Interventions for treating fractures of the patella in adults

Abstract

Background

Fractures of the patella (kneecap) account for around 1% of all human fractures. The treatment of these fractures can be surgical or conservative (such as immobilisation with a cast or brace). There are many different surgical and conservative interventions for treating fractures of the patella in adults. This is an update of a Cochrane Review first published in 2015.

Objectives

To assess the effects (benefits and harms) of interventions (surgical and conservative) for treating fractures of the patella in adults.

Search methods

We searched CENTRAL (2020, Issue 1), MEDLINE, Embase, LILACS, trial registers and references lists of articles to January 2020.

Selection criteria

We included randomised controlled trials (RCTs) or quasi‐RCTs that evaluated any surgical or conservative intervention for treating adults with fractures of the patella. The primary outcomes were patient‐rated knee function, knee pain and major adverse outcomes.

Data collection and analysis

At least two review authors independently selected eligible trials, assessed risk of bias and cross‐checked data extraction. Where appropriate, we pooled results of comparable trials.

Main results

We included 11 small trials involving 564 adults (aged 16 to 76 years) with patella fractures. There were 340 men and 212 women; the gender of 12 participants was not reported. Seven trials were conducted in China and one each in Finland, Mexico, Pakistan and Turkey. All 11 trials compared different surgical interventions for patella fractures. All trials had design flaws, such as lack of assessor blinding, which put them at high risk of bias, potentially limiting the reliability of their findings. No trial reported on health‐related quality of life, return to previous activity or cosmetic appearance. The trials tested one of seven comparisons. In the following, we report those of the main outcomes for which evidence was available for the three most important comparisons.

Four trials (174 participants) compared percutaneous osteosynthesis versus open surgery. Very low‐quality evidence means that we are uncertain of the findings of no clinically important difference between the two interventions in patient‐rated knee function at 12 months (1 study, 50 participants) or in knee pain at intermediate‐term follow‐up at eight weeks to three months. Furthermore, very low‐quality evidence means we are uncertain whether, compared with open surgery, percutaneous fixation surgery reduces the incidence of major adverse outcomes, such as loss of reduction and hardware complications, or results in better observer‐rated knee function scores.

Two trials (112 participants) compared cable pin system (open or percutaneous surgery) versus tension band technique. The very low‐quality evidence means we are uncertain of the findings at one year in favour of the cable pin system of slightly better patient‐rated knee function, fewer adverse events and slightly better observer‐rated measures of knee function. There was very low‐quality evidence of little clinically important between‐group difference in knee pain at three months.

Very low‐quality evidence from two small trials (47 participants) means that we are uncertain of the findings of little difference between biodegradable versus metallic implants at two‐year follow‐up in the numbers of participants with occasional knee pain, incurring adverse events or with reduced knee motion.

There was very low‐quality and incomplete evidence from single trials for four other comparisons. This means we are uncertain of the results of one trial (28 participants) that compared patellectomy with advancement of vastus medialis obliquus surgery with simple patellectomy; of one quasi‐RCT (56 participants) that compared a new intraoperative reduction technique compared with a standard technique; of one quasi‐RCT (65 participants) that compared a modified tension band technique versus the conventional AO tension band wiring (TBW) technique; and of one trial (57 participants) that compared adjustable patella claws and absorbable suture versus Kirschner wire tension band.

Authors' conclusions

There is very limited evidence from nine RCTs and two quasi‐RCTs on the relative effects of different surgical interventions for treating fractures of the patella in adults. There is no evidence from trials evaluating the relative effects of surgical versus conservative treatment or different types of conservative interventions.

Given the very low‐quality evidence, we are uncertain whether methods of percutaneous osteosynthesis give better results than conventional open surgery; whether cable pin system (open or percutaneous surgery) gives better results than the tension band technique; and whether biodegradable implants are better than metallic implants for displaced patellar fractures.

Further randomised trials are needed, but, to optimise research effort, these should be preceded by research that aims to identify priority questions.

Plain language summary

Treatments for broken kneecaps in adults

Broken kneecaps (patella fractures) account for 1% of all fractures. There are many treatments for these fractures and they can be treated with surgery or conservatively (any treatment where surgery is not used). Conservative interventions include cast immobilisation, knee brace and immobilisation by traction. Surgery can be open, percutaneous (minimally invasive surgery that uses small incisions) or arthroscopic (using two mini‐incisions with the help of an internal camera). The implants used to fix the fracture can be metallic or non‐metallic implants, and can be wires, screws, plates, threads, strings, suture buttons, external fixators, rods, nails and combinations of these.

Aim of review

This review aimed to evaluate the effects of different methods for treating kneecap fractures in adults, with or without surgery. The main outcomes we were interested in were patient‐rated knee function, knee pain and complications (adverse events).

Search results and quality of the evidence

We searched the scientific literature to January 2020 and found 11 relevant studies with 564 participants. Participants in these studies were aged between 16 and 76 years. There were 340 men and 212 women; the gender of 12 participants was not reported. Seven trials were conducted in China and one each in Finland, Mexico, Pakistan and Turkey. All 11 studies compared different types of surgery or surgical devices. Thus, we found no studies comparing different types of conservative treatment or surgery versus conservative treatment.

The 11 studies made seven comparisons. We judged the evidence available for each comparison was of very low quality. This was mainly because all the trials had design flaws that put them at high risk of bias and the studies were also small with few events.

What the included studies found

None of the studies reported on health‐related quality of life, return to previous activity or cosmetic appearance.

We report here the findings for the three most important comparisons.

Four studies compared methods of percutaneous fixation (surgery using small incisions to insert the fixation devices) with open surgery (involving wide incisions). One study found no important difference between the two methods in patient‐rated knee function at 12 months. Pooled data showed little difference between groups in knee pain at around two to three months. There were fewer adverse events in the percutaneous group and better observer‐rated knee function scores at 12 months.

Two studies compared cable pin system (open or percutaneous surgery) with tension band technique. These found slightly better patient‐rated knee function at one year, fewer adverse events and slightly better observer‐rated measures of knee function in the cable pin group. There was little important difference between the two groups in knee pain at three months.

Two studies comparing biodegradable (non‐metallic) versus metallic implants found little difference in reported outcomes (knee pain, adverse events and knee motion) between the two groups. Neither study reported patient‐rated function.

There was very low‐quality and incomplete evidence from single trials for four other comparisons.

The very low‐quality and incomplete evidence from single trials testing the four other comparisons of different surgical methods meant that we are uncertain of the results for these.

Conclusions

The very low quality of the evidence for the three key comparisons and four other comparisons of difference methods of surgery means that we are uncertain of the findings. Thus, the available evidence is insufficient to draw firm conclusions about the best method of treatment for kneecap fractures. Further research is warranted and should be preceded by research to determine which questions should be prioritised.

Summary of findings

Summary of findings 1. Percutaneous patellar osteosynthesis versus open surgery for treating fractures of the patella in adults.

Percutaneous patellar osteosynthesis compared with open surgery for treating fractures of the patella in adults
Patient or population: adults with fractures of the patella Settings: hospital (tertiary care) Intervention: percutaneous patellar osteosynthesis (minimally invasive surgery that uses small incisions and devices or implants as tension band, screws and others) Comparison: open surgery (traditional surgery that uses wide incisions and traditional devices or implants as tension band, screws, wires and others)
Outcomes	Illustrative comparative risks* (95% CI)		Relative effect (95% CI)	No of participants (studies)	Quality of the evidence (GRADE)	Comments
	Assumed risk	Corresponding risk
	Open surgery	Percutaneous patellar osteosynthesis
Patient‐rated knee function score assessed using the Lysholm Knee Scoring Scale (0–100; 100 = best function). Follow‐up: long term (12 months)	The mean Lysholm score in the open surgery group was: 89.8 points	The mean Lysholm Knee Scoring Scale in the percutaneous surgery group intervention group was 3.40 higher (0.26 lower to 6.54 higher)	—	52 (1 study)	⊕⊝⊝⊝ Verylowa,b,c	An injury‐specific MCID for this outcome is not available. One MCID for anterior cruciate ligament reconstruction is 10 points. While the percutaneous group had higher Lysholm scores at 12 months, the between group difference is clinically unimportant.
Anterior knee pain measured using VAS: 0 to 10 cm (10 cm = worst score) Follow‐up: intermediate (range 8 weeks to 3 months)	The mean pain VAS score ranged across the open surgery groups from 2.6 to 4.5	The mean pain in the percutaneous surgery groups was 1.88 lower (1.42 to 2.34 lower)	—	183 (4 studies)	⊕⊝⊝⊝ Verylowa,b,c	This is not a clinically important effect as it is less that an MCID of 3.0 cm (a clinically important difference in pain severity that corresponds to a person's perception of adequate pain control in an acute setting).
Major adverse outcome: total events Follow‐up: long term (12 months)	Study population		RR 0.26 (0.14 to 0.46)	145 (3 studies)	⊕⊝⊝⊝ Verylowa,b,c	The reported individual events differed among the 3 studies and thus the derivation of this outcome (number of participants with > 1 adverse events) varied across studies. Adverse events reported were failure of treatment reduction requiring change in procedure, failure of treatment (loss of reduction), infection, hardware complications, delayed wound healing and soft tissue irritation symptoms.
	584 per 1000d	152 per 1000 (82 to 269)
Observer‐rated measures of knee function Knee function scores (see comment) Follow‐up: long term (12 months)	The mean difference in observer‐rated knee function in the percutaneous surgery groups was 1.01 standard deviations higher (0.08 to 1.95 higher)		SMD 1.01 (0.08 to 1.95)	124 (3 studies)	⊕⊝⊝⊝ Verylowa,c,e	Two studies used the Böstman score (0 to 30; 30 = best function) and used the Knee Society Clinical Rating System (0 to 100; 100 = best function). Heterogeneity was high (I² = 83%). An SMD of 0.2 represents a small difference, 0.5 a moderate difference and 0.8 a large difference. Thus, based on this general rule, there is a large difference between the 2 groups. However, at 1.0 and 0.5, the MDs in Böstman scores are unlikely to be clinically important. The substantial heterogeneity (I² = 83%) reflected the more extreme results of 1 trial.
Health‐related quality of life	See comment	See comment	—	—	—	None of the trials for this comparison evaluated health‐related quality of life.
Return to previous activity	See comment	See comment	—	—	—	None of the trials for this comparison evaluated return or time to return to previous activities.
Cosmetic appearance	See comment	See comment	—	—	—	None of the trials for this comparison evaluated cosmetic appearance.
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence Interval; MCID: minimal clinically important difference; RR: risk ratio; SMD: standardised mean difference; VAS: visual analogue scale (or score).
GRADE Working Group grades of evidence High quality: further research is very unlikely to change our confidence in the estimate of effect. Moderate quality: further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low quality: further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low quality: we are very uncertain about the estimate.

^aDowngraded two levels for very serious risk of bias. All studies had methodological flaws; all were susceptible to detection bias in the measurement of self‐reported outcomes due to lack of blinding of the participant and observers. All studies were at high risk of performance and reporting biases.
^bTotal number of events or number of participants (or both) were small, and thus we downgraded the quality of the evidence by one level for serious imprecision (reflecting the small sample size or wide confidence interval (or both)).
^cDowngraded one level for indirectness because of the poorly defined and poor quality outcome measures and for the percutaneous versus open surgery comparison, where novel self‐developed techniques were used in the percutaneous surgery group.
^dThe estimated risk in the control groups was a mean.
^eDowngraded one level for serious inconsistency. There was considerable heterogeneity (I² more than 75%).

Background

The patella has several important functions. It acts as protective shield for the knee joint, augments the strength of the quadriceps muscle and has a cosmetic function. These functions can be damaged by the occurrence of fractures.

Description of the condition

The patella or kneecap is one of the three bones that form the knee joint. It is the largest sesamoid bone (i.e. bone embedded within a tendon; in this case, the patellar tendon) in the human body and has a rounded triangular shape. The cartilage‐covered underside of the patella articulates with the two lower ends (condyles) of the femur (the thigh bone). The patella acts as a protective shield for the knee joint, augments the effect of the quadriceps muscle during knee extension (straightening) and has a cosmetic function (Anderson 1978; Insall 2006).

Fractures of the patella account for approximately 1% of all skeletal fractures (Court‐Brown 2006; Gwinner 2016; Larsen 2016). In a large study conducted in Denmark, Larsen 2016 reported an average incidence of patellar fractures of 13.1 per 100,000 per year between 2005 and 2014. Previously we observed that these fractures occur mainly in people aged between 20 and 50 years and around twice as often in men than women (Sayum Filho 2012). This was based on summaries in the literature and remains a current understanding for some authors (e.g. Gwinner 2016). However, both old and more recent epidemiological studies have reported different population characteristics, including age distributions that differ according to sex. For example, Boström 1972, which studied 422 people over 16 years in Sweden who had fractures between 1959 and 1964, reported the mean age of 42 years for males and 54 years for females, with a male to female ratio of 57% to 43%. Larsen 2016, reporting on 756 patellar fractures occurring between 2005 and 2014 in Denmark, reported mean ages of 46 years for males and 61 years for females, with a male to female ratio of 44% to 56%. Both Byun 2019, reporting on 1596 people with patellar fractures in Korea from 2003 to 2017, and Larsen 2016 found that fractures of the patella in the older people (aged 60 years or more) and women are increasing. In older people, particularly women, the fractures commonly result from low‐energy trauma, such as falls from standing height or less, and can be considered fragility fractures (Larsen 2016). Only 2% to 7% of fractures of the patella are open (Dy 2012; Insall 2006; Muller 1991). Fractures of the patella can be caused by two different mechanisms: direct impact (compression injury: e.g. resulting from falling on the knee) and indirect impact (traction injury: e.g. from sudden eccentric contraction of quadriceps muscle) (Sylvain 2020). However, some surgeons consider fractures associated with patellar dislocations as a third mechanism. The most common causes of patellar fractures are falls onto the knee, blows and sudden severe muscle pulls (Muller 1991).

