Surgical versus non-surgical management of lateral compression type-1 pelvic fracture in adults 60 years and older: the L1FE RCT

Pelvic fractures

Pelvic fractures can be life-threatening and people can experience significant problems with mobility following the injury. 1 They are associated with a total inpatient mortality rate of 9% and an all-cause mortality rate within 3 months of fracture of 13%. 2 All-cause mortality following pelvic fracture is approximately 50% at 3 years. 3 Mortality for fragility fractures of the pelvis is reported to be around a third after 1 year. 4

The incidence of pelvic fractures in the UK between 2008 and 2012 in men and women over 50 was 2.0 and 7.4/10,000 person-years (based on a retrospective observational study using data from the Clinical Practice Research Datalink, which is shown to be broadly representative of the UK population). 5 People can experience this injury as a result of a high-energy impact such as a road traffic accident or, in the case of older adults (generally 65 years of age and older), as a result of a low-energy fall from a standing height. 1 The UK age-specific incidence of pelvic fractures (based on a single centre) has increased from 39.6/100,000 (95% confidence interval: 31.8 to 48.1) in 1997 to 71.6/100,000 (58.4 to 81.0) in 2007–8 amongst people 65 years and older; 84% of these had pubic rami fractures. 6

Lateral compression type-1 (LC-1) fractures are a common fragility fracture of the pelvis in older adults and are associated with low bone density. 7 They typically involve a fracture of the pubic ramus, which patients experience as groin pain when they mobilise. There is usually also a ‘buckle’ fracture to the sacrum posteriorly, which is felt as low-back or buttock pain when moving the legs. The likelihood of an LC-1 fragility fracture increases with age and is more common in women. 3,8,9

Lateral compression type-1 fractures are often painful, particularly when the patient moves. This leads to reduced mobility, usually over a 1- to 2-month period, although it is estimated that 25% of patients experience pain for up to 5 years following the injury. 10 The ability of patients to mobilise following an LC-1 fragility fracture varies: some can mobilise and walk, albeit with some degree of pain, whereas others struggle to walk due to the pain. Patients who do not manage to get back walking due to ongoing pain are at greater risk of immobility-related complications. 11 These can include respiratory tract infections, urinary tract infections, pressure sores and venous thromboembolic events such as deep vein thrombosis or pulmonary embolism. 11,12 These individuals are also at risk of irreversible muscle wasting (systemic sarcopenia), loss of confidence and permanently decreased levels of independence, often leading to increased care requirements. Inability to return to independent living can result in utilisation of intermediate care or residential facilities. 13,14 Some patients do not regain their pre-injury level of walking or their prior independence with activities of daily living due to their loss of confidence and muscle strength/conditioning. 8,9,15,16 Additionally, individuals with LC-1 fractures have reported emotional stress, family strain, employment and financial difficulty, sleep disturbance and anxiety. 17

The current standard care for lateral compression type-1 fragility fractures

In the UK, standard treatment for LC-1 fractures is to ‘mobilise as pain allows’, prescribing pain relief and getting patients up within a few days of injury with physiotherapy input and getting them to mobilise with an assistive device until the fracture heals. 11,18,19 This is unlike fragility neck of femur fractures (hip fractures), which are usually treated surgically, and there is evidence that early surgery (within 1–2 days) is associated with reduced risk of death and pressure sores. 20 Despite LC-1 fractures being similarly disabling for some patients in terms of pain and immobility, while also occurring in the same patient group as hip fractures, to date, it has not been shown whether or not patients with LC-1 fragility fractures would have a better recovery with surgery than with non-surgical management.

Surgical fixation

Traditional pelvic implants carry poor ‘bite’ or ‘purchase’ in low-quality osteoporotic bone around the pelvis and surgeons have been reluctant to offer surgery to patients with LC-1 fragility fractures. External fixators, consisting of pins inside the pelvis connected to bars and clamps outside of the skin, are cumbersome, poorly tolerated and carry a high incidence of pin-site infections and soft-tissue problems. 21 The alternative is surgical fixation of the back of the pelvis with ilio-sacral screws;9 however, in the majority of elderly patients, these screws carry poor ‘purchase’ in osteoporotic bone, leading to ineffective fracture stabilisation and persistence of pain. 11

A more recent approach is use of an anterior pelvic fixation device that resembles a traditional external fixator, in that it has screws that are secured into the pelvic bone, and these are connected by a metal bar across the front of the patient. This is an internal fixation device (INFIX). Unlike traditional external fixation devices, it is fitted internally, sitting entirely underneath the patient’s skin, with no external metalwork visible. INFIX has two potential benefits over external fixation: it is less cumbersome and less inconvenient to patients, compared with pins, clamps and bars protruding out of the skin. It also does not have pin sites (where the bone pins exit through the skin), which make traditional external fixation very susceptible to local infection. The INFIX technique involves percutaneous placement of screws in the pelvic bone and connects them with a bar under the skin. 22 The pelvic bone where the screws are placed is generally strong and easy to visualise intraoperatively, even in highly osteoporotic bone. Although a proportion of implants need to be removed, this is usually done as a day case procedure. INFIX is used in younger patients with high-energy LC-1 fractures. It is now a well-described technique with a number of peer-reviewed series confirming its safety. 23 It is therefore a widely practiced, rather than ‘novel’, technique and is technically straightforward to carry out.

