Evidence library

REBOA did not reduce mortality; the Bayesian analysis showed a high probability of harm, with more deaths, many occurring early from bleeding.

Both groups improved substantially, with no statistically significant difference in the primary outcome at 52 weeks between replacement and arthrodesis.

Pain at 3 days was equivalent between the bandage and rigid immobilisation groups in both intention-to-treat and per-protocol analyses.

Reconstruction gave a statistically significantly better KOOS4 at 18 months than rehabilitation. Around two-fifths of the rehabilitation group went on to have reconstruction within 18 months, so most avoided surgery in that period.

There was no significant difference in WOMAC at 18 months; single-stage patients had better function early after surgery, and single-stage revision was less costly.

There was no significant difference in PRWE at 12 months between K-wires and casting, though a minority of cast patients later needed surgery for loss of position.

Disability at 18 weeks did not differ significantly between injection and microdiscectomy, and injection was the cheaper first-line option on economic evaluation.

Plaster casting was not superior to a removable brace: ankle function at 16 weeks did not differ significantly between groups, and secondary analyses were consistent.

There was no significant difference in ankle function at 16 weeks between cast and brace, and no clinically important differences at other time points; casting was not superior to functional bracing.

Progressive exercise was not superior to a single advice session at 12 months. Injection gave a short-term benefit at 8 weeks but not at 12 months.

There was no statistically significant or clinically important difference in OMAS at 16 weeks between brace and cast, nor in secondary analyses.

Replaced PRISMA 2009 with an expanded 27-item checklist, an abstract checklist and revised flow diagrams for both new and updated reviews, reflecting advances in how studies are identified, selected, appraised and synthesised.

There was no clinically important difference in PRWE at 12 months between plating and casting, and no clinically important differences in quality of life.

Non-inferiority of 6 weeks was not shown: persistent infection was more frequent with 6 weeks than 12 weeks, with the difference most marked in patients treated by DAIR.

Tourniquet use was associated with a higher risk of serious adverse events and slightly more early postoperative pain, with no meaningful gains in function and only a modest shortening of operating time.

Tourniquet use was associated with a higher risk of serious adverse events (such as venous thromboembolism and infection) and slightly worse early postoperative pain, with little or no functional benefit.

There was no clinically or statistically significant difference in PRWE at one year between early fixation and casting. Complications potentially linked to surgery were more frequent in the fixation group, whereas the cast group needed occasional later fixation for non-union.

Accelerated surgery did not significantly reduce either mortality or the composite of major complications at 90 days. Some secondary outcomes, such as delirium and time to mobilisation, favoured accelerated surgery.

There was no clinically important difference in DASH at 12 months; bracing had a higher non-union rate, with some patients later converted to surgery.

There was no significant difference in ATRS at 9 months between the brace and cast groups, and re-rupture rates were similar. The brace was not shown to be clinically superior.

All three groups improved substantially; differences between them did not reach clinical importance. Capsular release carried more risk and cost, and manipulation under anaesthesia was the most cost-effective.

PRP did not improve heel-rise endurance or patient-reported outcomes compared with placebo at 24 weeks.

Arthroscopy produced significantly better HOS-ADL scores at 8 months than physiotherapy and activity modification, with a difference around the prespecified minimum clinically important difference.

There was no significant difference in the rate of secondary procedures at 24 months. THA gave a small functional and quality-of-life advantage that was unlikely to be clinically important, while dislocation was more common after THA.

There was no significant difference in Oxford Knee Score at 5 years, with similar reoperation and complication rates; partial knee replacement was more cost-effective and showed small advantages in some secondary measures.

Quality of life at four months did not differ significantly between implants, and any difference was below the minimum clinically important threshold. Mortality and mobility were also similar.

There was no clinically relevant difference in pain between decompression and diagnostic arthroscopy at two years.

Both surgical arms improved slightly more than no treatment, but decompression gave no clinically important benefit over arthroscopy alone. The difference between the two surgical groups was not significant.

