Preoperative intravenous iron for anaemia in elective major open abdominal surgery: the PREVENTT RCT

Article history

The research reported in this issue of the journal was funded by the HTA programme as project number 10/104/06. The contractual start date was in September 2012. The draft report began editorial review in November 2019 and was accepted for publication in July 2020. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The HTA editors and publisher have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the draft document. However, they do not accept liability for damages or losses arising from material published in this report.

Copyright statement

Copyright © 2021 Richards et al. This work was produced by Richards et al. under the terms of a commissioning contract issued by the Secretary of State for Health and Social Care. This is an Open Access publication distributed under the terms of the Creative Commons Attribution CC BY 4.0 licence, which permits unrestricted use, distribution, reproduction and adaption in any medium and for any purpose provided that it is properly attributed. See: https://creativecommons.org/licenses/by/4.0/. For attribution the title, original author(s), the publication source – NIHR Journals Library, and the DOI of the publication must be cited.

2021 The authors

Scientific background

Anaemia is common in patients undergoing major surgery. Observational and database studies have suggested that both anaemia and blood transfusion are associated with increased patient risk and worse outcomes following surgery. 1

Intravenous iron can produce a rapid rise in haemoglobin (Hb). Small studies and case series have shown that, if intravenous iron is applied in the preoperative setting, patients could have their anaemia corrected by the time of surgery. This may reduce the need for blood transfusion and improve patient outcomes.

In 2016, National Institute for Health and Care Excellence (NICE) guidelines recommended that patients undergoing surgery with an expected blood loss of ≥ 500 ml should be screened for anaemia at least 2 weeks prior to surgery, and recommended treatment with oral or intravenous iron therapy. 2 These guidelines were endorsed and supported by the 2018 Frankfurt consensus on patient blood management,3 and recently the NHS England Commissioning for Quality and Innovation scheme4 for 2020–21 set targets for 60% of patients to be screened and treated for iron-deficiency anaemia (IDA) before major surgery. Nevertheless, these initiatives were based on very low quality evidence and no large randomised controlled trial (RCT) has shown superiority of intervention with preoperative iron therapy.

It is not known whether or not intravenous iron given to patients before major surgery can correct anaemia and reduce the associated risk to the patient at operation and in the postoperative period.

Why is the study needed now?

The use of intravenous iron has been studied in a variety of clinical settings. Meta-analyses show proven efficacy in renal failure,32 cardiac failure,33 inflammatory bowel disease34 and women’s health. 35,36 Overall, end points from clinical trials have focused on change in Hb levels or correction of anaemia rather than the effect of intravenous iron on blood transfusion or patient outcomes. 37

Anaemia is common in all areas of surgical practice, with database and observational data linking preoperative anaemia to an associated increased patient risk. Risk has primarily focused on the need for blood transfusion, with secondary end points including associated risk of adverse events (AEs) in most organ systems [acute kidney injury (AKI), acute neurocognitive decline, cardiac risk, wound infection, etc.] as well as increased length of ICU or hospital stay, and increased overall mortality.

Major abdominal surgery includes upper gastrointestinal, hepatobiliary and pancreatic, colorectal, urological, vascular and gynaecological surgery. These are often lengthy and complex operations. The patient population is often older and patients frequently have comorbidities. Although laparoscopic surgery has increased in the last decade, with an associated reduction in blood transfusion use,38–43 large open operations remain a significant burden to patients and also to health-care utilisation. Interventions to reduce surgical risk would benefit patient outcomes and also reduce health-care costs.

The cause of anaemia in the preoperative setting is also not fully understood. It is frequently related to blood loss from the underlying condition, such as bowel cancer, or secondary to inflammation, leading to a state of FID. Therefore, it is highly likely that intravenous iron would be efficacious in this setting to increase Hb levels and to be an effective treatment for preoperative anaemia.

If intravenous iron can correct anaemia in a timely manner before major surgery, this may reasonably be predicted to improve patients’ energy levels and reduce fatigue, consequently helping patients to recover from surgery. In turn, this may be associated with a reduction in complications, faster recovery and reduced length of hospital stay.

Study design

This was a randomised, double-blind, parallel-group, placebo-controlled, multicentre, Phase III study with 1 : 1 randomisation to either intravenous iron or placebo. All patients were treated according to protocol and received double-blind intravenous iron therapy or placebo. All patients were followed for 6 months from date of operation. As the active and placebo infusion fluids cannot be matched in appearance, unblinded study personnel not otherwise involved in the study or in patient management were responsible for investigational drug administration. PREVENTT is reported in accordance with Consolidated Standards of Reporting Trials (URL: www.consort-statement.org/).

Aim

This study aimed to assess whether or not giving a single dose of intravenous iron to patients with anaemia prior to major abdominal surgery reduced the need for transfusion or the likelihood of death in the perioperative period. In addition, the effect of the intervention on postoperative complications, hospital stay, re-admission to hospital and quality-of-life outcomes was assessed.

Important changes to methods

The initial working protocol included an additional hospital visit for patients for assessment before their operation. This additional visit was not acceptable to patients, with failure to recruit patients in the first 4 months of the trial. In a protocol amendment, the timing of the preoperative assessments was changed such that these took place at the patient’s admission for their operation, avoiding the need for any additional visits to the hospital for the purpose of the trial. The amended trial protocol was relaunched at the start of 2014.

After this, there were only four further protocol amendments during the course of the trial, including reducing the timeline to surgery from 14 days to 10 days, revising the description for major surgery, adjusting the diagnosis of anaemia in line with the World Health Organization’s definitions (120 g/l for women and 130 g/l for men), and removing the need for preoperative liver function testing and updates as a result of revised Medicines and Healthcare products Regulatory Agency guidelines (see Appendix 1).

