EBP projects

N316 EBP Paper Part 1 Helpful Hints

N316 EBP Paper Part 2 Helpful Hints

N316 EBP Paper Part 3 Helpful Hints

O R I G I N A L A R T I C L E

A multicentre, randomised controlled clinical trial evaluating the effects of a novel autologous, heterogeneous skin construct in the treatment of Wagner one diabetic foot ulcers: Interim analysis

David G. Armstrong1 | Dennis P. Orgill2 | Robert Galiano3 | Paul M. Glat4 |

Lawrence Didomenico5 | Alexander Reyzelman6 | Robert Snyder7 |

William W. Li8 | Marissa Carter9 | Charles M. Zelen10

1Department of Surgery, University of Southern California, Keck School of Medicine, Los Angeles, California 2Division of Plastic Surgery, Brigham and Women's Hospital, Boston, Massachusetts 3Division of Plastic Surgery, Northwestern University Feinberg School of Medicine, Chicago, Illinois 4Drexel University, Philadelphia, Pennsylvania 5Lower Extremity Institute of Research and Therapy, Youngstown, Ohio 6Center for Clinical Research, San Francisco, California 7Clinical Research Barry University SPM, Brand Research Center, Barry University, Miami, Florida 8The Angiogenesis Foundation, Cambridge, Massachusetts 9Strategic Solutions, Bozeman, Montana 10Department of Medical Education, The Professional Education and Research Institute (PERI), Roanoke, Virginia

Correspondence Charles M. Zelen, DPM, Department of Medical Education, The Professional Education and Research Institute, 222 Walnut Ave., Roanoke, VA 24016. Email: [email protected]

Funding information Polarity TE, Grant/Award Number: 002

Abstract

We desired to carefully evaluate a novel autologous heterogeneous skin con-

struct in a prospective randomised clinical trial comparing this to a standard-

of-care treatment in diabetic foot ulcers (DFUs). This study reports the interim

analysis after the first half of the subjects have been analysed. Fifty patients

(25 per group) with Wagner 1 ulcers were enrolled at 13 wound centres in the

United States. Twenty-three subjects underwent the autologous heterogeneous

skin construct harvest and application procedure once; two subjects required

two applications due to loss of the first application. The primary endpoint was

the proportion of wounds closed at 12 weeks. There were significantly more

wounds closed in the treatment group (18/25; 72%) vs controls (8/25; 32%) at

12 weeks. The treatment group achieved significantly greater percent area

reduction compared to the control group at every prespecified timepoint of

4, 6, 8, and 12 weeks. Thirty-eight adverse events occurred in 11 subjects (44%)

in the treatment group vs 48 in 14 controls (56%), 6 of which required study

removal. In the treatment group, there were no serious adverse events related

to the index ulcer. Two adverse events (index ulcer cellulitis and bleeding)

were possibly related to the autologous heterogeneous skin construct. Data

from this planned interim analysis support that application of autologous het-

erogeneous skin construct may be potentially effective therapy for DFUs and

provide supportive data to complete the planned study.

K E Y W O R D S

biological products, diabetic foot, randomised controlled trial, ulcer, wound healing

Received: 17 February 2021 Revised: 28 March 2021 Accepted: 31 March 2021

DOI: 10.1111/iwj.13598

This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any

medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made.

© 2021 The Authors. International Wound Journal published by Medicalhelplines.com Inc (3M) and John Wiley & Sons Ltd.

64 Int Wound J. 2022;19:64–75.wileyonlinelibrary.com/journal/iwj

1 | INTRODUCTION

Diabetic foot ulcers (DFUs) cost Medicare $6.2–18.7 bil- lion each year and have a devastating annual impact on the economy of United States, with an annual burden of over $50 billion.1,2 Approximately 1.5 million Americans have DFUs, which contribute to 130 000 annual lower- extremity amputations.3,4 A real-world analysis of 62 964 DFUs registered in the US Wound Registry found that their healing rate at 12 weeks was only 30.5%.5 A meta- analysis of DFUs treated in trials with standard of care revealed a 12-week closure rate of 24%.6 Biological skin substitutes are commonly used as adjunctive therapy to improve wound closure.7,8 However, most products are quite costly and require multiple applications. Split- thickness skin grafting (STSG) can contribute new healthy tissue to the wound bed but has a failure rate of approximately 30% when applied to DFUs as a conse- quence of poor graft take by the chronic wound bed, the presence of diabetes, vascular insufficiency, other com- orbidities, and/or bacterial contamination.9-14 As many DFUs are treated in the outpatient setting, another disad- vantage of skin grafting is that it involves a surgical pro- cedure in the operating room.

A novel autologous heterogeneous skin construct (AHSC) created from a small harvest of full thickness, healthy skin may be safe and effective as adjunctive ther- apy in treating complex and refractory wounds.15-24 AHSC is composed of small multicellular segments and contains the endogenous regenerative cellular populations of healthy skin that promote wound closure, so that a single application can regenerate full-thickness, functionally polarised skin on the wound bed.20-25 The manufacturing process of the AHSC retains the endogenous regenerative cellular populations associated with wound healing pre- sent within hair follicles, glands, and the interfollicular epidermis, facilitating engraftment optimisation and wound closure.24 AHSC is not cultured ex vivo, but rather it is expeditiously returned to the provider to be adminis- tered topically over a clean, debrided, viable wound bed and covered with common nonadherent, nonabsorbent dressings in the outpatient setting. The AHSC conforms nicely to the wound and over days forms small skin islands that expand and coalesce across the entire wound bed to close the wound, rather than initiating epithelialisation solely from the wound margin.20,21 In a pilot study of 11 patients with DFUs extending up to the tendon, bone, or capsule, 10 patients closed within 8 weeks of a single application of AHSC, with the mean percent area reduc- tion (PAR) for all wounds at 4 weeks at 83%.24 A larger, controlled trial was needed to confirm these initial findings in DFUs. A planned interim analysis of the first 50 of the 100 patients of a randomised controlled trial (RCT) was

performed to compare the effects of AHSC to standard of care in the treatment of Wagner 1 DFUs.

2 | METHODS

2.1 | Study design and population

This was a planned interim analysis of the first 50 patients of a prospective, multicentre, RCT evaluating wound clo- sure rates of DFUs treated in an outpatient setting. Thir- teen wound care centres in the United States participated in this study. The null hypothesis was the proportion of wounds closed at 12 weeks, after up to 12 weeks of AHSC and standard of care or standard of care alone, would be equal for groups 1 (AHSC + standard of care) and 2 (con- trol). Formally, H0: I1–I2 = 0; HA: I1–I2 = D1 ≠ 0, where I1 was the proportion of wounds closed in group 1, I2 was the same metric for group 2, D1 was the differ- ence (I1–I2); assuming the alternative hypothesis and sta- tistical test used was chi square/Fisher exact test. The primary endpoint was the percentage of index ulcers closed at 12 weeks. Complete closure was defined when 100% epithelialisation without drainage was first observed, followed by a closure confirmation visit

Key Messages

• This interim analysis of an ongoing, random- ised controlled trial evaluated a single applica- tion of autologous heterogeneous skin construct (AHSC) as adjunctive therapy to standard of care in Wagner 1 diabetic foot ulcers compared to standard of care alone in 50 initial subjects.

• There were significantly more wounds closed in the AHSC group (18/25; 72%) compared to the control group (8/25; 32%) (P = .005) and a significantly greater percent area reduction in the AHSC group compared to the control group at each prespecified timepoint of 4 weeks (79% vs 24%, P = .0002), 6 weeks (83% vs 44%, P = .004), 8 weeks (87% vs 47%, P = .002), and 12 weeks (88% vs 50%, P = .012), respectively.

• In the AHSC group, there were no serious adverse events related to the index ulcer or determined to be related to AHSC treatment.

• These data support continuation of the planned study

ARMSTRONG ET AL. 65

2 weeks later. Secondary endpoints included the PAR at 4, 6, 8, and 12 weeks; changes in wound quality-of-life (W-QOL short questionnaire, with each question scored on a scale of 0 = “not at all” to 4 = “very much”); reduced pain (based on the Visual Analogue Scale [VAS], with 0 = no pain and 10 = worst possible pain); improve- ments in peripheral neuropathy by Semmes Weinstein monofilament test; and incidence of adverse events (AEs) and complications.

The sample size was determined to be 102 (51 in each group) to achieve 89% power to detect a difference between the group proportions of 0.3. The proportion in the AHSC group was assumed to be 0.3 under the null hypothesis and 0.6 under the alternative hypothesis. The proportion in the control group was 0.3. The test statistic used was the two-sided Z test with pooled vari- ance. The significance level of the test was targeted at 0.05. The significance level actually achieved by this design was 0.05. Unblinded interim analysis was per- formed after 50 subjects completed the study in order to assess subject outcomes between the groups and to recalculate the sample size for the primary endpoint. This study was conducted according to the principles expressed in the Declaration of Helsinki, and the Insti- tutional Review Board Advarra (Columbia, MD) approved the study protocol. The study protocol was registered on clinicaltrials.gov (NCT03881254).

Adult patients with a Wagner 1 DFU that did not involve the tendon, muscle, or bone, provided that it was below the aspect of the medial malleolus, were screened for study participation. Table 1 details complete inclusion and exclusion criteria. Eligible patients provided their written informed consent and were enrolled into the study. During their first screening visit, their demo- graphics and medical history were recorded; a complete physical examination was performed; laboratory tests were taken; the index ulcer was assessed for infection and pain; adequate perfusion was confirmed; Semmes Weinstein monofilament test for peripheral neuropathy was performed; subjects answered the W-QOL short questionnaire; sharp debridement of the index ulcer was performed as needed; the wounds were dressed with standard of care; and offloading was initiated.

Two weeks after the initial screening visit, subjects returned to undergo the same assessments to check for any changes in their health, ulcer healing status, and eli- gibility. Randomisation occurred if the ulcer did not change in size greater than 30% and still met eligibility. The Organisation1 (City2, State2) used a block size of 10 for randomisation (5 sheets of paper with a standard- of-care assignment and 5 with an AHSC assignment). Each sheet was inserted into an opaque envelope that was sealed. The study coordinator shuffled the envelopes,

while under observation by the principal investigator and staff. After repeating the process 10 times, the envelopes were sent to the study sites, ensuring that site investiga- tors were blinded to the randomisation method and treat- ment assignment. The site investigators enrolled the subjects into the study and were aware of the study group following randomisation.

2.2 | AHSC preparation, application, and follow-up

Following randomisation, standard of care was applied to both groups, and the AHSC group underwent the skin harvest procedure. Standard of care included offloading of the DFU (CAM boots or total contact casting, if the subject's foot was too large for a CAM boot, or per the provider's discretion), appropriate sharp or surgical debridement, collagen alginate and appropriate wound care covering, including 4 × 4 gauze pads, foam, and a multilayer compression ban- daging system comprised a soft roll layer, an elastic layer, and a cohesive bandage layer (Dyna-Flex, KCI, St. Paul, MN).In the AHSC group, a 1 × 2 cm full- thickness harvest of healthy skin was excised from the index limb of each subject using sterile technique and local. The provider sutured closed the harvest site. The harvest was shipped overnight to a Food and Drug Administration–registered biomedical manufacturing facility (PolarityTE, Salt Lake City, UT) and used to manufacture the AHSC (Product, Organisation3). The AHSC was returned to the provider within 48 hours of tissue harvest and applied to the wound within 4 days after the harvesting procedure. The AHSC was shipped and stored at 4�C before application.

On the day of the application procedure, the wound was cleaned and sharply debrided, if required. The AHSC was spread evenly across the wound bed. Next, the wound was dressed with a silicone dressing covered by an absorbent foam dressing (DermaFoam, DermaRite Industries, North Bergen, NJ). A three-layer compression bolster was then applied. Dressings were changed weekly, and wounds continued to be offloaded. At the third follow-up visit, a nonadherent contact layer (Adaptic Touch, KCI) replaced the silicone dressing. After the AHSC was applied and the wound was addressed, a time-out procedure undertaken by the on- site study team confirmed the application of the subject's own harvested construct to the index ulcer.

Subjects in both groups had weekly follow-up visits and dressing changes with standard of care for up to 12 weeks. At each visit, wound sites (including the har- vest sites in the AHSC group) were assessed for healing

66 ARMSTRONG ET AL.

status, pain, and infection; the index ulcer was measured and assessed for graft take; and AEs were reported. A licensed provider who did not treat the index ulcer first performed an initial, blinded wound closure assessment of the wound in-person. Once considered healed by the blinded investigator, the wound images were forwarded to a group of university plastic surgeon adjudicators who determined if the wound was healed within 24 hours of receiving the photographs. If two-thirds of the adjudica- tors agreed that the wound had closed, then the subject returned for a closure confirmation visit 2 weeks later. At the end-of-study visit, W-QOL was also assessed, and a Semmes-Weinstein monofilament test was administered for peripheral neuropathy.

2.3 | Data collection and analysis

Data were stored in an Excel database. The statistical analysis was performed using PASW 27 (IBM, Chicago, IL). Blinded, interim analysis was first performed, and coding for treatment was then applied to the analysis involving comparison of groups.

The intent-to-treat (ITT) and safety populations com- prised randomised subjects who received at least 1 treat- ment. All analyses used the ITT approach. The last observation carried forward principle that was used with regard to missing area data at study visits. Study variables were summarised as means and SDs for continuous vari- ables as well as medians for nonnormal data. Categorical

TABLE 1 Patient inclusion and exclusion criteria

Inclusion criteria

• At least 18 years old • Presence of a Wagner 1 DFU that did not extend through the dermis or subcutaneous tissue and did not involve the tendon, muscle,

or bone, provided that it was below the aspect of the medial malleolus • If two or more Wagner 1 DFUs were present, then the index ulcer was the largest ulcer and the only one evaluated in the study. Any

other ulceration must have been 2 cm distant from the index ulcer.

• Index ulcer was present for at least 4 weeks • Index ulcer was a minimum of 1.0 cm2 and a maximum of 25 cm2 at screening visit and did not reduce/increase in area by 30% or

more after 14 days of standard of care prior to first treatment visit

• Index ulcer had been offloaded for ≥14 days prior to randomisation

• Index ulcer had a clean granular base, was free of necrotic debris, and appeared to healthy, vascularised tissue at time of AHSC placement

• Affected foot had adequate circulation as documented by a dorsal transcutaneous oxygen measurement or a skin perfusion pressure measurement of ≥30 mmHg, or an ankle brachial index of ≥0.7 and ≤1.2, or arterial Doppler with a minimum of biphasic flow within 3 months of treatment

• Women of childbearing age were willing to use contraception during the study and undergo pregnancy tests

• Patient understood and was willing to participate in the study, could comply with the weekly visits and follow-up, and provided written informed consent.