Both groups improved, but hip arthroscopy gave significantly better iHOT-33 scores at 12 months than Personalised Hip Therapy, a difference judged to be clinically meaningful.

Both groups improved, but there was no significant difference between surgery and physical therapy at 2 years; many patients allocated to physical therapy crossed over to surgery, limiting interpretation.

Physical therapy was non-inferior to arthroscopic partial meniscectomy for patient-reported knee function over 24 months. Some physiotherapy patients crossed over to surgery for persistent symptoms.

Robotic TKA was associated with significantly less early postoperative pain and analgesia requirement, faster attainment of straight leg raise and range of motion, and shorter hospital stay.

There was no significant difference in disability at 12 months, and no significant reduction in deep infection with negative pressure wound therapy.

Romosozumab-then-alendronate significantly reduced new vertebral, clinical and hip fractures compared with alendronate alone.

There was no significant difference in Oxford Shoulder Score, re-tear rate or complications between arthroscopic and open repair.

Disability at 6 months did not differ significantly between plate and nail. Nailing gave significantly better function at 3 months.

The panel made a strong recommendation against arthroscopic surgery for almost all patients with degenerative knee disease, as any benefit was very small and short-lived.

Pregabalin did not significantly reduce leg pain compared with placebo at 8 weeks or at 1 year, and caused more adverse events such as dizziness.

No significant difference in DASH or other functional scores at any point; the operative group had an unacceptably high complication rate and the trial was stopped early for this reason.

Reoperation at 24 months did not differ significantly between implants, but avascular necrosis was significantly more common with the sliding hip screw. Exploratory subgroups suggested possible benefit from the sliding hip screw in smokers and in displaced or basicervical fractures.

Close contact casting gave ankle function equivalent to surgery on OMAS at 6 months. Casting had fewer infection and wound problems but more radiological malunion, and a minority of casting patients converted to surgery after early loss of reduction.

Return to work did not differ between groups, but early weight bearing gave better ankle range of motion and early functional outcome at 6 weeks, without an increase in complications.

Romosozumab significantly reduced new vertebral fractures at 12 months, with the benefit maintained after switching to denosumab; non-vertebral fracture reduction was not significant.

There was no clinically relevant or significant difference in KOOS4 at 2 years, and the exercise group had better thigh muscle strength in the short term.

Simple decompression and transposition gave similar clinical improvement, with more wound complications after transposition; evidence comparing endoscopic with open release and for conservative measures was of low quality, although advice on avoiding prolonged elbow flexion and pressure may help mild cases.

LIPUS did not improve functional recovery or speed radiographic healing compared with sham; the trial was stopped early for futility.

The posterior approach significantly reduced Trendelenburg gait and stem malposition compared with the lateral approach, with non-significant trends towards less dislocation and heterotopic ossification; neither approach gave better function.

There was no significant difference in Oxford Shoulder Score between surgical and non-surgical treatment over two years.

There was no significant difference in 24-hour or 30-day mortality, but the 1:1:1 group achieved haemostasis more often and had fewer deaths from exsanguination in the first 24 hours, without more complications.

TKR plus non-surgical treatment produced significantly greater improvement in pain, function and quality of life at 12 months, but with more serious adverse events; most in the non-surgical group improved and many had not undergone TKR by 1 year.

No procedure showed clear superiority over another, including over trapeziectomy alone, for pain, function, quality of life or adverse events; most trials had unclear risk of bias.

Surgery was associated with a lower risk of recurrent dislocation, but the evidence was of low or very low quality and any difference in patient-reported function was uncertain.

Joint-preserving procedures are indicated before femoral head collapse, with good mid- to long-term results in several series; once collapse has occurred, total hip arthroplasty gives excellent long-term outcomes, whereas resurfacing and hemiresurfacing have performed poorly.

The redesigned pathway, with leaflet-supported direct discharge of minor stable injuries, did not increase unplanned ED reattendance, shortened consultations where backslabs were replaced by removable splints, and reduced unnecessary face-to-face fracture clinic review.

