The prevention and of diabetic foot ulcers: a review of clinical effectiveness studies Eva Kaltenthaler, C. Jane Morrell, Andrew Booth, Ron L. Akehurst School of Health and Related Research, University of Sheffield, UK Diabetic foot ulcers are associated with considerable morbidity, mortality and costs to the NHS. This paper describes a review of publications reporting randomised controlled trials examining the clinical effectiveness of interventions for diabetic foot ulcers for 1986-1996. Only 23 studies were identified that described randomised controlled trials and, of these, only 4 fulfilled three criteria shown to influence reliability of results: adequate description of randomisation, double blinding and description of withdrawals and drop-outs. cost-effectiveness studies conducted alongside trials were identified from the searches. This shortage of rigorous trials highlights the need for more well designed research in the prevention and of diabetic foot ulcers to determine clinical effectiveness as well as relative cost-effectiveness. Journal of Clinical EffectivenessVol.3. 3, 1998: 99-104 Financial Times Professional Ltd 1998 Introduction Foot ulceration is a common complication of diabetes which can lead progressively to: gangrene; amputation of the toes, feet or legs; or death. It is associated with considerable mortality and morbidity as well as loss of income and poor quality of life. Three major contributory factors to the incidence are: neuropathy, peripheral vascular disease and poorer socioeconomic circumstances. 1 It has been estimated that approximately 50% of people with foot ulcers have both vascular disease and neuropathy present 1 and the extent of both vascular disease and neuropathy must be investigated in order to facilitate optimal. 2 In the case of vascular disease, revascularisation will be necessary for healing to take place, while in the case of neuropathy, relief from pressure is necessary for healing to occur. Foot ulceration occurs when one or both of these factors are combined with physical stresses to the foot. 3 The risk of ulceration is greater for males than females and increases with age, non-insulin-dependent diabetes mellitus (NIDDM) and duration of diabetes. 3 It has been estimated that at some time, approximately 15% of people with diabetes will suffer from foot problems. 2 Studies have also shown that about 20% of people with diabetes who are hospitalized have foot problems. 4 Between 20 and 45% of lower limb amputations in the UK can be attributed to diabetes. 5 Diabetic foot ulcers are difficult to classify and Eva Kaltenthaler BSc, MSc, PhD, C.Jane Morrell BSc, HV, MPhil, Andrew Booth BA, MSc, DipLib, ALA, Ron L. Akehurst BSc, School of Health and Related Research (ScHARR), University of Sheffield, Regent Court, 30 Regent Street, Sheffield S1 4DA, UK. Correspondence to: E.K. describe. A system developed to overcome these difficulties, known as the Wagner Classification System, 2 divides diabetic foot ulcers into six grades of severity as described in Table 1. The number of well-conducted trials to determine the effectiveness of prevention or s for diabetic foot ulcers is limited. There is also a need to determine which interventions are the more cost-effective. The purpose of this paper is to review critically the evidence on the effectiveness of interventions (including prevention) for diabetic foot ulcers. Method To identify the studies used in this review, the following major databases were searched: CINAHL, Cochrane library, EMBASE, Healthstar, Medline, Pharmacoeconomics and Outcomes News, NEED and Datastar. Other methods included searching the Internet and hand searching. The criteria for inclusion were: articles which were written between 1986 and 1996; articles written in English; articles relating specifically to diabetic foot ulcers; Table 1 1994 2 ) Wagner classification of diabetic foot ulcers (Shenaq et al., Grade 0: open ulcers, but bony deformities and/or hyperkeratoses that increase risk of ulcer formation Grade 1: Ulceration extending into dermis Grade 2: Ulceration penetrating dermis and extending into tendon and/or joint capsule Grade 3: Ulceration extending to bone with or without osteomyelitis Grade 4: Localized gangrene Grade 5: Gangrene involving major part of the foot J. Clin. Effect. Volume 3 Number 3 September 1998 99