Exclusion criteria

• Active osteomyelitis, cellulitis, soft tissue infection, or active Charcot's arthropathy of the affected foot involving or near the index ulcer site, or on the same limb as the index ulcer within 30 days prior to randomisation

• Index ulcer was suspicious of cancer

• History of radiation at the index ulcer site

• History of >2 weeks treatment with immunosuppressants (including systemic corticosteroids), cytotoxic chemotherapy, or application of topical steroids to the index ulcer surface within 1 month prior to screening, or who were anticipated to require such medications during the study

• Evidence of unstable HIV, hepatitis B, or hepatitis C • On an investigational drug or therapeutic device within 30 days of screening • Index ulcer was previously treated or needed to be treated with any prohibited therapies such as chlorhexidine or collagenase • Presence of any condition which seriously compromised the patient's ability to complete the study or had a known history of poor

adherence with medical treatment • In the opinion of the investigator, the patient was noncompliance with offloading or index ulcer dressing prior to randomisation • Pregnant or breastfeeding • Presence of diabetes with poor metabolic control as documented with an HbA1c ≥12.0 within 30 days of randomisation • Presence of end-stage renal disease as evidenced by serum creatinine of greater than 3.0 mg/dL within 120 days of randomisation

Abbreviations: AHSC, autologous homologous skin construct; DFU, diabetic foot ulcer.

ARMSTRONG ET AL. 67

variables were presented as counts and proportions or percentages. Statistical testing between groups at baseline was carried out to examine the success of randomisation. For categorical variables, chi-squared or Fisher exact tests were performed, and for continuous variables indepen- dent t tests or Mann-Whitney tests were used (depending on variable normality) to test for statistical differences.

The PAR for the index ulcer at X weeks was calcu- lated as ([AI – AXW]/AI)×100, where AI is the area of the index wound at randomisation and AXW is the area at X weeks. When AHSC was applied twice, area data by week was based on data associated with the first AHSC application, followed by the second AHSC application, and then follow-up.

The primary endpoint (proportion of wounds closed at 12 weeks) between study groups was analysed using chi square.

Secondary endpoints between study groups were analysed by chi-squared or Fisher exact tests for categori- cal variables, while independent t tests or Mann–Whitney tests were used to test for statistical differences for contin- uous variables depending on outcome variable normality, which was examined using the Wilks-Shapiro test. The exception was PAR at 2, 4, 6, 8, and 12 weeks, which was analysed using general linear mixed modelling (GLMM) with repeated measures (no random effects). Two-sided P values <.05 were considered significant.

Summary statistics were used as inputs to calculate the conditional statistical power for all endpoints based on a final N of 100 using PASS13 software (NCSS, LLC, Kaysville, UT).

All AEs were categorised as “serious” or “not serious” and assessed for severity (mild, moderate, severe, or life- threatening) and relationship to the AHSC product and harvesting and placement procedures (not related, possi- bly related, probably related, or definitely related).

3 | RESULTS

Study recruitment began on April 2, 2019, and all sub- jects exited the trial by June 20, 2020. This interim analy- sis covers the 79 patients screened for eligibility and the 50 subjects (63%) who were enrolled (Figure 1). One sub- ject (4%) was withdrawn from the AHSC group due to development of respiratory illness and sepsis, whereas 6 subjects (24%) were withdrawn in the control group due to 1 subject being incarcerated and 5 having AEs occur that required study removal (Figure 1). Table 2 summarises patient demographics and medical history with no significant differences between groups. Three subjects in the AHSC and 6 in the control group had missing HbA1c data. The index ulcer was treated with

multiple therapies prior to study enrolment, with similar treatments applied to both groups, except for antibiotics, which were administered significantly more to the con- trol group (P = .023) (Table 2).

All 25 subjects in the AHSC group underwent the AHSC harvest and application procedure, but 2 subjects required a second AHSC application due to loss of the first application requiring a second tissue harvest. The proximal medial calf was the most common harvest site (17/27, 63%). Upper medial thigh and proximal lateral leg were harvested for the remainder of the cases. Nine AHSC constructs (33%) were applied 2 days after harvest, 17 (63%) after 3 days, and 1 (4%) after 5 days.

3.1 | Closure rates

There were statistically significantly more wounds closed in the AHSC group (18/25; 72%) compared to the control group (8/25; 32%) at 12 weeks (P = .005). Closure rates through week 12 are shown in Figure 2. Based on these data and using the 2-side Z test with pooled variance, the projected statistical power for 100 subjects was 98.8%.

Table 3 and Figure 3 summarise the PAR data through 12 weeks. The GLMM model would not always converge when a random intercept model was incorpo- rated into a factorial fixed effects model with 4 levels (PAR at the 4 time periods) no matter what covariance matrix was selected. Removing the random effects and using the simpler model with an unstructured correla- tions covariance matrix resulted in a worse fit but similar to other covariance matrices (−2LL or BIC); however, for treatment, a significant effect was observed (P = .013). Based on these data, the projected statistical power for 100 subjects was 90+%. Representative images of wound closure are shown in Figure 4.

All harvest sites remained closed following primary closure and fully healed within 12 weeks except for in 1 subject who was withdrawn from the trial before healing could be confirmed.

3.2 | Safety analysis

There were 86 AEs allocated to 25 subjects. The AHSC group had 38 AEs allocated to 11 subjects (44%), while the control group had 48 AEs allocated to 14 subjects (56%). The overall AE rate was 1.5 for the AHSC group and 1.9 for the control group.

There were 13 SAEs, 7 in the AHSC group and 6 in the Control group. In the AHSC group, 1 subject had 3 SAEs (congestive heart failure, dyspnea episode that was a symptom of SARS-CoV-2 infection, and his index

68 ARMSTRONG ET AL.

Assessed for eligibility (n = 79)

Excluded (n = 29)a

♦ Ineligible ulcer area (22.4%) ♦ Ulcer area increased/decreased by

≥30% 2 weeks after screening (19.0%)

♦ Declined to participate (10.3%) ♦ Treatment site had soft tissue infection

or gangrene (12.1%)

♦ Noncompliance with offloading or dressing (6.9%)

♦ Not a Wagner 1 DFU (6.4%) ♦ HbA1c ≥12.0 (3.4%) ♦ End stage renal disease (3.4%) ♦ COVID-19 concerns (3.4%) ♦ Hospitalized with COVID-19 (3.4%) ♦ Unable to measure ulcer area (3.4%) ♦ Subject did not show up for visit (3.4%) ♦ Ineligible wound area (22.4%) ♦ Prohibited therapies applied to index ulcer

(1.1%)

♦ Osteomyelitis/bone infection, cellulitis, active Charcot’s arthropathy of the index limb (1.1%)

Analysed (n = 25)

Lost to follow-up (n = 0)

Withdrawn (n = 1)

Allocated to AHSC intervention (n = 25)

♦ Received allocated intervention (n = 25)

Lost to follow-up (n = 0)

Withdrawn (n = 6)

♦ Incarcerated (n = 1) ♦ Adverse Events (n = 5) ♦ Osteomyelitis and a non-STEMI (n = 1) ♦ Tunneling of study wound (n = 1) ♦ Secondary ulcer, prescribed antibiotics,

and major protocol deviation (n = 1)

♦ Died after sepsis with possible pneumonia and acute cerebrovascular accident (n = 1)

♦ Multiple fractures after vehicle crash (n = 1)

Allocated to Standard of Care (n = 25)

♦ Received allocated intervention (n = 25)

Analysed (n = 25)

Allocation

Analysis

Follow-Up

Randomized (n = 50)

Enrolment

FIGURE 1 Legend on next page.

ARMSTRONG ET AL. 69

wound required cauterisation following admission for his congestive heart failure and during the admission, his wound was debrided against protocol by the non-trail site-admitting service, while the patient was on anti- coagulation), while another subject had 2 SAEs (sepsis, related to a hepatitis A infection, and cellulitis of the right leg, which was not related to the index ulcer). Two other subjects had 1 SAE each: an upper gastrointestinal bleed and an acute kidney injury. In the control group, 1 subject had 4 SAEs over a 3-week period, beginning with the development of left foot cellulitis related to the index ulcer, followed by acute osteomyelitis, which required surgery; severe sepsis occurred after the surgical procedure, but it quickly resolved. A separate SAE also occurred in a control group subject during this time period (non–ST-segment elevation myocardial infarc- tion). Another control group subject developed a soft tis- sue infection related to the index ulcer, which was treated with sharp debridement and antibiotics and resolved after 7 weeks.

In the AHSC group, there were no product-related SAEs. There were two AEs that were possibly related to the treatment of the index ulcer: 1 infection of the study right heel DFU and a bleeding episode of the study ulcer located on the plantar aspect of the 5th metatarsal head, right foot. Only 2 AEs (pain and cel- lulitis) occurred in the harvest site, both in the AHSC group. There were 7 index ulcer infections (including cellulitis) in the control group compared to 1 in the AHSC group. There were 4 non-index ulcer infections in both groups. The AHSC group had 7 other complica- tions reported, including 1 for the index ulcer, while the Control group had 17 other complications, includ- ing 4 for the index ulcer. There were 24 other causes of AEs in the AHSC group versus 20 in the Control Group.

3.3 | Other secondary endpoints

The mean (SD) difference in the W-QOL scores between week 1 and week 12 visits was 0.1 (0.8) in the AHSC group vs 0.6 (1.2) in the control group (P = .09).

The mean (SD) difference in pain scores between week 1 and week 12 visits was 0.7 (1.6) in the AHSC group and 0.5 (1.6) in the control group (P = .48).

The mean (SD) difference in Semmes-Weinstein scores between week 1 and week 12 visits was 0.1 (1.4) in

the AHSC group and 0.4 (2) in the control group (P = .16).

4 | DISCUSSION

A traditional method of tissue reconstruction for Wagner 1 ulcers is a skin graft once the wound has been cleaned and granulating.20 However, a skin graft requires techni- cally demanding surgical procedure with careful postop- erative care, which is not easily available in many wound care centres. It is further complicated because neuropa- thy increased the risk of infection, endothelial dysfunc- tion, and overall higher graft failure rate compared to other wound types and locations..9-14,26,27 There are many investigators developing biological ulcer products with the goal of creating an ideal cost effective wound dressing that when applied to wounds will assist with healing without the complexities of surgical intervention.26,28 In a small pilot study, AHSC applied just once to DFUs in the outpatient setting was able to close 10/11 (91%) of index ulcers by 12 weeks.24 In our current study, we analysed the outcome data of the initial 50 patients as part of a planned interim analysis of a larger, ongoing RCT. These data support that adjunctive AHSC appears to facilitate greater DFUs closure compared to standard of care alone. The AHSC 12-week closure rates were sig- nificantly superior to the controls (72% vs 32%, P = .005) and allow us to project statistical power for 100 subjects in this ongoing trial at 99%. The AHSC 12-week DFU clo- sure rate of 72% in this interim analysis is a stark contrast to the mean closure rate reported in an analysis of 26 DFU RCTs, whereby only 38% of wounds healed at 12 weeks.5 In our study, 92% of subjects required only 1 application of AHSC. Additionally, all harvest sites remained closed following primary closure at the time of harvest and the harvest procedure was tolerated well by all participants. The occurrence of AEs and SAEs was similar between the AHSC and control groups, and only 2 AEs were possibly related to the study product. Nota- bly, in the AHSC group, there were no SAEs related to the index ulcer, whereas 2 subjects in the control group had 4 SAEs related to the index ulcer, including 1 subject that developed cellulitis followed by acute osteomyelitis requiring surgical incision and drainage. Statistical signif- icance between groups for AEs and SAEs was not included in the interim analysis predefined statistical analysis plan, but the occurrence of more index ulcer-

FIGURE 1 Patient flow diagram. The superscript letter “a” indicates that when multiple exclusion criteria applied, a weighted figure was applied so that percentages for each criterion added up to 100%. AHSC, autologous homologous skin construct, DFU, diabetic foot ulcer;

non-STEMI, non–ST-segment elevation myocardial infarction

70 ARMSTRONG ET AL.

TABLE 2 Patient demographics and medical history

Variable AHSC group (n = 25) Control group (n = 25) P

Patient age (years) 61.6 (10.3) 59.3 (13.5) .51

BMI 32.3 (7.6) 33.4 (7.5) .59

Sex

Male 18 (72) 17 (68) .76

Female 7 (28) 8 (32)

No. of comorbidities 9.6 (3.3) 10.8 (6.2) .40

Creatinine 1.4 (0.6) 1.3 (0.5) .37

HbA1c

Baseline 7.1 (1.4) 7.7 (1.7) .16

End of study 7.1 (1.6)a 8.0 (1.3)b .059

Wound area (cm2) 4.3 (4.2); median: 3.6; IQR: 3.2 3.3 (4.3); median: 1.8; IQR: 1.4 .19

Wound age (weeks) 25.3 (31.4); median: 15.3; IQR: 19 22.1 (22.6); median: 14.0; IQR: 20 .57

DFU location

Plantar 21 (84) 21 (84) 1.00

Dorsal 4 (16) 4 (16)

DFU location

Toe 4 (16) 5 (20)

Forefoot 10 (40) 13 (52) .16

Midfoot 9 (38) 2 (8)

Heel 2 (8) 4 (16)

Ankle 0 (0) 1 (4)

No. of debridements prior to enrolment 9.0 (3.8); median: 9; IQR: 6 10.6 (4); median: 10; IQR: 8 .17

Frequency of comorbidities

Hypertension 23 (92) 22 (88) .64

Peripheral arterial/vascular disease 3 (12) 4 (16) 1.00

Heart disease (any type) 3 (12) 6 (24) .46

Gastroesophageal reflux disease 6 (24) 3 (12) .46

Hyperlipidaemia 15 (60) 14 (56) .77

Renal disease 3 (12) 3 (12) 1.00

Venous insufficiency 3 (12) 1 (4) .61

Prior lower extremity amputation (any kind) 10 (40) 10 (40) 1.00

Mental disorder (any) 7(28) 10 (40) .37

Treatments up to 1 year prior

Debridements 14 (56) 16 (64) .56

Wraps or offloading 12 (48) 10 (40) .57

Negative pressure wound therapy 0 (0) 2 (8) .49

Cellular and/or tissue-based product 1 (4) 2 (8) .55

Collagen or oxidised regenerated cellulose 8 (320 6 (24) .53

Antibacterial dressing 4 (16) 3 (12) .68

Nonactive dressing 8 (32) 14 (56) .087

Antibiotics (any route) 1 (4) 8 (32) .023

Note: Continuous variables are reported as means (SD) and categorical variables as counts (percentage). Abbreviations: AHSC, autologous homologous skin construct; BMI, body mass index; DFU, diabetic foot ulcer; IQR, interquartile range. an = 22. bn = 19.