Kinematically aligned knees had significantly better pain relief, Oxford, WOMAC and Knee Society scores and greater flexion at 2 years than mechanically aligned knees.

Tranexamic acid was associated with significantly lower transfusion rates, with no increase in thromboembolic events, acute renal failure or other complications.

There was no clinically relevant or statistically significant difference in PRWE at 3, 6 or 12 months, nor in quality of life or complication numbers.

There was no significant difference in Kerr-Atkins score at 2 years, and the operative group had more complications and reoperations.

The cast excluding the thumb produced significantly greater extent of union on CT at 10 weeks; overall union was almost universal and there were no significant differences in motion, grip, wrist scores or pain.

All groups improved, with no significant difference between PRP, glucocorticoid and saline at 3 months. Glucocorticoid gave better pain relief at 1 month and reduced tendon thickness and Doppler activity.

At 5 years there was still no significant difference in KOOS4, other patient-reported outcomes or radiographic osteoarthritis between groups. About half of the rehabilitation-first group had still not needed reconstruction.

Both groups improved, with no significant difference between partial meniscectomy and sham surgery for any primary outcome at 12 months, and no significant difference in subsequent knee surgery or serious adverse events.

Methylprednisolone significantly improved symptoms at 10 weeks compared with placebo and modestly reduced the rate of surgery at one year, but most patients in all groups still went on to have surgery.

Patients given corticosteroid had significantly worse 1-year recovery and higher recurrence than those given placebo injection. Adding physiotherapy to steroid did not offset this harm.

Plate fixation markedly reduced non-union and gave slightly better DASH and Constant scores at 1 year, but these functional differences disappeared once patients with non-union were excluded.

Recurrence was significantly and substantially more frequent and earlier after needle fasciotomy than after limited fasciectomy, although PNF had faster recovery and fewer complications.

No significant difference in patient-reported outcome between techniques; the double-incision group had a small gain in flexion strength. Lateral antebrachial cutaneous nerve neuropraxia was more frequent with the single anterior incision.

There were no significant differences in pain, grip or pinch strength or movement between the three techniques; good pain relief was maintained over time, although pinch strength declined regardless of procedure.

Percutaneous and open release both resolved triggering in all cases and were similarly effective, while steroid injection had lower cure rates even when a second injection was allowed.

Only around half of binders sat at trochanteric level, and placement too high was the commonest error. In open-book injuries, high binders left a significantly wider symphyseal gap than correctly placed ones.

Plating gave better early PRWE and DASH scores and better grip strength throughout, but by 12 months there was no significant difference in range of motion, pain, PRWE or DASH; anatomical reconstruction did not translate into better motion or daily function.

There was no significant difference in Oxford Knee Score or any other outcome at 5 years, including reoperation and later patella-related surgery, and total healthcare costs did not differ between resurfacing and non-resurfacing.

Treatment within 1 hour, and between 1 and 3 hours, significantly reduced bleeding deaths, whereas treatment after 3 hours appeared to increase the risk of death from bleeding.

86% of evaluable patients responded, including every patient assessed histologically (near-elimination of giant cells); serious adverse events were uncommon and none was judged treatment related.

Uncemented THA had significantly lower revision-free survival than cemented THA, mainly because of poorer uncemented cup performance; uncemented stems did better than cemented stems, though unrecognised intraoperative femoral fractures may cause early stem failure, and the commonest modern uncemented cups performed similarly to cemented cups.

There was no significant difference in KOOS4 change at 2 years between the two strategies; around 40% of the rehabilitation-first group had delayed reconstruction, so most avoided surgery without a worse outcome.

Re-rupture rates were low and not significantly different between groups, and there was no significant difference in strength, function or range of movement. Surgery carried more complications, mainly wound-related.

The PRP group had significantly more patients with successful pain and function outcomes at 1 year; the steroid group improved early and then declined.

Primary repair gave significantly better Constant and ASES scores, pain-free abduction and pain relief at one year. Roughly a fifth of the physiotherapy group went on to secondary repair because of insufficient improvement.