Participants

Setting

Forty-six hospitals in the UK participated: University College London Hospitals NHS Foundation Trust; Royal Free London NHS Foundation Trust; Royal Cornwall Hospitals NHS Trust; Royal Devon and Exeter NHS Foundation Trust; Royal Marsden NHS Foundation Trust; The Hillingdon Hospitals NHS Foundation Trust; Swansea Bay University Health Board – Morriston Hospital; York Teaching Hospital NHS Foundation Trust; Dorset County Hospital NHS Foundation Trust; Maidstone and Tunbridge Wells NHS Trust; Newcastle Hospitals – Freeman Hospital; Southmead Hospital Bristol; The Royal London Hospital – Barts Health NHS Trust; Sheffield Teaching Hospital – Northern General Hospital; University Hospital Southampton NHS Foundation Trust; University Hospital of North Staffordshire NHS Trust; University Hospitals Bristol NHS Foundation Trust; Royal Sussex County Hospital; St James’s University Hospital; Guy’s and St Thomas’ NHS Foundation Trust; Central Manchester University Hospitals NHS Foundation Trust; Blackpool Teaching Hospitals; West Suffolk NHS Foundation Trust; Royal Surrey County Hospital; Wythenshawe Hospital; James Cook University Hospital; Broomfield Hospital – Mid Essex Hospital Trust; Royal Liverpool and Broadgreen University Hospitals NHS Trust; Salford Royal NHS Foundation Trust; The County Hospital, Wye Valley NHS Trust; Northampton General Hospital; Imperial College Healthcare NHS Trust; John Radcliffe Hospital – Oxford University Hospitals; Queen’s Medical Centre, Nottingham University Hospitals NHS Trust; Aintree University Hospital NHS Foundation Trust; Queen Elizabeth Hospital, NHS Gateshead; Royal Infirmary of Edinburgh; The Pennine Acute Hospitals; Norfolk and Norwich University Hospital; Peterborough and Stamford Hospitals; Russells Hall Hospital, Dudley; King’s College Hospital; Liverpool Women’s NHS Foundation Trust; Basildon University Hospital; Countess of Chester Hospital; Hinchinbrook Hospital – North West Anglia Foundation Trust; and Cheltenham and Gloucester Hospital, Gloucestershire Hospitals NHS Foundation Trust.

Randomisation

Randomisation was by a secure web-based service through the Clinical Trials Unit (CTU) at the London School of Hygiene & Tropical Medicine (LSHTM), provided by Sealed Envelope (London, UK). Randomisation used minimisation, taking into account baseline Hb level (< 100 g/l or ≥ 100 g/l), age (< 70 or ≥ 70 years), centre and operation type (major/major+/complex). Patients were randomised to receive either active treatment (intravenous iron as ferric carboxymaltose 1000 mg) or placebo. The web-based database allocated the participant a unique trial identification number and their identification details were entered onto the trial patient identification log < 100 g/l or ≥ 100 g/l kept in the investigator site file. Once this number was assigned to a patient, it was not reused, even in the case of participant withdrawal from the study.

Randomisation was performed at the trial sites by the unblinded member of staff who was delegated this responsibility by the principal investigator (PI) only, as evidenced by documentation in the delegation log. Each unblinded member of staff was trained in the use of the web-based randomisation service at the site initiation visit, and was then provided with their own individual password and personal identification number (PIN) to access the service.

The blinded staff did not have access to the randomisation system, and, therefore, remained blinded to the treatment allocated.

Blinding and unblinding

Interventions

Patients who conformed to all eligibility criteria and provided written informed consent were randomised to receive either intravenous iron or placebo 10–42 days before the planned date of their surgery, as described in Randomisation.

Administration of the IMP was carried out in a hospital in line with local protocols. The study medication was administered to patients by the unblinded member of staff. An intravenous line was sited for drug administration Following this, it was advised that the skin along the donor vein be wiped using an iodine swab to help maintain the blinding. The patient was shielded from seeing preparation of the study drug, drug administration, disconnection and removal of the intravenous line, as described above. Patients were monitored for AEs or signs of hypersensitivity during and for at least 30 minutes following the administration of the treatment.

In the intravenous iron group, 1000 mg of ferric carboxymaltose (Ferinject) was administered as an intravenous infusion (100 ml n/saline) over a minimum of 15 minutes using a black infusion kit.

In the placebo group, normal saline was administered as an intravenous infusion (100 ml n/saline) over a minimum of 15 minutes using a black infusion kit.

Adverse events occurring in connection with the administration of study medication were recorded. In the event of a patient having an allergic reaction or signs of intolerance during study drug administration, the investigator managed this in accordance with local protocol and submitted a completed serious adverse event (SAE) form to LSHTM within 24 hours of the event.

Data management

Monitoring and site visits

The conduct of the study was supervised by specifically trained monitors from the LSHTM CTU. A trial-specific monitoring plan was established following a risk assessment and full details are available in the PREVENTT Monitoring SOP. The trial was monitored according to this agreed plan.

Baseline assessment

Baseline assessments were planned to coincide with the routine hospital schedule such as outpatient, endoscopy or pre-assessment clinic attendance. The ‘baseline assessments’, including laboratory tests for the purposes of the PREVENTT trial, were the same as those for routine clinical care in pre assessment before major surgery, so as to follow routine clinical practice and surgical/anaesthetic pathways where possible. For convenience, patients could undergo ‘baseline’ assessments, randomisation and trial drug administrations at the same attendance.

Assessments included the following:

• checking conformance with inclusion/exclusion criteria, including laboratory tests taken within the prior 4 weeks [full blood count (FBC), urea and electrolytes (UE), LFT where clinically indicated as part of pre assessment, iron studies, estimated glomerular filtration rate (eGFR), CRP, thyroid function tests, vitamin B₁₂ and folate].

• documentation of past medical history

• vital signs (blood pressure, pulse rate, body weight, height and temperature)

• 12-lead ECG (electrocardiography)

• additional blood samples for central laboratory analysis (FBC, iron studies, TSAT and total iron binding capacity)

• HRQoL questionnaires

• documentation of health resources used (HRU); patient diary issued.

Follow-up

Patients were initially followed up during their stay in hospital, up to discharge.

Safety assessments

Patient and public involvement

The Trial Steering Committee (TSC) comprised two lay members; one had ulcerative colitis and had undergone two major open laparotomies with a total colectomy and ileorectal anastomosis, and the other had several gynaecological procedures cumulating in a total abdominal hysterectomy. The lay members attended TSC meetings as full, voting members and contributed to the protocol; specifically, they prioritised the secondary end-point selection and also reviewed the patient information sheet, patient poster and diaries. The lay members were also very helpful in advising the Project Management Group (PMG) about how to approach patients for recruitment into the trial.

Definition of the end of the trial

The end of the study is defined as the date at which the last patient completed their last study visit.

Co-primary end points

The trial had two co-primary end points:

1. risk of the composite end point of blood transfusion or death from randomisation until 30 days following the index operation

2. blood transfusion rate (including repeat transfusions) from randomisation until 30 days following the index operation.