Existing evidence

A systematic review found no robust evaluations of the effectiveness of internal fixation with INFIX in patients with osteoporotic LC-1 fractures. 24 The review identified five case series, with four being retrospective. Participants were aged 64 or over and most had sustained their injury from a low-energy fall. A variety of fixation techniques were used. Of the 225 patients in the five studies, most had internal devices, with 25 having external fixation; most patients had more than one type of fixation.

In the single series evaluating INFIX alone, 19 of the 29 patients had LC-1 fractures. 25 Six patients had anterior fixation with INFIX alone and the remaining 23 had INFIX with additional internal fixation. Postoperatively, 22 of the 29 (76%) returned to their pre-injury walking ability, and a further 6 patients had some deterioration but could still walk. Complications included chronic pain (n = 3, 10.3%) and painful lateral femoral cutaneous nerve hyperaesthesia (n = 8, 27.5%) and, less commonly, infections, implant loosening, pneumonia and thrombosis.

Our search of clinicaltrials.gov in advance of our study identified an ongoing trial in the USA of surgical versus non-surgical management of patients aged between 18 and 80 with LC-1, LC-2 and LC-3 pelvic fractures in 130 participants. The aim of this trial is to determine which patients would benefit from early surgical stabilisation. 26 There is also a Research for Patient Benefit funded feasibility study of surgical versus non-surgical management of LC-1 pelvic fractures in patients with non-fragility fractures. 27

The existing evidence on the effectiveness of using an INFIX device to stabilise LC-1 fractures is limited to one case study with 29 participants. Surgeons are already carrying out INFIX surgeries in the absence of extensive evidence. There is a need to collect more evidence and there are currently no other clinical trials focusing on the effectiveness of surgical fixation with INFIX in adults with fragility fractures.

Inclusion criteria

• Patients aged 60 years or older.

• An LC-1 pelvic fracture, arising from a low-energy fall from standing height or less.

• Patients unable to mobilise independently to a distance of around 3 m and back due to pelvic pain (or perceived pelvic pain) 72 hours after injury. Use of a walking aid and verbal guidance were permitted; however, physical assistance was not.

Exclusion criteria

• Unable to perform surgery within 10 days of injury.

• Surgery was contraindicated due to soft-tissue concerns, or because the patient was not fit for anaesthetic (spinal or general).

• Patients who were non-ambulatory or required physical assistance to walk, prior to their injury (use of a walking aid was permitted).

• Concomitant injury or polytrauma that impedes mobilisation.

• Fracture configurations not amenable to internal fixation using INFIX, with or without ilio-sacral screws.

Due to the coronavirus disease 2019 (COVID-19) pandemic, and emerging evidence that infection with COVID-19 may increase the risk of mortality in patients undergoing surgery,28 the following exclusion criterion was added when the study reopened following a pause to recruitment: patients who tested positive for COVID-19 within 72 hours of admission (applicable only where testing was standard of care).

Informed consent

Once eligibility was confirmed, hospital research staff sought written informed consent from patients who had mental capacity to provide it. This study also included recruitment of patients who lacked capacity, and in this instance, consultee agreement or consent was sought in line with legal requirements. 29,30 Consent or consultee agreement was sought for follow-up beyond the duration of the trial to allow the possibility of future long-term follow-up including the use of routinely collected Hospital Episode Statistics (HES) and Office of National Statistics (ONS) data.

Primary outcome

The primary outcome measure was originally intended to be the average health-related quality of life (HRQoL), over 6 months, assessed by the patient-reported outcome measure, EQ-5D-5L utility score. However, due to the poor recruitment and the resulting low sample size, the average EQ-5D-5L utility score over the follow-up period was not feasible. Instead, the EQ-5D-5L utility scores were reported descriptively at each follow-up time point.

EQ-5D-5L was collected at baseline [for the day of baseline data collection and 1 week prior to injury (adapted with permission)], 2 weeks, 6 weeks, 12 weeks and 6 months post-randomisation time points, as well as at the optional 12-month follow-up point where relevant.