Prevention and of diabetic foot ulcers articles dealing with the following interventions: prevention, multi-disciplinary education and support, s including topical applications, dressings, surgery, antibiosis, growth substances, hyperbaric oxygen, drug therapy, wound grafting, footwear and contact casts; randomised, controlled trials of interventions for diabetic foot ulcers. Papers dealing with other types of wounds alongside diabetic foot ulcers, screening methods, precursors to diabetic foot ulcers and complications associated with diabetic foot ulcers were not included. Subject terms used in the searches included: Foot-Ulcer, Diabetic-Foot, Diabetes-Mellitus. Text words included: Diabet*, foot or feet, ulcer*. For information on effectiveness, studies were restricted to those identified as clinical trial, review, meta-analysis, major clinical study or randomised clinical trial. Very few randomised controlled trials were identified from the studies retrieved. In those randomised controlled trials which were identified (n=23), the following points were assessed: sample size; the extent to which blinding was used; the intervention; outcome measures such as: healing, ulcer size, symptoms, costs and patient acceptability. The information gathered formed the basis for an overview of randomised controlled trials of diabetic foot ulcer interventions, rather than a systematic review as time constraints did not permit it to be fully comprehensive. Although only articles written in English were included in the review, many of these papers were from non-english speaking countries, representing a different perspective on interventions. Results A total of 412 references were identified of which 117 were obtained and entered into the study database. There was some duplication of the references identified from the various databases. In total 79 references describing trials were used and of these 23 (29%) described randomised controlled trials. The interventions described include: prevention through health education, therapeutic shoes, dressings, drug therapy, antibiotics, topical s, wound grafting, growth factors, contact casting, hyperbaric oxygen and surgery. These studies are outlined in Table 2. The quality of these 23 trials was assessed firstly on whether the method of randomisation was described, secondly whether the study was double blinded and Table 2 Summary of diabetic foot ulcer clinical effectiveness studies Study Type of Treatments compared Sample size Outcome measure and result Follow-up period Ahroni et al., 1993 6 Apelqvistet al., 1990a 7 Apelqvist et al., 1990b 8 Blackman et al., 1994 9 Chantelau et al., 1996 10 Clever and Dreyer, 1995 11 DiMauro et al., 1991 12 Doctor et al., 1992 13 Donaghue et al., 1996 14 Faglia et al., 1996 15 Drug therapy Oral antibiotics Topical Hyperbaric oxygen Systemic hyperbaric oxygen therapy (s-hbot) Moist dressing regime (calcium alginate gel) Dry dressing regime (fine mesh gauze) Adhesive zinc oxide tape Adhesive occlusive hydrocolloid dressing Selective seratonin Polymeric (semi-permeable) wet to dry saline dressing Amoxicillin and davulanic acid Polyurethane gel dressing Polyurethane foam dressing Lyophilised collagen Hyaluronic acid medicated gauze Hyperbaric oxygen therapy Conventional Collagen-alginate dressing Saline-moistened dressing s-hbot Conventional n=20 n=19 total n = 40 n=11 n = 7 n = 20 n = 20 total n = 20 total n = 30 n = 45 n =17 n = 35 n = 33 Complete healing: no difference in healing rates between two groups 50% reduction in size: 14/21 improved in 5 weeks 6/21 improved in 5 weeks (NS) Complete healing and 50% reduction in size: no significant difference between two groups Ulcer size: 35±16% of baseline 105±28% of baseline (P< 0.03) Complete healing: 6 healed 10 healed (NS) Mean healing time:25.19±23.52 days 20.43±14.74days(NS) Mean healing time: 32.4±8.6 days 49.0±11.0 days (P< 0.001) Bacterial cultures: positive cultures decreased from 19 to 3 Positive cultures decreased from 16 to 12 (P< 0.05) Complete healing and mean time to complete healing: 24 (55%), 5.9 weeks (patients preferred) 9(53%), 5.3 weeks (NS) Major amputation: 3 (8.6%) underwent amputation 11 (33%) underwent amputation (P= 0.016) 128 days 35 days 3 months 2 months 20 days 16 weeks t stated t stated 8 weeks t stated 100 J. Clin. Effect. Volume 3 Number 3 September 1998