A blood transfusion event was defined as transfusion of any volume of 1 unit of blood or blood product. When more than 1 unit of packed red cells or any other blood product was intended to be received contiguously, this was regarded as a single blood transfusion. The blood transfusion rate was defined as the number of blood transfusions divided by the total patient time at risk.

The PREVENTT TSC trial sites were selected after compliance with NHS Blood and Transplant guidelines. 3,44 Patient blood management and transfusion practice was assessed in two independent audits during recruitment.

Key secondary end points

• Total number of units of blood or blood components transfused between randomisation and 30 days postoperatively (also at 6 months postoperatively as other secondary end point), excluding large blood transfusions. A large blood transfusion was defined as a single transfusion consisting of ≥ 4 units of blood or blood products.

• Days alive and out of hospital (DAOH) from the date of the planned surgery until 30 days post index operation.

• Postoperative complications (from index operation to date of discharge) using CD classification.

• HRQoL outcome:

  • Multidimensional Fatigue Inventory (MFI) questionnaire total score at 8 weeks postoperatively (also at the 10-day assessment and 6 months postoperatively as other secondary end points).

  • EuroQol-5 Dimensions, five-level version (EQ-5D-5L) questionnaire total score at 8 weeks postoperatively (also at the 10-day assessment and 6 months postoperatively as other secondary end points).

  • Single question outcome measure (SQOM) at 8 weeks postoperatively (also at the 10-day assessment and 6 months postoperatively as other secondary end points).

Other secondary end points

• Change in Hb levels from randomisation to (1) day of index operation (prior to surgery), (2) 8 weeks and (3) 6 months post index operation.

• Correction of anaemia (Hb level of ≤ 120 g/l for women/Hb level of ≤ 130 g/l for men) at day of index operation.

• Risk of blood transfusion or death, excluding large blood transfusions, from randomisation to (1) 30 days and (2) 6 months post index operation. A large blood transfusion is defined as a single transfusion consisting of ≥ 4 units of blood or blood products.

• POMS outcomes at 3, 5, 7 and 14 days following the index operation. Outcomes consist of the presence of morbidity defined by the domains of POMS (e.g. gastrointestinal or cardiovascular).

• ICU and total hospital LOS from date of index operation until discharge.

• Re-admission to hospital at (1) 8 weeks and (2) 6 months post operation.

• All-cause mortality from randomisation to (1) 8 weeks and (2) 6 months post index operation.

• Health economics outcomes:

  • change in health-care resource utilisation from baseline to 6 months post operation

  • change in calculated NHS and societal costs from baseline to 6 months post operation

  • change in quality-adjusted life-years (QALYs) from baseline to 6 months post operation

  • cost-effectiveness, measured in terms of the incremental cost per percentage reduction in patients receiving blood transfusions and incremental cost per QALY gained, using data from baseline to 6 months post operation.

• Safety and related efficacy outcomes:

  • Large blood transfusion from randomisation to 30 days post index operation

  • Any reaction or side effect from trial therapy

  • Any reaction or side effect from blood or blood product (transfusion reaction)

  • SAEs and suspected unexpected serious adverse events (SUSARs)

  • Development of perioperative AKI

  • Concomitant medications

  • Vital signs. Change from randomisation to (1) day of index operation (prior to surgery), (2) 8 weeks and (3) 6 months post index operation

  • Laboratory results. Change from randomisation to (1) day of index operation (prior to surgery), (2) 8 weeks and (3) 6 months post index operation. Of particular interest will be changes in eGFR, creatinine, serum phosphate, ferritin and TSAT levels.

Subgroup analysis

A number of subgroup analyses for both co-primary end points were predefined; these were age (< 70 and ≥ 70 years), baseline Hb level (< 100 g/l and ≥ 100 g/l), sex (male and female), body mass index (< 30 kg/m² and ≥ 30 kg/m²), ferritin (< 100 and ≥ 100), TSATS (< 20% and ≥ 20%) and type of operation (major, major+ and complex). The subgroups were analysed by inclusion of an interaction term between treatment group and the subgroup in the relevant regression model.

Change in end points over the time of the trial

There were no changes to the co-primary end points of the trial following publication of the trial protocol. DAOH was added as a secondary outcome.

Sample size

The sample size requirement was calculated for the composite co-primary end point of blood transfusion or death by 30 days post operation. Assumptions for the sample size calculations were based on data from the pilot study, observational trials and audits carried out previously.

The anticipated risk of blood transfusion in the control group was estimated as approximately 40%. On the basis that the trial would have a type 1 error rate of 5% and an estimated 5% loss to follow-up, an estimated sample size of 500 patients (250 in each group) will have 90% power to detect an absolute reduction in risk of 14% [equivalent to a 35% relative risk reduction, risk ratio (RR) = 0.65] in the treatment group or approximately 80% power to detect an absolute reduction of 12% (30% relative reduction).

The above sample size should also provide similar or greater power for the second co-primary end point of blood transfusion rate, which includes repeat transfusions.

Laparoscopic surgery

PREVENTT was planned for major open abdominal surgery, which represented high-risk surgery for blood transfusion and also patient risk. 39–43 Cochrane reviews have shown that laparoscopic surgery is associated with a lower risk of blood loss, improved postoperative recovery, reduced patient complications and reduced length of hospital stay. In 2010, at initial grant submission, it was envisaged that colorectal surgery patients would be the leading population recruited into PREVENTT; however, there has been a steady rise in laparoscopic surgery over the last 10 years. By 2017, the majority of procedures (63%, SD 18%) in the 162 centres in the national bowel cancer audit45 were being performed laparoscopically. 45 Although this means that one-third of abdominal surgeries that were performed by open laparotomy, recruiting centres found that most operations were listed as an intended laparoscopic procedure.

The TSC reviewed trial recruitment in detail and recalculated the power analysis to include laparoscopic surgery; aside from changing the protocol halfway through the study, the anticipated actual recruitment would need to double, such that two patients undergoing laparoscopic surgery would be needed for every one patient planned to undergo open surgery. Therefore, the TSC decided not to include patients who were due to undergo laparoscopic surgery.

Statistical methods

Other secondary end points

Ethics considerations

Ethics approval for the study in the UK was given by the East of England – Cambridgeshire and Hertfordshire Research Ethics Committee on 5 November 2012 (reference number 12/EE/0445). The trial was registered with a National Clinical Trial number of NCT01692418.

The trial was overseen by three committees: the TSC, the DSMC and the PMG.