ARMSTRONG ET AL. 71

related SAEs and index ulcer infections in the control group suggests that earlier wound closure and a higher rate of wound closure with AHSC adjunctive treatment may avoid wound-related complications.

The manufacturing process of the AHSC retains the endogenous regenerative cellular populations associated with wound healing present within hair follicles, glands, and the interfollicular epidermis, facilitating engraftment optimisation and wound closure.24 The resulting

FIGURE 2 Weekly closure rates. AHSC, autologous homologous skin

construct

TABLE 3 Mean (SD) percentage area reduction at weeks 4, 6, 8, and 12

Week AHSC group Control group

4 78.6 (35.6) 24.0 (106.5)

6 83.2 (40.9) 43.8 (102)

8 86.6 (39.6) 47.2 (89.9)

12 88.2 (39.1) 49.6 (101.4)

FIGURE 3 Weekly percentage area reduction values. AHSC, autologous

homologous skin construct

72 ARMSTRONG ET AL.

construct has a high surface area-to-volume ratio, facili- tating cellular sustenance from plasmatic imbibition in the DFU wound bed during the first 48 hours prior to inosculation and vascularisation.24,29,30 Consequently, a single application of AHSC can quickly regenerate healthy tissue and close DFUs, which has significant cost implications (to be further explored in the final trial analysis).

There were no significant differences in W-QOL or the Semmes-Weinstein test between groups. This is notable as patients were required to undergo a small tissue harvest for the AHSC treatment. The harvest site procedure did not significantly negatively impact their W-QOL scores, which may have been balanced by faster wound closure with AHSC treatment. The lack of significant difference in the Semmes-Weinstein test may be due to the prevalence and severity of neuropathy present in both patient groups that cannot be corrected with topical treatments alone.

The results of this interim analysis are limited by the ongoing nature of the trial. However, the purpose of this interim analysis was to determine conditional statistical power for all study endpoints. A further study limitation is that there was no follow-up period after 12 weeks or fol- lowing wound closure beyond 2 weeks. This RCT is also

limited by its lack of blinding, which, given the interven- tion, was not possible. For blinding to have occurred, all patients would have had to undergo the harvest site proce- dure, which would not be ethically justified in the control group. However, wound closure was assessed in person by nontreating blinded study personnel and further confirmed by a blinded adjudication panel of three plastic surgeons using high-resolution digital photography.

This interim analysis of data from 50 patients enrolled in a larger, ongoing RCT demonstrated that a single, topi- cal application of the AHSC facilitated DFU closure. The results of this analysis confirm our previous power analy- sis and are encouraging to complete the planned study.

ACKNOWLEDGEMENT PolarityTE provided a grant to complete this clinical trial.

CONFLICT OF INTEREST This study was funded through a research grant from Pol- arityTE; provided to the Professional Education and Research Institute (PERI), which Charles M Zelen, DPM, is medical director. David Armstrong, DPM, MD, PhD, received research funds from PERI to serve as Principal Investigator for this trial and to design and administrate

FIGURE 4 Representative images of AHSC-treated patients, at the time of

randomisation (baseline), AHSC

deployment, during follow-up (interim

closure), and at closure confirmation

visit. AHSC, autologous homologous skin

construct

ARMSTRONG ET AL. 73

the trial and also assist with the writing and review of the manuscript. Dennis Orgill, MD, PhD, received research funds to serve as a validating/adjudicating plastic surgeon to review study photos and assist with the writing and review of the manuscript. Robert Galiano, MD, received research funds to serve as a vali- dating/adjudicating plastic surgeon to review study photos and assist with the writing and review of the manuscript. Paul Glat, MD, received research funds to serve as a validating/adjudicating plastic surgeon to review study photos and assist with the writing and review of the manuscript. Lawrence Didomenico, DPM, received research funds and served a site investigator for this trial and assisted with the writing and review of the manuscript. Alexander Reyzelman, DPM, received research funds and served a site investigator for this trial and assisted with the writing and review of the manuscript. Robert Snyder, DPM, received research funds and served a site investigator for this trial and assisted with the writing and review of the manuscript. Marissa Carter, PhD, received research funds to provide the statistical analysis plan and provide the statistical analysis for this trial and assist with writing of the result section of the manuscript. William W Li, MD, received research funds to serve as the medical monitor and assisted with the writing and review of the manu- script. Charles M. Zelen, DPM, is the medical director of the PERI and his company received research funds to administrate the clinical trial and write the paper for publication. There are no other conflicts of interest with any of the authors in relationship to this study or with regard to PolarityTE. IRB conflict of interest statements are on file with PERI.

DATA AVAILABILITY STATEMENT The data that support the findings of this study are avail- able on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.

ORCID Charles M. Zelen https://orcid.org/0000-0001-5682- 7056

REFERENCES 1. Nussbaum SR, Carter MJ, Fife CE, et al. An economic evalua-

tion of the impact, cost, and Medicare policy implications of chronic nonhealing wounds. Value Health. 2018;21:27-32.

2. Driver VR, Fabbi M, Lavery LA, Gibbons G. The costs of dia- betic foot: the economic case for the limb salvage team. J Am Podiatr Med Assoc. 2010;100:335-341.

3. Armstrong DG, Boulton AJM, Bus SA. Diabetic foot ulcers and their recurrence. N Engl J Med. 2017;376:2367-2375.

4. U.S. Department of Health and Human Services. Centers for Disease Control and Prevention. National Diabetes Statistics Report 2020 https://www.cdc.gov/diabetes/pdfs/data/statistics/ national-diabetes-statistics-report.pdf Accessed 22 July 2020.

5. Fife CE, Eckert KA, Carter MJ. Publicly reported wound healing rates: the fantasy and the reality. Adv Wound Care (New Rochelle). 2018;7:77-94.

6. Margolis DJ, Kantor J, Berlin JA. Healing of diabetic neuro- pathic foot ulcers receiving standard treatment. A Meta- Analysis Diabetes Care. 1999;22(5):692-695.

7. Gordon AJ, Alfonso AR, Nicholson J, Chiu ES. Evidence for healing diabetic foot ulcers with biologic skin substitutes: a sys- tematic review and meta-analysis. Ann Plast Surg. 2019;83(4S) Suppl 1:S31-S44.

8. Santema TB, Poyck PPC, Ubbink DT. Skin grafting and tissue replacement for treating foot ulcers in people with diabetes. Cochrane Database Syst Rev. 2016;2:CD011255.

9. Ramanujam CL, Han D, Fowler S, Kilpadi K, Zgonis T. Impact of diabetes and comorbidities on split-thickness skin grafts for foot wounds. J Am Podiatr Med Assoc. 2013;103:223-232.

10. Donegan RJ, Schmidt BM, Blume PA. An overview of factors maximizing successful split-thickness skin grafting in diabetic wounds. Diabet Foot Ankle. 2014;5:24769.

11. Ramanujam CL, Stapleton JJ, Kilpadi KL, Rodriguez RH, Jeffries LC, Zgonis T. Split-thickness skin grafts for closure of diabetic foot and ankle wounds: a retrospective review of 83 patients. Foot Ankle Spec. 2010;3:231-240.

12. Thourani VH, Ingram WL, Feliciano DV. Factors affecting suc- cess of split-thickness skin grafts in the modern burn unit. J Trauma. 2003;54:562-568.

13. Reddy S, El-Haddawi F, Fancourt M, et al. The incidence and risk factors for lower limb skin graft failure. Dermatol Res Pract. 2014;2014:582080.

14. Rose JF, Giovinco N, Mills JL, Najafi B, Pappalardo J, Armstrong DG. Split-thickness skin grafting the high-risk dia- betic foot. J Vasc Surg. 2014;59:1657-1663.

15. Armstrong DG, Orgill DP, Galiano RD, Glat PM, Carter MJ, Zelen CM. Open-label venous leg ulcer pilot study using a novel autolologous homologous skin construct. Plast Reconstr Surg Glob Open. 2020;8(7):e2972.

16. Feldman MJ, McLawhorn MM, Han J, et al. A prospective, multicenter, pilot trial of a novel homologous skin construct on deep partial-thickness and full-thickness burns. Ann Burns Fire Disasters. 2020;33:191-197.

17. Johnson ON, Nelson M, Estabrooke I, Sopko N, Swanson EW. Successful treatment of war zone traumatic lower extremity wound with exposed tendons using an autologous homologous skin construct. Cureus. 2020;12:1-7.

18. Berg A, Kaul S, Rauscher GE, Blatt M, Cohn S. Successful full- thickness skin regeneration using epidermal stem cells in trau- matic and complex wounds: initial experience. Cureus. 2020;12: e10558.

19. Murphy GA, Woelfel SL, Armstrong DG. What to put on (and what to take off) a wound: treating a chronic neuropathic ulcer with an autologous homologous skin construct, offloading and common sense. Oxford Med Case Rep. 2020;2020:259-263.

20. Granick MS, Baetz NW, Labroo P, Milner S, Li WW, Sopko NA. In vivo expansion and regeneration of full-thickness functional skin with an autologous homologous skin construct:

74 ARMSTRONG ET AL.

clinical proof of concept for chronic wound healing. Int Wound J. 2019;16:841-846.

21. Patterson CW, Stark M, Sharma S, Mundinger GS. Regenera- tion and expansion of autologous full-thickness skin through a self-propagating autologous skin graft technology. Clin Case Rep. 2019;7:2449-2455.

22. Isbester K, Wee C, Boas S, Sopko N, Kumar A. Regeneration of autologous, functional, full thickness skin with minimal donor site contribution using autologous homologous skin construct. Plast Surg Case Stud. 2020;6:2513826X1989881.

23. Mundinger GS, Armstrong DG, Smith D, et al. Autologous homologous skin constructs allow safe closure of cutaneous wounds: a retrospective, non-controlled, multi-centered case series. Plast Reconstr Surg Glob Open. 2020;8(5):e2840.

24. Armstrong DG, Orgill DP, Galiano R, et al. Complete wound closure following a single topical application of a novel autolo- gous homologous skin construct: first evaluation in an open- label, single-arm feasibility study in diabetic foot ulcers. Int Wound J. 2020;17(5). https://doi.org/10.1111/iwj.13404.

25. Wong VW, Levi B, Rajadas J, Longaker MT, Gurtner GC. Stem cell niches for skin regeneration. Int J Biomater. 2012;2012:926059.

26. Colazo JM, Evans BC, Farinas AF, Al-Kassis S, Duvall CL, Thayer WP. Applied bioengineering in tissue reconstruction, replacement, and regeneration. Tissue Eng Part B Rev. 2019;25: 259-290.

27. McCartan B, Dinh T. The use of split-thickness skin grafts on diabetic foot ulcerations: a literature review. Plast Surg Int. 2012;2012:715273.

28. Chlup�ac J, Filov�a E, Bac�akov�a L. Blood vessel replacement: 50 years of development and tissue engineering paradigms in vascular surgery. Physiol Res. 2019;58(Suppl 2):S119.

29. Martinson M, Martinson N. A comparative analysis of skin substitutes used in the management of diabetic foot ulcers. J Wound Care. 2016;25(Supp 10):S8-S17.

30. Januszyk M, Rennert RC, Sorkin M, et al. Evaluating the effect of cell culture on gene expression in primary tissue samples using microfluidic-based single cell transcriptional analysis. Microarrays. 2015;4:540-550.

How to cite this article: Armstrong DG, Orgill DP, Galiano R, et al. A multicentre, randomised controlled clinical trial evaluating the effects of a novel autologous, heterogeneous skin construct in the treatment of Wagner one diabetic foot ulcers: Interim analysis. Int Wound J. 2022;19: 64–75. https://doi.org/10.1111/iwj.13598

ARMSTRONG ET AL. 75

1Seidel D, et al. BMJ Open 2020;10:e026345. doi:10.1136/bmjopen-2018-026345

Open access

Negative pressure wound therapy compared with standard moist wound care on diabetic foot ulcers in real- life clinical practice: results of the German DiaFu- RCT

Dörthe Seidel ,1 Martin Storck,2 Holger Lawall,3,4 Gernold Wozniak,5 Peter Mauckner,6 Dirk Hochlenert,7 Walter Wetzel- Roth,8 Klemens Sondern,9 Matthias Hahn,10 Gerhard Rothenaicher,11 Thomas Krönert,12 Karl Zink,13 Edmund Neugebauer14,15

To cite: Seidel D, Storck M, Lawall H, et al. Negative pressure wound therapy compared with standard moist wound care on diabetic foot ulcers in real- life clinical practice: results of the German DiaFu- RCT. BMJ Open 2020;10:e026345. doi:10.1136/ bmjopen-2018-026345

► Prepublication history and additional material for this paper are available online. To view these files, please visit the journal online (http:// dx. doi. org/ 10. 1136/ bmjopen- 2018- 026345).

Received 28 August 2018 Revised 15 January 2020 Accepted 16 January 2020

For numbered affiliations see end of article.

Correspondence to Ms Dörthe Seidel; Doerthe. Seidel@ uni- wh. de

Original research

© Author(s) (or their employer(s)) 2020. Re- use permitted under CC BY- NC. No commercial re- use. See rights and permissions. Published by BMJ.

AbstrACt Objectives The aim of the DiaFu study was to evaluate effectiveness and safety of negative pressure wound therapy (NPWT) in patients with diabetic foot wounds in clinical practice. Design In this controlled clinical superiority trial with blinded outcome assessment patients were randomised in a 1:1 ratio stratified by study site and ulcer severity grade using a web- based- tool. setting This German national study was conducted in 40 surgical and internal medicine inpatient and outpatient facilities specialised in diabetes foot care. Participants 368 patients were randomised and 345 participants were included in the modified intention- to- treat (ITT) population. Adult patients suffering from a diabetic foot ulcer at least for 4 weeks and without contraindication for NPWT were allowed to be included. Interventions NPWT was compared with standard moist wound care (SMWC) according to local standards and guidelines. Primary and secondary outcome measures Primary outcome was wound closure within 16 weeks. Secondary outcomes were wound- related and treatment- related adverse events (AEs), amputations, time until optimal wound bed preparation, wound size and wound tissue composition, pain and quality of life (QoL) within 16 weeks, and recurrences and wound closure within 6 months. results In the ITT population, neither the wound closure rate (difference: n=4 (2.5% (95% CI−4.7% – 9.7%); p=0.53)) nor the time to wound closure (p=0.244) was significantly different between the treatment arms. 191 participants (NPWT 127; SMWC 64) had missing endpoint documentations, premature therapy ends or unauthorised treatment changes. 96 participants in the NPWT arm and 72 participants in the SMWC arm had at least one AE (p=0.007), but only 16 AEs were related to NPWT. Conclusions NPWT was not superior to SMWC in diabetic foot wounds in German clinical practice. Overall, wound closure rate was low. Documentation deficits and deviations from treatment guidelines negatively impacted the outcome wound closure.

trial registration numbers NCT01480362 and DRKS00003347.

bACkgrOunD More than 400 million people worldwide suffer from diabetes,1 2 and about 15% of all these patients will develop a diabetic foot ulcer

strengths and limitations of this study

► The DiaFu study included patients with diabetic foot ulcers both with peripheral neuropathy and periph- eral arterial occlusive disease, which corresponds to the typical mixed patient population in real- life clinical practice. This allows a general statement on effectiveness and safety of negative pressure wound therapy (NPWT) in the typical medical care situation, but including patients with peripheral artery occlu- sive disease and clinical signs of inflammation (sus- pected infection) had a potentially negative effect on the treatment outcome wound closure.