Prevention and of diabetic foot ulcers Original article Table 2 (continued) Study Type of Treatment compared Sample size Outcome measure and result Follow-up period Gentzkow et al., 1996 16 Grayson et al., 1994 17 Leslie et al., 1988 18 Lipsky et al., 1990 19 Litzelman et al., 1993 20 Maloneetal., 1989 21 Mueller et al., 1989 22 Pinzur et al., 1995 23 Richard et al., 1995 24 Steed et al., 1992 25 Steed et al., 1995 26 Steed et al., 1996 27 Uccioli et al., 1995 28 Wound grafting Intravenous antibiotics Hyperbaric oxygen Oral antibiotics Health education Health education 4 groups: 3 receiving Dermagraft and 1 receiving a non-adherent interface Ampicillin/sulbactam Imipenem/cilastatin Topical hyperbaric oxygen hyperbaric oxygen Cephalexin Clindamycin Health education programme education Health education programme education Contact casting Total contact casting Surgery Growth factor Growth factor Topical Growth factor Therapeutic shoes Traditional dressing Disarticulation one stage technique Disarticulation Wagner two stage technique bfgf CT-102 RGD Peptide Matrix Topical saline and standard rhpdgf Therapeutic shoes Patient's own shoes Total n = 50 48 episodes 48 episodes Total n = 28 n = 29 n = 27 n =191 n = 205 n = 103 n = 100 n = 21 n=19 n = 13 n= 8 Total n = 17 Total n= 13 n = 40 n = 25 Total n= 118 n = 33 n = 36 Complete healing: in 50% (n = 12) of highest dosage group; 8% (n = 13) in control group (P = 0.03) Bacterial culture: 81 % cured 67% eradicated 85% cured 75% eradicated (NS) Bacterial culture and ulcer size: no significant difference between two groups Bacterial culture and wound healing: 21 (72%) infection cured, 9 (33%) healed 21 (78%) infection cured, 10(40%) healed (NS) Presence of lower extremity abnormality: 59% reduction in risk in education group (P = 0.05) Amputation rate and ulcer: 7 of 177 limbs (4%), 8/177 (4.5%) ulcer; 21 of 177 limbs (12%) (P< 0.005), 26/177 (15%) ulcer (P< 0.005) Complete healing: 19 of 21 ulcers in 42±29 days 6 of 19 ulcers in 65±29 days (P< 0.05) Complete healing: 9/13 healed (69.2%) 5/8 healed (62.5%) Complete healing: 3 of 9 ulcers healed (33%) 5 out of 8 ulcers healed (63%) (NS) Complete healing: 5 of 7 ulcers healed by week 15(71%) 1 of 6 ulcers healed by week 20 (17%) (P<0.05) Complete healing and >50% ulcer closure: 14 healed (35%), 30 (75%) > 50% ulcer closure 2 healed (8%), 12 (48%) >50% ulcer closure (P = 0.03) 10 weeks Complete healing: 46.8% healed 25% healed (P= 0.01) Ulcer relapse: 27.7% 58.3% (P = 0.009) 12 weeks Mean follow up period: 1 year 14 days 3 months Intervention lasted one year, followup after that not stated 2 years t stated t stated 18 weeks 20 weeks 10 weeks 20 weeks 1 year finally whether a description was provided of withdrawals and drop-outs from the study. 29 This simple critical appraisal checklist described by Jadad et al. uses only three points to assess the quality of randomised controlled trials. Other, more complicated critical appraisal checklists are available. 30 However, it was felt no trials of diabetic foot ulcers would have been included in a more rigorous list of criteria. Table 3 summarizes the characteristics of the 23 studies. When applying the criteria to assess quality of studies, as outlined by Jadad et al., 29 only four of the randomised controlled trials identified provided adequate information, 8,10,17,26 highlighting the lack of well-designed and well-reported trials of interventions for diabetic foot ulcers. Apelqvist et al., 8 as described in Table 2, examined a total of 40 patients and found no significant difference between the groups receiving serotonin antagonist and placebo. Grayson et al. 17 looked at a total of 96 episodes using two different intravenous antibiotic regimes and again found no significant difference between the two regimes. Chantelau et al. 10 described a study comparing an oral antibiotic regime with a placebo in a total of 44 patients. There were no significant differences between the two groups. This study had a very short follow-up period. Steed et al. 26 compared an RGD (arginineglycine-aspartic acid) peptide matrix (n = 40) with control using standard of topical saline and debridement (n = 25). The results showed an improved healing rate and total number of ulcers healed in the intervention group. J.Clin. Effect. Volume 3 Number 3 September 1998 101