The TSC had overall responsibility for the scientific integrity and quality of the trial. This involved: ensuring the trial was conducted to the standards set out in the guidelines for Good Clinical Practice; ensuring adherence to protocol as far as possible; responsibility for overall patient safety; and considering new relevant information arising throughout the duration of the trial. The TSC also had responsibility to consider any recommendations made by the DSMC. The TSC met annually throughout PREVENTT to monitor the progress and quality of the trial, to review the recruitment rate and consider protocol amendments.

The DSMC had the responsibility to ensure the safety of patients in the trial. The DSMC was the only group to review interim analyses broken down by treatment group during the conduct of the trial. The DSMC performed interim safety analyses annually, but there were no interim efficacy analyses. The interim reports were semi-unblinded (i.e. summaries were presented by group A and group B) and contained details of patient recruitment, demographic and baseline characteristics, details of the intervention, primary safety end points, primary efficacy end point and other end points identified by the DSMC including AEs and SAEs. The chairperson of the DSMC reported directly to the chairperson of the TSC.

The PMG was responsible for the day-to-day running of the trial, meeting fortnightly during the setting up of PREVENTT and the early stages of recruitment, and then approximately monthly for the remainder of the trial.

Timelines before surgery

In July 2015, screening data from 4979 patients were collected over a 3-month period from the active sites. Overall, 25% of those patients screened were eligible for PREVENTT, 47% were not anaemic and 28% were not undergoing major open elective surgery. Of those eligible, the timeline of 10–42 days before operation was not possible for 11% of patients, and 6% of patients did not have the relevant blood results for screening. These two factors, patients not having a Hb result or not being screened within the timelines for inclusion in the trial, were the biggest reasons for recruitment failure. There were several factors that contributed to this.

Preoperative pathways did not adequately allow for the recognition, diagnosis and treatment of anaemia before the planned operation date. Most patients did not have a recorded set of blood results (either in the notes or on the hospital system) at the time of surgical review. The surgical pathway was dictated by the 2-week rule for diagnosis to treatment in cancer surgery. The surgical decision for operation, and, therefore, the operation date, was predominantly made at the final multidisciplinary meeting, with the date for surgery often in the following week. Preoperative assessment clinics were frequently the first time that a FBC was performed to assess for anaemia, and this preoperative assessment clinic appointment was often within the 10 days before surgery.

Surgical practice and clinical care have changed significantly in the UK over recent years. In the last decade, hospitals and NHS trusts reorganised patient care with centralisation to high-volume centres, with the intent to improve outcomes. Consequently, cancer services such as upper gastrointestinal, hepatobiliary and urological surgery often involved patients being referred between hospitals. Sharing of clinical data between hospitals and NHS trusts was variable and presented a challenge to local research teams. Often, patients referred to tertiary hospitals were identified without full clinical data or laboratory results available.

Individual patient factors were not a major problem. Among the 384 patients eligible at that point, 159 were recruited, representing 41% of the potential population, a reasonable achievement for any surgical trial. Among those who were not recruited, the main reasons were that patients felt that there was too much going on, they did not want to make additional hospital visits, they had concerns over paperwork, they did not want additional tests and they did not want to receive the placebo.

National Institute for Health and Care Excellence guidance

In November 2015, NICE guidance on blood transfusion practice was released. 2 This was supported by the NICE Quality Standard 1382 in December 2016, which stated:

People should have their haemoglobin levels checked at least 2 weeks before surgery, if possible and necessary for the procedure they are having. If they have iron-deficiency anaemia, they should be offered iron supplementation. Oral iron should be offered initially, and started at least 2 weeks before surgery. If oral iron is not appropriate, intravenous iron should be considered.

Reproduced with permission from NICE Guideline (NG24). 2 © NICE 2015 Blood Transfusion. Available from www.nice.org.uk/guidance/ng24. All rights reserved. Subject to Notice of rights. NICE guidance is prepared for the National Health Service in England. All NICE guidance is subject to regular review and may be updated or withdrawn. NICE accepts no responsibility for the use of its content in this product/publication

National Institute for Health and Care Excellence guidelines2 also recommended that the blood transfusion threshold should be lowered to a Hb level of < 70 g/l (80 g/l in patients with ischaemic heart disease). This is also a significant change from the previous transfusion threshold of 90 g/l (which was the inclusion level for the trial) and may have contributed to the lower blood transfusion rate seen in this trial (29% compared with the predicted rate of 40%).

Through 2014 to 2016 there was steady recruitment to PREVENTT, with a median number of nine patients (range 2–18 patients) per month over 36 months. Following publication of the NICE guidance,2 there was considerable communication from sites with concerns about the need to follow this guidance and give iron therapy, and the lack of equipoise in the use of preoperative intravenous iron. There was also a steady drive in the UK and throughout Europe to increase preoperative intravenous iron clinics. 23,48,49

The overall impact of the 2015 NICE guidance was significant;2 several centres withdrew from PREVENTT and patient recruitment reduced to a median of 7 (range 4–15) patients per month over the subsequent 21 months through 2017 and 2018 (p = 0.01).

Owing to the difficulties with recruitment, the TSC decided to halt recruitment in September 2018 after 487 patients (compared with the 500 that were planned) had been enrolled and randomised to enable follow-up to be completed within the trial period. The effect of recruiting 13 fewer patients than planned was considered to be negligible, with sample size calculations indicating a small loss of power (88% vs. 90%) and loss to follow-up having been less than anticipated in the initial power calculations.

Patient flow

Between 6 January 2014 and 28 September 2018, 487 patients were randomised into PREVENTT (243 placebo, 244 intravenous iron) (Figure 1). Six patients (two placebo, four intravenous iron) did not receive their intended randomised treatment. In total, eight patients withdrew consent during follow-up. Three patients (two placebo, one intravenous iron) withdrew consent between randomisation and the planned operation, and five patients (two placebo, three intravenous iron) withdrew consent between surgery and the 6-month visit. Twenty-three patients did not undergo their planned operation (13 placebo, 10 intravenous iron). Of these, 10 patients’ operations were cancelled because of a clinical plan change, seven patients were deemed unfit for surgery and six patients had disease progression. A total of 228 patients in the placebo group and 233 patients in the intravenous iron group underwent their planned operation.

Eight patients (three placebo, five intravenous iron) died between surgery and the 8-week visit, and a further 14 patients (seven in each group) died between the 8-week and 6-month visits. Follow-up rates were very good, with only five patients (three placebo, two intravenous iron) who had not died or withdrawn consent missing the 6-month visit. Vital status at 6 months was known for 236 (97.1%) and 238 (97.5%) patients in the placebo and intravenous iron groups, respectively.