The EQ-D-5L is a validated generic patient-reported outcome measure (www.euroqol.org), including validation in patients with hip fractures and orthopaedic patients with cognitive impairment. The descriptive system has five health domains (mobility, self-care, usual activities, pain/discomfort and anxiety/depression) with five response options for each domain (no problems, slight problems, moderate problems, severe problems and extreme problems/unable to do). In addition, it has a health status visual analogue scale (VAS) which measures self-rated health with endpoints ranging from ‘the best health you can imagine’ to ‘the worst health you can imagine’. The EQ-5D-5L was scored according to the User Guide. 31 The measure is easily completed and can be completed by proxy (which was important for our clinical population), and it can also be scored for those who die during follow-up. EQ-5D-5L data were collected either in patient-completed questionnaires (for participants with capacity) or in proxy questionnaires (for those who lacked capacity). EQ-5D-5L proxy version 2 was used where the caregiver (the proxy) is asked to rate how he or she (the proxy) thinks the patient would rate his or her own HRQoL, if the patient were able to communicate it.

Secondary outcomes

• Self-rated health using the EQ-5D-5L VAS

  The second component of the EQ-5D-5L instrument is scored as a number ranging from 0 to 100 where 0 indicates the worse imaginable health and 100 indicates the best imaginable health. The EQ-5D-5L VAS score was collected from patients at baseline (for the day of baseline data collection and 1 week prior to injury), 2 weeks, 6 weeks, 12 weeks, 6 months and (if the patient was followed up at this time) 12 months post randomisation.

• Physical function using the eight-item Patient-Reported Outcome Measures Information System (PROMIS) lower extremity function (LEF) (mobility) – Short Form. 32

  Lower extremity function is an extremely important outcome domain for people with a LC-1 fracture, due to the impact of the injury on ability to mobilise. Measured at baseline, 2 weeks, 6 weeks, 12 weeks and 6 months (by proxy where required).

• Physical function using Timed Up and Go (TUG)

  This test, used routinely in clinical practice, assesses walking speed, mobility, balance and fall risk. 33,34 The TUG test involves measuring the time it takes for a patient to rise from a chair, walk 3 m, turn 180 degrees and return to sitting in the chair. The outcome of this test is the time in seconds taken to complete the test and lower times indicate better mobility. In addition, the number and type of walking aids used by the patient to complete the test was recorded. This measure was undertaken at the 12-week follow-up only in the clinic setting (there was no attempt to perform this where a remote visit was undertaken).

• Global mental health using the PROMIS scale v1.2 – Global Health Mental 2a subscale

  This is a two-question subscale on global mental health (GMH). Inclusion of this subscale was highly commended by our patient and public involvement (PPI) group. Measured at baseline, 2 weeks, 6 weeks, 12 weeks and 6 months (by proxy where required).

• Pain experienced from their pelvis using a VAS

  A unidimensional measure of pain intensity in adults. 35 This is a validated measure for acute and chronic pain in adults where scores are recorded by the patient placing a handwritten mark on a 10-cm line that represents a continuum from ‘No pain’ to ‘Worst imaginable pain’. 36 This outcome was collected from people with capacity only, at baseline, 2 weeks, 6 weeks, 12 weeks, 6 months, and an optional 12-month follow-up for those recruited early within the study.

• Delirium using the Abbreviated Mental Test Score

  Abbreviated Mental Test Score is a short, verbal test widely used in clinical practice to screen for confusion and dementia. 37,38 Recent data confirm its validity in emergency admissions in older adults within UK hospitals. 37 Measured at baseline, 2 weeks and at 12 weeks. Repeat assessment at 12 weeks was to confirm whether delirium was temporary or represented a permanent change.

• Delirium using the 4AT Rapid Assessment Test for Delirium

  4AT Rapid Assessment Test for Delirium is a short, practical instrument validated for detecting delirium, routinely used in clinical practice. 39,40 Postoperative delirium is a known complication for the elderly, particularly those with dementia. Therefore, its use as an outcome measure was to monitor this potential adverse effect of surgery. The strengths of the 4AT Rapid Assessment Test for Delirium are that it can be used on patients who are drowsy or agitated (which is common after surgery), it does not require specialist training and it takes less than 2 minutes to complete. Measured at baseline, 2 weeks and at 12 weeks. Repeat assessment at 12 weeks was to confirm whether delirium was temporary or represented a permanent change.

• Displacement of the pelvis

  A radiological assessment of the pelvis was to be performed at the 12 weeks’ visit for all participants and was to investigate whether there is clinically significant displacement of the pelvic ring; the week 12 investigator case report form (CRF) included a yes/no question asking if there has been clinically significant displacement of the pelvic ring. These X-rays would be standard care for patients undergoing surgical fixation but may be over and above what is routine practice for patients being managed non-surgically. To allow for scheduling around local restrictions and capacity within imaging departments arising from the impacts of the COVID-19 pandemic, the 12-week X-ray could be performed up to 6 months post randomisation.

• Mortality

  Collected at 6 months (and 12 months for those participants who agreed to this additional follow-up).

