treat basis [77]. Normothermic therapy A single published study was described as an interim analysis of a non-blinded RCT of wound ( normothermic ) warming (under an occlusive dressing), but did not describe the intended primary endpoint [78]. No significant benefit was observed. Magnet therapy A single RCT [79] was well designed but failed to recruit sufficient numbers and also had a high dropout rate. It was underpowered and no data were presented. Complex intervention including laser therapy Chiglashvili and Istomin [80] compared improvement in wound surface and in a non-blinded study of two, non-randomized groups and reported that outcome was improved by a complex intervention involving the administration of antioxidant and immunomodulatory agents, combined with laser therapy. No statistical analysis was presented. Discussion The causes of diabetic foot ulcers are complex and the factors that delay their are poorly understood. Management is generally supportive and requires the input of different health care professionals, using different interventions often over a prolonged period. It follows that the main aim of management (ulcer ) is unlikely to be responsive to any single treatment and it may therefore be difficult to demonstrate the effectiveness of an intervention in studies that use measures of ulcer as the primary endpoint. Moreover, the complex pathogenesis of chronic foot ulcers means that different causative factors may be dominant in different individuals and if the benefit of an intervention is limited to one particular type of ulcer, limb or person, it may be masked in a study of the relatively large numbers needed for an RCT. Assessment of surgical interventions poses particular difficulties in that they are generally adopted when other therapies have failed, and thus in a selected population. It may also be ethically difficult to undertake a study which involves randomization of patients to surgical and non-surgical groups. Trial design therefore poses considerable problems in this field, and the absence of evidence of effectiveness from robust studies does not necessarily mean that an intervention will not be effective in some people. Nevertheless, the reports selected for review here should be sufficient to indicate whether or the interventions studied are likely to be effective in relatively

17 Interventions for Diabetic Foot Ulcers S135 Table 7. Stem cell therapy (including G-CSF) Reference Study design Study population and characteristics Gough 1997 [63] RCT Study quality: 9/9 Patients with foot ulcers complicated by soft tissue infection Intervention group 20 Control group 20 Followed for 7 days Lost to follow-up: 0 de Lalla 2001 [64] RCT Study quality: 4/9 Patients all with osteomyelitis Intervention group 20 Control group 20 Followed for 6 months Lost to follow-up: 4 Yonem 2001 [65] RCT Study quality: 3/9 Patients with ulcers Wagner grade 2 complicated by soft tissue infection (inflammation >2 cm) Intervention group 15 Control group 15 Lost to follow-up: nil Kastenbauer 2003 [66] RCT Study quality: 7/9 Patient with foot ulcers complicated by cellulitis Intervention group 20 Control group 17 Followed for 10 days Lost to follow-up: 0 Huang 2005 [67] RCT Study quality: 4/9 Patients with ischaemic ulcers Intervention group 14 Control group 14 Followed for 3 months Lost to follow-up: 0 G-CSF, granulocyte-colony stimulating factor; RCT, randomized controlled trial. Intervention and control conditions G-CSF administered sc daily for 7 days versus saline injections sc G-CSF sc and conventional treatment versus conventional treatment alone G-CSF given sc versus standard treatment for 10 days G-CSF sc daily for 10 days versus saline sc IM administration of autologous monocytes following G-CSF sc for 5 days versus IV administration of prostaglandin E2 Outcome category Results: primary outcome + statistics Ulcer 4 healed in the intervention group versus 0 controls (p = 0.09) Cure, improvement of infection, failure, amputation No significant differences were reported Duration of hospital admission, time to infection resolution and proportion of amputation Duration of hospital admission 26.9 days in the 28.3 controls (NS) Amputation 13.3% in the 20% controls (NS) Time to resolution 23.6 days in the intervention group versus 22.3 controls (NS) Ulcer volume reduction Reduction in ulcer volume in 59% in the intervention group versus 35% controls (NS) Ulcer 14/18 healed in the Intervention group versus 7/18 controls (p = 0.016) Level of evidence SIGN Comments on weaknesses 1++ This was primarily a study of the eradication of infection andnotpoweredforulcer Short duration of intervention 1 All dropouts were in the intervention group The use of composite endpoints makes interpretation difficult 1 No data regarding rate No information given on blinding 1+ Primary endpoint was eradication of infection (study not powered for volume reduction) 1 The primary endpoint was improvement of limb ischaemia Ulcers were analysed instead of patients

18 S136 R. J. Hinchliffe et al. Table 8. Bioengineered skin and skin grafts Reference Study design Gentzkow 1996 [69] RCT Study quality: 6/9 Naughton 1997 [70] RCT Study quality: 3/9 Marston 2003 [71] RCT Study quality: 5/9 Study population and characteristics Patients with non-ischaemic plantar foot ulcers Intervention groups: 12,14,11 Control group 13 Followed for 12 weeks Lost to follow-up: Patients with non-ischaemic plantar neuropathic ulcers of duration >2 weeks and area >1 cm 2 Intervention group 139 Control group 142 Followed for 12 weeks Lost to follow-up: 46 (17.4%) 245 patients with non-ischaemic plantar neuropathic ulcers of duration >2 weeks and area >1 cm 2 Intervention group 130 Control group 115 Lost to follow-up : 46 (19%) Intervention and Pcontrol conditions Group 1: application of 1 piece of dermal fibroblast culture weekly Group 2: 2 pieces every 2 weeks Group 3: 1 piece every 2 weeks Controls: saline-moistened gauze Dermal fibroblast culture weekly for 8 weeks versus standard care Dermal fibroblast culture weekly for up to eight treatments versus conventional therapy Outcome category Proportion with ulcer Healing at 12 weeks Healing at 12 weeks, time to Results: primary outcome + statistic Group 1: 50% Group 2: 21% Group 3: 18% Controls: 8% (Group 1 versus controls, p < 0.05) 38.5 healed in the 31.7% controls (NS) 30% healed in the 18% controls (p = 0.023) RR = 1.6 Time to : p = 0.04 in favour of the intervention group Level of evidence SIGN Comments on weaknesses 1+ The percentage of controls at 12 weeks was very low 1 Per protocol analysis. The data were also reanalysed on the basis of perceived metabolic inactivity of some batches of dermal fibroblast culture Short ulcer duration before study 1+ 90% of the patients were male, suggesting selection bias No raw data on time to Short ulcer duration before study