Protocol deviations

In addition to the patients described above who did not have the trial treatment or undergo their planned operation, 46 patients (26 placebo, 20 intravenous iron) had their surgery outside the prescribed timelines. Of these, 13 operations took place within 10 days of the trial treatment and 33 operations took place > 42 days after treatment. A further 20 patients underwent surgery that was not classified as major open abdominal surgery (12 placebo, eight intravenous iron). Of these, six patients underwent surgery that was open and close (i.e. abandoned), 13 patients had operations that were performed laparoscopically and one patient had a minor operation.

There were no cases reported of patients being unintentionally unblinded. At the time of surgery, patients were asked if they thought that they knew which treatment they had received. Overall, 47% of patients reported not knowing, 27% thought that they had received intravenous iron (of whom 58% actually did receive iron) and 25% thought that they had received placebo (of whom 60% actually did receive placebo).

FIGURE 1.

Enrolment, randomisation and follow-up. a, Patient has blood transfusion before withdrawal or loss to follow-up and is, therefore, included in the analysis of the co-primary end points. ITT, intention to treat.

Baseline characteristics

Overall, median (IQR) age was 66 (54–72) years and 267 (54.8%) patients were female. Most (87.9%) patients were white, 6.8% were African Caribbean and 4.9% were Asian. Most patients were American Society of Anesthesiologists grade II (61.0%) or III (25.6%). Comorbidities included that 182 patients had hypertension (37.4%), 75 (15.4%) were diabetic, 46 (9.4%) had had a previous heart attack, stroke or transient ischaemic attack, 76 (15.6%) reported renal problems and 64 (13.1%) reported respiratory problems. Half of patients had never smoked and 41 (8.5%) were current smokers. The two treatment groups were well balanced, with no major differences in any of the baseline characteristics (Table 1).

History of iron deficiency was reported by 139 (28.5%) patients. The groups were well matched with respect to potential aetiological factors that may have an impact on iron levels by reducing absorption (e.g. hiatus hernia or acid reflux), causing iron deficiency (bleeding tendency), or through inflammation (arthritis or inflammatory bowel disease). Similarly, the groups were similar with respect to the proportions that received preoperative chemotherapy or radiotherapy, which had been undertaken in 110 (22.6%) patients before their planned operation.

There was no difference in medications being taken preoperatively that may affect iron therapy or bleeding; overall, 95 (19.6%) patients were taking iron tablets, 42 (8.6%) patients were on anticoagulation therapy and 57 (11.7%) patients were on antiplatelet therapy (two on dual antiplatelets), and five patients were on both anticoagulation and antiplatelet therapy.

Operation details

Of the 487 patients randomised, 461 (95%) underwent surgery (228 placebo, 233 intravenous iron). The median (IQR) time from randomisation to surgery was 15 (12–22) days and was similar in the two groups.

The planned date of operation was postponed for 18 patients (3.7%); for 15 patients, this was due to lack of bed availability.

The two groups were well balanced in terms of the complexity and type of surgery undertaken, with the most common surgeries being upper gastrointestinal (34%), gynaecological (30%) and colorectal (15%) (Table 2). Anaesthetic time and total procedure time were similar in the two groups. Overall median (IQR) total procedure time (anaesthesia, preparation and surgery) was 250 (175–355) minutes and median (IQR) total operation time (knife to skin to drapes removed) was 165 (110–277) minutes. Cell salvage was used in 21 (5%) patients (11 placebo, 10 intravenous iron) with the median (IQR) volume being reinfused 400 ml (200–658 ml).

Of the 461 patients who underwent their planned surgery, 286 (62%) went to ICU with median (IQR) ICU LOS of 2 (2–4) days. Overall, the median total hospital LOS was 9 (IQR 6–14) days. Three patients died following surgery without being discharged. Both ICU LOS and total hospital LOS were similar in the two groups.

Eighteen patients (3.9%) were returned to theatre, for the following documented reasons: bleeding (two patients), surgical airway (three patients), major wound revision (seven patients), surgical complicatons (three patients) and other (three patients).

Efficacy of intravenous iron

At randomisation, Hb levels were well balanced between the placebo and intravenous iron groups: mean (SD) 111.0 g/l (11.9 g/l) and 111.2 g/l (11.8 g/l), respectively.

At the time of surgery, mean Hb was significantly higher in the intravenous iron group (113.5 g/l vs. 108.2 g/l) with the difference being 4.7 g/l, 95% CI 2.7 to 6.8 g/l; p < 0.0001 (from ANCOVA model adjusting for baseline Hb). Anaemia was corrected in 42 (21%) patients in the intervention group compared with 21 (10%) patients in the placebo group (RR 2.06, 95% CI 1.27 to 3.35 g/l; p = 0.002).

Haemoglobin levels were not significantly different in the immediate postoperative days, but were significantly higher in the intravenous iron group at 8 weeks (mean difference 10.7 g/l, 95% CI 7.8 to 13.7 g/l; p < 0.0001) and at 6 months (mean difference 7.3 g/l, 95% CI 3.6 to 11.1 g/l; p < 0.001) (Figure 2).

FIGURE 2.

Haemoglobin levels of the trial participants by treatment group and visit. BL and OP were measured by the central laboratory; all other measurements are from local laboratories. D2–3, D4–5, D6–7 and D14 measurements are available only for patients still hospitalised at that time. BL, baseline prerandomised treatment; D, day post operation (e.g. D2–3 = day 2 or 3 post operation); IV, intravenous; OP, day of operation.

Co-primary end points

A total of 474 out of 487 (97.3%) patients (237 in each group) were included in the intention-to-treat analysis for the two co-primary end points. Overall, from randomisation to 30 days following their index operation, 136 patients (67 placebo, 69 intravenous iron) received at least one blood transfusion or died. There was no difference in the risk of transfusion or death at 30 days between the treatment groups (RR 1.03, 95% CI 0.78 to 1.37; p = 0.84; absolute risk difference +0.8%, 95% CI –7.3% to 9.0%). A total of 216 transfusion episodes (111 placebo, 105 intravenous iron) occurred between randomisation and 30 days following the index surgery. There was no difference between the groups in the rates of blood transfusion (rate ratio 0.98, 95% CI 0.68 to 1.43; p = 0.93; absolute rate difference 0.00, 95% CI –0.14 to 0.15) (Table 3). Excluding large blood transfusions did not alter these results.