Complications and adverse events

Lateral cutaneous nerve injury was an adverse event of special interest (AESI), and information on this was collected on an adverse event (AE) form. Patients were also asked about this in the 2-week, 6-week, 12-week and 6-month questionnaires, as well as in the 12-month questionnaires for those who agreed to this additional follow-up.

Information on expected complications, including additional surgery, was collected in hospital at 2 weeks, at 12 weeks and at discharge (if after 2 weeks). Expected complications that were recorded included (but were not limited to): neurological complications, deep wound infection [using Centres for Disease Control (CDC) and Prevention definition],41 superficial infection (using CDC definition), rehospitalisation, re-operation (including removal of implant) and skin problems.

We collected data for the AESI and any unexpected AEs that were related to treatment for the original injury. We collected AE data from the point of randomisation up to 6 months post randomisation for all patients and up to 12 months post randomisation for patients that agreed to this additional time point. All AEs were listed on the appropriate AE or serious adverse event (SAE) CRF for routine return to York Trials Unit (YTU).

The planned process for SAEs was that they were entered onto the SAE reporting form and forwarded to YTU within 24 hours of the investigator becoming aware of them. Once received, causality and expectedness were to be confirmed by the CI. SAEs that were deemed to be unexpected and related to the trial were to be notified to the Research Ethics Committee (REC) and sponsor within 15 days and reported to the Trial Steering Committee (TSC) and Data Monitoring and Ethics Committee (DMEC) at the soonest meeting.

Resource use

Information on resource use throughout patients’ hospital stays and at discharge was collected to assess the impact on the NHS as part of the economic evaluation. Data collected in CRFs included length of hospital stay, medication, surgery details and details of therapy during rehabilitation. The 2-week and late discharge CRFs also collected details on any aids or adaptations required and any change of place of residence (e.g. own home to residential care home) at discharge relative to baseline. Resource use data were also to be collected in the 12-week patient questionnaire, from patients with capacity only. This included information on any readmittance to hospital, outpatient care received, any additional medications, aids or adaptations since discharge and return to work.

Patient preference data

Patient preference data were collected for patients invited to participate in the study. Patient preference data were collected on the baseline CRF for patients who did consent to the study and on an optional, separate patient preference form for those who did not. The baseline CRF asked which treatment the patient would have preferred or if they had no preference. The patient preference form asked what the patient’s preferred study treatment was: ‘No preference’, ‘Non-operative management’ or ‘Surgical fixation’. The strength of their preference was also assessed on a scale of 1–10 (with 1 corresponding to no preference and 10 corresponding to very strong preference). Also, the patient was asked how effective they thought each treatment would be: ‘Very ineffective’, ‘Fairly ineffective’, ‘Can’t decide’, ‘Fairly effective’, ‘Very effective’. The patient preference form given to patients who did not consent to participate in the study asked them to record their reason for non-consent.

Data management

An electronic management system was used to track participant recruitment and study status as well as CRF returns. Data from CRFs were processed by administrative personnel at YTU. Data were verified through cross-checking of the data against the hard copy of the CRF. A Validation Plan for the CRFs was written by the trial coordinators and trial statistician in consultation with the YTU Data Manager. The Plan included detailed coding for the CRFs and data query resolution rules/procedures. Quality Control was applied at each stage of data handling to ensure that all data were reliable and processed correctly.

Data were handled in accordance with the Data Protection Act 2018, GDPR legislation, the latest Directive on Good Clinical Practice, and local policy.

All personal information collected about enrolled participants was stored electronically in a secure environment at the University of York, with permissions for access in line with standard operating procedures (SOPs). All paper records containing personal information such as consent forms and consultee declaration forms were stored safely in a separate compartment of a locked cabinet.

Clinical information was only looked at by responsible individuals from the study team, the Sponsor, the NHS Trust or from regulatory authorities, where it was relevant to the patient taking part in this research as he or she agreed to at the time of consent or consultee declaration.

Once randomised, patients were assigned a participant ID number. This was used on all CRFs, and individual participants were only to be referred to by their participant ID number to maintain patient confidentiality. All study data were completely anonymised for any analyses, reports and publications.

Overview

The originally planned primary analysis was a mixed-effects linear regression model, with EQ-5D-5L scores at 2 weeks’, 6 weeks’, 12 weeks’ and 6 months’ follow-up as the dependent variable, adjusting for baseline EQ-5D-5L, randomised group and other pertinent baseline characteristics as fixed effects. The plan was to control for potential clustering at hospital site level by including it in the model as a random effect and to account for the correlation of scores within patients over time by means of an appropriate covariance structure. The secondary outcomes were to be analysed by similar mixed-effects linear regression models. Mortality was planned to be analysed using a logistic regression model.