Prevention and of diabetic foot ulcers Table 3 Quality assessment of diabetic foot ulcer randomised controlled trials Study Randomisation method Double blind Description of withdrawals and drop-outs AhronietaL, 1993 6 Apelqvist et al., 1990a 7 Apelqvist et al., 1990b 8 Blackman et al., 1994 9 Chantelau et al., 1996 10 Clever and Dreyer, 1995 11 DiMauro et al,. 1991 12 Doctor et al., 1992 13 Donaghue et al., 1996 14 Faglia et al., 1996 15 Genzkow et al., 1996 16 Grayson et al., 1994 17 Leslie et al., 1988 18 Lipsky et al., 1990 19 Litzelman et al., 1991 20 Malone et al., 1989 21 Mueller et al., 1989 22 Pinzur et al., 1995 23 Richard et al., 1995 24 Steed et al., 1992 25 Steed et al., 1995 26 Steed et al., 1996 27 Uccioli et al., 1995 28 method described method described Cards in sealed envelopes method described Computer generated randomisation code method described method described method described method described Table of random numbers Study numbers randomised into 4 groups Computer generated randomisation code Independently with table of random numbers method described Health care terms were randomised By social security number Independent random sampling By social security number method described method described Prearranged randomisation order method described method described (blinded assessment) (blinded assessment by surgeon) (blinded assessment) (blinded assessment) Discussion The two studies describing randomised trials of prevention strategies 20,21 did not fulfil all three criteria as it was not possible to blind the research teams or patients. However, these studies were of good quality and used a large sample size. The results of Litzelman et al. 20 showed a significant reduction in risk (P< 0.05) in the intervention group and Malone et al. 21 reported a significantly lower rate of amputation in their intervention group. Therefore, prevention remains a promising intervention worthy of more research. Evidence-based optimal healing requires good evidence from well-designed trials. However, other factors which need to be considered include: patients' health-related quality of life; condition-specific symptoms such as pain in ischaemic diabetic foot ulcers; patients' tolerance, compliance, satisfaction and preference for specific s. Diabetic foot ulcers are associated with a variety of other complications of diabetes and rarely occur as the only complication of diabetes. There was very little mention of patients' views on their or of the effect of the on their health-related quality of life in many of the papers reviewed. Costs of interventions Many interventions have high financial costs, yet few studies mention the cost per or episode of care. Among the most expensive s for diabetic foot ulcers are the surgical techniques such as debridement, revascularisation and amputation. 31,32 Only one randomised controlled trial describing a surgical technique was identified in this review. 23 Growth factors, skin grafts and RGD peptide matrix may also prove to be extremely expensive although there is little information available on the costs of these interventions. Although these interventions may be extremely expensive they may be useful from patients' perspective due to their effectiveness in healing. One study which examined the effectiveness of prevention programmes 20 did report costs and concluded that this could prove to be more cost-effective than intervening when foot ulcers had developed. Prevention programmes would need to focus on the cost per ulcer avoided over a prescribed time interval. Where there is no difference in outcome between groups, for example, as reported by Grayson et al., 17 the of choice would be the one associated with the lower financial costs. These authors gave no suggestion of comparative costs so no clear conclusion can be