Typically, the diagnosis of patellar fracture is made from the patient's history, physical examination and plain x‐ray. Magnetic resonance imaging or computed tomography are generally required only for more subtle injuries (Helfet 2003). Nowadays, preoperative computed tomography has gained an important role as a diagnostic tool for patellar fractures because it gives an exact assessment of the fracture and helps in the decision making for the correct treatment (Henrichsen 2018; Pesch 2019). The physical signs of a patellar fracture are deformity, inability to extend the knee, tenderness and, in some cases, a palpable gap. Most fractures are transverse (horizontally across the patella). Complications include haemorrhage (bleeding) into the knee joint and serious tears of the surrounding soft tissues of the joint (Catalano 1995; Levack 1985).

The classifications for describing these fractures are based on mechanism of injury, the degree of displacement, the fracture pattern or a combination of these descriptors. A commonly used classification based on fracture pattern includes the following descriptors: transverse, vertical, stellate (comminuted, i.e. broken into several pieces), apical, marginal and osteochondral (Insall 2006; Sturdee 2002). The fracture pattern often influences the choice of treatment.

Description of the intervention

Fractures of the patella can be treated conservatively or surgically (Insall 2006; Lotke 1981). Treatment approaches are often based on practitioner preferences and patient characteristics. Surgery should be avoided in people with high preoperative risk or in those with joint ankylosis, prior failed extensor mechanism or those who are non‐ambulatory (Melvin 2011).

Conservative intervention involves immobilisation of the leg in nearly full extension for five to six weeks using a long‐leg plaster or other type of cast or a brace. The patient is often allowed to partially weight‐bear during this time and most people will use crutches to ambulate. The leg is kept straight until fracture consolidation (healing) is evident on x‐rays (Torchia 1996). Typically, conservative treatment is used for people with less severe fractures, such as non‐displaced (still in place) fractures.

Surgical treatment generally entails reducing (realigning) the displaced fragments and fixing these together with some combination of screws, pins and wires. One fixation method developed in the 1950s by the Arbeitssgemeinschaft fur Osteosynthesefragen/Association for the Study of Internal Fixation (AO/ASIF) is the anterior tension band principle, which offers a stable construct for some fracture types (e.g. transverse fractures), allowing early mobilisation (Günal 1997; Helfet 2003). Two other surgical techniques used since the 1960s are partial and complete patellectomy (Appel 1993; Insall 2006). These entail the partial or total removal of the patella. Typically, surgical intervention is considered for fractures that have greater than 2 mm of articular displacement or 3 mm of fragment separation, in comminuted fractures with displacement of the articular surface, in osteochondral fractures with displacement into the joint, in marginal or longitudinal fractures with comminution or displacement, and in any case in which the integrity of the extensor mechanism has been lost. The objectives of surgical treatment are to obtain accurate reduction of the fracture and the joint surface, provide stable fixation to allow early range of motion (ROM) and restore the knee‐extensor mechanism.

Surgery can be percutaneous (minimally invasive surgery that uses small incisions to insert fixation devices), conventional (open) or arthroscopically assisted (with two mini‐incisions and an arthroscope) (Cerciello 2017). There are many options for skin incisions, methods of internal fixation, types of internal and external implants, and new implants such as special angle stable plate fixation, titanium cable purse string suture and suture buttons (Bukva 2019; Yang 2018). Skin incisions can be transversal, longitudinal, medial longitudinal, lateral longitudinal, percutaneous, open and arthroscope portals. Techniques described for internal fixation include tension band wiring and screw fixation. The material used for fixation may be metallic or biodegradable (Melvin 2011). The external implants can be metallic external fixators or carbon external fixators. The internal implants may be metallic or non‐metallic plates (locked or not locked), wires, metallic or non‐metallic screws, cannulated or non‐cannulated screws, threads (biodegradable or non‐biodegradable), titanium strings, nails and others (Muller 2019; Pesch 2019; Siljander 2017; Yang 2018).

How the intervention might work

Conservative intervention is usually chosen when there is integrity of the extensor mechanism (the leg can be straightened out), good preservation of articular congruity (joint surfaces fit correctly together) and lack of displacement between bone fragments. Where these conditions are not met, surgical intervention for patellar fracture has been advocated as, hypothetically, it enables the restoration of the joint anatomy, thus avoiding deformity, incongruity and an unsatisfactory outcome. However, key disadvantages of surgery include potential migration and breakage of wires and pins, erosion of the bone, failure of the fixation methods, loss of reduction, a painful or unsightly scar, development of arthralgia and arthritis or need of a second surgery to remove the fixation hardware (Helfet 2003; Smith 1997).

The advantages of the conservative interventions are avoidance of hospitalisation, surgery and anaesthesia. The disadvantages are a longer period of immobilisation, possible loss of reduction and possible stiffness resulting from fibrous adhesions in the joint (Muller 1991).

Fracture fixation may be achieved in various ways. Several implants are available and the choice should be directed by the fracture pattern. The most common method is a tension band technique; however, this is only possible after the fracture is converted into a two‐part fracture. Additional fracture fragments can be fixed using supplementary small‐fragments screws (3.5 mm) or mini‐fragments screws (less than 3.5 mm), locked or non‐locked plates, wires, threads, external fixators, rods, nails and other devices (Baran 2009; Berg 1997; Fortis 2002; Muller 2019; Wurm 2018).

In addition, evaluation of the extent of comminution is crucial; infrequently, a partial or total patellectomy is performed, such as in cases of severe and irreparable comminution. The patient should be informed of the possible need for fragment excision, which might cause loss of strength of the extensor mechanism (Donken 2009; Günal 2001; LeBrun 2012).

Why it is important to do this review

This is an update of a Cochrane Review first published in 2015 (Sayum Filho 2015). Based on evidence from five small trials at high risk of bias, Sayum Filho 2015 concluded that there was very limited evidence about the relative effects of different surgical interventions for treating fractures of the patella in adults and there was no evidence about the relative effects of surgical versus conservative treatment or different types of conservative interventions.

There are many surgical or conservative interventions for treating patella fractures in adults, each having advantages and disadvantages. Although the undisplaced acute fractures used to be treated conservatively, the current treatment of some types of patellar fracture is more invasive, involving surgery. There is limited evidence to guide the clinical practice on management of patellar fractures and, except for Sayum Filho 2015, other reviews we have identified have not restricted their included studies to randomised controlled trials (RCTs) (Dy 2012; Heusinkveld 2013a). Since our last review, the literature shows that new implants and new surgical techniques have been developed and used (Henrichsen 2018; Yang 2018). Furthermore, in the literature, there are new trials and reviews about interventions for treating fractures of the patella in adults (Pesch 2019).

Objectives

To assess the effects (benefits and harms) of interventions (surgical and conservative) for treating fractures of the patella in adults.

Methods

Criteria for considering studies for this review

Types of studies

We included RCTs that evaluated interventions (surgical and conservative) for treating fractures of the patella in adults. Quasi‐RCTs (method of allocating participants to a treatment that is not strictly random, e.g. by date of birth, hospital record number, alternation) were also eligible for inclusion.

Types of participants

We included trials with adults with one (unilateral) or two (bilateral) acute patella fractures. We included trials with other knee injuries if separate data for participants with an acute patella fracture were available.

Types of interventions

We considered all surgical and conservative interventions used for treating patella fractures.

Surgical interventions included internal fixation using anterior tension band, screw fixation, wires in cerclage, plates, external fixators, threads, Kirschner (K)‐wires, strings, rods, nails and partial or total patellectomy. Conservative interventions included cast, plaster cast and brace. The broad comparisons of interest were: different types of conservative interventions; surgical versus conservative interventions; and different types of surgical interventions.

Types of outcome measures

We sought the following outcome measures.

Primary outcomes

• Patient‐rated, preferably validated, knee function scores (e.g. Knee Injury and Osteoarthritis Outcome Score (KOOS) (Roos 1998), Lysholm Knee Questionnaire (Lysholm 1982), and Tegner Activity Scale (Tegner 1985)). A specific scoring scheme for patellofemoral disorders is the Kujala score (Kujala 1993).

• Anterior knee pain. Preference was given to reports of pain measured using validated pain scales, such as visual analogue scale (VAS) and numeric rating scale (NRS).

• Major adverse outcomes (e.g. infection, venous thromboembolism) as well as treatment failure requiring secondary unplanned intervention (e.g. operation or reoperation for unresolved non‐union or loss of reduction; problems with internal fixation devices).

Timing of primary outcome measurement

We extracted outcome data at the following time periods: short‐term follow‐up (up to six weeks following treatment); intermediate‐term follow‐up (more than six weeks and up to six months after the end of treatment) and long‐term follow‐up (greater than six months after the end of treatment).

Secondary outcomes

• Observer‐rated measures of knee function (e.g. knee range of motion (ROM), knee strength) including those measured with clinician‐rated knee scoring systems (e.g. Knee Society Clinical Rating System (Insall 1989)).

• Health‐related quality of life scores (e.g. 36‐item Short Form (SF‐36) (Ware 1992)).

• Return to previous activities (sports, manual labour, etc.) (including time to return).

• Cosmetic appearance including deformity.

• Hardware removal.

Search methods for identification of studies

Electronic searches

We searched the Cochrane Central Register of Controlled Trials (CENTRAL) (2020, Issue 1) in the Cochrane Register of Studies (CRS web), MEDLINE (Ovid MEDLINE In‐Process & Other Non‐Indexed Citations) (1946 to 14 January 2020), Embase (1980 to 15 January 2020) and Latin American and Caribbean Health Sciences Literature (LILACS) (1982 to 14 January 2020). We searched the World Health Organization International Clinical Trials Registry Platform (WHO ICTRP; apps.who.int/trialsearch/) and the ClinicalTrials.gov registry to 15 January 2020 for ongoing and recently completed studies. For this update, the searches were limited to 2014 onwards to capture only those records published since the previous search. We applied no restrictions based on language.

At the time of the search, CENTRAL was fully up‐to‐date with all records from the Cochrane Bone, Joint & Muscle Trauma Group's Specialised Register and so it was not necessary to search this separately.

Details of the search strategies used for previous versions of the review are given in Sayum Filho 2015.

In MEDLINE, a subject‐specific strategy was combined with the sensitivity‐maximising version of the Cochrane Highly Sensitive Search Strategy for identifying randomised trials (Lefebvre 2011) (see Appendix 1). Search strategies for the CENTRAL, Embase and LILACS are also shown in Appendix 1.

Searching other resources

We searched reference lists from relevant articles, reviews and textbooks for possible studies, and contacted experts in the field. 

Data collection and analysis

Selection of studies

Two review authors (JS and ML) selected and assessed potentially eligible studies for inclusion in the review using a prepiloted form. We resolved any disagreements by discussion and, if necessary, adjudication by a third review author (MJ). We obtained translations of studies published in languages other than English.

Data extraction and management

Two review authors (JS and JB) used a prepiloted data extraction form to independently collect data including methods, participants, interventions and outcomes. Any disagreements were resolved by a third review author (FM). Two review authors (JS and FM)) entered data into Review Manager 5 (Review Manager 2014). We sent requests to trial authors for additional information or data.

Assessment of risk of bias in included studies

Two review authors (JS and ML) independently assessed the included trials for risk of bias using the domain‐based evaluation described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2017). We assessed the following domains.

• Random sequence generation.

• Allocation concealment.

• Blinding of participants and personnel.

• Blinding of outcome assessment.

• Incomplete outcome data.

• Selective reporting.

• Other bias (e.g. major baseline imbalance; inappropriate influence of funders; and risk of bias associated with inexperience of surgeons and other care providers with the interventions, and differences in rehabilitation).

Each of these criteria was explicitly judged as low risk of bias, high risk of bias or unclear risk of bias (either lack of information or uncertainty over the potential for bias). We resolved disagreements between review authors regarding the risk of bias for domains by consensus.

Measures of treatment effect

We calculated risk ratios (RRs) together with 95% confidence intervals (CIs) for dichotomous outcomes. We expressed continuous outcome data as mean differences (MDs) with 95% CIs. We used the standardised mean difference (SMD) where studies used different scales to measure the same continuous outcome.

Unit of analysis issues

The unit of randomisation for the included trials was the individual participant. We were alert to potential unit of analysis issues such as those relating to reporting of outcomes at different times or where participants had multiple complications. In the Effects of interventions section (Percutaneous patellar osteosynthesis versus open surgery: major adverse outcomes), we highlighted the one instance where we may have inadvertently presented data from one trial for total complications rather than total number of participants with complications, but we suggest that the problem is unlikely to have serious implications in this case.

Dealing with missing data

We performed an intention‐to‐treat analysis to include all participants randomised to any intervention. When there was insufficient information relevant to estimate effects, such as number of participants, means, measures of uncertainty (standard deviation or error), or number of events and participants, we tried to contact the main authors of the included trials.

Where it was impossible to acquire adequate data for a forest plot (e.g. means and standard deviations), we present data in the text.

Assessment of heterogeneity

We assessed the heterogeneity of estimate effects between the included studies by visual inspection of the forest plot (analysis) along with consideration of the Chi² test for heterogeneity and the I² statistic (Higgins 2003; Higgins 2011).

We quantified the possible magnitude of inconsistency (i.e. heterogeneity) through studies using the I² statistic as follows: 0% to 40% might not be important; 30% to 60% may represent moderate heterogeneity; 50% to 90% may represent substantial heterogeneity and 75% to 100% very substantial ("considerable") heterogeneity (Deeks 2017). In cases of considerable heterogeneity (defined as I² ≥ 75%), we explored the data further by comparing the characteristics of individual studies and conducting subgroup analyses. 

Assessment of reporting biases

We had planned to draw funnel plots of primary outcomes to assess the potential publication bias; however, the small number of included trials and different comparisons precluded this form of analysis.

We assessed the presence of small‐study bias in the overall meta‐analysis by checking whether the random‐effects estimate of the intervention effect showed more benefit than the fixed‐effect estimate (Sterne 2017). We also assessed outcome reporting bias by comparing results extracted from published journal reports with results from other sources (e.g. correspondence) and by checking trial registrations (at the WHO International Clinical Trials Registry Platform) or published protocols.

Data synthesis

When considered appropriate, we pooled results of comparable groups of trials using the fixed‐effect model and 95% CIs. More rarely, we also used the random‐effects model where there was clear diversity in clinical or methodological characteristics.