Analyses of the co-primary end points adjusting for age, baseline Hb and type of surgery did not materially change the results (adjusted RR for blood transfusion or death to 30 days 1.06, 95% CI 0.81 to 1.38; p = 0.69 and adjusted rate ratio for transfusion episodes 0.99, 95% CI 0.69 to 1.42; p = 0.96). Similarly, per protocol analyses did not materially alter the results (RR for blood transfusion or death to 30 days 1.04, 95% CI 0.77 to 1.41; p = 0.79 and rate ratio for transfusion episodes 0.98, 95% CI 0.67 to 1.44; p = 0.93).

Key secondary end points

Other secondary end points

Length of intensive care unit and hospital stay

There was no difference in ICU LOS. Median (IQR) ICU LOS was 1 (0–3) day in the placebo group, and 2 (0–3) days in the intravenous iron group (p = 0.11). Overall, 75% of patients had their operation on the day of admission and there was no difference in total hospital LOS, with the median (IQR) total hospital LOS being 9 (7–14) days in the placebo group and 9 (5–14) days in the intravenous iron group (p = 0.14). The distribution of total hospital LOS by treatment group is shown in Figure 3. For three patients who died before discharge (two in the placebo group/one in the intravenous iron group), LOS was calculated as the number of days from date of admission for surgery until date of death.

FIGURE 3.

Distribution of total hospital LOS for patients undergoing surgery in (a) the placebo group; and (b) the intravenous iron group.

Predefined subgroup analyses

Predefined subgroup analyses for the co-primary end points were performed for age (< 70 vs. ≥ 70 years), sex (male vs. female), body mass index (< 30 km/m² vs. ≥ 30 kg/m²), operation type (complex major operation, major, major+), Hb (< 100 g/l vs. ≥ 100 g/l), ferritin (< 100 ng/ml vs. ≥ 100 ng/ml) and TSATs (< 20% vs. ≥ 20%). No interactions were observed (Table 10).

Power calculation

The initial power calculation in the PREVENTT protocol1 was based on 19 observational studies on the use of intravenous iron in surgical patients, with a total of 3043 patients and an average increase in Hb of 9.6 g/l. The combined results showed a reduction in blood transfusion from 39.6% in the control groups (n = 1127) to 29.6% in those who received intravenous iron (n = 657).

In a prospective observational pilot study for PREVENTT, we reviewed all patients (n = 718) undergoing major surgery at University College Hospital (UCH) [now University College London Hospital (UCLH)] who spent 48 hours or more in hospital, with an average LOS of 6 days. Average Hb was 133 g/l, but 154 (21%) patients were anaemic. Blood transfusion occurred in 64 (41.5%) patients with anaemia, who received a total of 190 units, compared with 58 (10.2%) patients without anaemia, who received 121 units.

A crude projection of results from the above data in a population of 250 patients with anaemia before surgery, where intravenous iron successfully increased preoperative Hb by 10 g/l in 80% of patients, would result in only 114 of the 250 patients being anaemic at the time of operation [i.e. we expected that intravenous iron would correct anaemia in 136 (54.4%) patients]. Consequently, the projected results would have been that intravenous iron reduced the frequency of blood transfusion from 98 patients to 59 patients.

These projections were not realised in PREVENTT. The intervention of intravenous iron did not result in a 10 g/l rise in Hb levels in 80% of patients, and the average Hb rise was significantly lower; only 21.1% of patients had their anaemia corrected before their operation, less than half than the number that was expected (54.4%).

This may reflect those patients who did not have iron deficiency or where the diagnosis of FID was incorrect. The impact of inflammation on FID and on bone marrow suppression merits further mechanistic investigation.

Efficacy

The Hb rise was less than anticipated and may be the reason that there was no difference in the co-primary end points. For there to be a reduction in transfusion, the difference in operative Hb required between the two groups would need to have been much greater. The rise in Hb levels seen in PREVENTT was lower than expected, with a mean difference between intravenous iron and placebo of 4.7 g/l (95% CI 2.7 to 6.8 g/l) over a median 2-week period. Anaemia was corrected in only 42 (21.1%) patients before operation, with no difference seen between the intention-to-treat and per protocol analyses. In the IVICA trial,50 patients had colorectal cancer and were randomised an average of 21 (15–34) days before surgery, with a mean Hb level rise of 5.0 g/l with oral iron and 15.5 g/l with intravenous iron. This rise was lower in the Australian trial,19 in which preoperative intravenous iron patients were treated a median (IQR) of 8 (6–13) days before their operations and their Hb level rose from a baseline mean (SD) of 107 g/l (13 g/l) to 115 g/l (13 g/l). In a meta-analysis of 9004 patients in 65 studies of the role of iron to treat anaemia, intravenous iron resulted in higher Hb levels than control (mean difference 1.04, 95% CI 0.52 to 1.57 g/l; I² = 93%; χ² test for heterogeneity; p < 0.00001). 1 However, there was considerable heterogeneity (I² = 93%), with a point estimate of the mean difference in Hb levels ranging from −7 g/l to 30 g/l.

The results from PREVENTT may have been affected by the use of oral iron in 19.5% of patients. In the Cochrane analysis,13 oral iron increased Hb levels (mean difference 9.1 g/l, 95% CI 4.8 to 13.5 g/l; I² = 67%; χ² test for heterogeneity; p = 0.0004) but with significant heterogeneity (I² = 67%), with point estimates of the mean difference in Hb levels ranging from 2.0 to 22 g/l higher in the oral iron group than controls. When comparing intravenous iron to oral iron, Hb concentration was higher in the intravenous iron group (mean difference 0.53, 95% CI 0.31 to 0.75), which is comparable to that seen in PREVENTT.

Another reason could be that PREVENTT included only those with a Hb level of > 90 g/l. In the pilot studies, only 5% of those with anaemia had a Hb level of < 90 g/l, and these patients often had more comorbidities and represented a generally sicker population of patients, so this may have added confounders to the analysis. Overall, the average Hb at recruitment in PREVENTT was not dissimilar to previous RCTs, but in many preoperative clinics in the UK the greatest use of intravenous iron is often for those with a Hb level of < 100 g/l. 54 By excluding severely anaemic patients, we may have excluded those who could have benefited the most. However, in the predefined subgroup analysis of patients with a Hb level of < 100 g/l, there was no clinically significant difference in the preoperative Hb rise compared with those with a Hb level of > 100 g/l and there was no interaction for either of the co-primary end points.