drawn about the more efficient option. Chantelau et al. 10 concluded that antibiotics are not necessarily needed when treating less severe diabetic foot ulcers. This could have considerable cost implications in avoiding unnecessary antibiotics. However, as stated previously a small sample size was used and caution is needed in interpreting the results. Study design There were a number of common problems with the studies identified. Many had a very small sample size (and only one study 15 described the sample size calculation method used) and did not include any pre-existing power calculations based on expected differences in healing rates. Only one study referred to the problem of small sample size 10 and the potential for a type II statistical error, that is, missing a true difference. Many studies compared the following variables to illustrate the homogeneity between the intervention and control groups: age, sex, severity of diabetes, extent of neuropathy and vascular disease. These studies were probably confounded in view of the small number of patients involved and the wide range of important variables for which they were unable to control. Apelqvist et 102 J. Clin. Effect. Volume 3 Number 3 September 1998
Prevention and of diabetic foot ulcers Original article al. 8 suggested that since no significant difference was found between the two groups in their study, the trial should be repeated with a larger group of patients matched for distal blood pressure levels. Randomising an appropriate number of patients into each group, based on a pre-determined sample size calculation, would have maximized the chance of distributing all the important variables equally between the comparison groups and therefore determining a more meaningful result. Although it may be difficult to obtain sufficient numbers of patients to reach the pre-determined sample size, this problem could be overcome with multi-centre trials. pooling of data was attempted through metaanalysis as the studies were too dissimilar and the data therefore not homogeneous. In order to conduct a meta-analysis of studies at least three studies comparing the same intervention are needed. 33 Blinding Only seven of the 23 randomised controlled trials were double blinded. These included studies of wound applications such as growth factors, antibiotics or drugs where the was not externally obvious. It would be difficult to achieve successful blinding to some s, such as hyperbaric oxygen, where both the patient and clinician would be aware of the interventions. For this reason, many studies described comparison of two interventions rather than an intervention compared to a placebo. Only five studies were defined as including a placebo with sample sizes ranging from 13-40 patients for the whole study. Outcome measures Studies reported a range of outcome measures, including number of patients with complete healing within a defined time period and mean healing time. It is not possible to compare the results of studies where one reports time to reduce an ulcer size, 8,10,26 another reports the proportion of patients completely healed 8 or clinically cured 17 at a certain follow-up time and another reports number of ulcers healed in mean number of days 22 as the primary outcome. Moreover often no definition is given of the term used as outcome measure. Standardization of the outcomes monitored, such as complete healing, would allow comparison of results with other research. Follow-up interval The studies that fulfilled the criteria reported varying follow-up periods such as 20 days, 10 10 weeks, 26 3 months 8 or a median of 36 and 57 months. 17 Often length of follow-up, number of drop-outs and reasons for drop-outs are not reported. A standard follow-up period of, for example, 12 weeks would facilitate comparison of outcome with other studies. Where the follow-up period is short, the incidence of recurrence is hidden, since diabetic foot ulcers often reappear. Results presented with short follow-up periods may therefore give false impressions of effectiveness. Research priorities The studies which did show promising results included s such as lyophilised collagen, 12 RGD peptide matrix, 26 total contact casting, 22 growth factors, 25,27 wound grafting, 16 polymeric dressings, 9 therapeutic shoes, 28 systemic hyperbaric oxygen therapy for severe foot ulcers, 15 as well as prevention. 