Subgroup analysis and investigation of heterogeneity

There were insufficient data to perform the planned subgroup analyses by age (people older than 65 years compared with those under 65 years), type of fracture (transverse versus comminuted) or timing of surgery (immediate versus delayed (from one week to four weeks after injury)). In future updates, should these analyses be possible, we will test whether subgroups were statistically significantly different from one another using the test for interaction provided in Review Manager 5 (Review Manager 2014).

Sensitivity analysis

There were insufficient data to conduct the planned sensitivity analyses examining various aspects of trial and review methodology, including the effects of missing data and study risk of bias relating to inadequate allocation concealment and lack of outcome assessor blinding.

Summary of findings and assessment of the certainty of the evidence

We used the GRADE approach to assess the quality (synonymous with 'certainty' in this context) of the evidence related to each of the key outcomes listed in Types of outcome measures (Schünemann 2011; Section 12.2). There were four levels of evidence quality: 'high', 'moderate', 'low' or 'very low'. Quality may be downgraded due to study limitations (risk of bias), imprecision, indirectness, inconsistency or publication bias. We presented the main results of the percutaneous patellar osteosynthesis versus open surgery comparison in Table 1, which provides key information concerning the quality of the evidence, the magnitude of effect of the interventions examined and the sum of available data on the main outcomes. We produced the 'Summary of findings' table using Review Manager 5 (Review Manager 2014). We identified the following key outcomes for presentation in our 'Summary of findings' table: patient‐rated knee function (long‐term follow‐up), anterior knee pain (intermediate‐term follow‐up), major adverse outcomes, observer‐rated measures of knee function (long‐term follow‐up), health‐related quality of life (long‐term follow‐up), return to previous activity and cosmetic appearance (long‐term follow‐up).

Results

Description of studies

See Characteristics of included studies, Characteristics of excluded studies, and Characteristics of ongoing studies tables.

Results of the search

We updated the search from May 2014 to January 2020. We screened 394 records from the following databases: CENTRAL (54), MEDLINE (67), Embase (176), LILACS (78), WHO ICTRP (15) and ClinicalTrials.gov (four). We did not identify any potentially eligible studies from other sources. Once duplicates had been removed, we had 343 records. We excluded 320 records based on titles and abstracts. We obtained the full text of the remaining 23 records and linked any references pertaining to the same study under a single study ID.

After screening and study selection, we included six new studies, which were reported in seven articles (Afsar 2014; Hai‐feng 2015; Lin 2015; Shao 2019; Tian 2015; Wei 2019). Furthermore, we excluded 13 studies, and identified two ongoing studies (ChiCTR1800016105; NCT03445819). Additionally, we found another report for Mao 2013, a previously included study.

Overall, our review now includes 11 included trials, 22 excluded studies and two ongoing studies. There are no studies awaiting classification. Figure 1 shows a flow diagram summarising the study selection process.

1.

Study flow diagram.

Included studies

Details of the 11 studies are in the Characteristics of included studies table. No studies published their study protocol before their results. Lin 2015, however, reported conducting a pilot study involving 14 participants. Two trials were reported at different times, without reference to the earlier article (Mao 2013; Wei 2019). As explained in the Characteristics of included studies table, we used data from the first report of Mao 2013 rather than from the later report in Yong‐liang 2015. Conversely, we used the later report of Wei 2019, reporting results for 56 participants, rather than the interim report of results for 32 participants (Wei 2017). Ten trials were reported in English, although additional reports were in Chinese for Mao 2013 and Wei 2019; and one trial was reported in Chinese only (Hai‐feng 2015).

We sent requests seeking additional information or data to authors of all studies, but received no responses.

Design and setting

Nine studies were RCTs and two were quasi‐RCTs. All studies were single‐centre trials with two intervention groups. Seven studies were set in China (Chen 1998; Hai‐feng 2015; Lin 2015; Mao 2013; Shao 2019; Tian 2015; Wei 2019) and the other four set in Finland (Juutilainen 1995), Mexico (Luna‐Pizarro 2006), Pakistan (Afsar 2014), and Turkey (Günal 1997).

Sample sizes

The 11 trials enrolled 564 participants; with outcome data available for a maximum of 500 participants (88.6%).

Participants

Age and gender

All participants were adults; where stated, their ages ranged between 16 and 76 years. The mean ages ranged from 28 years in Günal 1997 to 52 years in Lin 2015. Of the 564 participants, 212 (37.6%) were women, 340 (60.3%) were men and the sex of 12 was not reported. Afsar 2014 reported the lowest percentage of women (6.1%) and Juutilainen 1995 the highest percentage of women (70%). The characteristics for the individual studies are listed below.

• Afsar 2014 (reported characteristics for 65/108 participants randomised): mean age 31.5 years;  61 men and four women.

• Chen 1998 (38 participants): mean age 46 years; 27 men and 11 women.

• Günal 1997 (28 participants): mean age 28 years; 16 men and 12 women.

• Hai‐feng 2015 (57 participants): mean age 38.5 years; 30 men and 27 women.

• Juutilainen 1995 (10 participants): mean age 48 years; two men and seven women (gender of one excluded participant was not reported).

• Lin 2015 (reported characteristics for 52/63 randomised participants): mean age 52 years; 28 men and 24 women.

• Luna‐Pizarro 2006 (53 participants): mean age 47 years; 30 men and 23 women.

• Mao 2013 (40 participants): mean age 42 years; 25 men and 15 women.

• Shao 2019 (38 participants): mean age 41 years; 25 men and 13 women.

• Tian 2015 (73 participants): mean age 45 years; 27 men and 46 women.

• Wei 2019 (56 participants): mean age 39 years; 29 men and 27 women.

Type/classification of fractures

Afsar 2014 and Wei 2019 included transverse fractures and comminuted fractures. Chen 1998 and Juutilainen 1995 included participants with transverse or oblique displaced fractures. Günal 1997 and Hai‐feng 2015 included only comminuted fractures. Lin 2015 included only transverse fractures and used the AO classification and anatomical descriptive classification. Luna‐Pizarro 2006 used the AO classification (Muller 1991), but also described complete articular transverse fractures. Mao 2013 included participants with transverse displaced fractures. Shao 2019 and Tian 2015 included only transverse fractures. All trials excluded people with open fractures.

Interventions

All 11 trials compared different surgical interventions. We grouped the included trials into seven comparisons based on the method of surgical technique or implant.

• Biodegradable (biotension band wiring plus screws or plugs made of biomaterials that were degradable) versus metallic implants (tension band wiring) (Chen 1998; Juutilainen 1995). There were follow‐up data for 47 participants (23 with biodegradable implant and 24 with metallic implant).

• Patellectomy with advancement of vastus medialis obliquus (VMO) versus simple patellectomy (Günal 1997). There were follow‐up data for 28 participants (12 with patellectomy with advancement of VMO versus 16 with simple patellectomy).

• Percutaneous patellar osteosynthesis (minimally invasive surgery that used small incisions and devices or implants as tension band, screws and others) versus open surgery (traditional surgery that used wide incisions and traditional devices or implants as tension band, screws, wires and others) (Lin 2015; Luna‐Pizarro 2006; Mao 2013; Shao 2019). Lin 2015 compared closed reduction and percutaneous fixation with cannulated screws (CRCF) versus open reduction and internal fixation with tension band wiring (ORTF) of transverse patellar fractures. Luna‐Pizarro 2006 compared the percutaneous patellar osteosynthesis system (a new device to use the tension band technique using a percutaneous method) versus open surgery involving the tension band technique. Mao 2013 compared the cable pin system (CPS; a combination of interfragmentary screws and the tension band wire percutaneously) versus open surgery involving the tension band technique. Shao 2019 compared the CPS (a combination of interfragmentary screws and the tension band wire) inserted percutaneously versus the same cable pin in open surgery. There were follow‐up data for 174 participants (90 with percutaneous patellar osteosynthesis and 84 with open surgery) of the 194 randomised into the four trials.

• A new intraoperative reduction technique (three‐dimensional strapping reduction technique) versus traditional (towel clamp reduction technique) (Wei 2019). There were follow‐up data for 56 participants (28 in each group).

• A modified tension band technique versus conventional AO tension band wiring (TBW) technique (Afsar 2014). There were follow‐up data for 65 participants (33 with modified tension band technique and 32 with conventional AO TBW technique).

• Adjustable patella claws and absorbable suture (new device) versus K‐wire tension band (KTB) in the repair of comminuted patellar fractures (Hai‐feng 2015). There were follow‐up data for 57 participants (29 with adjustable patella claws and absorbable suture and 28 with KTB).

• Cable pin system (open or percutaneous surgery) versus tension band technique (Mao 2013; Tian 2015). There were follow‐up data for 112 participants (54 CPS (open or percutaneous surgery) and 58 with tension band technique).

Outcomes

Primary outcomes

Patient‐rated knee function scores

Günal 1997 measured patient‐rated knee function; however, this was part of a non‐validated score described by Levack 1985. Hai‐feng 2015, Tian 2015, and Wei 2019 evaluated participants using the Hospital for Special Surgery (HSS) scoring standard; and Lin 2015 used the Lysholm Knee Scoring Scale.

Anterior knee pain

Afsar 2014 reported pain using a VAS, but did not report the scale used. Chen 1998 graded pain in three categories: no more than slight or occasional pain, occasional to moderate pain or constant pain. Günal 1997 graded pain as no pain, minimal pain during activity or constant severe pain. Juutilainen 1995 dichotomised pain as no pain or pain. Lin 2015, Luna‐Pizarro 2006, Mao 2013, and Shao 2019 measured pain using a VAS (from 0 (no pain) to 10 (most intense pain)). Hai‐feng 2015, Tian 2015, and Wei 2019 did not assess pain.

Major adverse outcomes

All studies reported major adverse outcomes. Afsar 2014 reported loss of reduction, wound infection, prominence of hardware, avascular necrosis, refracture and post‐traumatic arthritis. Chen 1998 reported infection, redisplacement, delayed wound healing and complications of bioabsorbable implants. Günal 1997 reported heterotopic ossification, lateral subluxation of patellar tendon and quadriceps tendon rupture. Hai‐feng 2015 reported delayed wound healing, infection, loosening of hardware, hardware breakage, skin irritation and frictional articular surface pain. Juutilainen 1995 reported infection, radiographic loss of reduction, refracture and complications of bioabsorbable implants. Lin 2015 assessed postoperative displacement, infection, painful hardware, tension band loosening or migration of the hardware, and reoperation rate. Luna‐Pizarro 2006 reported infection, painful hardware and redisplacement. Mao 2013 assessed loss of reduction, migration of hardware and broken wires. Shao 2019 reported defect of cartilage surface, arthritic changes, breaking or backing out of pin and cable rupture detected through radiological examinations. Tian 2015 reported incision infection, deep infection, K‐wires protruded through the skin, loosened K‐wires, displacement of the fragments, migration and breakage of the wires. Wei 2019 reported on bone non‐union, infection and broken nails.

Secondary outcomes

Observer‐rated measures of knee function

Afsar 2014 evaluated degree of extension lag and quadriceps power. Using a non‐validated score, Chen 1998 evaluated loss of movement in three categories: loss of movement of less than 15 degrees at the knee, loss of movement of 15 degrees to 30 degrees at the knee and loss of more than 30 degrees at the knee. Günal 1997 assessed limitation of activity, loss of quadriceps strength and functional assessment using a non‐validated score described by Marshall 1977 (a functional assessment scale used for recovery after knee ligament injuries, where patients are asked to duck‐walk, run on the spot, jump on one leg and squat). Juutilainen 1995 reported loss of movement in two categories: normal or restricted range of movement of the knee. Lin 2015 reported knee ROM and flexion. Luna‐Pizarro 2006 reported the outcome by goniometry for flexion and extension and used the Knee Society Clinical Rating System. Mao 2013 assessed knee ROM and Böstman score (Böstman 1981). Shao 2019 assessed knee ROM and Böstman score (Böstman 1981): the total score was stratified as follows: less than 20, unsatisfactory; 20 to 27, good; 28 to 30, excellent. Hai‐feng 2015, Tian 2015, and Wei 2019 did not report observer‐rated measures of knee function.

Health‐related quality of life scores

None of the included trials evaluated health‐related quality of life scores.

Return to previous activity

None of the included trials evaluated return or time to return to previous activities.

Cosmetic appearance

None of the included trials evaluated cosmetic appearance.

Hardware removal

Five trials reported removal of hardware, either routine or for symptoms (Juutilainen 1995; Lin 2015; Luna‐Pizarro 2006; Mao 2013; Tian 2015). However, only Lin 2015 reported that hardware removal was indicated in cases of painful hardware, skin irritation and postoperative fracture displacement before healing.

Other outcomes

Ten trials evaluated other outcomes that were not included in this review. Afsar 2014 evaluated quadriceps diameter in comparison with the contralateral limb. Chen 1998 evaluated time to union and radiographic results (graded as poor, fair or good) and wound healing. Hai‐feng 2015 evaluated incision length, surgical time and fracture healing. Shao 2019 evaluated radiographic results and mean operating time. Juutilainen 1995 evaluated radiographic results (graded as poor or good). Lin 2015 evaluated radiographic union (fracture healing time in months). Luna‐Pizarro 2006 evaluated surgical time. Mao 2013 evaluated wound healing, operating time and use of intraoperative fluoroscopy. Tian 2015 evaluated incision length, surgical time, intraoperative blood loss (with pneumatic tourniquet) and fracture healing in radiological examinations. Wei 2019 reported surgical time, intraoperative time of fluoroscopy and fracture healing in radiological examinations.

Timing of outcome measurements

The studies varied in timing of follow‐up. Afsar 2014 had a length of final follow‐up of 12 months. Chen 1998 reported follow‐up to 32 months. Günal 1997 conducted a mean follow‐up of 4.2 years (at least three years of final follow‐up). Hai‐feng 2015 and Shao 2019 reported final follow‐up data until two years. Juutilainen 1995 presented the length of final follow‐up until two years. Lin 2015 reported until 12 months. Luna‐Pizarro 2006 and Mao 2013 conducted follow‐up to two years. Tian 2015 presented final follow‐up data until 29 months. Wei 2019 reported a length of final follow‐up of 15 months.

For the studies that reported multiple time points, we considered short‐term follow‐up as up to six weeks following treatment; intermediate‐term follow‐up as more than six weeks and up to six months after the end of treatment; and long‐term follow‐up as greater than six months after the end of treatment.