Originally, it was intended that statistical analysis would generally be conducted using two-sided hypothesis tests with a 5% significance level. However, due to poor recruitment no formal statistical hypothesis testing has been undertaken, as the achieved sample size is too low for the conclusions of these tests to be meaningful. As a result, all analyses of the primary and secondary outcome measures are descriptive in nature. Analyses are conducted on an intention-to-treat (ITT) basis, with patients being analysed in the groups to which they were randomised.

Baseline data

Baseline data of the outcome measures and information on patient characteristics at baseline are presented descriptively by randomised group and overall in a table. For continuous measures, the mean, median, SD, first and third quartiles, minimum and maximum are reported. For categorical outcome measures, the data are presented as outcome measures and percentages.

Primary analysis

Analysis of the EQ-5D-5L utility index, the primary outcome which measures HRQoL, was descriptive in nature with no formal hypothesis testing due to the small sample size. The mean and median of the EQ-5DL-5L utility index score at each follow-up period in the study are reported. In addition to the mean and median, the following summary statistics are reported for the EQ-5D-5L utility index score at follow-up period: SD, first quartile, third quartile, minimum, maximum, number of observations and number of missing observations. All reported summary statistics are given at both a treatment group and overall level.

Sensitivity and subgroup analyses

No sensitivity analyses were undertaken. Originally there was to be a subgroup analysis exploring how patient’s knowledge and experience of their allocated treatment affected the results of their primary outcome measure; this was going to be achieved by including an interaction between patient preference and treatment group in the linear mixed-effects model for EQ-5D-5L utility index score. However, this subgroup analysis was not undertaken, as there was no statistical modelling of the primary outcome.

Secondary outcomes

A descriptive analysis of secondary outcomes was undertaken due to the low sample size. There was no formal hypothesis testing or statistical modelling. All summary statistics are reported on both a treatment arm and overall level for the secondary outcomes. These include mean, SD, median, minimum, maximum for continuous variables and counts and proportions for categorical variables.

Complications and adverse events

Economic evaluation

Due to the trial stopping early, it was not possible to undertake the planned economic evaluation as outlined in the study protocol.

The planned economic evaluation was to be conducted from the recommended NHS and personal social services (PSS) perspective according to NICE guidance. 45 Data were collected on the costs and outcomes of each trial participant. Participants were asked to complete economic resource use questionnaires at 12 weeks and 6 months as well as at the optional 12-month time point. This included hospital (e.g. inpatient, outpatient, A&E) and community and social care resource used; and for the purposes of secondary analysis, costs associated with lost productivity and out-of-pocket costs. Hospital forms were designed to collect information on the cost of surgery (e.g. time in theatre, staff time, consumables and devices, nights in hospital after the procedure), complications, physiotherapy and removal of devices. Relevant UK unit costs, such as NHS Reference costs and Personal Social Services Research Unit (PSSRU) costs of health and social care, would have been applied to each resource item to value total resource use in each group.

For the planned economic evaluation, the raw EQ-5D-5L scores at baseline (today and 1 week prior to injury), 2 weeks, 6 weeks, 12 weeks and 6 months post randomisation according to domain would have been displayed, in order to examine the movements between levels for each domain according to group. The overall difference in EQ-5D-5L index scores between the two groups would have been examined through regression methods, consistent with the model selected in the statistical analysis. The EQ-5D-5L health states would have been valued using the mapping function developed by van Hout et al. (2012) in accordance with NICE recent recommendation (www.nice.org.uk/Media/Default/About/what-we-do/NICE-guidance/NICE-technology-appraisal-guidance/eq5d5l_nice_position_statement.pdf). Quality-adjusted life-years (QALYs) would have been calculated by plotting the utility scores at each of the four time points and estimating the area under the curve. 46 For the analysis, regression methods would have been used to express the incremental cost per QALYs gained. Results would have been presented using incremental cost-effectiveness ratios (ICERs) and cost-effectiveness acceptability curves (CEACs) generated via non-parametric bootstrapping. The pattern of missing data would have been analysed and handled by means of multiple imputation (MI) methods. 47 A range of sensitivity analyses would have been conducted to test the robustness of the results using different scenarios, including probabilistic sensitivity analysis. We proposed to undertake longer-term modelling if this had been deemed appropriate (i.e. there was a non-dominant situation in the trial-based evaluation). To do this, we would have undertaken a secondary analysis to explore how the differences observed during the trial evolved beyond the study. For this projection, we would have used a decision-modelling approach to extrapolate the cost-effectiveness data observed in the study to a lifetime horizon. The analysis would have been based on a combination of observed in-trial cost and HRQoL and projections of life expectancy.

If the planned economic evaluation had gone ahead, full analyses would have been detailed in a pre-specified health economics analysis plan (HEAP) that would have been signed off by the trial management team and oversight committees.

