Ischaemic or borderline-perfusion wounds

• All three ischaemic wounds healed completely, with no further wound-edge necrosis and no reported pain during therapy (p.2–3).
• SPP values were low (Case 1: 25; Case 2: 30; Case 3: 20 mmHg) yet outcomes were good (p.2, Table 1).
• Duration of LPN was 21–28 days (median 24 days), similar to NPWT durations reported for non-ischaemic wounds (p.3).
• Two patients (Cases 1 and 3) had a split-thickness skin graft after good granulation; Case 2 healed by wound contraction after LPN was stopped (p.2).
• Authors caution the 50 mmHg setting may not suit all ischaemic wounds — patients with even lower SPP may still develop necrosis, so close observation is required (p.4).

CAT: OCEBM 4 · Three patients in a single arm cannot support the comparative claim that 50 mmHg is safer than 125 mmHg, and that is exactly the claim this paper is used for. GRADE starts Low for observational data and is downgraded twice for extreme imprecision and for very serious risk of bias: no control arm, no blinding, outcomes judged clinically by the treating surgeons, and a hypothesis that was formed before the cases and confirmed by them. Very low is the floor and there is nothing to raise it. What is actually here, with every denominator: 3 of 3 patients healed completely; 0 of 3 developed further wound-edge necrosis; 0 of 3 reported pain during therapy. Skin perfusion pressures were 25, 30 and 20 mmHg, all below the 40 mmHg cut-off the unit uses (sensitivity 72 per cent, specificity 88 per cent). Ulcer sizes were 50 x 25, 50 x 22 and 60 x 18 mm. Therapy ran 28, 23 and 21 days. Two of 3 were closed with a split-thickness skin graft and 1 of 3 healed by contraction 14 days after therapy stopped. Arithmetic that does not reconcile: the paper states 'the median duration of LPN was 24 days'. The median of 21, 23 and 28 is 23. Twenty-four is the mean. The paper mislabels its only summary statistic, and 24 days is the number this library and the downstream protocol have carried. The rationale is borrowed, not demonstrated: the physiological premise — that 125 mmHg raises tissue pressure enough to occlude capillaries at an ischaemic wound edge — is taken from the Kairinos work this library holds, and no tissue pressure, perfusion or oximetry was measured in any of these three patients at any point. The authors concede the limit themselves: 50 mmHg 'may not suit all ischaemic wounds', patients with lower SPP may still necrose, and close observation is required.

Figures: Figures checked

• In all three configurations tissue pressure increased proportionately to applied suction (p < 0.0005); no negative pressures were ever recorded (p.1, p.4).
• Circumferential dressings produced significantly larger pressure increments than noncircumferential (p < 0.0005) or cavity (p < 0.0005) groups; the latter two did not differ (p = 0.269) (p.1, p.5).
• Over 48 h the raised pressure fell significantly (circumferential p = 0.04; noncircumferential p < 0.0005), but only 3 of 10 wounds dropped below their pre-dressing baseline; soonest was 24 h (p.1, p.7).
• In 2 of 10 wounds (both circumferential) tissue pressure rose further over time, and 3 of 10 stayed ≥10 mmHg above baseline at 48 h (p.7).
• Because normal capillary perfusion pressure is only 10–35 mmHg, the authors argue a sustained ≥10 mmHg rise could occlude capillaries in tissue with already-compromised perfusion (p.7).

CAT: OCEBM 5 (mechanistic reasoning) / GRADE not applicable — the outcome is interstitial pressure in mmHg, a surrogate, and no patient-important effect estimate exists for GRADE to act on. The JBI Critical Appraisal Checklist for Quasi-Experimental (non-randomised experimental) Studies is applied in its place. · GRADE does not apply and assigning it a certainty level would be a category error. There is no comparator, no clinical event, no risk, no ratio and no confidence interval anywhere in this paper. Its outcome is a pressure transducer reading. GRADE grades certainty in an estimate of an effect on an outcome that matters to a patient; this study produces none, so the correct answer is a reasoned refusal, and the previous entry of 'OCEBM 3 / GRADE Low, start low cohort, +1 large effect' was wrong on every element. The finding, with its denominators: tissue pressure rose proportionately to applied suction in all three dressing configurations, and a negative (hypobaric) tissue pressure was never recorded in any of the 15 tests. Circumferential dressings raised pressure more than non-circumferential (p less than 0.0005) and more than cavity dressings (p less than 0.0005); the latter two did not differ (p = 0.269). Over 48 hours at −125 mmHg the raised pressure declined significantly (circumferential beta = −0.21, p = 0.04; non-circumferential p less than 0.0005) but only 3 of 10 wounds fell below their own baseline, the soonest at 24 hours, 2 of 10 rose further, and 3 of 10 remained at least 10 mmHg above baseline at 48 hours. Because normal capillary perfusion pressure is 10 to 35 mmHg, a sustained 10 mmHg rise is mechanistically sufficient to occlude capillaries in tissue that is already marginal. Two qualifications the paper's own text supplies and the abstract does not. First, the abstract prints p less than 0.0005 for all three groups while the results print p less than 0.005 for the non-circumferential and cavity groups. Second, one of the five cavity wounds, a septic spinal dehiscence, showed no significant rise at all (p = 0.51), so the claim holds at group level in 3 of 3 configurations but at wound level in 4 of 5 cavity wounds.