20,21 Although three of the trials 8,10,17 fulfilled the Jadad 29 criteria, they nevertheless showed no clinical benefit of the intervention over control. The one trial with significant results used only a 10-week follow-up, and reported percentage wound closure only. 26 These results should be interpreted as preliminary findings due to the limitations already described. More rigorous trials in the areas of total contact casting, growth factors, wound grafting and prevention are needed before sound conclusions may be drawn about clinical effectiveness of s. Conclusions and recommendations The and prevention of diabetic foot ulcers is extremely complex. Many factors contribute to the formation of diabetic foot ulcers and these must be taken into account when determining appropriate interventions. There is currently very little evidence upon which clinicians can base their clinical decisions for the of people with diabetic foot ulcers. This literature review of s for diabetic foot ulcers identified very few randomised controlled trials and a poor overall quality of research in the area. This highlights the need for more rigorous research to determine clinical effectiveness of s. As described above, several interventions show promise such as lyophilised collagen, RGD peptide matrix, total contact casting, growth factors, wound grafting, polymeric dressings and prevention. We recommend complete healing as the most appropriate outcome measure of diabetic foot ulcer s. We also recommend a follow-up period of 12 weeks with a long-term follow-up of one year in order to monitor recurrence and enable calculation of wound-free time during the one-year period as a standard outcome. In addition to monitoring clinical outcomes, more attention should be paid to assessing patients' healthrelated quality of life, acceptability, satisfaction, compliance and tolerance of. Of equal importance is the need for prospective randomised controlled trials, alongside economic evaluations, to identify the relative efficiency or cost-effectiveness of s for diabetic foot ulcers. References 1. Ward J D. The cost of diabetic foot problems. Pharmacoeconomics 1995; 8: 55-57 2. Shenaq S M, Klebuc M J, Vargo D. How to help diabetic patients avoid amputation. Prevention and management of foot ulcers. Postgraduate Medicine 1994; 96: 177-180, 183-186, 191-192 3. Boulton A J M. End-stage complications of diabetic neuropathy: foot ulceration. Canadian Journal of Neurological Sciences 1994; 21: S18-S22 J. Clin. Effect. Volume 3 Number 3 September 1998 103
Prevention and of diabetic foot ulcers 4. Bouter K P, Storm AJ, de Groot R R, Uitslager R, Erkelcns D W, Diepersloot RJ. The diabetic foot in Dutch hospitals: epidemiological features and clinical outcome. European Journal of Medicine 1993; 2: 215-218 5. Murray H J, Boulton A J. The pathophysiology of diabetic foot ulceration. Clinics in Podiatric Medicine and Surgery 1995; 12:1-17 6. Ahroni J H, Boyko E J, Pecoraro R E. Diabetic foot ulcer healing: extrinsic vs. intrinsic factors. Wounds: a Compendium of Clinical Research and Practice 1993; 5: 245-255 7. ApelqvistJ, Larsson J, Stenstorm A. Topical of necrotic foot ulcers in diabetic patients: a comparative trial of DuoDerm and MeZinc. British Journal of Dermatology 1990a; 123: 787-792 8. Apelqvist J, Castenfors J, Larsson J. Ketanserin in the of diabetic foot ulcer with severe peripheral disease. International Angiology 1990b; 9: 120-124 9. Blackman J D, Senseng D, Quinn L, Mazzone T. Clinical evaluation of a semipermeable polymeric membrane dressing for the of chronic diabetic foot ulcers. Diabetes Care 1994; 17: 322-325 10. Chantclau E, Tanudjaja T, Altenhofer F, Ersanli Z, Lacigova S, Metzger C. Antibiotic for uncomplicated neuropathic forefoot ulcers in diabetes: a controlled trial. Diabetic Medicine 1996; 13: 156-159 11. Clever H U, Dreyer M. Comparing two wound dressings for the of neuropathic diabetic foot ulcers. Proceedings of the 5th European Conference on Advances in Wound Management. London: MacMillan Magazines 