Barnaby Reeves

Barnaby Reeves, Professor of Health Services Research, Bristol Trials Centre, University of Bristol

Health services researcher and trialist, carrying out Health Technology Assessment primarily in hospital settings. Surgical interventions are a special interest, hence also the validity and applicability of non-randomized studies.

Current collaborations include: large multi-centre trials of gastric band vs bypass to treat morbid obesity, adalimumab to treat autoimmune uveitis, aspirin to prevent cardiovascular events in patients admitted with pneumonia; research on the feasibility of surgical closure of severe pressure ulcers, surgical strategies to prevent the development of parastomal hernia.

Co-convenor of Cochrane Collaboration Non-Randomized Studies Methods Group, advising on the methods for systematic reviews that include non-randomized studies.

Presentation at The Society of Tissue Viability 2023 Conference

Effectiveness of surgical reconstruction to treat full-skin thickness pressure ulcers: evidence and feasibility of research


After attending this session, persons will be able to:

  • Frequency of surgical reconstruction of full-skin thickness pressure ulcers in England
  • Evidence of the effectiveness of surgical closure compared to conservative treatment
  • State-of-the-art method for emulating a randomised trial with data from usual care


Background – Case series describe surgical reconstruction (SR) to close severe (full-skin thickness) pressure ulcers (SPUs). The percentage of SPUs treated by SR is unclear and the effectiveness of SR compared to non-surgical treatment (except for debridement; NST) has not been evaluated.

Objectives – Describe the frequency of SR and compare health outcomes in patients who have had SR with patients having NST.

Design – Retrospective cohort study using Hospital Episode Statistics (HES) from 01/04/2011 to 30/09/2018, emulating a randomised trial of SR vs NST.

Participants – Patients >18 years admitted to an English NHS hospital (index admission) with a diagnosis of SPU (ICD-10 codes L89.2, L89.3, L89 or L89.X).

Intervention and comparator – SR and NST were defined by presence/absence of a reconstructive flap (OPCS-4 code: S17-S20, S22, S24-S27).

Outcomes – Time to subsequent HES admission with SPU diagnosis (primary); time to any subsequent HES admission; duration of index admission hospital stay; all-cause mortality at 6 months; SR after discharge from index admission.

Results – The cohort comprised index admissions for 291,268 patients with a SPU. A minimum of 325, and a maximum of 1,018, patients had SR during the index admission, i.e. 43 to 136 per year. Ten NHS trusts accounted for half the SRs. Patients who had SR had a mean age between 50 and 60 years, about 70% were male and 50-75% had a cause of impaired mobility assigned on admission, i.e. injury or neurodegenerative disease. The emulation compared 325 in a SR group with 1388 in a NST group. SR extend time to a subsequent admission with a SPU diagnosis and time to any further admission (neither effect statistically significant). However, the NST group may not have been comparable.

Conclusions – In England, SR is carried out too infrequently for future research on its effectiveness currently to be feasible. The comparative analyses gave some relevant insights but had limited validity. If evaluation of SR is a priority, stakeholders should need to increase access to SR.

The SIPS study is funded by the NIHR HTA Programme (Project number NIHR127850). The views expressed are those of the authors and not necessarily those of the NHS, the NIHR HTA or the Department of Health.