Excluded studies

We excluded 22 studies (13 new in this update) because they did not meet our inclusion criteria; see Characteristics of excluded studies table. All 22 studies had an irrelevant study design, which included five systematic reviews.

Ongoing studies

Our search found two ongoing RCTs; see Characteristics of ongoing studies table.

ChiCTR1800016105 appears to be single‐centre study set in China. The trial aims to compare the effect on rotation restriction of the knee joint between modified tension band and traditional tension band in the treatment of patellar fracture. The trial target population is 148 participants.

NCT03445819 is a multicentre study, set in three centres in Canada, that aims to compare surgery (open reduction and internal fixation) with non‐surgical treatment. It plans to enrol 84 participants.

Risk of bias in included studies

All trials had methodological flaws rendering them at high risk of bias (see Figure 2 and Figure 3).

2.

'Risk of bias' summary: review authors' judgements about each risk of bias item for each included study.

3.

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Allocation

Afsar 2014 and Wei 2019 were at high risk of bias relating to random sequence generation. Afsar 2014 was described as consecutively allocated with reference to "non‐probability consecutive sampling"; and Wei 2019 was described as "according to admission date". We considered both studies quasi‐RCTs. Six studies were at low risk of bias (Günal 1997; Hai‐feng 2015; Lin 2015; Mao 2013; Shao 2019; Tian 2015): Günal 1997 reported that random sequence generation was performed by drawing lots; Hai‐feng 2015 described using a random table method; Lin 2015, Mao 2013, and Tian 2015 used a computer random number generator; and Shao 2019 used a coin flip. Chen 1998, Juutilainen 1995, and Luna‐Pizarro 2006 did not provide details of random sequence generation and were at unclear risk of bias.

Since Chen 1998, Günal 1997, Hai‐feng 2015, Juutilainen 1995, Lin 2015, Shao 2019, and Tian 2015 did not describe their methods of allocation concealment, we assessed them at unclear risk of bias. Both Luna‐Pizarro 2006 and Mao 2013 used sealed envelopes but the former, which we judged at unclear risk of bias, provided no other details of safeguards. We judged Mao 2013 at low risk of bias because the sealed envelopes were also sequentially numbered and opaque. The two quasi‐RCTs were at high risk of this bias (Afsar 2014; Wei 2019).

Blinding

We judged all trials at high risk of performance and detection bias. As the trials all compared surgical interventions, it was not possible to blind treatment providers. No trials included sham surgery or blinded the participants. While it may have been possible to blind outcome assessors, none of the trials mentioned assessor blinding.

Incomplete outcome data

We considered trials at low risk of bias if more than 80% of participants completed the follow‐up, missing outcomes data were balanced in number across intervention groups and an intention‐to‐treat analysis was reported for the primary outcomes. As Chen 1998, Günal 1997, Hai‐feng 2015, Juutilainen 1995, Lin 2015, Luna‐Pizarro 2006, Mao 2013, Tian 2015, and Wei 2019 met these criteria, we judged them at low risk of attrition bias. No participants were lost to follow‐up in Chen 1998, Günal 1997, Hai‐feng 2015, Shao 2019, and Wei 2019. Juutilainen 1995 and Mao 2013 reported that only one participant was lost to follow‐up. Lin 2015 reported that missing outcome data were balanced in numbers across the groups, with similar reasons for missing data across groups: 11/63 (17.4%) participants were missing from follow‐up at 12 months. Luna‐Pizarro 2006 reported that missing outcome data were balanced in numbers across intervention groups, with similar reasons for missing data across groups (8/53 (15%) participants were missing from follow‐up at 24 months). We judged Afsar 2014 at high risk of attrition bias because 43/108 (40%) participants were lost during follow‐up.

Selective reporting

Except for Lin 2015, all trials were at high risk of reporting bias because one or more outcomes of interest were incompletely reported or not reported.

Lin 2015 was at unclear risk of bias because, despite some missing outcomes, it was registered prospectively (in the Chinese Clinical Trial Register) and a pilot was performed beforehand.

Other potential sources of bias

We judged three trials at unclear risk of 'other' bias (Chen 1998; Günal 1997; Juutilainen 1995). The authors from Chen 1998 and Juutilainen 1995 did not provide sufficient information about the surgeons and care providers nor sufficient details of rehabilitation after surgery. Günal 1997 reported only the baseline characteristics of age and gender.

We judged Afsar 2014, Hai‐feng 2015, Luna‐Pizarro 2006, Mao 2013, Shao 2019, Tian 2015, and Wei 2019 at high risk of other bias because of a potential conflict of interest reflecting the lack of independent scrutiny as each trial tested novel self‐developed techniques. Notably, this involved a new device (Cable pin), albeit that the comparison was percutaneous versus open surgery. In Luna‐Pizarro 2006, this also involved a locally designed device and Wei 2019 involved a new reduction technique. Furthermore, Tian 2015 did not provide sufficient information about the surgeons and care providers, and about the tension band technique (modified TBW).

We considered that Lin 2015 was at low risk of other bias because it provided sufficient information about the surgeons and care providers and sufficient details of rehabilitation after surgery.

Effects of interventions

See: Table 1

The included studies only compared different methods of surgical intervention. We describe the following seven comparisons in turn:

• biodegradable versus metallic implants;

• patellectomy with advancement of vastus medialis obliquus versus simple patellectomy;

• percutaneous patellar osteosynthesis versus open surgery;

• a new intraoperative reduction technique (three‐dimensional strapping reduction technique) versus traditional (towel clamp reduction) technique;

• a modified tension band technique versus conventional AO tension band wiring technique;

• adjustable patella claws and absorbable suture (new device) versus K‐wire tension band; and

• CPS versus tension band technique (open or percutaneous surgery).

None of the trials reported health‐related quality scores, return to previous activity or cosmetic appearance including deformity.

The minimal clinically important difference (MCID) is an useful tool to determine whether a difference in a continuous outcome score is clinically meaningful (Katz 2015). We have not identified a topic‐specific MCID for pain measured using the VAS after treatment for fractures of the patella. Danoff 2018 reported that the MCID for VAS (0 to 100; 100 = worst pain) for total knee arthroplasty was –22.6 mm for (TKA) pain improvement and 29.1 mm for worsening. A more general result from Lee 2003, which included only people from an adult emergency department, was a mean reduction in VAS of 30.0 mm, which represents a clinically important difference in pain severity that corresponded to patients' perception of adequate pain control. We consider that defining MCID based on adequate analgesic control rather than minimal detectable change may be more appropriate for future analgesic trials, when effective treatments for acute pain exist. Hence, in our review, we use 30 mm (3.0 cm) as MCID for pain because this is generally used and is a reasonable threshold.

We rated the quality (certainty) of the evidence for all reported outcomes and for all comparisons as very low, which means we are very uncertain about the estimates.

Biodegradable versus metallic implants

Two trials (48 participants) compared biodegradable implants with metallic tension bands for treating patella fractures (Chen 1998; Juutilainen 1995).

Patient‐rated knee function scores

Neither trial measured patient‐rated knee function scores.

Anterior knee pain

Both trials assessed pain. Pooled data showed no evidence of a difference between the two groups at long‐term follow‐up (range: 12 to 24 months) in participants with pain (1/23 with biodegradable implants versus 2/24 with metallic tension bands; RR 0.56, 95% CI 0.05 to 5.62; Analysis 1.1). All three participants with pain in Chen 1998 had 'occasional pain'. No participants in Juutilainen 1995 reported pain at long‐term follow‐up.

1.1. Analysis.

Comparison 1: Biodegradable versus metallic implants, Outcome 1: Anterior knee pain at long‐term follow‐up (presence of pain)

Major adverse outcomes

Both trials assessed major adverse outcomes. The sole case of 'treatment failure' was a refracture after new trauma in the biodegradable group in Juutilainen 1995. Other adverse outcomes were an infection in the metallic tension bands group and two cases of delayed wound healing in the biodegradable group (all in Chen 1998). Neither trial found evidence of complications of biodegradable implants (aseptic swelling or sinus tract formation), although Chen 1998 observed that tissue reaction may have contributed to delayed wound healing. Overall, there was no evidence of a difference between groups (3/24 with biodegradable implants versus 1/24 with metallic tension bands; RR 2.19, 95% CI 0.35 to 13.66; Analysis 1.2).

1.2. Analysis.

Comparison 1: Biodegradable versus metallic implants, Outcome 2: Major adverse outcomes

Observer‐rated measures of knee function

Both trials assessed knee range of motion (ROM). Pooled data showed similar numbers in the two groups had reduced knee ROM at long‐term follow‐up (range: 12 to 24 months) (2/23 with biodegradable implants versus 2/24 with metallic tension bands; RR 1.11, 95% CI 0.17 to 7.09; Analysis 1.3). All four participants in Chen 1998 had loss of knee motion of 20 degrees or more. All participants in Juutilainen 1995 had regained a good ROM at long‐term follow‐up.

1.3. Analysis.

Comparison 1: Biodegradable versus metallic implants, Outcome 3: Observer‐rated measures of knee function

Hardware removal

Juutilainen 1995 reported that all metallic implants were removed one year after the primary operation.

Secondary outcomes not reported 

Juutilainen 1995 did not measure health‐related quality of life scores, return to previous activities or cosmetic appearance including deformity.

Patellectomy with advancement of vastus medialis obliquus versus simple patellectomy

One trial (28 participants) compared patellectomy with VMO advancement versus simple patellectomy for treating comminuted fractures of the patella (Günal 1997). Both interventions entailed open surgery. Long‐term follow‐up was 3 to 5.5 years

Patient‐rated knee function scores

Günal 1997 reported subjective patient‐rated knee function using part of the non‐validated Levack score (Levack 1985). They found all participants in the VMO advancement group had good results (scoring over 50 points or a 100‐point score) at long‐term follow‐up (range: 3 to 5.5 years) compared with only 69% of those in the simple patellectomy group (12/12 with advancement of VMO versus 11/16 with simple patellectomy; RR 1.42 favouring VMO advancement, 95% CI 1.01 to 2.01; Analysis 2.1).

2.1. Analysis.

Comparison 2: Patellectomy with vastus medialis obliquus (VMO) advancement versus simple patellectomy for comminuted fractures, Outcome 1: Patient‐rated knee function score

Anterior knee pain

The difference in the number of participants with pain at long‐term follow‐up also favoured the VMO advancement group (5/12 with advancement of VMO versus 13/16 with simple patellectomy; RR 0.51, 95% CI 0.25 to 1.04; Analysis 2.2).

2.2. Analysis.

Comparison 2: Patellectomy with vastus medialis obliquus (VMO) advancement versus simple patellectomy for comminuted fractures, Outcome 2: Anterior knee pain at long‐term follow‐up (presence of pain)

Major adverse outcomes

There were no cases of heterotopic ossification and quadriceps rupture in either group. The only adverse event reported was a patellar tendon subluxation in the simple patellectomy group (0/12 with advancement of VMO versus 1/16 with simple patellectomy; RR 0.44, 95% CI 0.02 to 9.85; Analysis 2.3).

2.3. Analysis.

Comparison 2: Patellectomy with vastus medialis obliquus (VMO) advancement versus simple patellectomy for comminuted fractures, Outcome 3: Major adverse outcome

Observer‐rated measures of knee function

Günal 1997 found better results in favour of the patellectomy with the VMO advancement with respect to the number of participants who had unlimited activity at long‐term follow‐up (RR 4.44 favouring VMO advancement, 95% CI 1.55 to 12.71) and number of participants with no loss of quadriceps strength (RR 10.90 favouring VMO advancement, 95% CI 2.36 to 50.42; Analysis 2.4). In terms of specific activities, all participants in both groups could perform a 'duck‐walk' and squat fully. However, in the simple patellectomy group, three participants could not run on the spot and four could not jump on one leg (see Analysis 2.4).

2.4. Analysis.

Comparison 2: Patellectomy with vastus medialis obliquus (VMO) advancement versus simple patellectomy for comminuted fractures, Outcome 4: Observer‐rated measures of knee function at long‐term follow‐up

Secondary outcomes not reported 

Günal 1997 did not measure health‐related quality of life scores, return to previous activities, cosmetic appearance including deformity or hardware removal.

Percutaneous patellar osteosynthesis versus open surgery

Four trials (194 participants) compared percutaneous patellar osteosynthesis with open surgery for treating patella fractures (Lin 2015; Luna‐Pizarro 2006; Mao 2013; Shao 2019). Each trial tested different devices or implants in percutaneous or open surgery.

Patient‐rated knee function scores

Lin 2015 (data from 52 participants) reported patient‐rated knee function scores using the Lysholm Knee Scoring Scale (0 to 100; 100 = best function). Lysholm scores were higher in, and thus favoured, the percutaneous group at three months (MD 5.40, 95% CI 2.38 to 8.42), six months (MD 5.20, 95% CI 2.17 to 8.23); and, although smaller, at 12 months (MD 3.40, 95% CI 0.26 to 6.54) (Analysis 3.1). However, these differences are unlikely to be clinically important; the 95% CIs in each case were less than 10 points; which is the MCID for the Lysholm score reported for anterior cruciate ligament reconstruction (Nwachukwu 2017).

3.1. Analysis.

Comparison 3: Percutaneous patellar osteosynthesis (PPO) versus open surgery, Outcome 1: Patient‐rated knee function score (Lysholm‐range: 0 to 100 points; 100 = better)

Anterior knee pain

Pooled data for knee pain (VAS: 0 to 10: 10 = worse score) showed lower scores in the percutaneous patellar osteosynthesis group at short‐term (range 4 to 6 weeks: MD –1.89, 95% CI –2.32 to –1.46; I² = 54%; 3 studies, 131 participants), intermediate‐term (range 8 weeks to 3 months: MD –1.88, 95% CI –2.34 to –1.42; I² = 39%; 4 studies, 183 participants), and at six months (MD –0.40, 95% CI –0.74 to –0.06; I² = 78%; 2 studies, 92 participants) (Analysis 3.2). However, none of these are clinically important differences as all were less than the 3.0 (30 mm) MCID. Lin 2015 found no significant or clinically important difference between groups at long‐term follow‐up (12 months: MD –0.60, 95% CI –1.44 to 0.24; 52 participants).