► The study does not provide any information on the effectiveness of NPWT in specific patient groups.

► In this health services research study, hospitals and outpatient facilities were selected by means of a qualification checklist, and clinical investiga- tors were obliged to provide patients with the best clinical practice in compliance with all relevant di- agnostic and treatment guidelines, but there was no active monitoring of the implementation of these guidelines.

► To ensure the best quality of local wound treatment and to achieve optimal baseline conditions, the study sites were trained for both NPWT and standard moist wound care, but treatment application was at the discretion of the clinical investigators.

► A high number of missing endpoint documentations, premature termination of NPWT and unauthorised therapy changes negatively impacted the treatment outcome wound closure and may have led to bias in the results.

2 Seidel D, et al. BMJ Open 2020;10:e026345. doi:10.1136/bmjopen-2018-026345

Open access

(DFU) during their lifetime.3 4 Approximately 50%–70% of all lower limb amputations are due to diabetes.4 DFUs represent complex chronic wounds with a major impact on patients’ morbidity, mortality and quality of life (QoL). Beside an optimal diabetes and infection control, pressure- relieving strategies and restoring pulsatile blood flow, effective local wound care is part of the holistic approach necessary to optimally treat patients with DFUs. Only a few modern moist wound dressings and topical agents have been convincingly shown to achieve higher wound closure rates compared with traditional wet gauze dressings in patients with diabetic foot wounds.5 Also, for other ulcer types, there is uncertainty as to which dress- ings and topical agents are most effective for treatment.6 Negative pressure wound therapy (NPWT) is an inno- vative treatment option and one of the most commonly used and well- established technologies with the aim to promote wound healing.7 The first use of vacuum sealing was described in 1993 by Fleischmann et al,8 and the commercially available product was developed later in the 1990s.9 10 Positive effects of NPWT on wound healing have been suggested in various basic studies.10 11 At the time of planning the DiaFu study, the clinical evidence largely consisted of clinician perception, case reports and series, small cohort studies and weakly powered or low- quality randomised trials that documented broad use of NPWT in various clinical settings and constituted a substantial number of publications but an overall small amount of evidence.12–15 Two randomised controlled trials (RCTs) performed by Armstrong and Lavery16 and Blume et al17 provided a solid basis for planning a study.

In the recent years, a specific review for the use of NPWT in diabetic foot wounds performed by Dumville et al in 2013,18 an assessment in the home setting by Rhee et al in 201419 and a health technology assessment partic- ularly issued for the evaluation of NPWT for managing DFUs20 in 2014, as well as the most recent work of Liu et al in 201721 22 all concluded that although NPWT may have a positive effect, the trials that have been performed have methodological flaws and sufficient, unbiased evidence of whether wounds heal better or worse with NPWT than with conventional treatment is still missing.

In Germany, the issue of evidence for effectiveness and safety of NPWT in acute and chronic wounds was first addressed in 2002 when the German Federal Joint Committee (German: Gemeinsamer Bundesausschuss (G- BA)) needed to decide whether NPWT could be reim- bursed without restrictions in outpatient care.

Finally, in 2007 taking into account all available evidence the G- BA decided that the benefits of the treat- ment method NPWT should be evaluated in a so- called model project. The project was intended to include the conduct of clinical studies for which the G- BA defined basic requirements. This essentially concerned a study hypothesis that supports G- BA’s overall question if NPWT can be reimbursed in German outpatient care without any limitation, the selection of a comparator that represents the current treatment standard in Germany,

and implementation of all measures to ensure a sufficient certainty of the results.

Following the announcement of the G- BA, the German statutory health insurance funds initiated an overall project through a European tender. The DFU was chosen to be the representative for chronic wounds in an RCT comparing NPWT and standard moist wound care (SMWC) in clinical practice.

MethODs Aim of the study The aim of the DiaFu study was to evaluate whether the effectiveness and safety of NPWT is superior to SMWC in German real- life clinical practice.

study design The DiaFu study was a multicentre, randomised controlled clinical superiority trial with blinded assessment of wound closure, wound size and wound tissue qualities using photographs. This German national study was conducted both in hospital departments and outpatient facilities with a special qualification for diabetic foot care. Study sites were selected based on their qualifications and experiences using a prestudy qualification checklist and annual quality reports of the respective institution (if available). Study treatment was allowed to be started both in inpatient and outpatient care and should be continued outpatient whenever possible. More detailed information on the study design can be found in the study protocol publication that is available open access.23

Patient and public involvement Patients were not involved in the design, recruitment or conduct of the study. The results of this study will not be disseminated directly to study participants.

Participants Following a pragmatic approach with the aim to include a patient population best representing real- life clinical practice, inclusion and exclusion criteria were selected based on manufacturers' contraindications and US Food and Drug Administration (FDA) warnings, the necessity to exclude patients in need of protection and who are unable to give their consent, and the intention to avoid general study- related and treatment specific influences on the results.

Adult patients (age >18 years) with at least 4- week- old chronic DFUs corresponding to Wagner 2–4 were screened for study participation by the local investigators. Before inclusion, the study protocol required either a debridement or, if necessary, an amputation of foot parts, or a thorough wound cleansing, depending on the indi- vidual needs of the patients. Thus, chronic diabetic foot wounds after adequate wound pretreatment as well as postsurgical amputation wounds below the upper ankle joint were eligible for inclusion. The initially planned minimum ulcer age of 6 weeks was reduced to 4 weeks

3Seidel D, et al. BMJ Open 2020;10:e026345. doi:10.1136/bmjopen-2018-026345

Open access

during the course of the study. As in clinical practice, the assessment of patients’ suitability for a specific wound therapy with the aim of complete wound closure and (due to randomisation) for both study treatment arms (NPWT and SMWC) was at the discretion of the treating physicians (clinical investigators of the study). Particular attention was to be paid to the diagnosis and therapy of concomitant diseases.

Patients estimated to be at risk of non- compliance with study requirements, with wounds with necrotic tissue present that could not be removed by debridement or amputation, with exposed blood vessels within or directly surrounding the wound not possible to be suffi- ciently covered or with an increased risk of bleeding with haemodynamic consequences (mainly relevant for poste- rior tibial artery dorsalis pedis artery), and outpatients receiving anticoagulation therapy or suffering from a high- grade impaired clotting function with a heightened risk of bleeding with haemodynamic consequences were excluded from the DiaFu study. The use of NPWT devices on the study wound within 6 weeks prior to study start represented an exclusion criterion in order to demon- strate a clear therapeutic effect of each treatment arm.

Written informed consent was obtained from every participant after being informed about all aspects of the trial, and before randomisation and any trial- related procedure. As the statutory health insurance funds provided integrated care contracts for outpatient NPWT, only patients who were members of a participating health insurance fund were allowed to be enrolled.

randomisation and masking Patients were randomly allocated to the treatment arms in a 1:1 ratio using a computer- generated list located on a centralised web- based tool. The randomisation list consisted of permuted blocks of variable length which were randomly arranged. Patients were stratified by study site and by Wagner- Armstrong stage within each site (<Wagner- Armstrong stage 2C and ≥Wagner- Armstrong stage 2C). The randomisation lists were generated with the help of a self- created Java program and integrated into the study database. Each registered investigator received individual access to the randomisation tool via the study website but without knowledge of future treat- ment assignment, which provided adequate allocation concealment. The investigators were responsible for adequately implementing the assigned therapy. Due to the physical differences between the treatment regimens, it was not possible to blind either participant or physician to the treatment assignment. Verification of complete wound closure was performed by independent, blinded assessment of wound photographs. Determination of wound size and percentage wound tissue quality was also performed by central, blinded outcome assessors based on the wound photographs using the Wound Healing Analyzing Tool (W.H.A.T.). The determination of suffi- cient wound bed conditioning and the indication for surgical closure was carried out by the treating physician,

as in clinical practice. The treating physician was not blinded to treatment allocation.

Procedures Basic data were collected for all patients considered for study participation during screening and were updated during the randomisation visit. Patients received an exten- sive examination of overall health status, specific diabetes associated disorders and relevant influence factors on wound healing during screening with an update at the randomisation visit. Neuropathy and vascular diagnos- tics were performed according to the German National Health Care Guidelines for Type 2 Diabetes Foot Compli- cations.24 After anamnesis and general diagnostics (phys- ical examination), this care guideline recommends the following further vascular diagnostics: ankle–arm index (‘Ankle- Brachial- Index’) and additional assessment of the Doppler frequency spectrum (due to the possible falsi- fying of the results by Media sclerosis) and, if necessary, additional hydrostatic toe pressure measurement (pole test) or a transcutaneous oxygen measurement (tcPO

2 ),

duplex sonography to determine the extent and distri- bution pattern of a potential peripheral artery occlusive disease (PAOD) (including the lower leg arteries if neces- sary). In case of inconclusive findings, contrast agent- enhanced MR angiography and intra- arterial digital subtraction angiography were considered. No detailed examination results of the vascular diagnostics but the final diagnosis of PAOD and critical limb ischaemia (CLI) were to be documented in the electronic case report form (eCRF) by the clinical investigators. Infection diagnosis comprised clinical evaluation and laboratory testing. In case of suspected diabetic foot osteomyelitis, a probe to bone test and a stepwise approach to imaging modalities were applied in order to confirm the clinical diagnosis and to determine the best treatment regimen for the study participants.

Before randomisation and start of study treatment, all patients underwent one or more of the following no longer than 6 hours before randomisation: amputa- tion, debridement or thorough wound cleansing. Study therapy was allowed to be started either in- hospital or as outpatient and was intended to be continued in outpa- tient care whenever possible.

In the intervention arm commercially available CE- marked NPWT devices of the manufacturers Kinetic Concepts Incorporated (KCI) and Smith & Nephew (S&N) were used in the discretion of the clinical inves- tigator according to clinical routine and manufacturers’ instructions.23 Intermittent and continuous NPWT was allowed to be used with the negative pressure to be adapted as recommended for the dressing applied (V.A.C.-Granufoam Black or Silver; V.A.C.-White Foam; Renassys–F/P; Renassys–G) and adapted to the wound needs. Recommendations for use are available on the manufacturers’ websites. As part of the European tender for the overall project, the German statutory health insur- ance funds awarded lots for the provision of the medical

4 Seidel D, et al. BMJ Open 2020;10:e026345. doi:10.1136/bmjopen-2018-026345

Open access

products by the respective manufacturers. Germany was divided into four supply areas. During the award proce- dure, S&N received one lot and KCI three lots. Thus, devices and consumables of S&N were used for the north and northern east region of Germany, and for the rest of Germany, the therapy systems of KCI were used. Within the study, NPWT was required to be used for wound bed preparation in order to achieve at least 95% gran- ulation of the wound area. After optimal preparation of the wound, complete closure could be achieved either by secondary intention with dressings or by surgical closure with subsequent removal of the suture.

Control therapy was defined as any SMWC according to local clinical standards and guidelines.25 26 Healthcare providers were obligated to provide patients with best practice. In the control arm, it was permitted to apply any local wound treatment standard used in the respec- tive study site that did not have an experimental status or was NPWT. To ensure the best quality of local wound treatment, the study sites were trained for both the inter- vention arm by the manufacturers and the control arm by the German Society for Wound Healing and Wound Treatment, which provided parts of its curriculum and experienced instructors.

The maximum study treatment time was 16 weeks after randomisation. Study visits needed to be performed at week 1, 3, 5, 12 and 16, and in the event of end of treat- ment, hospital discharge, wound closure and for wound closure confirmation after a minimum of 14 days. Study participants were followed up until 6 months after rando- misation. The initially planned follow- up period of 12 months was reduced to 6 months in the course of the study. The amendment to the study protocol was endorsed by the Ethics Committee and immediately communicated to all participating study sites.

Outcomes The primary outcome was wound closure (100% epithe- lialisation of the wound, no drainage, no suture material and no need for wound dressing or adjuvants) within the maximum study treatment period of 16 weeks. Wound closure could be achieved both by healing by secondary intention and by delayed primary closure. Complete closure of the wound needed to sustain for a minimum of 14 days and to be confirmed by independent blinded observers using wound photographs.

Secondary outcomes were wound closure after 6 months, time until optimal preparation of the wound bed (a minimum of 95% granulation), amputations and resections, wound size and wound tissue composition, pain and QoL within 16 weeks, and recurrence within 6 months. The initial planned secondary endpoint time until wound closure within 6 months was abandoned during the course of the study. It was found that a time- to- event survey was not possible outside the active study treatment period. This was mostly due to the fact that after this 16- week period, weekly study visits were no longer an

obligation, and further patient care was no longer bound to the study site.

Minor and major amputations were assessed sepa- rately, whereas the disarticulation at the midtarsal joint (Chopart’s amputation) was considered still to be minor. Wound size and wound tissue composition (percentage of granulation tissue, fibrin and necrosis) were moni- tored at each study visit. QoL was measured using the questionnaire Euro Quol 5D (EQ5D) at inclusion, end of the maximum treatment time or end of the therapy and at the 6- month follow- up visit. At each study visit participants were asked to provide their assessment of wound- associated pain on a numerical rating scale (0–10). The incidence of serious adverse events (SAEs) within 6 months and the incidence of device- related and treatment- related adverse events (AEs) occurring within 16 weeks or until wound closure confirmation were safety endpoints of this trial.

statistical analysis Sample size calculation was performed using the expected difference between wound closure rates in both treat- ment arms based on information extracted from previ- ously published studies by Armstrong and Lavery16 and Blume et al.17 We assumed a complete wound closure rate of 45% for NPWT and 30% in the SMWC group, resulting in a minimum difference of 15% after a treatment time of 16 weeks. Based on a type 1 error of α=0.05 and a type 2 error of β=0.2 (corresponding to a power of 80%), a total sample size of 162 patients per group was calculated. The computer program of Dupont and Plummer was used for sample size calculation.27

We performed all analyses based on a modified intention- to- treat (ITT) population that includes all randomised participants who have a valid baseline and at least one valid post baseline wound assessment. As a secondary approach a per- protocol (PP) analysis was performed excluding patients with any serious protocol deviations, like temporary changes from SMWC to NPWT, permanent wound treatment changes or without valid documentation until wound closure confirmation or end of maximum treatment time (EOMTT). Safety data are presented on an ‘as treated’ basis. Subgroup analysis is presented for small versus large wound subpopulations. There was no interim analysis.