From the limited data collected, we have provided a descriptive analysis of resource use. Data of resource use during the patient’s stay in hospital up to and including discharge (which encompasses length of hospital stay, medication, rehabilitation and surgery details) and since their discharge from hospital (which includes re-admission to hospital, outpatient care received, any additional medications, aids or adaptions since discharge and return to work) are analysed descriptively by comparing summary statistics from each treatment arm. Descriptive statistics include counts and proportions for categorical variables and mean, SD, median, minimum, maximum for continuous variables (or discrete variables which have a large range of values).

Patient preference at baseline

Five participants in the non-surgical management group provided data on their preferences at baseline; two had no preference and three had a preference for INFIX surgical fixation.

EQ-5D-5L VAS

Table 7 reports the EQ-5D-5L VAS score by intervention group and overall for baseline, 2 weeks, 6 weeks, 12 weeks, 6 months and 12 months post randomisation, including the number.

PROMIS lower extremity function score

Appendix 3, Table 14 provides details of the number of missing data items for this measure and Table 8 provides a summary of the scores over time.

Timed Up and Go test

One participant randomised to the surgical fixation group provided TUG data at week 12. This participant did not use any walking aids and completed the test in 18 seconds.

PROMIS global mental health

Table 9 reports the PROMIS GMH 2a score by intervention group and overall for baseline, 2 weeks, 6 weeks, 12 weeks, 6 months post randomisation, including the number.

Pain visual analogue scale

One participant in the surgical fixation group recorded the pain VAS score across all time points up to 6 months. The score was 99 (100 = worst imaginable pain) at baseline moving to 52 across weeks 2, 6 and 12 and then reducing to 5 at 6 months. In the non-surgical management group, five participants recorded the pain VAS at baseline, three participants at week 2 and week 6, none at week 12, two participants at month 6 and one participant at month 12. The average scores across time points was 63 (SD 34) at baseline, 54 (SD 10) at week 2, 26 (SD 24) at week 6, one participant rated their pain as 1 and the other 34 at 6 months and at 12 months it was rated as 1 (0 = no pain).

Abbreviated Mental Test Score

Table 10 reports the AMTS by intervention group and overall for baseline, 2 weeks and 12 weeks, including the number.

4AT rapid test for delirium

Table 11 reports the 4AT score for intervention group and overall for baseline, 2 weeks and 12 weeks post randomisation, including the number.

Complications and adverse events

There were no recorded clinically significant displacements of the pelvic ring during the trial. There was one participant who died during the trial who was randomised to the non-surgical management group. No responses were received at week 2 for expected complications but 9 of the 11 randomised patients had data at discharge (5 non-surgical management and 4 surgical fixation). Of those nine patients, one person in the surgical group had acute renal failure with no other expected complications noted. At week 12, five responses were received (one non-surgical management and four surgical fixation) and one person in the surgical group had a urinary tract infection and delirium. One participant in the surgical fixation group had a single non-SAE within 6 months of randomisation which was considered unexpected but not related to the trial intervention. Another participant in the surgical fixation group had a lateral cutaneous nerve injury which was an AESI.

Resource use

Appendix 4, Tables 15–21 provide details of resource use during hospital stay in relation to physiotherapy rehabilitation, medication use, surgery, length of hospital stay, walking aids provided on discharge, location of discharge and medication provided to take home. Of those providing data, all participants received a physiotherapy leaflet and therapy. The average number of sessions in the non-surgical management group was 9.2 (SD 5) and in the surgical fixation group was 8.6 (SD 5.4) and for the average session duration was 31.5 (SD 8.2) and 41.3 (SD 5.4) respectively. Most participants received pharmacological venous thromboembolism (VTE) prophylaxis across both groups, but no other medication was reported being used during the hospital stay. All five participants randomised to surgery received surgical fixation with INFIX. Most participants received a general anaesthetic and had a local infiltration. In theatre the lead surgeon was a consultant with a registrar/speciality trainee or associate specialist as the first assistant present. In the non-surgical management group two patients had their length of hospital stay recorded and these were 7 and 14 days. In the surgical fixation group three patients had their length of stay recorded and the median was 24 days with a minimum of 23 and maximum of 39 days. No data were returned for the number of walking aids patients were discharged with in the surgical fixation group and in the non-surgical management group one patient was discharged with a walking frame and one with a wheelchair. The discharge location was recorded for one participant in the non-surgical management group (rehabilitation unit) and three participants in the surgical fixation group (live alone, live alone but with support from carers and early supported discharge). All participants with data were discharged with antibiotics (across both groups) and of all those in the surgical fixation group received analgesia and two of the four in the non-surgical management group received analgesia.