Figures: Figures checked

• Circumferential NPWT reduced hand perfusion by a mean 40 ± 11.5% at −400 mmHg and 17 ± 8.9% at −125 mmHg (both p < 0.0005); no hand showed increased perfusion (p.1, p.4).
• The perfusion drop was significantly greater at −400 than −125 mmHg (p < 0.015) (p.1, p.4).
• Noncircumferential NPWT lowered TcPO2 in all 12 legs: 7.35 ± 7.4 mmHg at −400 mmHg and 5.10 ± 7.4 mmHg at −125 mmHg (both p < 0.0005); the −400 vs −125 difference was not significant (p = 0.07) (p.1, p.4).
• Tissue oxygen after suction-off remained significantly below the pre-suction baseline (p = 0.02 and 0.03) (p.5).
• The authors propose laser Doppler misreads compression: as vessels narrow, blood velocity spikes (equation of continuity) so Doppler "perfusion units" rise even while true perfusion falls, reconciling the conflicting literature (p.6, p.7, p.8).

CAT: Same dataset as chapters 6 and 7 of the Kairinos 2011 UCT thesis — see the rationale. OCEBM 5 (mechanistic experiment on healthy volunteers). RoB 2 applied to the randomised -400 against -125 mmHg comparison, giving SOME CONCERNS to High; the suction-versus-control contrast is appraised on its own terms as a paired physiological experiment. GRADE is not applicable — the outcomes are radioactivity counts and transcutaneous oxygen tension in intact healthy skin, which are surrogates for nothing that has been validated against a wound outcome. · First, the double-counting warning, because it matters more than the GRADE. This is the same dataset as chapters 6 and 7 of Kairinos 2011. Same volunteers, same hands, same legs, same numbers. A motivation or a review that cites the thesis alongside this paper as though they were two sources is counting ten volunteers twice. The two are not interchangeable even as a single source: the thesis prints p less than 0.005 where this paper's abstract prints p less than 0.0005, and the thesis gives the -125 mmHg oximetry reduction as 5.0 mmHg (SD 4.67) where the abstract gives 5.10 plus or minus 7.4. GRADE is refused for a stated reason: the outcome is a surrogate, and not a validated one. Gamma-camera counts under a foam sandwich on an intact healthy hand, and transcutaneous oxygen tension on the shin of a seated volunteer, are physiological quantities measured over minutes. No healing, granulation, infection, graft take or amputation outcome exists here, and no study has established that a 17% fall in hand counts predicts anything about a wound. Rating certainty in an effect on a surrogate that has never been linked to a patient outcome would give the number a standing it has not earned. The findings themselves are internally consistent and point one way. Circumferential NPWT reduced perfusion by 40 plus or minus 11.5 percent at -400 mmHg and 17 plus or minus 8.9 percent at -125 mmHg, with no hand showing an increase and the blinded nuclear medicine reporter calling 19 of 20 scans asymmetric. Transcutaneous oxygen fell in all 12 LEGS, and stayed below the pre-suction mean after the suction was switched off. Consistency across two independent modalities, in every single subject, is the strength of this paper. Its weakness is that the stated precision is not supported, and the details are in the appraisal field below: three different p-values are printed for one comparison, and two of the three group means do not reproduce from the paper's own tables.

Figures: Figures checked

• §4 Wound care, Recommendation 6 (Grade 2B) — the NPWT recommendation, verbatim (the guideline restarts its numbering in each of its five sections, so an unqualified "Recommendation 6" is ambiguous — §1 Recommendation 6 is a different statement, recommending against prophylactic arterial revascularisation. Always cite the section. Verified against the source 2026-07-29): "We suggest the use of negative pressure wound therapy (NPWT) for chronic diabetic foot wounds that do not demonstrate expected healing progression with standard or advanced wound dressings after 4 to 8 weeks of therapy."
• The supporting evidence cited is the multicentre RCT of 342 patients (Blume — Blume 2007), in which NPWT was as safe as and more efficacious than advanced moist wound therapy: faster healing to closure, significantly fewer secondary amputations, and significantly fewer home-care therapy days.
• Further cited evidence: reduced time to complete healing and reduced duration and frequency of hospital admission; decreased healing time and improved quality of life; increased rate of granulation tissue appearance; reduced length of hospitalisation with reduced amputation rates and functional residual extremity; and reduced time to granulation with clearing of bacterial infection.
• Retrospective reimbursement-claims analysis showed reduced amputation numbers in NPWT groups against traditional therapies regardless of wound depth, and decreased resource utilisation through fewer nursing visits.
• The guideline explicitly qualifies its own recommendation on cost: "Consideration of high cost of NPWT products and access to trained personnel for application of NPWT dressings should be weighed in choosing this treatment modality."
• §4 Wound care, Recommendation 5 (Grade 1B) — the stronger and, for a vascular unit, the more important one: re-evaluation of vascular status, infection control and off-loading is recommended to ensure optimisation before adjunctive wound therapy is started. Note the asymmetry: the prerequisite is a Grade 1 recommendation while NPWT itself is only Grade 2.
• Adjunctive therapies are graded comparatively. NPWT and PDGF (becaplermin) both sit at Grade 2B; the guideline states evidence was "most supportive for hyperbaric oxygen therapy". Corrected 2026-07-29: this page previously called HBOT a pharmacological adjunct. The commissioned review covered three adjunctive categories — hyperbaric oxygen, arterial pump devices and pharmacologic agents — and HBOT is not in the pharmacologic group.

AGREE II: Recommended — Scope 78% · Stakeholders 61% · Rigour 71% · Clarity 100% · Applicability 54% · Independence 25%

Figures: Figures checked