The superiority hypothesis was tested in parallel for the wound closure rate and the time to wound closure within 16 weeks. Incidence of complete wound closure was analysed using Fishers’ exact test comparing the two treatment arms. Time to complete wound closure was compared between the two treatment arms using a log- rank test. The method of Bonferroni- Holm was used for adjustment of the α-error for parallel confirmatory testing of both primary endpoints. Missing values have been incorporated as censored values.

During study planning, the following concomitant diseases and therapeutic measures with a possible influence on the primary study outcome wound closure (confounders)

5Seidel D, et al. BMJ Open 2020;10:e026345. doi:10.1136/bmjopen-2018-026345

Open access

were identified: presence of neuropathy (sensation loss according to the Perfusion, Extent, Depth, Infection and Sensation (PEDIS) classification system28); presence of diabetic neuropathic osteoarthropathy (anatomical classifi- cation according to Sanders and Frykberg29 and progression stages according to Levin30); Wagner31 grading of the ulcer; presence of peripheral arterial occlusive disease (Ruther- ford classification for chronic limb ischaemia32); chronic venous insufficiency (Widmer I–III33); presence of extreme foot deformities and malpositions of toes, foot or the entire limb; untreated or therapy- refractory inflammation in the wound area; chronic anaemia; heel necrosis; presence of a lymphedema; infection; heightened glycated haemo- globin level; dialysis; application of hyperbaric oxygen or normothermal therapy; application of recombinant or autologous growth factors to the study wound; and applica- tion of skin or dermal substitutes and with living cells that produce growth factors. These covariates thought to influ- ence wound closure were analysed for their effect on the two primary endpoints. Covariates were excluded from the analysis if the number of missing values was too high. First, the relevant covariates were tested by means of a univariate analysis with regard to their effect on wound closure rate and time without consideration of the treatment arms. If there was a significant influence, the frequency of occur- rence in the treatment arms was analysed. Secondary, multivariate analyses were performed for both primary endpoints, taking into account treatment assignment and including all relevant covariates. The multivariate analysis of the primary endpoint wound closure rate was performed with binary logistic regression to describe the influence of the independent covariates (regressors) on the dependent dichotomous variable wound closure. The multivariate analysis of the primary endpoint time to wound closure was performed using a ox regression model.

Safety and secondary endpoints were analysed using conventional univariate testing.

Within an a- priori- planned subgroup analysis, the ITT population was divided into a group of small wounds and a group of large wounds based on the wound surface area documented during the randomisation visit. Wounds smaller than or equal to the total median wound surface (483 mm²) were assigned to the subgroup ‘small wounds’. Patients with wound surface areas larger than the median value were assigned to the subgroup ‘large wounds’. Since no citable scientific definition of a large wound was available at the time of study planning and the clinical experts involved could not make a decision, the median of all wounds was chosen as the criterion for the division into the two subgroups. Confirmatory analysis of primary and secondary endpoints was repeated for the subgroups.

Missing values for the following outcome parameters were replaced using the last observation carried forward (LOCF) method: wound closure rate, wound size and wound tissue quality, recurrence and amputation. The outcome parameters time to wound closure and time until optimal preparation of the wound bed did not require data replacement, since missing values are included in the

analysis as right- censored values. If wound closure was not confirmed to be closed after a minimum of 14 days, the wound wass considered as an unsustained wound closure. All missing QoL values (EQ5D) were replaced with the overall QoL assessment (visual analogue scale), if available. If there was no QoL assessment, there was no replacement. For missing values of the demographic and baseline char- acteristics, which are necessary for the estimation of the regression coefficients, no replacement was performed. IBM SPSS Statistics (V.23) was used for all analyses.

This study is registered with ClinicalTrials. gov and in the German Clinical Trial Registry.

A data monitoring committee was formed to oversee overall study performance and safety.

role of the funding source Through a European tender, the study was initiated by a consortium of 19 statutory German health insurance funds, which provided integrated care contracts for all study partic- ipants and for up to 7000 patients with acute and chronic wounds in Germany, defined basic rules for study design based on the requirements of the German authorities; and provided a critical review of the study protocol and the final report. The study was funded by the manufacturers KCI (Acelity) and S&N. Both companies provided the NPWT devices and associated consumable supplies in the assigned regions of Germany as well as all necessary support and information about the used material. The manufacturers had no role in study design, data collection, data analysis, data interpretation or writing of the report. All authors had full access to all of the data (including statistical reports and tables) in the study and take full responsibility for the accu- racy of the data analysis.

results Between 23 December 2011 and 12 August 2014, 386 patients were enrolled and randomly assigned to receive NPWT (181) or SMWC (187) in the DiaFu study (figure 1) in overall 40 study sites, which recruited minimum 1 patient and maximum 76 patients. Thirteen clinical inves- tigators randomised more than 10 patients. Twenty- three study sites enrolled only between one and four patients. Most of these study sites refused further study participa- tion due lack of time and staff for adequately performing the documentation. In the further course of the trial research nurses were hired by the independent scientific institute overseeing the trial in order to support the docu- mentation in the study sites whenever needed.

Demographics and relevant baseline characteristics of the DFU are presented in table 1 and online supple- mentary table 1. Baseline characteristics of the patients in the NPWT and the SMWC arm are similar in the ITT population without any relevant difference between the treatment arms.

The baseline of the identified factors possibly influ- encing wound closure is shown in table 2.

6 Seidel D, et al. BMJ Open 2020;10:e026345. doi:10.1136/bmjopen-2018-026345

Open access

Figure 1 Trial profile (CONSORT); CONSORT, Consolidated Standards of Reporting Trials; NPWT, Negative Pressure Wound Therapy; SMWC, Standard Moist Wound Care.

Details on revascularisation performed before study start are shown in table 3.

results for the primary outcome wound closure in the Itt population In the ITT population, there was no significant differ- ence between the treatment arms for either wound closure rate (table 4) or time to complete wound closure (p=0.244, log- rank test; figure 2) within 16 weeks. Begin- ning in week 5, the number of study participants with open wounds in the NPWT arm was lower than in the SMWC arm (figure 2). However, after 16 weeks, the difference between the treatment arms was only 2.5% (95% CI −4.7% – 9.7%) (table 4). Wounds treated with

NPWT were approximately at the same risk of remaining open as wounds treated with SMWC (RR 0.97 (95% CI 0.89−1.06)).

Since the cumulative number of patients with open wounds was more than 70% after 16 weeks, we could not calculate medians for the time to wound closure.

results for the secondary outcomes in the Itt population Only one recurrence of the foot wound after complete, sustained and confirmed closure was documented for one study participant in the NPWT arm (table 4). Study participants treated with NPWT were at slightly higher risk for a recurrence than participants treated with SMWC 0.96 (95% CI 0.87−1.04).

7Seidel D, et al. BMJ Open 2020;10:e026345. doi:10.1136/bmjopen-2018-026345

Open access

Table 1 Demographics and baseline characteristics of the ITT population

Demographics of the study population and baseline parameters of the DFU in the ITT population

Total n=345 (100%)

NPWT n=171 (49.6%)

SMWC n=174 (50.4%)

Male Female

267 of 345 (77.4%) 78 of 345 (22.6%)

133 of 171 (77.8%) 38 of 171 (22.2%)

134 of 174 (77.0%) 40 of 174 (23.0%)

Age (years) (n=345), mean (SD) 67.8 (11.9) 67.6 (12.3) 68.1 (11.5)

Height (n=340) (in cm), mean (SD) 174.1 (12.4) 173.4 (14.6) 174.8 (9.9)

Weight (n=335) (in kg), mean (SD) 93.3 (22) 92.7 (21.5) 93.8 (22.6)

Localisation of the ulcer

Regio calcanea Dorsum pedis Planta pedis Metatarsalia Phalanges distales Phalanges mediales Phalanges proximales Hallux Digitus pedis II Digitus pedis III Digitus pedis IV Digitus minimus

39 (11.3%) 20 (5.8%) 56 (16.2%) 147 (42.6%) 64 (18.6%) 28 (8.1%) 40 (11.6%) 42 (12.2%) 22 (6.4%) 14 (4.1%) 20 (5.8%) 25 (7.2%)

17 (9.9%) 13 (7.6%) 30 (17.5%) 73 (42.7%) 31 (18.1%) 14 (8.2%) 21 (12.3%) 24 (14%) 10 (5.8%) 7 (4.1%) 7 (4.1%) 12 (7.0%)

22 (12.6%) 7 (4.0%)

26 (14.9%) 74 (42.5%) 33 (19%) 14 (8.0%) 19 (10.9%) 18 (10.3%) 12 (6.9%) 7 (4.0%) 13 (7.5%) 13 (7.5%)

Type of ulcer

Primary ulcer Recurrence

279 of 342 (80.9%) 63 of 342 (18.3%)

136 of 170 (79.5%) 34 of 170 (19.9%)

143 of 172 (82.2%) 29 of 172 (16.7%)

Duration of ulcer (days)

n Mean (SD) Median (IQR) Min – Max

335 189.7 (360.2)

83 (136) 0–4468

168 217.1 (458.1)

81 (140) 0–4468

167 162.1 (220)

85 (132) 0–1826

Wound surface area at randomisation (mm2)

Mean (SD) Median (IQR) Min- Max

1101 (2543) 491 (1079) 12–40 773

1060 (1536) 550 (1217) 20–13 188

1141 (3247) 471 (1007) 12–40 773

Wound surface area at randomisation for small wounds (mm2)

n Mean (SD) Median (IQR) Min- Max

173 213 (136) 188 (220) 12–484

83 212 (138) 176 (220) 20–484

90 213 (135) 196 (222) 12–471

Wound surface area at randomisation for large wounds (mm2)

n Mean (SD) Median (IQR) Min- Max

172 1995 (3377) 1276 (1482) 491–40773

88 1860 (1805) 1364 (1242) 520–13188

84 2135 (4474) 1242 (1708) 491–40773

Data are number (n) and percentage (%), mean and standard deviation (SD), median and interquartile range (IQR), and minimum – maximum (min – max). ‘n=’ is stating the number of patients with actual available information. Based on the median wound surface area of all included patients, the wounds were divided into an a priori planned subgroup of large (median wound surface area ≤484 mm² and a subgroup of small wounds (median wound surface area >484 mm²). DFU, diabetic foot ulcer; ITT, intention to treat; NPWT, negative pressure wound therapy; SMWC, standard moist wound care.

After 6 months, the number of study participants with closed wounds was higher in the SMWC arm than in the NPWT arm (36 of 174 (20.7 %) vs 24 of 171 (14.0 %)), but the difference was not significant (p=1.00).

The time until optimal preparation of the wound for further treatment to achieve a complete epithelialisation (min 95% granulation tissue) was significantly shorter for patients treated with NPWT (p=0.008) (table 5).

In the ITT population, wound surface area and wound volume were similar at baseline (table 1) and decreased

continuously during the study treatment time of 16 weeks in both treatment arms (online supplementary tables 2 and 3). The values are largely scattered. Measurements derived from the blinded photo analysis using the W.H.A.T. were smaller than the values documented by the clinical investigators.

Wound tissue composition (online supplementary table 4) was similar in both treatment arms at baseline. Granu- lation tissue values increased during the study treatment period of 16 weeks and fibrin values decreased, with

8 Seidel D, et al. BMJ Open 2020;10:e026345. doi:10.1136/bmjopen-2018-026345

Open access

Table 2 Baseline of the identified factors possibly influencing wound closure in the ITT population

Confounders at baseline in the ITT population

Total n=345

NPWT n=171

SMWC n=174

Presence of neuropathy (sensation loss according to the PEDIS classification system)

250 of 334 (72.5%) 125 of 166 (73.1%) 125 of 168 (71.8%)

Presence of a diabetic neuropathic osteoarthropathy 61 (17.7%) 30 (17.5%) 31 (17.8%)

Wagner grading of the ulcer 1: superficial ulcer of skin or subcutaneous tissue 2: ulcers extend into tendon, bone, or capsule 3: deep ulcer with osteomyelitis, or abscess 4: gangrene of toes or forefoot 5: midfoot or hindfoot gangrene

6 (1.7%)

225 (65.2%) 85 (24.6%) 26 (7.5%) 3 (0.9%)

2 (1.2%)

110 (64.3%) 45 (26.3%) 13 (7.6%) 1 (0.6%)

4 (2.3%)

115 (66.1%) 40 (23%) 13 (7.5%) 2 (1.1%)

Peripheral arterial occlusive disease (PAOD) PAOD with critical limb ischaemia*

244 of 345 (70.7%) 26 of 243 (10.7%)

121 of 171 (70.8%) 15 of 121 (12.4%)

123 of 174 (70.7%) 11 of 122 (9.0%)

No chronic venous insufficiency (CVI) CVI Widmer I CVI Widmer II CVI Widmer III

259 of 302 (75.1%) 25 of 302 (7.2%) 12 of 302 (3.5%) 6 of 302 (1.7%)

132 of 150 (77.2%) 11 of 150 (6.4%) 3 of 150 (1.8%) 4 of 150 (2.3%)

127 of 152 (73.0%) 14 of 152 (8.0%) 9 of 152 (5.2%) 2 of 152 (1.1%)

Presence of extreme foot deformities and malpositions of toes, foot or the entire limb

59 of 342 (17.1%) 26 of 170 (15.2%) 33 of 172 (19.0%)

Untreated or therapy- refractory inflammation in the wound area 15 of 343 (4.3%) 7 of 170 (4.1%) 8 of 173 (4.6%)

Presence of a heel necrosis 23 of 342 (6.7%) 10 of 168 (5.8%) 13 of 174 (7.5%)

No lymphoedema Primary lymphoedema Secondary lymphoedema

282 of 340 (81.7%) 12 of 340 (3.5%) 46 of 340 (13.3%)

139 of 167 (81.3%) 5 of 167 (2.9%)

23 of 167 (13.5%)

143 of 173 (82.2%) 7 of 173 (4.0%)

23 of 173 (13.2%)

Clinical signs of inflammation (suspected infection) 159 of 344 (46.1%) 83 of 170 (48.5%) 76 of 174 (43.7%)

Local wound swab as part of the clinical routine 248 of 343 (71.9%) 126 of 170 (73.7%) 122 of 173 (70.1%)

Detection of germs within the local wound swab 205 of 247 (59.4%) 104 of 125 (60.8%) 101 of 122 (58.0%)

Haemoglobin n Mean (SD)

177 of 345

9.5 (3.2)

86 of 171 9.6 (3.1)

91 of 174 9.4 (3.3)

Haemoglobin A1c n Mean (SD)

32 of 345 15.6 (18.3)

13 of 171 16.8 (16.7)

19 of 174 14.7 (19.6)

Requiring dialysis 29 of 343 (8.4%) 15 of 170 (8.8%) 14 of 173 (8.0%)

Application of skin or dermal substitutes and with living cells that produce growth factors

0 of 341 (0%) 0 of 169 (0%) 0 of 172 (0%)

Findings, diagnoses and procedures documented by the investigators are presented. Data are number (N), percentage (%), mean and standard deviation (SD), and minimum – maximum (min – max). *Critical limb ischemia was defined as persistant pain at rest with regular analgesia for a period of two weeks while nerve function is maintained or the occurence of ulceration or gangrene of the foot or toes with a systolic blood pressure of the ankle below 50 mmHg or a systolic toe pressure below 30 mmHg or tcPO

2 <20 mmHg.