Appendix 4, Tables 22–26 provide details of resource use in relation to participants’ pelvic fracture since discharge from hospital, at 12 weeks post randomisation for physiotherapy, outpatient care, community care, private treatment, medication use, walking aids bought or received since discharge and days off work and unpaid activities missed. One participant in the surgical fixation group received five sessions of NHS physiotherapy in the community which lasted on average 15 minutes. One participant in the non-surgical management group received five sessions of private physiotherapy which lasted on average 45 minutes. One participant in the surgical fixation group received two orthopaedic care appointments in relation to their pelvic fracture since discharge from hospital and attended the emergency department once. They also had a single GP home visit and three GP telephone contacts. One participant in the non-surgical management group had a GP telephone contact, a visit from the district nurse and received a four-wheel walker and a shower and sink seat since discharge from hospital, at 12 weeks post randomisation. One participant in the surgical fixation group reported being unable to do unpaid activities (e.g. chores, shopping) for 91 days.

Barriers to recruitment

Actions taken to address recruitment issues

Impact of the COVID-19 pandemic on recruitment

The global pandemic had a very strongly destabilising effect on the L1FE trial which was already struggling, as the barriers to recruitment, mentioned above, were greatly exacerbated. Even after the trial restart, research teams were still depleted and in conjunction with research and development teams were required to prioritise which studies to open again. Admissions avoidance teams were strengthened, in terms of personnel and community resources, and orthopaedic clinicians were encouraged to take a more non-operative treatment approach to fracture management in general. This made it counterintuitive for clinicians to encourage patients into a trial that had operative treatment as one of the arms, where previously the management would have been entirely non-operative. Patients and relatives also became noticeably more reluctant to consider surgery within the confused mix of information and emotions that surrounded (and continue to surround) COVID-19.

Due to the patient population and the nature of the injury, a high proportion of participants lacked capacity to consent for themselves. From the study outset, comprehensive materials and guidance were in place for the inclusion of people who lacked capacity or had fluctuating levels of capacity due to injury-related delirium. Prior to the COVID-19 pandemic, approximately half of participants were patients without mental capacity to consent for themselves and obtaining consultee agreement was not a problematic aspect of the study. However, when the study reopened to recruitment in March 2021, feedback from research staff was that it had become difficult to obtain personal consultee agreement from the next of kin, as this had to be done remotely due to restrictions in place on family visits to hospital. All discussions with consultees were documented in the medical records of the patient, and where agreement was obtained in principle, they were to be posted a copy of the consultee agreement form and written information. Consultees were asked to return a signed copy of the consultee agreement form using a pre-paid envelope. When site staff contacted next of kin by telephone, they found that they were, understandably, wishing to use the contact with the hospital to obtain further details about the status of their hospitalised relative rather than discussing a research study. Due to their workload at the time, exacerbated by redeployment and prioritisation of more urgent studies such as vaccine trials, research staff said they found it difficult to find the time that these intensive discussions with next of kin (and preparation for the call) required. They also felt that next-of-kin were reluctant to engage in these discussions, when the person on the other side of the phone was someone they had not met or who was not part of their relative’s regular care team. Site recruitment/research staff felt that when family were allowed into hospital, it was easier to build trust, as family members were more familiar with the team. However, when this took place remotely, it felt out of context and it was difficult to build rapport. There was a perception amongst staff that patients’ next of kin were also reluctant to agree inclusion on behalf of the patient when they were unable to see their relative, have a better idea of how they were doing or get a better sense of the study.

The Admissions Avoidance schemes were much more active during the COVID-19 pandemic, with additional resources made available to them resulting in fewer elderly patients with injuries from falls being managed in hospital. At some sites, Admissions Avoidance teams supported patients through alternative community pathways such as rehabilitation or respite, to prevent hospital admission or to discharge them in as timely a manner as possible, which could happen before the evaluation of ability to mobilise 72 hours following injury, required for the trial.

Additionally, there was a perception by recruiting staff and patients when the study reopened that the surgical option would prolong the patient’s hospital stay, worsening issues with presenting the study treatments in a balanced way and adversely impacting on patient preference and willingness to participate.

Furthermore, research staff at some sites reported that the COVID-19 pandemic had a large impact on the emotional and mental state of patients. They thought that patients were more fearful, wanting to go home rather than stay in hospital, or choosing to wait at home longer before deciding to come into hospital for treatment. Staff observed that restrictions on family visits had an immense impact on delirium and the associated confusion, which increased due to a lack of familiar faces. On the whole, this had an impact on what patients were prepared to put themselves through, or what their next of kin were willing to agree to.

It was reported by site research staff that there was reduced operating theatre space, site and bed capacity, and surgeon availability, as well as team capacity issues. During the peak of the COVID-19 pandemic, many research teams had been redeployed to alternative areas such as clinical care. At the time we reopened sites for recruitment to the L1FE study, most research staff members had returned to their previous roles; however, there were still issues with capacity due to the accumulated backlog. Numerous studies that had also paused due to the COVID-19 pandemic were also restarting, there were ongoing COVID-19 vaccine studies and new studies opening, all contributing to staff capacity problems.