3.2. Analysis.

Comparison 3: Percutaneous patellar osteosynthesis (PPO) versus open surgery, Outcome 2: Anterior knee pain (measured by VAS: 0 to 10 cm; 10 = worst score)

Major adverse outcomes

Pooled data from three trials showed a lower incidence of participants incurring an adverse event in the percutaneous group (11/73 with percutaneous patellar osteosynthesis versus 42/72 with open surgery; RR 0.26, 95% CI 0.14 to 0.46; I² = 0%; 145 participants; Analysis 3.3). Note, we could not confirm that the data from Luna‐Pizarro 2006 and Lin 2015 referred to the total number of participants with complications rather than total complications where participants may have had more than one complication. However, individual outcome data available for Mao 2013 show that, of the 13 participants who had complications in the open surgery group, six participants had more than one complication: four participants had hardware migration plus hardware irritation, one participant had broken wires plus loss of reduction and one participant had hardware migration plus loss of reduction. Shao 2019 only reported that until the last follow‐up visit, there was no defect of the articular surface, patellofemoral arthritic changes, breaking or backing out of pin, or cable rupture detected through radiological examinations.

3.3. Analysis.

Comparison 3: Percutaneous patellar osteosynthesis (PPO) versus open surgery, Outcome 3: Major adverse outcomes

Pooled data for individual complications also showed more individual complications (failure of treatment, loss of reduction and hardware complications) in the open surgery group than the percutaneous group (Analysis 3.3).

Observer‐rated measures of knee function

Three trials used two different observer‐rated composite measures of knee function (the Knee Society Clinical Rating System and Böstman score). Luna‐Pizarro 2006 used the Knee Society Clinical Rating System, while Mao 2013 and Shao 2019 used the Böstman score. Pooled data found better results for the percutaneous group at intermediate‐term follow‐up (range 8 weeks to 3 months: SMD 1.26, 95% CI 0.32 to 2.20; I² = 83%; 3 studies, 131 participants), at 12 months (SMD 1.01, 95% CI 0.08 to 1.95; I² = 83%; 3 studies, 124 participants) and at 24 months (SMD 0.62, 95% CI 0.18 to 1.06; I² = 0%; 2 studies, 84 participants) (Analysis 3.4). However, there was substantial statistical heterogeneity at the first two follow‐ups. There is no published MCID for the Knee Society Clinical Rating System and Böstman score. Of note though is that the differences between the groups in Böstman scores (0 to 30: 30 = best function) were fairly small, generally less than 2, for Mao 2013 and Shao 2019.

3.4. Analysis.

Comparison 3: Percutaneous patellar osteosynthesis (PPO) versus open surgery, Outcome 4: Observer‐rated measures of knee function

Knee range of motion

Pooled data for knee extension showed a small but clinically insignificant improvement in knee extension in favour of percutaneous patellar osteosynthesis at short‐term follow‐up (range: 4 to 6 weeks: MD 1.88 degrees, 95% CI 0.82 to 2.95; I² = 19%; 3 studies, 131 participants) and at intermediate‐term follow‐up (range: 7 weeks to 6 months: MD 2.28 degrees, 95% CI 0.15 to 4.42; I² = 81%; 3 studies, 131 participants). Lin 2015 found no evidence of a difference at 12 months (Analysis 3.5).

3.5. Analysis.

Comparison 3: Percutaneous patellar osteosynthesis (PPO) versus open surgery, Outcome 5: Knee range of motion: extension (degrees)

Pooled data for knee flexion showed a significant improvement in favour of percutaneous patellar osteosynthesis at short‐term follow‐up (MD 22.01 degrees, 95% CI 10.77 to 33.25; I² = 83%; 3 studies, 131 participants) and at intermediate‐term follow‐up (MD 24.02 degrees, 95% CI 7.22 to 40.82; I² = 93%; 4 studies, 183 participants) (Analysis 3.6). However, there was substantial heterogeneity at both follow‐ups; this mainly reflected the more extreme results of Luna‐Pizarro 2006. Long‐term follow‐up data for Mao 2013 and Lin 2015 also showed better results in favour of percutaneous patellar osteosynthesis at 12 months (MD 5.37 degrees, 95% CI 0.94 to 9.80; 92 participants; 2 studies; I² = 0%). Mao 2013 also reported results at 24 months (MD 8.30 degrees, 95% CI 1.47 to 15.13; 40 participants). The findings in favour of percutaneous fixation for the first two follow‐ups are potentially clinically important but both groups had impaired knee function. The differences at the long‐term follow‐up were smaller and unlikely to be clinically important.

3.6. Analysis.

Comparison 3: Percutaneous patellar osteosynthesis (PPO) versus open surgery, Outcome 6: Knee range of motion: flexion (degrees)

Hardware removal

Luna‐Pizarro 2006 reported on hardware removal due to pain or subcutaneous irritation at eight weeks and two years, and Mao 2013 reported on hardware removal due to pain, irritation and psychological reasons at two years. Lin 2015 only reported hardware removal in tables without any reasons or further explanations. Pooled data at two years showed a lower incidence of hardware removal in the percutaneous group (23/73 with percutaneous patellar osteosynthesis versus 44/72 with open surgery; RR 0.51, 95% CI 0.35 to 0.74; 3 studies, 145 participants), but there was substantially heterogeneity (I² = 84%) (Analysis 3.7).

3.7. Analysis.

Comparison 3: Percutaneous patellar osteosynthesis (PPO) versus open surgery, Outcome 7: Hardware removal (usually due to pain or subcutaneous irritation)

Secondary outcomes not reported 

None of the trials measured health‐related quality of life scores, return to previous activities or cosmetic appearance including deformity.

A new intraoperative reduction technique versus traditional intraoperative reduction technique

One trial (56 participants) compared a new intraoperative reduction technique with the traditional towel clamp reduction technique for treating fractures of the patella: 28 participants received a three‐dimensional strapping reduction technique and 28 received a traditional towel clamp reduction technique (Wei 2019). Both interventions entailed open surgery and are techniques for getting better reductions of patellar fractures.

Patient‐rated knee function scores

Wei 2019 reported patient‐rated knee function using the modified HSS knee score (range 0 to 100: 100 = best function). The HSS scores were higher with the new intraoperative reduction technique at six months (MD 4.30, 95% CI 1.82 to 6.78; Analysis 4.1). However, the between‐group difference was unlikely to be clinically important; although specific for total knee arthroplasty, an estimated MCID for the HSS knee score is 8.29 points (Singh 2013).

4.1. Analysis.

Comparison 4: A new intraoperative reduction technique versus a standard technique, Outcome 1: Patient‐rated knee function: Hospital for Special Surgery (0 to 100 points; 100 = better) (6 months)

Anterior knee pain

Wei 2019 did not assess anterior knee pain.

Major adverse outcomes

Wei 2019 reported there was no infection, non‐union or broken nails in either group.

Secondary outcomes not reported 

Wei 2019 did not measure observer‐rated measures of knee function, health‐related quality of life scores, return to previous activities, cosmetic appearance including deformity or hardware removal.

A modified tension band technique versus conventional AO tension band wiring technique

One quasi‐RCT (65 participants) compared a modified tension band technique versus a conventional AO TBW technique for treating fractures of the patella (Afsar 2014). Both interventions entailed open surgery.

Patient‐rated knee function scores

Afsar 2014 did not assess patient‐rated function scores.

Anterior knee pain

Afsar 2014 reported pain at 12 months' follow‐up measured using the VAS, but did not provide details of the range of the score. Assuming it is a 0 to 10 cm score, the results shown in Analysis 5.1 provide no evidence of a difference between the two groups.

5.1. Analysis.

Comparison 5: A modified tension band technique versus conventional AO tension band wiring technique, Outcome 1: Anterior knee pain (visual analogue score) (0 to 10; 10 = worst pain) (12 months)

Major adverse outcomes

Afsar 2014 reported frequency of complications in each group (loss of reduction at three months, wound infection, prominence of hardware, avascular necrosis, refracture at 12 months and post‐traumatic arthritis). There were fewer cases of individual complications in the modified technique group for all complications (Analysis 5.2). In particular, there were fewer cases in the modified technique group with loss of reduction at three months (3/32 with modified tension band technique versus 13/33 with conventional AO TBW technique; RR 0.24, 95% CI 0.07 to 0.76) and fewer cases with hardware prominence (3/32 with modified tension band technique versus 12/33 conventional AO TBW technique; RR 0.26, 95% CI 0.08 to 0.83).

5.2. Analysis.

Comparison 5: A modified tension band technique versus conventional AO tension band wiring technique, Outcome 2: Major adverse outcomes

Observer‐rated measures of knee function

Afsar 2014 indirectly assessed knee function by measuring ROM, quadriceps wasting, quadriceps power and extension lag. Some of the intermediate‐term and long‐term follow‐up results are presented in Analysis 5.3. None of the small between‐group differences in these outcome measures is likely to be of direct clinical importance.

5.3. Analysis.

Comparison 5: A modified tension band technique versus conventional AO tension band wiring technique, Outcome 3: Observer‐rated measures of knee function

Secondary outcomes not reported 

Afsar 2014 did not measure health‐related quality of life scores, return to previous activities, cosmetic appearance including deformity or hardware removal.

Adjustable patella claws and absorbable suture (new device) versus Kirschner wire tension band in the repair of comminuted patellar fractures

One trial (57 participants) compared a new device (adjustable patella claws and absorbable suture) with KTB for treating fractures of the patella (Hai‐feng 2015). Both interventions entailed open surgery.

Patient‐rated knee function scores

Hai‐feng 2015 reported patient‐rated knee function score using the HSS (0 to 100; 100 = best function). More participants in the new device group had an excellent HSS score of 85 or above (stable knee joint, no obstacles to functional activities) at two years: 28/29 with adjustable patella claws and absorbable suture versus 20/28 with KTB; RR 1.35, 95% CI 1.06 to 1.73; Analysis 6.1).

6.1. Analysis.

Comparison 6: Adjustable patella claws and absorbable suture (new device) versus Kirschner wire tension band for comminuted patellar fractures, Outcome 1: Patient‐rated knee function: 'excellent' Hospital for Special Surgery (2 years)

Anterior knee pain

Hai‐feng 2015 did not measure anterior knee pain.

Major adverse outcomes

Hai‐feng 2015 reported cases of delayed wound healing, infection, loosening, hardware breakage, skin irritation and frictional articular surface pain. There was no evidence confirming a difference between groups in the pooled results for total adverse events (1/29 with adjustable patella claws and absorbable suture versus 6/28 with KTB; RR 0.16, 95% CI 0.02 to 1.25) or for the individual complications (Analysis 6.2).

6.2. Analysis.

Comparison 6: Adjustable patella claws and absorbable suture (new device) versus Kirschner wire tension band for comminuted patellar fractures, Outcome 2: Major adverse outcomes

Secondary outcomes not reported 

Hai‐feng 2015 did not measure observer‐rated measures of knee function, health‐related quality of life scores, return to previous activities, cosmetic appearance including deformity or hardware removal.

Cable pin system versus tension band technique

Two trials (113 participants) compared CPS with tension band technique for treating patella fractures (Mao 2013; Tian 2015). The technique was open or closed.

Patient‐rated knee function scores

Tian 2015 reported the modified HSS knee score (0 to 100: 100 = best function) at one year after surgery. The CPS had better results than TBW (modified tension band group) in HSS score (MD 9.17, 95% CI 4.82 to 13.52; 73 participants; Analysis 7.1). As the 95% CI include the MCID of 8.29 points for the HSS (Singh 2013), the difference may be clinically important.

7.1. Analysis.

Comparison 7: Cable pin system versus tension band technique, Outcome 1: Patient‐rated knee function (Hospital for Special Surgery score: 0 to 100; 100 = best function) (1 year)

Anterior knee pain

Mao 2013 (40 participants) reported anterior knee pain measured using VAS (0 to 10: 10 = worst score). It found lower VAS results in the CPS group at short‐term follow‐up (1 month: MD –2.00, 95% CI –2.78 to –1.22), and at two intermediate‐term follow‐ups (3 months: MD –1.40, 95% CI –2.11 to –0.69; 6 months: MD –0.30, 95% CI –0.65 to 0.05; Analysis 7.2). However, all the 95% CIs were less than the MCID of 3.0 cm, and thus are unlikely to be clinically important.

7.2. Analysis.

Comparison 7: Cable pin system versus tension band technique, Outcome 2: Anterior knee pain (measured by visual analogue scale: 0 to 10 cm: 10 = worst score)

Major adverse outcomes

Mao 2013 reported on failure of treatment (loss of reduction), hardware complications, delayed wound healing and total adverse events. Tian 2015 reported infection, hardware complications and total adverse events. Pooled data from the two studies showed fewer overall complications in the CPS group (3/54 with CPS versus 22/59 with tension band technique; RR 0.14, 95% CI 0.05 to 0.44; 113 participants), which mainly reflected fewer participants in the CPS group with hardware problems (2/54 with CPS versus 18/59 with tension band technique; RR 0.14, 95% CI 0.04 to 0.49; 113 participants) (Analysis 7.3).

7.3. Analysis.

Comparison 7: Cable pin system versus tension band technique, Outcome 3: Major adverse outcomes

Observer‐rated measures of knee function

Mao 2013 reported on two observer‐rated measures of knee function: knee ROM and Böstman score (Böstman 1981). Mao 2013 found higher Böstman scores (0 to 30; 30 = best function) in the CPS group at intermediate‐term follow‐up (MD 2.40, 95% CI 1.01 to 3.79; 40 participants), at 12 months (MD 1.00, 95% CI 0.02 to 1.98; 40 participants) and at 24 months (MD 1.00, 95% CI 0.04 to 1.96; 39 participants) (Analysis 7.4). These differences may not be clinically important but we have not identified a published MCID for the Böstman score.

7.4. Analysis.

Comparison 7: Cable pin system versus tension band technique, Outcome 4: Observer‐rated measures of knee function: Böstman score (0 to 30: 30 = best function)

Knee range of motion

Only Mao 2013 (40 participants) reported knee ROM.