ITT, intention to treat; NPWT, negative pressure wound therapy; SMWC, standard moist wound care; tcPO 2 , transcutaneous oxygen measurement.

clinically documented values showing only minor differ- ences between treatment arms. The values for necrotic tissue were very low and did not differ relevantly between the treatment arms. The results of the W.H.A.T. evalua- tion for granulation and fibrin deviate markedly from the values documented by the clinical investigators.

Patients treated with NPWT were approximately at the same risk of undergoing an amputation or resec- tion like patients treated with SMWC (RR 0.99 (95% CI 0.65−1.50)) (table 6).

Overall, pain levels were very low and decreased further during the study treatment time (online supplementary table 5). The values hardly differ between the treatment arms at any observation time point.

At baseline, QoL (EQ5D) was significantly limited in both treatment arms (online supplementary table 6).

EQ5D levels were improved in both study participants reaching end of therapy as well as EOMT. On follow- up after 6 months, all patients still showed increased EQ5D levels in both treatment arms.

safety results The number of study participants with AEs was signifi- cantly higher in the NPWT arm (96 (56.1%)) than in the SMWC arm (72 (41.4%)) (p=0.007) but only 16 (10.2%) of the AEs in the NPWT arm were decided by the inves- tigators to have a definite relation to the medical device (table 7). The number of study participants with at least one AE documented to be serious (SAE) was not signifi- cantly different between the treatment arms (NPWT n=63 (36.8%); SMWC n=58 (33.3%); p=0.50) (table 7). None of the SAEs in the NWPT arm was documented as

9Seidel D, et al. BMJ Open 2020;10:e026345. doi:10.1136/bmjopen-2018-026345

Open access

Table 3 Revascularisations performed in the ITT population before study start

Revascularisation before study start in the ITT population Total

n=345 NPWT n=171

SMWC n=174

Performed revascularisation before study start Percutaneous transluminal angioplasty (PTA) PTA+stent Veins- bypass Polytetrafluoroethylene bypass Thromboendarterectomy and patch plastic

23 of 345 (6.7%) 13 of 23 (57.0%) 1 of 23 (4.0%)

5 of 23 (22.0%) 1 of 23 (4.0%) 2 of 23 (9.0%)

9 of 171 (5.3%) 6 of 9 (67.0%)

0 of 9 (0%) 2 of 9 (22.0%)

0 of 9 (0%) 0 of 9 (0%)

14 of 174 (8.0%) 7 of 9 (50.0%) 1 of 9 (7.0%)

3 of 9 (21.0%) 1 of 9 (7.0%)

2 of 9 (14.0%)

Revascularisation with influence on the wound 22 of 23 (96.0%) 9 of 9 (100%) 13 of 14 (93.9.0%)

Sufficient revascularisation result* Insufficient revascularisation result Revascularisation result not assessable

20 of 23 (88.0%) 2 of 23 (9.0%) 1 of 23 (4.0%)

7 of 9 (78.0%) 1 of 9 (11.0%) 1 of 9 (11.0%)

13 of 14 (93.0%) 1 of 14 (7.0%) 0 of 14 (0%)

Data are n and percentage (%). *Sufficient revascularisation result was defined as successful recanalisation of the tibial artery in which the foot lesion was located or, if it was technically impossible to recanalise the respective artery, achievement of an unhindered inflow into at least one of the tibial vessels. The evaluation of the revascularisation result was in the discretion of the attending physician. ITT, intention to treat; NPWT, negative pressure wound therapy; SMWC, standard moist wound care.

Table 4 Study participants with wound closure (wound closure rate) and the number of participants with recurrences (recurrence rate) in the ITT population

Wound closure and recurrence rate in the ITT population

Total n=345

NPWT n=171

SMWC n=174

Difference n %

(95%CI) p*

Patients with complete, sustained and confirmed wound closure within 16 weeks

n % (95% CI)

46 of 345 13.3

(9.8–17.8)

25 of 171 14.6

(9.5–21.6)

21 of 174 12.1

(7.5–18.4)

4 2.5

(−4.7 – 9.7) 0.53

Patients with recurrence of the diabetic foot wound after complete, sustained and confirmed closure within 6 months

n % (95% CI)

1 of 46 2.2

(0.1–12.1)

1 of 25 4

(0.1–22.3)

0 of 21 0

(0.0–14.3)

1 4

(−3.7 – 11.7) 1.00

Data show the number (N) of participants available for the analysis in total and for both treatment arms. Wound closures within the maximum study treatment time of 16 weeks and recurrences during the follow- up of 6 months are shown with the number (N), the percentage (%) of patients and the 95% Confidence Interval (CI). *F=Fishers’ exact test. ITT, intention to treat; NPWT, negative pressure wound therapy; SMWC, standard moist wound care.

definitely or possibly related to the medical device by the investigators. Nine of 171 (5.3%) study participants in the NPWT arm and 6 of 174 (3.5%) study participants in the SMWC arm died during the study.

secondary analyses and subgroups Of the factors possibly influencing the outcomes iden- tified during study planning, the covariate POAD was found to have significant influence on the endpoint time until wound closure (p=0.026, log rank test). The covariate clinical signs of inflammation (suspected infection) had a significant influence on the wound closure rate (p=0.012, χ2 test) in the univariate anal- ysis of the primary endpoints. However, both covariates

were almost equally represented in both treatment arms. Thus, the comparison of the treatment arms was not influenced by these confounders. Furthermore, the covariate suspected infection was found to be signifi- cantly associated with both wound closure rate (logistic regression; p=0.027) and time until wound closure (Cox regression; p=0.037) in the multivariate confounder analysis. Wound closure was significantly less likely in wounds with suspected infection (OR 0.38).

In the subgroup of large wounds (wound surface area at randomisation shown in table 1), wound closure rate within 16 weeks was significantly higher in the NPWT arm (13 of 88 (14.8 (7.4 to 22.2)%)) than in the SMWC

10 Seidel D, et al. BMJ Open 2020;10:e026345. doi:10.1136/bmjopen-2018-026345

Open access

Figure 2 Time until complete, sustained and verified wound closure in the ITT population. NPWT, negative pressure wound therapy; SMWC, standard moist wound care.

Table 5 Time until optimal preparation of the wound for further treatment (min 95% granulation tissue) in the ITT population

Time until optimal preparation of the wound bed (min 95% granulation tissue) within 16 weeks in the ITT population N

available values

Total n=183

NPWT n=100

SMWC n=83

Mean difference (95% CI)

p*

Mean (SD) 42.7 (39.0) 35.6 (34.6) 51.4 (42.6) 15.8 (4.6−27.0)

0.008 Median (IQR) 31 (64) 22.0 (48.0) 49.0 (53.6)

Min–Max 0–127 0–127 0–115

Data show the number (N) of participants available for the analysis in total and for both treatment arms. Time until optimal preparation of the wound is described with mean and standard deviation (SD); median and interquartile range (IQR); and minimum (min) and maximum (max). *Student’s t- test. ITT, intention to treat; NPWT, negative pressure wound therapy; SMWC, standard moist wound care.

arm (5 of 84 (6.0 (0.9 – 11.0)%)) (difference: n=8 (8.8 (−0.2 to 17.8)%), p=0.08). Study participants with large wounds had a lower risk of not achieving wound closure within 16 weeks when treated with NPWT (RR 0.91 (95% CI 0.82−1.0)) and achieved wound closure signifi- cantly faster in the NPWT arm than in the SMWC arm (p=0.027) (figure 3). The only recurrence occurred in the subgroup of large wounds. Both major amputations were performed in study participants with large wounds treated with NPWT.

In the subgroup of small wounds (wound surface area at randomisation shown in table 1), the time to reach 95% granulation tissue was significantly shorter for the patients treated with NPWT than for those treated with SMWC (p=0.005), but wound closure rate and time until wound closure within 16 weeks were not significantly different between the treatment arms (figure 4). Further

details of the subgroup analyses are presented in the online supplementary tables 7 and 8.

results for the primary and secondary outcomes in the PP population Demographics, relevant baseline characteristics and the results of the revascularisation before study start of the PP population are presented in online supplementary table 9. In the PP population, 14 of 44 study participants (31.8% (95% CI 18.1%−45.6%)) treated with NPWT and 19 of 110 participants (17.3% (95% CI 10.2%−24.3%)) treated with SMWC achieved complete, sustained and verified wound closure within 16 weeks, but the difference was not significant (5 (14.5% (95% CI −1.0% − 30.0%); p=0.053). Wounds treated with NPWT had a lower risk of remaining open after 16 weeks (RR 0.82 (95% CI 0.66−1.03)) than wounds treated with SMWC. Time to wound closure in the NPWT arm was significantly shorter than in the SMWC arm (p=0.004) (figure 5). After 6 months, wound closure rate in the SMWC arm (30 of 110 (27.3% (95% CI 18.9%−35.6%))) was higher than in the NPWT arm (11 of 44 (25.0% (95% CI 12.2−37.8))), but the difference was not significant (n=19 (2.3% (95% CI −13.0 − 17.6)); p=0.84). As in the ITT population, optimal wound bed preparation was achieved significantly faster in study participantss receiving NPWT (p<0.001). No recurrences occurred after complete, sustained and confirmed wound closure in the PP population. Neither the number of participants with amputations or resections nor the number of amputations or resections performed differed significantly between the treatment arms. No major amputations were performed in the PP population. Further details on the results for the PP population are presented in the online supplementary tables 10–16.

treatment compliance Twenty- nine (17.0%) participants in the NPWT arm had a temporary therapy change to SMWC (mean duration 20.5±21.6 days). In the SMWC arm, 17 (9.8%) partici- pants had a temporary therapy change to NPWT (mean duration 28.9±21.6 days). For only 2 of the 29 NPWT participants (6.9%) with a temporary therapy change to SMWC, the wound closure was achieved within 16 weeks,

11Seidel D, et al. BMJ Open 2020;10:e026345. doi:10.1136/bmjopen-2018-026345

Open access

Table 6 Study participants with amputations/resections and the number of amputations/resections performed in the ITT population

Amputations and resections in the ITT population

Total n=345

NPWT n=171

SMWC n=174

Difference (95% CI)

p

Study participants with amputation or resection

n (%) (95% CI)

71 (20.6%) (16.3−24.8)

35 (20.5%) (14.4 to 26.5)

36 (20.7%) (14.7−26.7)

1 (0.2%) (−19.0 − 18.6)

1.00 (F)

Total number of amputations and resections 102 45 57 12 0.89 (U)

Number of amputations and resections per study participant, n (%)

One event Two events Three events Four events Five events

49 (14.2%) 16 (4.6%) 4 (1.2%) 1 (0.3%) 1 (0.3%)

25 (14.6% 10 (5.8%)

0 (0% 0 (0%) 0 (0%)

24 (13.8%) 6 (3.4%) 4 (2.3%) 1 (0.6%) 1 (0.6%)

1 (0.8%) 4 (2.4%) 4 (2.3%) 1 (0.6%) 1 (0.6%)

Study participants with minor amputation 69 (20.0%) 33 (19.3%) 36 (20.7%) 3 (1.4%) 0.79 (F)

Study participants with major amputation 2 (0.6%) 2 (1.2%) 0 (0%) 2 (1.2%) 0.25 (F)

Data show the number (N) of participants, the percentage with the 95% CI, or the number of events accompanied with the respective percentage values in total and for both treatment arms. F, Fishers' exact test; ITT, intention to treat; NPWT, negative pressure wound therapy ; SMWC, standard moist wound care; U, Mann- Whitney U test.

whereas 16.2% (23 von 142) of the wounds of the NPWT participants without therapy change were completely closed.

A total of 57.3% (98 of 171) of the participants randomised to NPWT completed treatment before achieving a granulation surface of the wound of at least 95%. Fewer participants with this premature end of NPWT (4.7%, n=8) achieved a complete wound closure than participantss with no premature end of therapy (9.9, n=17). Mean NPWT duration until premature end of therapy was 28.5 days (SD 24.1), while a mean granulation area of 59.6% (SD 30. 5) was achieved. For 131 partici- pants (76. 6 %) in the NPWT arm less than the required three dressing changes per week were documented. Nine- teen participants (14. 5 %) with this protocol violation achieved a complete wound closure. Six (15.4%) of the 39 NPWT participants who received at least three therapy changes per week achieved a complete wound closure.

Documentation quality In the NPWT arm, 52 study participants and in the SMWC arm 43 participants were excluded from the PP popula- tion due to missing documentation until the EOMT or at wound closure confirmation (figure 1). In the eCRF, wound closure was documented for 96 patients (NPWT 56 of 171; SMWC 40 of 174), but only 46 participants (NPWT 25; SMWC 21) met all criteria for a complete, verified and sustained wound closure. For the wound closure visit, seven wound photographs (NPWT 7; SMWC 0) and for the wound closure confirmation visit four photographs (NPWT 3; SMWC 1) were missing. In addition, two of the

existing wound photographs for wound closure (NPWT 0; SMWC 2) and two photographs for wound closure confir- mation (NPWT 1; SMWC 3) were not assessable by the blinded observers due to serious quality issues. Further- more, 23 (NPWT 15; SMWC 8) existing and assessable wound photographs were not able to confirm wound closure and 3 (NPWT 1; SMWC 2) photographs were not able to confirm wound closure after 14 days.