1995; 201-203 12. DiMauro C, Ossino A M, Trefiletti M. Lyophilized collagen in the of diabetic ulcer. Drugs under Experimental and Clinical Research 1991; 17(7): 371-373 13. Doctor N, Pandya S, Supe A. Hyperbaric oxygen therapy in diabetic foot. Journal of Postgraduate Medicine 1992; 38:112-114 14. Donaghue V M, Chrzan J S, Rosenblum B I, Boston M A. A clinical evaluation of a collagen-alginate topical wound dressing in the management of diabetic foot ulcers. Presentation at 56th annual meeting and scientific sessions of the American Diabetes Association Number 2: June 9, 1996. http: //www.pharminfo.com/disease/ diabetes/ehlb/ehlbl_2.html 15. Faglia E, Favales F, Aldegi A et al. Adjunctive systemic hyperbaric oxygen therapy in of severe prevalently ischemic diabetic foot ulcer. Diabetes Care 1996; 19: 1338-1343 16. Gentzkow G D, Iwasaki S D, Hershon K S et al. Use of Dermagraft, a cultured human dermis, to treat diabetic foot ulcers. Diabetes Care 1996; 19: 350-354 17. Grayson M L, Gibbons G W, Habershaw G M et al. Use of ampicillin/sulbactam versus imipenem/cilastatin in the of limb-threatening foot infections in diabetic patients (published erratum appears in Clinical Infectious Disease 1994 Oct: 19(4): 820). Clinical Infectious Disease 1994; 18: 683-693 18. Leslie C A, Sapico F L, Ginunas V J, Adkins R H. Randomised controlled trial of topical hyperbaric oxygen for of diabetic foot ulcers. Diabetes Care 1988; 11:111-115 19. Lipsky B A, Pecoraro R E, Larson S A, Hanley M E, Ahroni J H. Outpatient management of uncomplicated lower-extremity infections in diabetic patients. Archives of Internal Medicine 1990; 150: 790-797 20. Litzelman D K, Slemenda C W, Langefeld C D et al. Reduction of lower extremity clinical abnormalities in patients with non-insulin-dependent diabetes mellitus: a randomised, controlled trial. Annals of Internal Medicine 1993;119:36-41 21. Malone J M, Snyder M, Anderson G, Bernhard V M, Holloway G A, Bunt T J. Prevention of amputation by diabetic education. American Journal of Surgery 1989; 158:520-524 22. Mueller M J, Diamond P T, Sinacore D R et al. Total contact casting in of diabetic plantar ulcers. Diabetes Care 1989; 12: 384-388 23. Pinzur M S, Smith D, Osterman H. Syme ankle disarticulation in peripheral vascular disease and diabetic foot infection: the one-stage versus two-stage procedure. Foot and Ankle International 1995; 16: 124-127 24. Richard J L, Parer-Richard C, Daures J P et al. Effect of topical basic fibroblast growth factor on the healing of chronic diabetic neuropathic ulcer of the foot. A pilot, randomised, double-blind, placebo-controlled study. Diabetes Care 1995; 18: 64-69 25. Steed D L, Malone J M, Goslen J B, Bunt J J, Holloway G A, Webster M W. Randomised prospective double-blind trial in healing chronic diabetic foot ulcers. Diabetes Care 1992; 15:1598-1604 26. Steed D L, Ricotta J J, Prendergast J J et al. Promotion and acceleration of diabetic ulcer healing by arginine-glycineaspartic acid (RGD) peptide matrix. Diabetes Care 1995; 18: 39-46 27. Steed D L, Donohoe D, Webster M W, Lindsley L. Effect of extensive debridement and on the healing of diabetic foot ulcers. Journal of the American College of Surgeons 1996; 183: 61-64 28. Uccioli L, Faglia E, Monticone G et al. Manufactured shoes in the prevention of diabetic foot ulcers. 1995; 18: 1376-1378 29. Jadad A R, Moore R A, Carroll D et al. Assessing the quality of reports of randomized clinical trials: is blinding necessary? Controlled Clinical Trials 1996; 17: 1-12 30. Moher D, Jadad A R, Nicol G, Penman M, Tugwell P, Walsh S. Assessing the quality of randomized controlled trials: an annotated bibliography of scales and checklists. Controlled Clinical Trials 1995; 16: 62-73 31. Eckman M H, Greenfield S, William C et al. Foot infections in diabetic patients: decision and costeffectiveness analyses. Journal of the American Medical Association 1995; 273: 712-720 32. Reiber G E. Diabetic foot care, financial implications and practice guidelines. Diabetes Care 1992; 15: 29-31 33. Eysenck H J. Problems with meta-analysis. In: Chalmers I, Altman D G (Eds). Sytematic Reviews. London: BMJ Publishing Group, 1995 104 J. Clin. Effect. Volume 3 Number 3 September 1998