Mao 2013 showed greater knee extension in the CPS group at one month (MD 2.00 degrees, 95% CI 0.72 to 3.28; 40 participants) and at intermediate‐term follow‐up (MD 0.90 degrees, 95% CI –0.20 to 2.00) (Analysis 7.5).

7.5. Analysis.

Comparison 7: Cable pin system versus tension band technique, Outcome 5: Knee range of motion: extension (degrees)

Knee flexion was also better in the CPS group at 1, 3, 12 and 24 months (24 months: MD 8.30 degrees, 95% CI 1.47 to 15.13) (Analysis 7.6).

7.6. Analysis.

Comparison 7: Cable pin system versus tension band technique, Outcome 6: Knee range of motion: flexion (degrees)

Hardware removal

Mao 2013 reported hardware removal due to pain, irritation and psychological reasons at two years. Tian 2015 reported that five participants had the implants removed in CPS group because of migration, breakage and patient request (without pain). Pooled data for hardware removal showed a lower incidence in the cable pin group (4/54 with CPS versus 20/59 with tension band technique; RR 0.23, 95% CI 0.09 to 0.55; 113 participants) (Analysis 7.7).

7.7. Analysis.

Comparison 7: Cable pin system versus tension band technique, Outcome 7: Hardware removal

Secondary outcomes not reported 

Neither trial measured health‐related quality of life scores, return to previous activities, or cosmetic appearance including deformity.

Discussion

This updated review now includes 11 trials, six more than the 2015 version, that recruited 564 participants with closed patella fractures.

Summary of main results

All nine RCTs and two quasi‐RCTs addressed only the effects (benefits and harms) of different surgical interventions for treating fractures of the patella in adults. The included studies were small with follow‐up data available for 500 participants. Seven studies were conducted in China. All 11 trials compared two different surgical interventions. The main results, restricted to the key outcomes, from the seven different treatment comparisons made across the 11 trials are summarised below. We begin with the three most important comparisons, each of which was tested by more than one trial.

Percutaneous patellar osteosynthesis versus open surgery

Four trials (194 participants, with follow‐up data for 174) compared percutaneous patellar osteosynthesis versus open surgery (Lin 2015; Luna‐Pizarro 2006; Mao 2013; Shao 2019). The majority of participants were men in all four trials.

The trials used different devices and techniques. Lin 2015 compared closed reduction and percutaneous fixation with cannulated screws (CRCF) versus open reduction and internal fixation with tension band wiring (ORTF) of transverse patellar fractures. Luna‐Pizarro 2006 compared the percutaneous patellar osteosynthesis system (a device to use the tension band technique in a percutaneous way) versus open surgery involving the tension band technique. Mao 2013 compared the cable pin system (a combination of interfragmentary screws and the tension band wire percutaneously) versus open surgery involving the tension band technique. Shao 2019 compared the cable pin system versus the cable pin in open surgery.

The evidence for seven key outcomes for this comparison is summarised in Table 1. All evidence for each outcome was very low quality, invariably for serious risk of bias and imprecision, which indicates our uncertainly in the results.

There is very low‐quality evidence from one trial (52 participants) of no clinically important difference between the two interventions in patient‐rated knee function measured using the Lysholm Knee Scoring Scale at 12 months.

There is very low‐quality evidence pooled from four trials (183 participants) of no clinically important difference between the two interventions in knee pain measured using the VAS at intermediate‐term follow‐up (eight weeks to three months).

There is very low‐quality evidence from three trials (145 participants) of fewer participants experiencing adverse events, mainly loss of reduction and hardware complications, after percutaneous surgery compared with open surgery.

There is very low‐quality evidence from three trials (124 participants) indicating better observer‐rated knee function scores after percutaneous fixation at 12 months' follow‐up; however, the clinical importance of the results is unknown.

None of the trials reported health‐related quality of life, return to previous activity or cosmetic appearance.

Thus, we are uncertain whether percutaneous patellar osteosynthesis compared with open surgery gives better results in patient‐rated knee function score, pain, adverse events and knee function scores.

Cable pin system (open or percutaneous surgery) versus tension band technique

Two trials compared CPS versus tension band technique for a maximum of 113 participants (Mao 2013; Tian 2015). Neither trial reported on health‐related quality of life, return to previous activity or cosmetic appearance. Very low‐quality evidence means we are uncertain of the findings of these trials. These include better HSS scores at one year (a patient‐rated measure of knee function), which probably include a clinically important difference, in the CPS group of one trial (73 participants); and statistically but not clinically important lower pain scores at three months in the CPS group. We are also uncertain of the findings favouring the CPS group of fewer major adverse outcomes, and slightly better observer‐related Böstman scores at 12 and 24 months in the CPS group.

Biodegradable (biotension band wiring plus screws or plugs made of biomaterials that are degradable) versus metallic implants (tension band wiring)

Two trials (48 participants) compared biodegradable implants with metallic tension bands for treating patella fractures (Chen 1998; Juutilainen 1995). Neither trial reported on patient‐rated knee function scores, health‐related quality of life, return to previous activity or cosmetic appearance. There were very few events for any binary outcome. Very low‐quality evidence means that we are uncertain of the findings of little difference between biodegradable versus metallic implants at two‐year follow‐up in the numbers of participants with occasional knee pain, incurring adverse events or with reduced knee motion.

Patellectomy with advancement of vastus medialis obliquus versus simple patellectomy

One trial (28 participants) compared patellectomy with VMO advancement versus simple patellectomy for treating comminuted fractures of the patella (Günal 1997). The trial did not report on health‐related quality of life, return to previous activity or cosmetic appearance. Very low‐quality evidence means that we are uncertain of the findings at four‐year follow‐up in favour of patellectomy with advancement of VMO surgery of better knee function, assessed as numbers of participants with a 'good' result based on a subjectively rated score, of fewer participants experiencing knee pain and more participants with unlimited activity. The only recorded adverse event was patellar tendon subluxation in the simple patellectomy group.

A new intraoperative reduction technique versus traditional technique

One quasi‐RCT (56 participants) compared a new intraoperative reduction technique with the traditional towel clamp reduction technique for treating fractures of the patella (Wei 2019). The study did not report on knee pain, health‐related quality of life, return to previous activity, cosmetic appearance or observer‐rated measures of knee function. Very low‐quality evidence means that we are uncertain of the findings of no clinically important difference, despite being statistically in favour of the novel technique, in patient‐rated knee function at six months. There was no infection, non‐union or broken nails in either group.

A modified tension band technique versus conventional AO tension band wiring technique

One quasi‐RCT (65 participants) compared a modified tension band technique versus a conventional AO TBW technique for treating fractures of the patella (Afsar 2014). The study did not report on patient‐rated knee function scores, health‐related quality of life, return to previous activity or cosmetic appearance. Very low‐quality evidence means that we are uncertain of the findings of little between‐group difference in pain at 12 months, or in the lower risk of loss in reduction or hardware prominence and the slightly better objectively monitored measures of knee function at 12 months in the modified tension band technique group.

Adjustable patella claws and absorbable suture (new device) versus Kirschner wire tension band in the repair of comminuted patellar fractures

One study (57 participants) compared a new device (adjustable patella claws and absorbable suture) with KTB for treating fractures of the patella (Hai‐feng 2015). The study did not report knee pain, health‐related quality of life, return to previous activity, cosmetic appearance or observer‐rated measures of knee function. Very low‐quality evidence means that we are uncertain on the finding of more participants in the new device group having an excellent patient‐rated knee function score at two years follow‐up. There was insufficient evidence on adverse events to draw conclusions; overall there were seven different adverse events (delayed wound healing, infection, loosening, hardware breakage, skin irritation and frictional articular surface pain).

Overall completeness and applicability of evidence

The review included 11 trials providing outcome data available for a maximum of 500 participants (88.6% of the 564 randomised). These compared different surgical treatments for patella fractures; we found no trials comparing surgical versus conservative interventions or trials comparing different conservative interventions. Furthermore, there was a maximum of four trials testing each of the seven comparisons of different surgical treatments. These four trials compared percutaneous versus open surgery but used different devices in the two intervention groups. Only five of these trials provided data for validated patient‐rated knee function. However, none of these trials reported health‐related quality of life, return to previous activity outcomes or cosmetic appearance including deformity, all outcomes we consider important for these patients. We received no response from the trial authors following our request for further details or data information for 10 of the 11 trials. Thus, the data are very limited in quantity and are incomplete.

Despite the small number of trials and participants in each trial, we consider the gender and the age distribution are sufficiently representative of the populations for which the interventions under test are considered in current practice. This also applies to the types of included fractures, the majority of which were transverse and all of which were closed. Only 2% to 7% of fractures of the patella are open (Dy 2012; Insall 2006; Muller 1991). The interventions applied in Afsar 2014, Günal 1997, Hai‐feng 2015, and in Wei 2019 were appropriate for the comminuted fractures.

The surgical techniques tested in this review ranged from traditional techniques such as open surgery, patellectomy, metallic tension band and modified metallic tension band to new percutaneous techniques and devices such as percutaneous screws, biodegradable implants, cable pins and bio‐tension band.

One trial tested a new reduction technique (three‐dimensional strapping) with the traditional reduction technique (towel clamp technique) thus showing another possibility for reduction of fracture of the patella.

The CPS is a popular technique and its use is increasing in all around the world (mainly in the East). Nowadays, the use of special locked patellar plates is becoming more common; however, we found no RCTs evaluating these.

Quality of the evidence

The quality of the evidence for all reported outcomes in all seven comparisons was very low, as assessed using the GRADE approach (Schünemann 2011). This means that we are very uncertain about the estimate of effect. Evidence was not available for many of the important outcomes, consistently, health‐related quality of life, return to previous activity outcomes and cosmetic appearance including deformity.

We downgraded all evidence for risk of bias that we considered serious or very serious. All evidence was susceptible to detection and performance biases for most outcomes in which blinding to the allocated intervention was not possible. Only one trial was rated at low risk of selection bias, and selective reporting biases were other reasons for downgrading.

Downgrading for imprecision was very common, reflecting the wide CIs, the small sample sizes and few events.

For some outcomes, we downgraded the quality of evidence for inconsistency reflecting substantial heterogeneity.

We downgraded for indirectness often reflecting poorly defined and poor‐quality outcome measures, but also for comparisons that used novel self‐developed techniques.

We did not downgrade for publication bias given that we had insufficient trials to explore this possibility.

Hence, the evidence is not robust for any of the seven comparisons and we can affirm that the numerical results of this review should be interpreted with caution and viewed as requiring confirmation from future RCTs of good methodological quality and adequate power.

Potential biases in the review process

We conducted this review following criteria and methods set out in a published protocol (Sayum Filho 2012). During the process of the review, we tried to minimise the effects of publication bias. We are confident that our search strategy was broad with no language restriction, and it has been maintained properly and regularly updated by the guarantor of review (JS). We believe that the comprehensive literature search used in this review has found relevant studies and minimised the likelihood of missing trials; we also handsearched conference proceedings and searched for ongoing and recently completed trials. Nevertheless, it is possible that we have missed some potentially eligible trials. We tried to contact authors of all included trials to obtain further information and missing data; however, this was unsuccessful.

We included two trials, each had two reports, the second of which reported a greater sample size. In both cases, the second report did not acknowledge the existence of the first. For Mao 2013, we excluded the findings of the second report, Yong‐liang 2015, because of an unlikely similarity in the baseline characteristics and results of the two populations. For Wei 2019, we used data from the second report, which seemed to reflect a genuine continuation of recruitment of this quasi‐RCT. We continue to have reservations about the conduct and validity of both these trials; their inclusion, however, does not distract from our findings of very low‐quality evidence for all comparisons.

Agreements and disagreements with other studies or reviews

In the first version of the review, we found two published non‐Cochrane systematic reviews that evaluated the surgical treatments for patellar fractures (Dy 2012; Heusinkveld 2013a), and discussed their findings in relation to ours (Sayum Filho 2015). Neither review restricted their results to RCTs only.

Our new search update resulted in the identification of four new non‐Cochrane systematic reviews that assessed the surgical treatments for fractures of the patella in adults (Camarda 2016; Matthews 2018; Zha 2017; Zhang 2018). No review restricted their results to solely RCTs or quasi‐RCTs.

Camarda 2016 evaluated the evidence for non‐metallic implants for patellar fracture fixation. It included nine mixed‐design studies, only two of which were RCTs (Chen 1998; Juutilainen 1995); both studies were included in Sayum Filho 2015. As the authors only summarised the results for the non‐metallic group, this review does not provide any additional insights of the comparison of non‐metallic versus metallic implants.

Matthews 2018 focused on comminuted patella in older people (aged 65 years or older) and examined the surgical options to avoid complications such as fixation failure and poor functional outcome. The 23 included studies (RCT, non‐RCT, prospective and retrospective cohort series and studies, case series, meta‐analysis and others) had 918 participants treated for patellar fracture. The authors concluded that older people with comminuted fractures have a higher likelihood of fixation failure and thus warrant special consideration. All five RCTs included in Matthews 2018 were included in Sayum Filho 2015 (Chen 1998; Günal 1997; Juutilainen 1995; Luna‐Pizarro 2006; Mao 2013). Matthews 2018 directed the reader to the Cochrane Review for "full assessment". The authors acknowledged the limitations of the evidence included in their review and raised the possibility of publication bias influencing the review findings. Other limitations of Matthews 2018 included the lack of a protocol, and the inclusion and pooling of evidence from mixed‐design studies.

Zha 2017 compared the therapeutic effect of CPS with KTB in the treatment of patella fractures among the Chinese Han population. Zha 2017 included 15 studies (RCTs, non‐RCTs, case‐control studies, retrospective studies and others) with 932 participants. The authors concluded that for treatment of patella fractures, among Chinese Han population, limited evidence suggests that the CPS is more suitable than the KTB when considering length of hospital stay, fracture healing time, flexion degree of knee six months after operation, incidence of postoperative complication and excellent‐good rate of the Böstman joint score. However, Zha 2017 recognised the lack of high‐quality evidence and proposed that more large‐scale RCTs are needed to validate the findings. Other limitations of Zha 2017 are the absence of a protocol and the inclusion and pooling of evidence from mixed‐design studies. Both RCTs included in Zha 2017 are included in our review (Mao 2013; Tian 2015).