DIsCussIOn The DiaFu study did not demonstrate significant superi- ority in wound closure rate or time to complete wound closure for neither NPWT nor SMWC. Wound closure rates were higher in the NPWT arm but did not signifi- cantly differ from those in the SMWC arm. Time to wound healing in the NPWT arm was lower than in the SMWC arm, while the difference between the treatment arms becomes statistically significant only in the PP popu- lation. Thus, with this study, we were not able to confirm our hypothesis that wound closure can be achieved more often and faster with NPWT than with SMWC when used in German real- life clinical practice. Previous RCTs, which were the basis for sample size calculation, showed a higher rate and a significant superiority in healing when using NPWT on amputation and chronic wounds,16 17 but the populations of these studies were different. Other than the Armstrong study, the DiaFu study did not exclude patients with venous insufficiency and included more than twice as many patients. The studies of Armstrong

12 Seidel D, et al. BMJ Open 2020;10:e026345. doi:10.1136/bmjopen-2018-026345

Open access

Table 7 Study participants with adverse events (AEs) and serious adverse events (SAEs) and the number of AEs and SAEs in the ITT population

AEs and SAEs in the ITT population

Total n=345

NPWT n=171

SMWC n=174

Difference (95% CI)

p

Study participants with at least one AE

n (%) (95% CI)

168 (48.7%) (43.4−54.0)

96 (56.1%) (48.7−63.6)

72 (41.4%) (34.1−48.7)

24 (14.7%) (4.3−25.1)

p=0.007 (F)

Study participants with one AE (n) Study participants with two or more AEs (n)

103 65

54 42

49 23

5 19

Total number of AEs (n) 269 167 102 65

AEs with relationship to the medical device

n available

Yes, n (%) Possible, n (%) No, n (%) Not assessable, n (%)

257 16 (6.2%) 13 (5.1%)

211 (82.1%) 17 (6.6%)

157 16 (10. 2%) 11 (7.0%)

117 (74.5%) 13 (8.3%)

100 0 (0%)

2 (2.0%)* 94 (94.0%)

4 (4.0%)

57 16 (10.2%)

9 (5%) 23 (19.5%)

9 (4.3%)

AEs with relationship to SMWC

n available

Yes, n (%) Possible, n (%) No, n (%) Not assessable, n (%)

185 2 (1.1%) 5 (2.7%)

163 (88.1%) 15 (8.1%)

110 0 (0%)

5 (4.5%) 96 (87.3%)

9 (8.2%)

75 2 (2.7%) 0 (0%)

67 (89.3%) 6 (8.0%)

35 2 (2.7%) 5 (4.5%) 29 (2.0%) 3 (0.2%)

AEs with relationship to the treatment procedure

n available

Yes, n (%) Possible, n (%) No, n (%) Not assessable, n (%)

244 10 (4.1%) 17 (7.0%)

191 (78.3%) 26 (10.7%)

148 6 (4.1%)

15 (10.1%) 111 (75.0%) 16 (10.8%)

96 4 (4.2%) 2 (2.1%)

80 (83.3%) 10 (10.4%)

52 2 (0.1%) 13 (8%)

31 (8.3%) 6 (0.4%)

Study participants with at least one SAE

n (%) (95% CI)

121 (35.1%) (30.0−40.1)

63 (36.8%) (29.6−44.1)

58 (33.3%) (26.3−40.3)

5 (3.5%) (−6.6 − 13.6) p=0.50 (F)

Study participants with one SAE (n) Study participants with two or more SAEs (n)

90 31

45 18

45 13

0 5

Total number of SAEs (n) 163 87 76 11

SAEs with relationship to the medical device

n available

Yes, n (%) Possible, n (%) No, n (%) Not assessable, n (%)

161 0 (0%) 0 (0%)

154 (95.7%) 7 (4.3%)

85 0 (0%) 0 (0%)

79 (92.9%) 6 (7.1%)

76 0 (0%) 0 (0%)

75 (98.7%) 1 (1.3%)

9 0 (0%) 0 (0%)

4 (5.8%) 5 (5.8%)

SAEs with relationship to SMWC

n available

Yes, n (%) Possible No, n (%) Not assessable, n (%)

121 1 (0.8%) 1 (0.8%)

113 (93.4%) 6 (5.0%)

64 0 (0%)

1 (1.6%) 57 (89.1%)

6 (9.4%)

57 1 (1.8%) 0 (0%)

56 (98.2%) 0 (0%)

7 1 (1.8%) 1 (1.6%) 1 (9.1%) 6 (9.4%)

SAEs with relationship to the treatment procedure

n available

Yes, n (%) Possible No, n (%) Not assessable, n (%)

156 4 (2.6%) 2 (1.3%)

140 (89.7%) 10 (6.4%)

84 0 (0%)

2 (2.4%) 74 (88.1%)

8 (9.5%)

72 4 (5.6%) 0 (0%)

66 (91.7%) 2 (2.8%)

12 4 (5.6%) 2 (2.4%) 8 (10.6%) 6 (6.7%)

Data show the number (n) and the percentage (%) in total and for both treatment arms. *No treatment change to NPWT has been documented. F=Fisher’s exact test (alpha=0.05). ITT, intention to treat ; NPWT, negative pressure wound therapy; SMWC, standard moist wound care.

13Seidel D, et al. BMJ Open 2020;10:e026345. doi:10.1136/bmjopen-2018-026345

Open access

Figure 3 Time until complete, sustained and verified wound closure for the subgroup of large wounds. NPWT, negative pressure wound therapy; SMWC, standard moist wound care.

Figure 4 Time until complete, sustained and verified wound closure for the subgroup of small wounds. NPWT, negative pressure wound therapy; SMWC, standard moist wound care.

Figure 5 Time until complete, sustained and verified wound closure in the PP population; NPWT, negative pressure wound therapy; SMWC, standard moist wound care.

and Blume excluded patients with Wagner stage 4; active Charcot; uncontrolled hyperglycemia and therapy with glucocorticoids, immunosuppressants or chemotherapy; and required proof of adequate perfusion. The DiaFu

study did not exclude patients with impaired perfusion but required adequate therapy of the circulatory disorder according to clinical practice guidelines. In the DiaFu study, we were able to show that the presence of PAOD at randomisation had a significant influence on the time to wound closure but not on the overall wound closure rate within the maximum study treatment time. The number of patients with critical limb ischaemia at baseline was low and differed only slightly between the treatment arms. As in clinical practice, in the DiaFu study, adequate treat- ment of concomitant diseases was mandatory. Invasive therapy of PAOD could be performed before initiation of wound therapy as well as during the study treatment period, if the wound needed pretreatment as a basis for the revascularisation procedure or if new or recurrent critical ischaemia occured.

The presence of clinical signs of inflammation (suspected infection) at randomisation had a significant effect on both, time to wound closure and wound closure rate within 16 weeks. Both covariates were equally repre- sented in the treatment arms, thus the differences in time until wound closure and wound closure rate were not affected by these confounders.

However, the probably most serious factors negatively influencing treatment and outcome are documentation deficiencies and deviations from treatment guidelines. Temporary therapy changes and premature therapy cessa- tion negatively impacted the patient relevant treatment outcome wound closure in study participants treated with NPWT. Missing study visits resulting in low numbers of complete endpoint documentations strongly affected the

14 Seidel D, et al. BMJ Open 2020;10:e026345. doi:10.1136/bmjopen-2018-026345

Open access

proof of the outcome wound closure in both, the NPWT arm and the SMWC arm.

Optimal preparation of the wound bed (95% granula- tion tissue) was achieved significantly earlier when using NPWT in the ITT and the PP population, but the overall rate of wound closures was low. Wound bed preparation and granulation tissue formation are important prereq- uisites for wound healing but are not a proof of treat- ment effectiveness and cannot serve as a basis for benefit assessment.

Although significantly more AEs were documented in the NPWT arm, only a small number of these events were related to the medical device according to the investi- gator’ s assessment. Mortality rates were very low in both treatment arms, and there was no significant difference between the treatment arms regarding amputations and resections performed during the study. Only two major amputations have been performed in patients with large wounds treated with NPWT. None of the treatments resulted in an additional impairment of the patients’ QoL during study treatment time or follow- up. Time until complete wound closure was significantly shorter with NPWT than with SMWC in the subgroup of large wounds, which indicates that NPWT has the potential to be a valu- able treatment option for this kind of wounds.

In the DiaFu study, methods against bias were imple- mented whenever possible in order to avoid bias that have been described by several systematic reviews,18–21 but blinding of study participants as well as attending physi- cians and nurses was not possible due to the nature of NPWT.

Not addressing and analysing all factors influencing the overall treatment outcome like targeted pressure relief, continuous infection control and adequate treat- ment of the underlying disease during the study treat- ment and observation period may be seen as a limitation of this healthcare research study. Study sites have been selected based on a self- disclosure by means of a qualifica- tion checklist and cross checks using quality reports. This ensured that all prerequisites were met for guideline- compliant patient care. Nevertheless, even in the applica- tion of NPWT, there were deviations from the standards.

In order to support the decision- making process of the German G- BA on general reimbursement of NPWT in German outpatient care, the real- life clinical practice DiaFu study included patients with chronic DFUs of neuropathic and angiopathic origin regardless of whether a simple wound cleansing, tissue debridement or even amputa- tion was necessary prior to application of wound therapy targeted to achieve complete wound closure. The study was performed without excluding concomitant diseases nega- tively impacting wound healing; with therapy application in the discretion of the attending physician; and with evalua- tion of patient relevant outcome. Thus, results can easy be generalised and applied in clinical practice settings. Anyway, shortcomings in data quality negatively impacted the study results, and statements about specific patient groups were not possible. A high number of study participants needed

to be excluded from the PP population (NPWT arm: 127 of 171 (74%), SMWC arm 64 of 174 (37%)). For most of these participants, documentation was lacking until the end of the maximum treatment period (total=88, NPWT=49, SMWC=39) (figure 1). In the primary analysis based on the ITT population, it was assumed that these patients did not achieve wound closure within 16 weeks’ study treat- ment and observation time (using the LOCF method, the open wound status was ‘carried forward’ until the end of the maximum treatment period. This may have led to a false negative bias in the outcome wound closure in the ITT population. Due to the high loss of patients and the difference in the number of participants excluded from the treatment arms, the validity of the PP analysis is very limited.

COnClusIOns NPWT was not superior to SMWC when evaluated in German real- life clinical practice. Missing compliance with therapy guidelines and poor documentation quality led to restrictions in achieving the patient- relevant endpoint complete wound closure and prevents a clear proof of effectiveness. The question if NPWT is supe- rior to SMWC for treating diabetic foot wounds remains unanswered due to the limitations of the DiaFu study. Although the study protocol required adequate moni- toring and therapy of the concomitant diseases, the presence of POAD and infection at randomisation had a significant influence on the outcome wound closure. Despite all limitations, NPWT showed a significant supe- riority in optimal wound bed preparation. This indicates that NPWT works according to its intended use and has a potential to be an effective treatment option. The results of the PP population suggest that without the negative impact of premature treatment cessation, temporary changes of the randomised therapy and partly incomplete documentation NPWT may be more effective for treating diabetic foot wounds than SMWC. In Germany, NPWT should be evaluated again after implementation of a suffi- cient, well- considered and widely accepted concept for quality control. In a future healthcare research study, the treatment outcome before and after the implementation of these quality measures should be evaluated, for which the results of this trial may serve as a basis.

Author affiliations 1Institut für Forschung in der Operativen Medizin (IFOM), Universität Witten/ Herdecke, Köln, Germany 2Klinik für Gefäß- und Thoraxchirurgie, Städtisches Klinikum Karlsruhe gGmbH, Karlsruhe, Germany 3Praxis für Herzkreislauferkrankungen, Ettlingen, Germany 4Innere Medizin, Max- Grundig Klinik, Bühlerhöhe, Germany 5Gefäßchirurgische Klinik, Knappschaftskrankenhaus Bottrop GmbH, Bottrop, Germany 6Innere Medizin, St. Remigius Krankenhaus Opladen, Leverkusen, Germany 7Gemeinschaftspraxis Schlotmann- Hochlenert- Zavaleta- Haberstock, Köln, Germany 8Chirurgische Praxis Wetzel- Roth, Buchloe, Germany 9Klinik für Innere Medizin/Diabetologie, Marien Hospital Dortmund- Hombruch, Dortmund, Germany 10Allgemein- und Viszeralchirurgie, Helfenstein Klinik, Geisslingen, Germany

15Seidel D, et al. BMJ Open 2020;10:e026345. doi:10.1136/bmjopen-2018-026345

Open access

11Chirurgische Praxis Rothenaicher, München, Germany 12Klinik für Gefäßchirurgie, Thüringen- Kliniken "Georgius Agricola" GmbH, Saalfeld, Germany 13Diabetes Klinik, Diabetes Zentrum Mergentheim, Bad Mergentheim, Germany 14Department fur Humanmedizin, Universität Witten/Herdecke, Witten, Germany 15Medizinische Hochschule Brandenburg -Theodor Fontane, Neuruppin, Germany