The use of intravenous iron therapy is for the treatment of iron deficiency. Our assumption was that iron deficiency would be the predominant cause for anaemia,14 whether AID or FID, which would respond to intravenous iron therapy as treatment. This could be a factor in the lack of efficacy seen because iron deficiency was not specifically diagnosed as being causal for anaemia in the eligibility criteria. Nevertheless, we did predefine a subgroup analysis in those patients with a ferritin level of < 100 ng/ml and TSAT < 20% in line with current guidelines. 55 At inclusion and randomisation, the majority of patients (76%) had TSAT < 20%, 57% had a ferritin level of < 100 ng/ml and 32% a ferritin level of < 30 ng/ml. There was no evidence of any interaction between treatment in those patients who were iron deficient and anaemic and these predefined subgroups for the co-primary end points of the study. It may be that the causal mechanism behind anaemia in the preoperative setting requires treatment with concurrent erythropoietin, as reported by Spahn et al. 51 in cardiac surgery. Erythropoietin is not licensed for such patients in the UK, but erythropoietin and intravenous iron have been recommended for anaemia in patients before orthopaedic surgery. 3

The lack of efficacy in terms of Hb rise in the treatment group meant that, in fact, the difference in Hb between the groups at the time of surgery was small. The transfusion of RBCs during the perioperative period is triggered by low Hb concentration, the trigger level of which may vary from institution to institution, and from practitioner to practitioner. Drops in Hb during surgery are due to either haemorrhage or haemodilution. Major haemorrhage will mostly prompt transfusion, often without measuring Hb, and this was the same in both groups, resulting in the decision to exclude large blood transfusion episodes from analyses. More minor haemorrhage or haemodilution will reduce Hb proportionally, and lower Hb levels are more likely to result in transfusion. Therefore, the fact that there was only a very small difference in Hb between the two groups may account for the lack of difference in transfusion rates and number of units of red cells transfused. In addition, we have shown that the major rise in Hb occurred postoperatively. This may account for the differences seen in re-admissions for complications in this period, as lower Hb is known to be associated with infection and other complications, and re-admission. 56 Therefore, it may be surmised that the only possible benefit demonstrated by the preoperative administration of intravenous iron is not during or immediately after surgery, but some weeks to months after surgery, where Hb rises more quickly in the patients treated initially with intravenous iron. This may also reflect the mechanism of anaemia, induced by blood loss during surgery. Oral iron is known to be ineffective in the first 4–6 weeks after surgery when hepcidin levels are high as part of the inflammatory response to surgery;57 hence, iron is not absorbed from the intestine, and red cell production is reliant on iron stores already in the liver, which were much greater in the patients treated with intravenous iron.

This trial has a number of important clinical implications. It would seem that there is no benefit in treating patients with this dose of intravenous iron in the immediate preoperative period. Future trials should examine the effect of treating patients with intravenous iron much earlier before surgery. However, this may be very challenging in the current NHS system, where preoperative assessment is carried out only 1 or 2 weeks before surgery, particularly for patients who have cancer, for whom surgery is a priority in terms of time and urgency. Another very interesting hypothesis generated from this study is whether or not the use of intravenous iron given in the postoperative setting, before discharge from hospital, results in an earlier correction of postoperative anaemia. This would be cheaper than giving it preoperatively because the patient would already be in hospital, being nursed and monitored in a hospital bed and probably with a cannula in situ. Such a trial should focus on the reduction of postoperative complications, longer-term, patient-oriented end points, and assessing whether or not reducing re-admissions to hospital for postoperative complications may be cost-effective. Other future trials might examine particular diagnoses, such as surgery for cancer, and focus on cancer outcomes, such as tolerance on adjuvant treatment or patient-reported outcomes, as well as rate of recurrence.

Generalisability

PREVENTT has addressed the common problem of patients presenting with anaemia before surgery.

The population studied was those patients undergoing major surgery, which would include major cancer and non-cancer surgery. Many of the patients were elderly and had comorbidities; therefore, this population represents those at highest risk of needing a blood transfusion and also at higher risk of postoperative complications.

The intervention was a total dose of intravenous iron of 1000 mg; this should be adequate to replenish a patient’s iron stores even in the presence of anaemia. The timing of the intervention, at least 10 days before an operation, should have provided adequate time for haematopoiesis. In addition, this is, realistically, the earliest feasible time for administration in the current UK preoperative pathway.

The control was a placebo and essentially standard care in the UK.

Outcomes were primarily risk of blood transfusion, but also included the important patient-related end points of postoperative complications, length of hospital stay and patient quality of life.

PREVENTT is, therefore, generalisable to current practice in the UK because if intravenous iron is clinically ineffective in this setting, it is unlikely to be effective in lower-risk or laparoscopic surgery. Similarly, the effect would be lessened in timelines closer to surgery. Indeed, the PREVENTT population is very similar to the target population for the NHS England Commissioning for Quality and Innovation framework, and consideration should be given to revising this framework.

The issue of restoring iron levels to improve recovery after surgery needs to be addressed in future research.

Overall evidence

We will add these data at a later date. The PREVENTT fellow who conducted a full systematic review and update of previous Cochrane review cannot complete this work until February 2021; thereafter, we will present the PREVENTT results in context.

Implications for health care

The primary results of the trial demonstrate no evidence of clinical benefit in giving intravenous iron preoperatively to patients undergoing major surgery (non-cardiac).

There was no evidence of impact in terms of blood transfusion rate or risk in the perioperative period, up to 6 months postoperatively, from giving intravenous iron to this group of patients preoperatively. Furthermore, there was no clear benefit demonstrated in terms of quality of life from this treatment, even at 6 months postoperatively. The current NHS England Commissioning for Quality and Innovation targets should be withdrawn.

There was, however, demonstrable improvement of the Hb level, in terms of correcting anaemia compared with placebo in both the preoperative and postoperative periods, and there were no negative effects seen or increases in AE reporting, particularly infection, with the use of intravenous iron. The post discharge increase in Hb levels and associated reduction in re-admissions for postoperative complications merit further research.

Recommendations for research

There will be further analyses carried out on the data obtained from the trial to assess the causality of anaemia in this patient population, particularly the study of levels of hepcidin and iron studies of serum samples, to assess the inflammatory states of patients and their relation to the study outcomes.