Zhang 2018 compared the efficacy of KTB fixation with other alternatives for patella fractures. Their results, based on data from nine mixed‐design studies (RCTs, non‐RCTs, case‐control studies, retrospective studies and others), included 949 participants treated for patellar fracture. The three RCTs included in Zhang 2018 are included in our review (Lin 2015; Mao 2013; Tian 2015). The authors concluded that alternative treatment strategies may be more effective for management of patella fractures than Kirchner tension band in reducing the incidence of complications, VAS score and increasing flexion degree, the Böstman joint function score and Lysholm score. However, the authors described several limitations in their review: several included studies were retrospectively performed, other studies were not randomised, there were small sample sizes, the non‐inclusion of studies of non‐metallic substitutes for Kirchner tension band and studies not published in English. As well as limitations, absence of a protocol and the inclusion and pooling of evidence from mixed‐design studies, shared by the above two reviews, we consider the decision by Zhang 2018 to pool different metallic tension bands techniques compared with different methods and devices for patellar osteosynthesis together is misguided.

Authors' conclusions

Implications for practice.

There is very limited evidence from randomised controlled trials (RCTs) or quasi‐RCTs on the relative effects of different surgical interventions for treating fractures of the patella in adults. There is no evidence from RCTs or quasi‐RCTs evaluating the relative effects of surgical versus conservative treatment or different types of conservative interventions.

Based on very low‐quality evidence from two or four small trial comparisons, we are uncertain whether methods of percutaneous osteosynthesis give better results than conventional open surgery; whether the cable pin system (open or percutaneous surgery) gives better results than the tension band technique; and whether biodegradable implants are better than metallic implants for displaced patellar fractures.

Based on very low‐quality evidence from single trial comparisons, we are uncertain of the relative effects of patellectomy with vastus medialis obliquus advancement versus simple patellectomy for comminuted patellar fractures; of a three‐dimensional strapping reduction technique versus a traditional towel clamp technique; of a modified tension band versus an AO tension band technique; and of adjustable claws and absorbable sutures versus a Kirschner wire technique.

Implications for research.

Despite the increase in the number of trials, our recommendations for research remain as in before in Sayum Filho 2015. We continue to believe that further RCTs (with adequate methodology) on interventions for treating fractures of the patella in adults are justified and necessary. As fracture of the patella is relatively common, adequately powered multicentre studies, with central and independent randomisation, should be developed. These future trials should meet the CONSORT criteria for the design and reporting of non‐pharmacological studies (Boutron 2008). In addition, future research should examine the effect of treatments, not only in trials comparing various interventions (surgical and conservative) with each other, but also trials comparing types of conservative treatments.

Further RCTs are needed, but, to optimise research effort, these should be preceded by research that aims to identify priority questions. We suggest that future research should consider current practice and differences in practice around the world to underpin multicentre, and preferably international, RCTs of high quality to address these priority questions. As well as for comparisons of different commonly used techniques of fixation, we recommend consideration be given to RCTs comparing different conservative interventions; for example, cylinder cast versus range of motion brace. Consideration is also needed for a specific focus in future trials on the increasingly important subgroup of fragility fractures of the patella occurring in older and mostly female patients.

In the meantime, in part as preparation for future trials of treatment interventions for patellar fractures, research is needed to identify or develop and then validate condition‐specific patient‐reported outcome measures; all future trials should collect validated knee function scores, pain outcomes, adverse events and cost outcomes. Systematic data collection at the short, intermediate and long term after treatment (i.e. during the first six weeks, six weeks to six months, and more than six months) is essential. Although the timing of final assessment will depend on practicalities, the interventions under investigation and the main outcomes of the trial, we suggest long‐term follow‐up should be at least one year.

What's new

Date	Event	Description
28 August 2020	New search has been performed	For this version of the review, we made the following changes. Revised and updated the 'Background'. Updated the search to January 2020. Included six new trials (Afsar 2014; Hai‐feng 2015; Lin 2015; Shao 2019; Tian 2015; Wei 2019). Upgraded the methodology, including assessment of risk of bias, use of GRADE to assess the certainty of the evidence and production of a 'Summary of findings' table. Changed the conclusions in line with the GRADE assessment.
28 August 2020	New citation required and conclusions have changed	Conclusions changed consistent with the new evidence from the inclusion of six new trials and four new comparisons; and the application of GRADE in the assessment of the quality of the evidence. Two new authors were added and three removed.

History

Protocol first published: Issue 2, 2012
Review first published: Issue 2, 2015

Appendices

Appendix 1. Search strategies

CENTRAL (CRS web)

The CENTRAL search was run in two stages: the first search was run in February 2019 and a second top‐up search was run in January 2020.

Search 1

1 MESH DESCRIPTOR Patella AND CENTRAL:TARGET (256)
2 MESH DESCRIPTOR Fractures, Bone AND CENTRAL:TARGET (1049)
3 MESH DESCRIPTOR Fracture Fixation EXPLODE ALL AND CENTRAL:TARGET (1414)
4 MESH DESCRIPTOR Fracture Healing AND CENTRAL:TARGET (457)
5 #2 or #3 or #4 (2372)
6 #1 and #5 (17)
7 (patell* near3 fracture*): AB,EH,KW,KY,MC,MH,TI,TO AND CENTRAL:TARGET (70)
8 #6 or #7 (75)
9 01/05/2014_TO_27/02/2019:CRSCREATED AND CENTRAL:TARGET (529882)
10 8AND9 (36)

Search 2 (top‐up search)

9. 27/02/2019_TO_15/01/2020:CRSCREATED AND CENTRAL:TARGET (342293)
10. #9 AND #8 (18)

MEDLINE (Ovid Online)

The MEDLINE search was run in two stages: the first search was run in February 2019 and a second top‐up search was run in January 2020.

Search 1

1 Patella/ (9473)
2 exp Fractures, Bone/ (173652)
3 exp Fracture Fixation/ (57699)
4 Fracture Healing/ (12363)
5 2 or 3 or 4 (183275)
6 1 and 5 (1538)
7 (patell* adj3 fracture*).tw. (1250)
8 6 or 7 (1916)
9 Randomized controlled trial.pt. (476026)
10 Controlled clinical trial.pt. (92894)
11 randomized.ab. (377174)
12 placebo.ab. (177577)
13 Drug therapy.fs. (2084259)
14 randomly.ab. (261703)
15 trial.ab. (391416)
16 groups.ab. (1628635)
17 9 or 10 or 11 or 12 or 13 or 14 or 15 or 16 (4036225)
18 exp Animals/ not Humans/ (4548295)
19 17 not 18 (3443925)
20 8 and 19 (140)
21 (201405* or 201406* or 201407* or 201408* or 201409* or 201410* or 201411* or 201412* or 2015* or 2016* or 2017* or 2018* or 2019*).ed,dt. (3926725)
22 20 and 21 (43)

Search 2 (top‐up search)

21 (201902* or 201903* or 201904* or 201905* or 201906* or 201907* or 201908* or 201909* or 201910* or 201911* or 201912* or 2020*).ed,dt. (1934147)
22 20 and 21 (24)

Embase (Ovid Online)

The Embase search was run in two stages: the first search was run in February 2019 and a second top‐up search was run in January 2020.

Search 1

1 Patella Fracture/ (1479)
2 Patella/ (8424)
3 exp Fracture/ (259266)
4 exp Fracture Treatment/ (100569)
5 exp Fracture Fixation/ (77783)
6 3 or 4 or 5 (295652)
7 2 and 6 (1003)
8 (patell* adj3 fracture*).tw. (1517)
9 1 or 7 or 8 (2736)
10 Randomized controlled trial/ (533029)
11 Clinical trial/ (950593)
12 Controlled clinical trial/ (458304)
13 Randomization/ (81018)
14 Single blind procedure/ (33741)
15 Double blind procedure/ (157211)
16 Crossover procedure/ (58040)
17 Placebo/ (329030)
18 Prospective study/ (497068)
19 ((clinical or controlled or comparative or placebo or prospective* or randomi#ed) adj3 (trial or study)).tw. (1162526)
20 (random* adj7 (allocat* or allot* or assign* or basis* or divid* or order*)).tw. (287317)
21 ((singl* or doubl* or trebl* or tripl*) adj7 (blind* or mask*)).tw. (224128)
22 (cross?over* or (cross adj1 over*)).tw. (98032)
23 ((allocat* or allot* or assign* or divid*) adj3 (condition* or experiment* or intervention* or treatment* or therap* or control* or group*)).tw. (395017)
24 RCT.tw. (30970)
25 10 or 11 or 12 or 13 or 14 or 15 or 16 or 17 or 18 or 19 or 20 or 21 or 22 or 23 or 24 (2762561)
26 Case Study/ or Abstract Report/ or Letter/ (1141984)
27 25 not 26 (2711985)
28 9 and 27 (316)
29 limit 28 to human (302)
30 (2014* or 2015* or 2016* or 2017* or 2018* or 2019*).dc,yr. (8254982)
31 29 and 30 (144)

Search 2 (top‐up search)

30 (2019* or 2020*).dc,yr. (2199468)
31 29 and 30 (32)

LILACS (Bireme iAH interface)

(mh:(Patella AND ("Fractures, Bone" OR "Fracture Fixation" OR "Fracture Healing"))) OR (tw:(fract* AND (tw:patell* OR tw:patel* OR tw:rotula))) (78)

ClinicalTrials.gov

fracture OR fractures | patella OR patellar | First posted from 1 January 2014 to 3 April 2019 (2)
fracture OR fractures | patella OR patellar | First posted from 3 April 2019 to 14 January 2020 (2)

WHO ICTRP

(patella OR patellar) AND (fracture OR fractures) 1 January 2014 and 3 April 2019 (14)
(patella OR patellar) AND (fracture OR fractures) 3 April 2019 and 14 January 2020 (1)

Data and analyses

Comparison 1. Biodegradable versus metallic implants.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1.1 Anterior knee pain at long‐term follow‐up (presence of pain)	2	47	Risk Ratio (M‐H, Fixed, 95% CI)	0.56 [0.05, 5.62]
1.2 Major adverse outcomes	2		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
1.2.1 Failure of treatment	2	48	Risk Ratio (M‐H, Fixed, 95% CI)	2.14 [0.11, 42.52]
1.2.2 Infection	2	48	Risk Ratio (M‐H, Fixed, 95% CI)	0.37 [0.02, 8.51]
1.2.3 Delayed wound healing	2	48	Risk Ratio (M‐H, Fixed, 95% CI)	5.53 [0.28, 107.96]
1.2.4 Total adverse events	2	48	Risk Ratio (M‐H, Fixed, 95% CI)	2.19 [0.35, 13.66]
1.3 Observer‐rated measures of knee function	2		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
1.3.1 Reduction in knee motion	2	47	Risk Ratio (M‐H, Fixed, 95% CI)	1.11 [0.17, 7.09]

Comparison 2. Patellectomy with vastus medialis obliquus (VMO) advancement versus simple patellectomy for comminuted fractures.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
2.1 Patient‐rated knee function score	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
2.1.1 Levack score – participants with good results at long‐term follow‐up	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
2.2 Anterior knee pain at long‐term follow‐up (presence of pain)	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
2.3 Major adverse outcome	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
2.3.1 Patellar tendon subluxation	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
2.3.2 Quadriceps rupture	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
2.4 Observer‐rated measures of knee function at long‐term follow‐up	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
2.4.1 Number of participants with unlimited activity	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
2.4.2 Number of participants with no loss of quadriceps strength	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
2.4.3 Number of participants who could run on spot	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
2.4.4 Number of participants who could jump on 1 leg	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected

Comparison 3. Percutaneous patellar osteosynthesis (PPO) versus open surgery.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
3.1 Patient‐rated knee function score (Lysholm‐range: 0 to 100 points; 100 = better)	1		Mean Difference (IV, Random, 95% CI)	Totals not selected
3.1.1 At intermediate‐term follow‐up (3 months)	1		Mean Difference (IV, Random, 95% CI)	Totals not selected
3.1.2 At intermediate‐term follow‐up (6 months)	1		Mean Difference (IV, Random, 95% CI)	Totals not selected
3.1.3 At long‐term follow‐up (12 months)	1		Mean Difference (IV, Random, 95% CI)	Totals not selected
3.2 Anterior knee pain (measured by VAS: 0 to 10 cm; 10 = worst score)	4		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
3.2.1 At short‐term follow‐up (4 weeks or 1 month)	3	131	Mean Difference (IV, Fixed, 95% CI)	‐1.89 [‐2.32, ‐1.46]
3.2.2 At intermediate‐term follow‐up (8 weeks to 3 months)	4	183	Mean Difference (IV, Fixed, 95% CI)	‐1.88 [‐2.34, ‐1.42]
3.2.3 At intermediate‐term follow‐up (6 months)	2	92	Mean Difference (IV, Fixed, 95% CI)	‐0.40 [‐0.74, ‐0.06]
3.2.4 At long‐term follow‐up (12 months)	1	52	Mean Difference (IV, Fixed, 95% CI)	‐0.60 [‐1.44, 0.24]
3.3 Major adverse outcomes	4		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
3.3.1 Failure of treatment: reduction was impossible requiring change in procedure	2	105	Risk Ratio (M‐H, Fixed, 95% CI)	0.99 [0.18, 5.52]
3.3.2 Failure of treatment (loss of reduction)	3	145	Risk Ratio (M‐H, Fixed, 95% CI)	0.15 [0.04, 0.54]
3.3.3 Infection	2	105	Risk Ratio (M‐H, Fixed, 95% CI)	0.16 [0.02, 1.31]
3.3.4 Hardware complications (irritation, broken or migration)	4	183	Risk Ratio (M‐H, Fixed, 95% CI)	0.29 [0.15, 0.56]
3.3.5 Delayed wound healing	1	40	Risk Ratio (M‐H, Fixed, 95% CI)	0.20 [0.01, 3.92]
3.3.6 Total adverse events	3	145	Risk Ratio (M‐H, Fixed, 95% CI)	0.26 [0.14, 0.46]
3.4 Observer‐rated measures of knee function	3		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only
3.4.1 Knee function scores (KSCRS and Böstman score) at intermediate‐term follow‐up	3	131	Std. Mean Difference (IV, Random, 95% CI)	1.26 [0.32, 2.20]
3.4.2 Knee function scores (KSCRS and Böstman score) at long‐term follow‐up (12 months)	3	124	Std. Mean Difference (IV, Random, 95% CI)	1.01 [0.08, 1.95]
3.4.3 Knee function scores (KSCRS and Böstman score) at long‐term follow‐up (24 months)	2	84	Std. Mean Difference (IV, Random, 95% CI)	0.62 [0.18, 1.06]
3.5 Knee range of motion: extension (degrees)	4		Mean Difference (IV, Random, 95% CI)	Subtotals only
3.5.1 Extension at short‐term follow‐up (4 weeks to 1 month)	3	131	Mean Difference (IV, Random, 95% CI)	1.88 [0.82, 2.95]
3.5.2 Extension at intermediate‐term follow‐up (8 weeks to 3 months)	3	131	Mean Difference (IV, Random, 95% CI)	2.28 [0.15, 4.42]
3.5.3 Extension at long‐term follow‐up (12 months)	1	52	Mean Difference (IV, Random, 95% CI)	‐0.70 [‐1.68, 0.28]
3.6 Knee range of motion: flexion (degrees)	4		Mean Difference (IV, Random, 95% CI)	Subtotals only
3.6.1 Flexion at short‐term follow‐up (4 weeks to 1 month)	3	131	Mean Difference (IV, Random, 95% CI)	22.01 [10.77, 33.25]
3.6.2 Flexion at intermediate‐term follow‐up (8 weeks to 3 months)	4	183	Mean Difference (IV, Random, 95% CI)	24.02 [7.22, 40.82]
3.6.3 Flexion at long‐term follow‐up (12 months)	2	92	Mean Difference (IV, Random, 95% CI)	5.37 [0.94, 9.80]
3.6.4 Flexion at long‐term follow‐up (24 months)	1	40	Mean Difference (IV, Random, 95% CI)	8.30 [1.47, 15.13]
3.7 Hardware removal (usually due to pain or subcutaneous irritation)	3		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
3.7.1 Hardware removal at intermediate‐term follow‐up (8 weeks)	1	53	Risk Ratio (M‐H, Fixed, 95% CI)	0.53 [0.21, 1.39]
3.7.2 Hardware removal at long‐term follow‐up (12 weeks to 2 years)	3	145	Risk Ratio (M‐H, Fixed, 95% CI)	0.51 [0.35, 0.74]
3.7.3 Total hardware removal	3	145	Risk Ratio (M‐H, Fixed, 95% CI)	0.53 [0.40, 0.70]