Acknowledgements The authors thank all investigators, nurses, patients and partners for supporting the study. At least one patient was included in the following facilities: HSK – Dr. Horst Schmidt Kliniken GmbH Klinik für Gefäßchirurgie Ludwig- Erhard- Straße 100 65199 Wiesbaden; Asklepios Westklinikum Hamburg Zentrum für Gefäßmedizin Suurheid 20 22559 Hamburg; Knappschaftskrankenhaus Bottrop Gefäßchirurgische Klinik Osterfelderstraße 157 46242 Bottrop; Städtisches Klinikum Karlsruhe Klinik für Gefäß- und Thoraxchirurgie Moltkestraße 90 76133 Karlsruhe; Gemeinschaftspraxis Schlotmann- Hochlenert- Zavaleta- Haberstock Merheimer Straße 217 50733 Köln; Klinikum Döbeln Abt. für Gefäßchirurgie Sörmitzer Straße 10 04720 Döbeln; Klinikum Bielefeld Mitte Klinik für Allgemeine Innere Medizin Teutoburger Straße 50 33604 Bielefeld; Klinikum Frankfurt/ Oder Klinik für Gefäßchirurgie Müllroser Chaussee 7 15236 Frankfurt/Oder; Weißeritztal- Kliniken GmbH Medizinische Klinik III Bürgerstraße 7 01705 Freital; Krankenhaus Porz am Rhein Klinik für Gefäßchirurgie Urbacher Weg 19 51149 Köln; St. Remigius Krankenhaus Opladen Innere Medizin An St. Remigius 26 51379 Leverkusen; Marien Hospital Dortmund- Hombruch Klinik für Innere Medizin/Diabetologie Gablonzstraße 9 44225 Dortmund; Zentrum für Chirurgie Klinik für Gefäß- und Endovascularchirurgie Theodor- Stern- Kai 7, Haus 23C/EG 60590 Frankfurt am Main; Facharzt für Chirurgie Thorax- Kardiovaskularchirurgie Hindenburgstraße 1 86807 Buchloe; Helfenstein Klinik Geisslingen Allgemein- und Viszeralchirurgie Eybstraße 16 73312 Geislingen/Steige; Paracelsus- Klinik am Silbersee Wundzentrum Hannover Oertzeweg 24 30851 Langenhagen; Klinikum Darmstadt Chirurgische Klinik III Grafenstraße 9 64283 Darmstadt; Ortenau Klinikum Offenburg- Ebertplatz Klinik für Allgemein-, Viszeral- und Gefäßchirurgie Ebertplatz 12 77654 Offenburg; Thüringen- Kliniken "Georgius Agricola" GmbH Klinik für Gefäßchirurgie Rainweg 68 07318 Saalfeld; Klinikum Dorothea Christiane Erxleben GmbH Klinik für Allgemein-, Viszeral- und Gefäßchirurgie Ditfurter Weg 24 06484 Quedlinburg; Franziskus- Krankenhaus Berlin Abt. für Innere Medizin Budapester Straße 15-19 10787 Berlin; Hegau- Bodensee Klinikum Radolfzell (HBK) Klinik für Innere Medizin Hausherrenstraße 12 78315 Radolfzell; Diabetologische Schwerpunktpraxis Dr. med. Hansjörg Mühlen & Partner Ruhrorter Straße 195 47119 Duisburg; Kliniken Maria Hilf Mönchengladbach Klinik für Gefäßchirurgie und Angiologie Sandradstraße 43 41061 Mönchengladbach; Städtisches Klinikum München/Bogenhausen Klinik für Endokrinologie, Diabetologie und Angiologie Englschalkingerstraße 77 81925 München; Gerhard Rothenaicher Facharzt für Chirurgie Cosimastraße 2 81927 München; Bürgerhospital Frankfurt am Main Interdisziplinäres Zentrum Diabetischer Fuß (DDG) Nibelungenallee 37- 41 60318 Frankfurt am Main; Gemeinschaftspraxis für Chirurgie und Gefäßmedizin Drs. Alter/ Pourhassan/Heim Klosterstraße 12 46145 Oberhausen; Ev. KH Königin Elisabeth Herzberge gGmbH Abt. für Kardiologie, Angiologie und Diabetologie Herzbergstraße 79 10365 Berlin; Städtisches Klinikum Neunkirchen gGmbH Abt. für Gefäßchirurgie & Phlebologie Brunnenstraße 20 66538 Neunkirchen; Westküstenklinikum Heide Klinik für Viszeral- und Gefäßchirurgie Esmarchstraße 50 25746 Heide/Holstein; Chir. Praxisgemeinschaft am Bayenthalgürtel Praxis Dr. med. Gerald Engels Bayenthalgürtel 45 50968 Köln; Malteser Krankenhaus – St. Franziskus- Hospital Medizinische Klinik I, Abt. für Diabetologie Waldstraße 17 24939 Flensburg; St. Marienkrankenhaus Siegen gGmbH Klinik für Gastroenterologie Kampenstraße 51 57072 Siegen; Krankenhaus Bietigheim Klinik für Innere Medizin, Kardiologie, Endokrinologie, Diabetologie und Internistische Intensivmedizin Riedstraße 12 74321 Bietigheim- Bissingen; Asklepios Kliniken Harburg Eißendorfer Pferdeweg 52 21075 Hamburg; Diabetologikum Ludwigshafen Diabetes- Schwerpunktpraxis Ludwigsplatz 9 67059 Ludwigshafen; Mariannen- Hospital Werl Abt. für Chirurgie Unnaer Straße 15 59457 Werl; Diabetes Klinik GmbH & Co KG Theodor- Klotzbücher- Straße 12 97980 Bad Mergentheim; Institut für Diabetesforschung Münster GmbH Hohenzollernring 70 48145 Münster. The study was initiated by a consortium of 19 statutory German health insurance funds represented by the AOK federal association (AOK- Bundesverband – AOK- BV), the association of alternative health insurance funds (Verband der Ersatzkrankenkassen – vdek) and the minors (Knappschaft). In order to guarantee outpatient care for all study participants without any restrictions, the contracting health insurance companies provided integrated care contracts for outpatient negative pressure wound therapy. A project advisory board was implemented to coordinate all processes and project partners. The board comprised two representatives each from the statutory health insurance funds, the management company and the sponsor as well as one

representative each from the participating medical device manufacturers (KCI and smith & nephew). Representing the contracting authority (statutory German health insurance funds) Dr. Gerhard Schillinger (AOK- BV) and Ute Leonhard (vdek) acted as contact persons for all aspects of the project. The management company “Gesundheitsforen Leipzig” has been entirely responsible for the logistics of the study. Central tasks of the management company included the recruitment of study sites and patients, the development of the IT infrastructure including the documentation, communication and invoicing software as well as the processing of all payments. The manufacturers Kinetic Concepts Incorporated (KCI) (Acelity) and smith & nephew provided the NPWT devices as well as support and training for the investigators and financed the study.The Private University of Witten/Herdecke gGmbH acted as the Sponsor of the trial and the Institute for Research in Operative Medicine with its former director Prof. E.A.M. Neugebauer, the current interim head Prof. Rolf Lefering and the head of the division for clinical research Dörthe Seidel was responsible for the scientific conception, the evaluation as well as the reporting and publication of the study. Prof. Dr. Rolf Lefering was responsible for the statistical planning and analysis. PD Dr. Peter Krüger was responsible for the data management of the study. Special thanks are going to Stefan Bauer, who supported the data management as well as the statistical analysis and reporting. We would like to thank Sophie Thorn, who checked the article as a native English speaker with regard to spelling and grammar.

Contributors DS was the principal coordinating investigator. She conceived the study, reviewed the scientific literature and was responsible for study design, data analysis, data interpretation, writing and reviewing of the report. She is the lead author and takes overall responsibility for this report. She affirms that the manuscript is an honest, accurate and transparent account of the study being reported; that no important aspects of the study have been omitted; and that any discrepancies from the study as originally planned (and, if relevant, registered) have been explained. MS and HL were study investigators and contributed to study design, data collection and interpretation and reviewed the report. GW, PM, WW- R and DH were study investigators and contributed to data collection and data interpretation and reviewed the report. KS, MH, GR, TK and KZ were study investigators and contributed to data collection and reviewed the report. EN contributed to study design and data interpretation and reviewed the report. All authors approved the final version of the report.

Funding Through a European tender, the study was initiated by a consortium of 19 statutory German health insurance funds, which provided integrated care contracts for all study participants and for up to 7000 patients with acute and chronic wounds in Germany; defined basic rules for study design based on the requirements of the German authorities; and provided a critical review of the study protocol and the final report. The study was funded by the manufacturers KCI and S&N. Both companies provided the NPWT devices and associated consumable supplies in the assigned regions of Germany as well as all necessary support and information about the used material. All authors had full access to all of the data (including statistical reports and tables) in the study and take full responsibility for the accuracy of the data analysis.

Disclaimer The manufacturers had no role in study design, data collection, data analysis, data interpretation or writing of the report.

Competing interests The German statutory health insurance companies commissioned the Witten/Herdecke University (UW/H) to plan, conduct, analyse and publish the study. DS is an employee of the UW/H. The study has been financed by the manufacturers KCI (Acelity) and S&N. DS received a consulting fee for the presentation of the study during an event organised by the manufacturer Hartmann. During study planning and conduct, EN was an employee of the UW/H. He was the director of the Institut für Forschung in der Operativen Medizin. The clinical investigators MS, HL, GW, PM, DH, WW- R, KS, MH, GR, TK and KZ received a case fee of 1000€ for each patient included in the DiaFu study in order to compensate for the additional organisational and especially the documentation effort during trial conduct. Furthermore, all investigators received compensation for travelling to the investigator meetings. The institutions of the investigators used integrated care contracts for NPWT during study conduct in order to provide best practice for the study participants during outpatient care. GW and WW- R are members of the scientific advisory board of the manufacturer KCI (now Acelity).

Patient consent for publication Not required.

ethics approval Ethical approval of the main ethical committee (EC): Ethical Committee of the University of Witten- Herdecke, has been fully granted without any conditions. Due to performing the trial according to § 23b MPG (German Medical Device Act), participating study sites in Germany only received a consultation for the main clinical investigator according to professional law by the respective EC. All

16 Seidel D, et al. BMJ Open 2020;10:e026345. doi:10.1136/bmjopen-2018-026345

Open access

investigators have been fully approved by the respective ECs. An evaluation of the study's content by ECs of participating study sites in Germany was not applicable. All study participants gave written informed consent prior to randomisation and any trial related procedure.

Provenance and peer review Not commissioned; externally peer reviewed.

Data availability statement Data are available on reasonable request. The datasets analysed for the results presented in this article are available from the corresponding author. Datasets are available in German language.

Open access This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY- NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non- commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non- commercial. See: http:// creativecommons. org/ licenses/ by- nc/ 4. 0/.

OrCID iD Dörthe Seidel http:// orcid. org/ 0000- 0002- 2287- 5217

reFerenCes 1 World Health Organization. Global report on diabetes, 2016.

Available: http://www. who. int/ diabetes/ global- report/ en/ 2 International Diabetes Federation. IDF diabetes atlas, 2015.

Available: www. diabetesatlas. org 3 Yazdanpanah L, Nasiri M, Adarvishi S. Literature review on the

management of diabetic foot ulcer. World J Diabetes 2015;6:37–53. 4 Leone S, Pascale R, Vitale M, et al. [Epidemiology of diabetic foot].

Infez Med 2012;20 Suppl 1:8–13. 5 Wu L, Norman G, Dumville JC, et al. Dressings for treating foot

ulcers in people with diabetes: an overview of systematic reviews. Cochrane Database Syst Rev 2015;55.

6 Norman G, Westby MJ, Rithalia AD, et al. Dressings and topical agents for treating venous leg ulcers. Cochrane Database Syst Rev 2018;6:CD012583.

7 Wu SC, Marston W, Armstrong DG. Wound care: the role of advanced wound- healing technologies. J Am Podiatr Med Assoc 2010;100:385–94.

8 Fleischmann W, Strecker W, Bombelli M, et al. [Vacuum sealing as treatment of soft tissue damage in open fractures]. Unfallchirurg 1993;96:488–92.

9 Argenta LC, Morykwas MJ. Vacuum- assisted closure: a new method for wound control and treatment: clinical experience. Ann Plast Surg 1997;38:563–77.

10 Morykwas MJ, Argenta LC, Shelton- Brown EI, et al. Vacuum- assisted closure: a new method for wound control and treatment: animal studies and basic Foundation. Ann Plast Surg 1997;38:553–62.

11 Morykwas MJ, Faler BJ, Pearce DJ, et al. Effects of varying levels of subatmospheric pressure on the rate of granulation tissue formation in experimental wounds in swine. Ann Plast Surg 2001;47:547–51.

12 Gregor S, Maegele M, Sauerland S, et al. Negative pressure wound therapy: a vacuum of evidence? Arch Surg 2008;143:189–96.

13 Peinemann F, Sauerland S. Negative- Pressure wound therapy: systematic review of randomized controlled trials. Dtsch Arztebl Int 2011;108:381–9.

14 Ubbink DT, Westerbos SJ, Evans D, et al. Topical negative pressure for treating chronic wounds. Cochrane Database Syst Rev 2008:CD001898.

15 Vikatmaa P, Juutilainen V, Kuukasjärvi P, et al. Negative pressure wound therapy: a systematic review on effectiveness and safety. Eur J Vasc Endovasc Surg 2008;36:438–48.

16 Armstrong DG, Lavery LA. Negative pressure wound therapy after partial diabetic foot amputation: a multicentre, randomised controlled trial. The Lancet 2005;366:1704–10.

17 Blume PA, Walters J, Payne W, et al. Comparison of negative pressure wound therapy using vacuum- assisted closure with advanced moist wound therapy in the treatment of diabetic foot ulcers: a multicenter randomized controlled trial. Diabetes Care 2008;31:631–6.

18 Dumville JC, Hinchliffe RJ, Cullum N, et al. Negative pressure wound therapy for treating foot wounds in people with diabetes mellitus. Cochrane Database Syst Rev 2013;10:CD010318.

19 Rhee SMet al. Negative pressure wound therapy technologies for chronic wound care in the home setting. Rockville (MD): Johns Hopkins University Evidence- based Practice Center, 2014.

20 CADTH Rapid Response Reports. Negative pressure wound therapy for managing diabetic foot ulcers: a review of the clinical effectiveness, cost- effectiveness, and guidelines. Ottawa (ON): Canadian Agency for Drugs and Technologies in Health, 2014.

21 Liu S, He C- Z, Cai Y- T, et al. Evaluation of negative- pressure wound therapy for patients with diabetic foot ulcers: systematic review and meta- analysis. Ther Clin Risk Manag 2017;13:533–44.

22 Liu Z, Dumville JC, Hinchliffe RJ, et al. Negative pressure wound therapy for treating foot wounds in people with diabetes mellitus. Cochrane Database Syst Rev 2018;10:CD010318.

23 Seidel D, Mathes T, Lefering R, et al. Negative pressure wound therapy versus standard wound care in chronic diabetic foot wounds: study protocol for a randomized controlled trial. Trials 2014;15:334.

24 Ärztliches Zentrum für Qualität in Der Medizin (Gemeinsame Einrichtung von Bundesärztekammer und Kassenärztlicher Bundesvereinigung) Im Auftrag von BÄK, K., AWMF, Nationale VersorgungsLeitlinie (NVL) Typ-2- Diabetes Präventions- und Behandlungsstrategien für Fußkomplikationen. 2010, Bundesärztekammer (BÄK) Arbeitsgemeinschaft Der Deutschen Ärztekammern, 2010. Available: http://www. baek. de; Kassenärztliche Bundesvereinigung (KBV); http://www. kbv. de; Arbeitsgemeinschaft der Wissenschaftlichen Medizinischen Fachgesellschaften (AWMF) http://www. awmf- online. de

25 Bauer H. Typ-2- Diabetes: Präventions- und Behandlungsstrategien für Fußkomplikationen Nationale Versorgungs Leitlinien; 2010.

26 Rüttermann M, Maier- Hasselmann A, Nink- Grebe B, et al. Local treatment of chronic wounds: in patients with peripheral vascular disease, chronic venous insufficiency, and diabetes. Dtsch Arztebl Int 2013;110:25–31.

27 Dupont WD, Plummer WD. Power and sample size calculations. A review and computer program. Control Clin Trials 1990;11:116–28.

28 Schaper NC. Diabetic foot ulcer classification system for research purposes: a progress report on criteria for including patients in research studies. Diabetes Metab Res Rev 2004;20 Suppl 1:S90–5.

29 Sanders LJ, Frykberg RG. Diabetic neuropathic osteoarthropathy: the Charcot foot. In: The high risk foot in diabetes mellitus. New York: Churchill Livingstone, 1991: 297–338.

30 Levin ME. Preventing amputation in the patient with diabetes. Diabetes Care 1995;18:1383–94.

31 Wagner FW. The diabetic foot. Orthopedics 1987;10:163–72. 32 Rutherford RB, Baker JD, Ernst C, et al. Recommended standards

for reports dealing with lower extremity ischemia: revised version. J Vasc Surg 1997;26:517–38.

33 Widmer LK, Plechl SC, Leu HJ, et al. [Venous diseases in 1800 employees. Basel Studies II]. Schweiz Med Wochenschr 1967;97:107–10.