Further follow-up in the cancer patients will also be carried out to assess longer-term follow-up in relation to giving iron to this group of patients.

Postoperative treatment of anaemic patients with intravenous iron also needs to be studied further to assess improved benefits.

Trial Steering Committee

Andrew Bradbury (Chairperson), Toby Richards (Chief Investigator), Trevor Burley, Shelley Van Loen, Stefan Anker, Andrew Klein, Iain Macdougall, Gavin Murphy, Martin Besser and Isabel Unsworth.

Observers: Tim Clayton, Tim Collier, Kimberley Potter, Sandy Abeysiri, Richard Evans, Rosemary Knight, Rebecca Swinson, Laura Van Dyck and Jane Keidan.

Data Safety and Monitoring Committee

Lorna Williamson (Chairperson), Angela Crook and John Pepper.

Unblinded statisticians supporting the DSMC were Joanna Dobson, Simon Newsome, Tom Godec and Matthew Dodd.

Project Management Group

Toby Richards, Laura Van Dyck, Richard Evans, Sandy Abeysiri, Ben Clevenger, Anna Butcher, Rebecca Swinson, Tim Collier and Kimberley Potter.

Study design and support

Stefan Anker, John Kelly, Steven Morris, John Browne, Jane Keidan, Michael Grocott, Marisa Chau and Rosemary Knight.

Trial statistician

Timothy Collier.

Randomisation service

Sealed Envelope.

Blood sample analysis

The Doctors Laboratory, London, UK.

Final report preparation

Megan Knight.

Contributing sites

Contributions of authors

Toby Richards (https://orcid.org/0000-0001-9872-4653) (Head of Department of Surgery) was the chief investigator and was involved in the trial concept and design, the conduct of the research, interpretation and reporting of results, and writing and reviewing the manuscript.

Ravishankar Rao Baikady (https://orcid.org/0000-0002-6827-9250) (Consultant Anaesthetist) was the Principal Investigator at the site with the highest recruitment levels, and was involved in the trial design, conduct of research, and reviewing the manuscript.

Ben Clevenger (https://orcid.org/0000-0001-5659-6323) (Consultant Anaesthetist) was a NIHR PREVENTT research fellow involved with the conduct of the research, the sites set-up and management, and reviewing the manuscript.

Anna Butcher (https://orcid.org/0000-0002-1454-9403) (Clinical Research Associate) was a NIHR PREVENTT research fellow involved with the conduct of the research, the sites set-up and management, and reviewing the manuscript.

Sandy Abeysiri (https://orcid.org/0000-0002-4258-2585) (Clinical Research Associate) was a NIHR PREVENTT research fellow involved with the conduct of the research, the sites set-up and management, and reviewing the manuscript.

Marisa Chau (https://orcid.org/0000-0003-2965-1870) (Clinical Project Manager) was a trial manager involved with the conduct of the research and produced the references for the manuscript.

Rebecca Swinson (https://orcid.org/0000-0002-4997-5998) (Trial Manager) was the trial manager and was involved in the trial set-up, site set-up, the conduct of the research and reviewing the manuscript.

Tim Collier (https://orcid.org/0000-0002-3095-277X) (Associate Professor, Medical Statistics) was the lead statistician and was involved in the trial design and conduct, analysing the data, and writing and reviewing the manuscript.

Matthew Dodd (https://orcid.org/0000-0002-6207-6604) (Research Fellow in Medical Statistics) was the unblinded trial statistician and a member of the Data Monitoring Committee, analysed the data and was involved in reviewing the manuscript.

Laura Van Dyck (https://orcid.org/0000-0002-5720-8633) (Data Manager) was the data manager and was involved in the conduct of the research and reviewing the manuscript.

Iain Macdougall (https://orcid.org/0000-0001-9098-8611) (Consultant Nephrologist) is an expert in clinical trials on intravenous iron, was a member of the TSC and was involved in writing and reviewing the manuscript.

Gavin Murphy (https://orcid.org/0000-0003-0976-1482) (Professor of Cardiac Surgery) is an expert in clinical trials on surgery and blood transfusion, was a member of the TSC and was involved in writing and reviewing the manuscript.

John Browne (https://orcid.org/0000-0003-3494-4336) (Director of the National Health Services Research Institute) is an expert in clinical trials and quality of life analysis, was a member of the TSC, was involved in the trial design, and in writing and reviewing the manuscript.

Andrew Bradbury (https://orcid.org/0000-0003-0530-9667) (Professor of Vascular Surgery) is an expert in clinical trials on surgery, was the chair of the TSC, and was involved in writing and reviewing the manuscript.

Andrew Klein (https://orcid.org/0000-0001-9913-6570) (Consultant Anaesthetist) is an expert in clinical trials on intravenous iron and blood transfusion, was a member of the TSC, and was involved in the trial concept and writing and reviewing the manuscript.

Publications

Richards T, Clevenger B, Keidan J, Collier T, Klein AA, Anker SD, Kelly JD. PREVENTT: preoperative intravenous iron to treat anaemia in major surgery: study protocol for a randomised controlled trial. Trials 2015;16:254.

Butcher A, Clevenger B, Swinson R, van Dyck L, Klein A, Richards T. PREVENTT trial: a snapshot of preoperative anaemia in the United Kingdom. J Clin Trials 2016;6:3.

Butcher A, Swinson R, Van Dyck L, Collier T, Richards T. The role of general practice in surgical trials. Br J Gen Pract 2017;67:157–8.

Richards T, Baikady RR, Clevenger B, Butcher A, Abeysiri S, Macdougall IC, et al. Preoperative intravenous iron to treat anaemia before major abdominal surgery (PREVENTT): a randomised, double-blind, controlled trial [published online ahead of print 4 September 2020]. Lancet 2020;396:1353–61.

Data-sharing statement

Research data will be made available to the scientific community with as few restrictions as possible so as to maximise the value of the data for research and for eventual patient and public benefit. All data requests should be submitted to the corresponding author for consideration.

Disclaimers

This report presents independent research funded by the National Institute for Health Research (NIHR). The views and opinions expressed by authors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, NETSCC, the HTA programme or the Department of Health and Social Care. If there are verbatim quotations included in this publication the views and opinions expressed by the interviewees are those of the interviewees and do not necessarily reflect those of the authors, those of the NHS, the NIHR, NETSCC, the HTA programme or the Department of Health and Social Care.

Recruitment by site

Recruitment by month

Recruitment curve