Both groups improved, with no significant difference in VISA-A between PRP and saline at 24 weeks.

Corticosteroid gave better early relief but its effect faded, whereas the PRP group had significantly more successful outcomes at 1 year.

TEA gave significantly better Mayo scores and shorter operations, and several ORIF patients were converted intraoperatively to TEA because fixation was not achievable.

A short period in a below-knee cast gave a faster recovery than tubular bandage, with the Aircast brace also beneficial; the Bledsoe boot did not show benefit. Differences narrowed by 9 months.

Denosumab significantly reduced new vertebral fractures (by about two thirds), with significant reductions in hip and non-vertebral fractures.

Collagenase achieved the primary end point in significantly more joints than placebo, with better results at the MCP than the PIP joint; adverse events were mostly local (swelling, bruising, skin tears), with rare tendon rupture.

Surgery gave significantly greater improvement in function and symptoms at 12 months, although both groups improved and the size of the difference was modest.

Pain relief, grip strength and satisfaction were broadly similar between the two salvage procedures; four-corner fusion had more complications, while proximal row carpectomy carried a higher risk of later radiocapitate arthritis.

Adding arthroscopic surgery gave no significant additional benefit in WOMAC score at 2 years over physical and medical therapy alone.

In as-treated analysis surgery gave significantly greater improvement in pain and function at 2 years; intention-to-treat analysis showed a significant benefit only for bodily pain because of crossover.

Reamed nailing reduced events in closed fractures, an effect largely due to fewer autodynamisations, with no clear difference in open fractures. The trial also showed that delaying reoperation for nonunion for at least 6 months reduces unnecessary procedures.

Early arthroscopic Bankart repair significantly reduced recurrent instability and improved shoulder-specific function compared with lavage.

Describes how THR transformed care of elderly arthritic patients, that both cemented and uncemented fixation can be durable, and how newer bearings, less invasive surgery and computer navigation aim to meet younger patients' demands, with cost-effectiveness governing future developments.

There was no significant difference in clinical scores or nerve conduction between transposition and simple decompression.

Packing was associated with rapid haemorrhage control, reduced post-procedure transfusion and low mortality from acute haemorrhage, with only a minority needing subsequent angioembolisation.

Zoledronic acid significantly reduced new clinical fractures (by about a third) and was associated with a significant reduction in all-cause mortality.

Proposes that successful healing needs all four elements together, so nonunion treatment should identify and address each deficient element rather than relying on a single adjunct.

Plate fixation gave significantly better Constant and DASH scores, faster union and lower rates of non-union and symptomatic malunion than sling treatment.

Drains gave no significant benefit for infection, haematoma or reoperation, but were associated with a significantly higher blood transfusion requirement; undrained wounds needed more dressing reinforcement.

Neither group had a major loss of reduction or an iatrogenic ulnar nerve injury; mild loss of reduction, radiographic angles, alignment and elbow motion did not differ significantly between lateral-entry and crossed pinning.

Zoledronic acid markedly and significantly reduced vertebral fractures (around 70%) and significantly reduced hip fractures (around 40%) versus placebo.

Early surgery produced faster relief of leg pain and faster perceived recovery, but by 1 year outcomes were similar between groups; a substantial proportion of the conservative group eventually had surgery.

Injection gave the best short-term results at 6 weeks, but at 1 year it was significantly worse than physiotherapy and wait and see, with high recurrence. Physiotherapy was better than wait and see early on, with little difference by 1 year.

Splinted children had better physical function at 2 weeks and less difficulty with daily activities such as bathing than those in casts, with similar pain and no refractures in either group.

Surgery plus radiotherapy significantly improved the proportion able to walk and the duration of walking, and more patients who were non-ambulant at presentation regained walking; corticosteroid and opioid needs were also lower.

Outcomes were similar, but transposition had significantly more complications.

Clinical outcomes did not differ significantly between the two operations, but complications were significantly more frequent after transposition.

Screw fixation gave substantially shorter times to union and return to sport, while casting had a high failure rate from non-union, delayed union and refracture.