The eligibility criteria

A key factor was that more patients than expected were able to mobilise independently or with supervision to 3 m and back at 72 hours post injury. This was an important eligibility criterion to avoid unnecessary surgeries being carried out on patients who would have been able to mobilise anyway. Patients who were still unable to mobilise at 72 hours were at risk of immobility-associated complications, and they were the ones who had the potential to benefit from surgical management. Reliable epidemiological data were not available prior to the trial on the proportion of over 60-year-olds with an LC-1 pelvic fracture able to mobilise at 72 hours post injury. We found that just over half of the screened patients were able to mobilise independently to 3 m and back at this stage. Based on feedback from sites, there was a concern that the ability to mobilise independently was being assessed differently across sites and across physiotherapists, with different levels of assistance provided. The protocol was amended to make explicit that use of a walking aid and verbal guidance were permitted, but physical assistance was not. A video was also developed to support research physiotherapists in their assessment of this inclusion criterion. There was a perception amongst some recruiting staff that the focus on patient mobilisation at 72 hours and an awareness that this would be assessed increased awareness of pain management, an important component of current standard care. There was anecdotal feedback from sites that these factors were related to lack of equipoise amongst some physiotherapists and a concern to help ensure patients avoided surgery. However, we recognise that the situation may simply be that more patients are able to independently mobilise at 72 hours post injury than anticipated. Opinions from site research staff on surgeon equipoise were mixed. While some reported that they did not believe this to be a problem, others reported examples which indicated that equipoise was an issue for some surgeons.

We had physiotherapist involvement in the trial team and consulted with physiotherapists at sites. Similarly, we had orthogeriatrician involvement. However, on reflection, the majority of our work at the grant application stage focused on surgeons and their willingness to be involved in the trial. If a trial in the future was under consideration, factors affecting equipoise in a wider range of healthcare professionals involved in the patient pathway should be considered at an earlier stage.

Patient pathway

It is likely that a reasonable number of potentially eligible patients were not being screened. At the design stage of the trial, based on a review of records from two of the centres, there were 312 patients admitted with low-energy fractures over 29 months (November 2013 to March 2016) aged 60 years or older. Therefore, the numbers being screened are lower than expected. It is not surprising to find this gap, given the complex admissions pathway for this group of patients. Patients often arrived through the emergency department, making it important for A and E nurses and radiographers to be aware of the study. Additionally, it was helpful if site recruitment/research staff regularly screened the PACS, to identify potential participants being admitted via the emergency department, as well as screening on-call admissions lists from orthopaedic and care of the elderly teams. Upon leaving the emergency department, patients could then be admitted to the Geriatric ward, Orthopaedic ward, Medical ward or an Admissions unit.

We were aware of this at the outset and each research team was briefed at the SIV about the need to identify patients for the trial beyond the orthopaedic department. Other resources and ongoing support were provided during the trial. However, this remained an ongoing issue at several centres. Permeating knowledge of the trial around multiple departments with regular repeated sessions to accommodate staff changeover (i.e. between shifts as well as staff turnover) was an ongoing challenge that many sites found resource-intensive and draining, particularly following the COVID-19 pandemic.

Consent from patients

Just over 30% of eligible patients (11 of 36) who were approached consented to participate in the trial (or a consultee agreed on their behalf). This compounded the impact of the high number of patients not meeting the eligibility criteria. The most common reason for eligible patients not consenting to take part was that they were unwilling to be randomised to a treatment (n = 10). It is difficult to determine the extent to which the consent rate could have been improved from the information we have available. Making a decision about participating in a trial where the interventions are very different in intensity and risks is potentially a daunting one for anyone and this is likely to be especially so in the population for the L1FE trial where the average age of participants was 83 years and people may be frail and/or have dementia.

COVID-19 pandemic

The internal pilot had been running for 7 months when recruitment was paused due to the COVID-19 pandemic in March 2020. The recruitment challenges were apparent by this stage and we used the pause in recruitment to develop strategies to address these, as outlined in Chapter 3. Despite adjustments and mitigations to address the various challenges, and a short extension to the pilot phase agreed by the funder, we found recruitment barriers to be exacerbated by the ongoing impacts of the pandemic when we reopened to recruitment in March 2021. Even aspects that had been functioning relatively well beforehand became difficult. In particular, admissions avoidance teams were strengthened and orthopaedic clinicians were encouraged to take a more non-operative treatment approach to fracture management in general. This made it particularly difficult for the L1FE trial where the existing standard care was non-operative management. It also became more difficult for staff at sites to protect time for the consent process for L1FE which was considered time intensive. Where patients lacked capacity and consultee agreement was required, the restrictions on family visits became a barrier to building trust and rapport with next of kin. Staff observed that restrictions on family visits had an immense impact on delirium and the associated confusion, which increased due to a lack of familiar faces. On the whole, this had an impact on what patients were prepared to put themselves through, or what their next of kin were willing to agree to.