Fusion produced only a small difference in Oswestry score of borderline significance, with no meaningful advantage in walking, while carrying surgical risk and greater cost.

Early fixation offered no lasting functional advantage over casting by one year, although some early benefits in function were seen. Non-unions in the cast group united after subsequent fixation, and surgery carried its own complications.

Significantly more surgical patients were ambulatory after treatment (84% versus 57%) and they retained walking ability for longer, with reduced steroid and opioid requirements.

Most nonunions united, and those that failed to heal had received significantly fewer and less concentrated progenitor cells, suggesting a dose threshold.

Radial nerve palsy complicated about 12% of humeral shaft fractures, most often middle and middle-distal transverse or spiral fractures; most recovered spontaneously with conservative care, and final outcomes were similar whether nerves were explored early or managed expectantly.

Most elbows regained concentric stability and a functional arc of motion, though reoperation for stiffness or complications was not rare.

Children over 8 years at onset with lateral pillar B or B/C border hips had significantly better outcomes with surgery, whereas younger children with these groups did well regardless of treatment, and lateral pillar C hips did poorly whatever the treatment.

Very few feet required extensive corrective surgery; relapses were strongly associated with poor adherence to foot abduction bracing.

Both groups improved at 2 years with no clinically important difference in clinical outcome; microfracture showed better SF-36 physical component improvement, and histology did not show clear superiority of ACI.

ACI gave better functional results and a higher proportion of excellent or good arthroscopic repairs than mosaicplasty at 1 year; mosaicplasty performed particularly poorly in patellar defects.

Introduced a five-level hierarchy, from high-quality RCTs (level I) to expert opinion (level V), applied separately to therapeutic, prognostic, diagnostic and economic studies.

The higher dose of rhBMP-2 significantly reduced secondary interventions, accelerated healing and reduced infection rates, particularly in Gustilo type III injuries.

Most patients achieved good or excellent hip scores with a low revision rate for acetabular component loosening, supporting acute arthroplasty as a viable option in selected patients.

Surgery gave significantly higher success rates than splinting at all follow-up points, and many patients in the splint group went on to request surgery.

Simple partial head fractures did well, but comminuted fractures with more than three fragments had a high rate of early failure, nonunion and poor results.

Preoperative radiotherapy roughly doubled wound complications (35% versus 17%, significant); tumour size and site were also risk factors, and overall survival was marginally better with preoperative treatment.

Surgery gave significantly less pain and better function and satisfaction at 12 months than orthoses or waiting. Orthoses gave only short-term symptom relief.

The large majority of fractures united with bracing, with non-union uncommon and residual angulation usually functionally and cosmetically acceptable.

Overall sensory recovery was similar between groups; conduits gave better two-point discrimination than end-to-end repair for small gaps and outperformed autograft for gaps of 8 mm or more.

Patients managed with external fixation were more severely injured, yet the staged approach was safe with low conversion complication rates, supporting its use in unstable patients.

Among patients completing the study, cure was significantly more frequent with the rifampicin combination than with ciprofloxacin alone.

Alendronate roughly halved new radiographic vertebral fractures and also significantly reduced hip and wrist fractures compared with placebo.

Useful elbow flexion was restored in the reported cases without significant loss of ulnar nerve function in the hand.

Most femoral condyle lesions had good or excellent clinical results with repair tissue resembling hyaline-like cartilage on biopsy; results for patellar lesions were noticeably poorer.

Showed that large acetabular corrections were achievable through a single approach with the posterior column left intact, allowing early mobilisation without external fixation.

Argued that stable (type A) injuries do well without surgery, whereas unstable injuries, particularly vertically unstable posterior ring disruptions, are associated with persistent pain and deformity and benefit from anatomical reduction and stable fixation.

Surgical repair gave no advantage over non-operative treatment.

Each tissue tolerates a different level of strain: granulation tissue tolerates high strain, cartilage intermediate and lamellar bone very little, so bone can only bridge the gap once strain falls below its tolerance.