Comparison 4. A new intraoperative reduction technique versus a standard technique.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
4.1 Patient‐rated knee function: Hospital for Special Surgery (0 to 100 points; 100 = better) (6 months)	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected

Comparison 5. A modified tension band technique versus conventional AO tension band wiring technique.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
5.1 Anterior knee pain (visual analogue score) (0 to 10; 10 = worst pain) (12 months)	1		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
5.2 Major adverse outcomes	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
5.2.1 Loss of reduction (2 weeks)	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
5.2.2 Loss of reduction (3 months)	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
5.2.3 Wound infection	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
5.2.4 Prominence of hardware	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
5.2.5 Avascular necrosis	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
5.2.6 Refracture (12 months)	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
5.2.7 Post‐traumatic arthritis	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
5.3 Observer‐rated measures of knee function	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
5.3.1 Quadriceps diameter wasting (6 months) (cm)	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
5.3.2 Quadriceps diameter wasting (12 months) (cm)	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
5.3.3 Extension lag (3 months) (degrees)	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
5.3.4 Extension lag (12 months) (degrees)	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
5.3.5 Quadriceps power (3 months) (% of other leg)	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
5.3.6 Quadriceps power (12 months) (% of other leg)	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected

Comparison 6. Adjustable patella claws and absorbable suture (new device) versus Kirschner wire tension band for comminuted patellar fractures.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
6.1 Patient‐rated knee function: 'excellent' Hospital for Special Surgery (2 years)	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
6.2 Major adverse outcomes	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
6.2.1 Frictional pain	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
6.2.2 Device loosening	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
6.2.3 Hardware breakage	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
6.2.4 Delayed wound healing	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
6.2.5 Total adverse outcomes	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected

Comparison 7. Cable pin system versus tension band technique.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
7.1 Patient‐rated knee function (Hospital for Special Surgery score: 0 to 100; 100 = best function) (1 year)	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
7.2 Anterior knee pain (measured by visual analogue scale: 0 to 10 cm: 10 = worst score)	1		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
7.2.1 Short‐term follow‐up (1 month)	1	40	Mean Difference (IV, Fixed, 95% CI)	‐2.00 [‐2.78, ‐1.22]
7.2.2 Intermediate‐term follow‐up (3 months)	1	40	Mean Difference (IV, Fixed, 95% CI)	‐1.40 [‐2.11, ‐0.69]
7.2.3 Intermediate‐term follow‐up (6 months)	1	40	Mean Difference (IV, Fixed, 95% CI)	‐0.30 [‐0.65, 0.05]
7.3 Major adverse outcomes	2		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
7.3.1 Failure of treatment (loss of reduction)	1	40	Risk Ratio (M‐H, Fixed, 95% CI)	0.20 [0.01, 3.92]
7.3.2 Infection	1	73	Risk Ratio (M‐H, Fixed, 95% CI)	0.57 [0.05, 6.05]
7.3.3 Hardware complications (irritation, broken or migration)	2	113	Risk Ratio (M‐H, Fixed, 95% CI)	0.14 [0.04, 0.49]
7.3.4 Delayed wound healing	1	40	Risk Ratio (M‐H, Fixed, 95% CI)	0.20 [0.01, 3.92]
7.3.5 Total adverse events	2	113	Risk Ratio (M‐H, Fixed, 95% CI)	0.14 [0.05, 0.44]
7.4 Observer‐rated measures of knee function: Böstman score (0 to 30: 30 = best function)	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
7.4.1 At intermediate‐term follow‐up (3 months)	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
7.4.2 At long‐term follow‐up (12 months)	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
7.4.3 At long‐term follow‐up (24 months)	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
7.5 Knee range of motion: extension (degrees)	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
7.5.1 Extension at short‐term follow‐up (1 month)	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
7.5.2 Extension at intermediate‐term follow‐up (3 months)	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
7.6 Knee range of motion: flexion (degrees)	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
7.6.1 Flexion at short‐term follow‐up (1 month)	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
7.6.2 Flexion at intermediate follow‐up (3 months)	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
7.6.3 Flexion at long‐term follow‐up (12 months)	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
7.6.4 Flexion at long‐term follow‐up (24 months)	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
7.7 Hardware removal	2	113	Risk Ratio (M‐H, Fixed, 95% CI)	0.23 [0.09, 0.55]

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Afsar 2014.

Study characteristics
Methods	Study design: quasi‐randomised prospective controlled trial (consecutively allocated) Duration of study: 18 March 2009 to 17 March 2012 Protocol finalised before recruitment of participants: not reported Details of trial registration: not reported Funding sources: not reported
Participants	Place of study: (Lady Reading Hospital) Private Medical Center of Peshawar, Pakistan Number of participants assigned: 108 Number of participants assessed: 65 Inclusion criteria: people presenting with clinical and radiological evidence of fracture of the patella secondary to trauma informed written consent Exclusion criteria: fractures with comminution of a degree not amenable to osteosynthesis open fracture associated long bone fractures of the same limb people with ASA III Age: conventional AO TBW technique: mean 31.58 (SD 8.55) modified technique: mean 31.34 (SD 6.74) Gender: only proportion (men:women): conventional AO TBW technique: 15.5:1; modified technique: 15:1 Classification of injury: no classification used. With radiological evidence of > 2 mm displacement between either of the comminuted fragments, people underwent TBW with cerclage. All patterns of fracture were included. Those fractures with comminution of a degree not amenable to osteosynthesis were excluded as they were planned for partial patellectomy.
Interventions	Timing of intervention: not reported Types of surgical interventions Conventional AO TBW technique (placement of a cerclage wire around the anterior surface of the patella apposing the comminuted segments followed by outwards‐inward placement of 2 parallel Kirchner's (K‐wires) extending from the superior to the inferior pole of the patella. Then, the wires were fixed in a figure of 8 configuration with a vertical orientation and a single twist around the K‐wires at corners. To finish the wires were cut. The wires were of a stainless‐steel variety). Modified technique (similar cerclage was placed followed by placement of parallel K‐wires, but the tension band was horizontally oriented with double twist around the K‐wires at corner and instead of simply cutting the K‐wires, they were bent to hook the tension band wire. The wires were stainless steel. Rehabilitation: in both groups, participants were immobilised for 3 weeks followed by prone hang exercises at 3 weeks and crutches, which were discontinued after 6 weeks.
Outcomes	Length of final follow‐up: all participants were advised follow‐up visits at 2 weeks, and 1, 3, 6 and 12 months Loss of follow‐up: of 108 participants, 28 with open fractures and 6 with closed fractures had associated fractures on the same limb and were excluded from the study. 9 participants were lost during follow‐up and were also excluded from the study. That left 65 participants for final analysis. Primary outcomes Patient‐rated knee function scores – not assessed Anterior knee pain – assessed using VAS Major adverse outcomes – frequency of complications (loss of reduction at 2 and 3 months, wound infection, prominence of hardware, avascular necrosis, refracture at 12 months, post‐traumatic arthritis) Secondary outcomes Observer‐rated measures of knee function – indirectly assessed measuring only: ROM, quadriceps wasting, quadriceps power and extension lag Health‐related quality of life scores – not assessed Return, including time to return, to previous activities – not assessed Cosmetic appearance – not assessed Other outcomes: quadriceps diameter in comparison to contralateral limb
Notes	We contacted the authors to obtain further information; however, they did not reply.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Method of generating the random sequence was not described, it was only described as consecutively allocated (we considered this study a quasi‐randomised controlled trial).
Allocation concealment (selection bias)	High risk	Description of the method of allocation is unhelpful but almost certainly the non‐probability consecutive sampling rules out randomisation.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Participants and personnel were probably not blinded.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Outcome assessors were not blinded.
Incomplete outcome data (attrition bias) All outcomes	High risk	43 (39%) participants were lost to follow‐up.
Selective reporting (reporting bias)	High risk	There was no protocol. Pain was only measured at 12 months (with VAS), the loss of reduction was only measured at 2 and 3 weeks. The complications (wound infection, prominence of hardware, avascular necrosis, refracture in 12 months and post‐traumatic arthritis) were reported in a very simple way, without details (only the quantities of them in each group were given). Extension lag was only measured at 2, and 3 and 12 months. Quadriceps power was only measured at 3 weeks, and 3 and 12 months. 9 participants were lost during follow‐up, with no more details.
Other bias	High risk	The authors did not provide sufficient information about the surgeons and care providers or sufficient details of rehabilitation after surgery. There was a potential conflict of interest reflecting the lack of independent scrutiny as the trial tested a new tension band technique.

Chen 1998.

Study characteristics
Methods	Study design: randomised controlled trial Duration of study: August 1994 to May 1997 Protocol finalised before recruitment of participants: not reported Details of trial registration: not reported Funding sources: not reported
Participants	Place of study: Changzheng Hospital, Shanghai, China Number of participants assigned: 38 (18 with biodegradable implant and 20 with metallic implant) Number of participants assessed: 38 (18 with biodegradable implant and 20 with metallic implant) Inclusion criteria: aged > 20 years closed, displaced patella fracture transverse or oblique fracture of the patella with 2 or 3 fragments seen in x‐rays Exclusion criteria: severe comminuted fractures Age: overall mean 46; range 20–76 years biodegradable group: mean 45; range 24–72 years metallic group: mean 47; range 20–76 years Gender: 27 men; 11 women Classification of injury: all fractures were transverse or oblique with 2 or 3 fragments (closed fractures)
Interventions	Timing of intervention: biodegradable group: mean 1.7, range 0–8 days metallic group: mean 1.8; range 0–7 days Types of surgical interventions Biodegradable group: this was a modification of the traditional TBW, but made of 2 self‐reinforced polyglycolide acid plugs and a polyester ligament Metallic group: the traditional tension band with 2 K‐wires and 1 metallic cerclage wire Rehabilitation: same in both groups, plastic splints were used for 3 weeks; weight‐bearing was allowed immediately postoperatively
Outcomes	Length of final follow‐up: mean 24; range 14 to 32 months biodegradable group: mean 23; range 14 to 32 months metallic group: mean 24; range 14 to 32 months Participants were also evaluated at 3 and 8 weeks and 3, 6 and 12 months after the surgery Loss of follow‐up: 0 (implied) Primary outcomes Patient‐rated knee function scores – not assessed Anterior knee pain – indirectly assessed using a non‐validated score Major adverse outcomes – delayed wound healing and infection Secondary outcomes Observer‐rated measures of knee function – indirectly assessed using a non‐validated score Health‐related quality of life scores – not assessed Return, including time to return, to previous activities – not assessed Cosmetic appearance – not assessed
Notes	We contacted the authors to obtain further information; however, they did not reply. The authors reported that outcomes were assessed at 3 and 8 weeks and at 3, 6 and 12 months after the surgery; however, they only described outcomes data at final follow‐up (long‐term follow‐up).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Method of generating the random sequence not reported.
Allocation concealment (selection bias)	Unclear risk	Whether and how allocation was concealed not reported.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Participants and personnel were probably not blinded.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Outcomes assessors were not blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	No loss to follow‐up.
Selective reporting (reporting bias)	High risk	Patient‐rated knee function and health‐related quality of life scores were not assessed. Anterior knee pain and observer‐rated measures of knee function were indirectly assessed using a non‐validated score. No details of the trial protocol available.
Other bias	Unclear risk	The authors did not provide sufficient information about the surgeons and care providers or sufficient details of rehabilitation after surgery.

Günal 1997.

Study characteristics
Methods	Study design: prospective randomised trial Duration of study: not reported Protocol finalised before recruitment of participants: not reported Details of trial registration: not reported Funding sources: none known
Participants	Place of study: Osmangazi University Hospital, Turkey Number of participants assigned: 28 (12 patellectomy with advancement of vastus medialis obliquus and 16 simple patellectomy) Number of participants assessed: 28 (12 patellectomy with advancement of vastus medialis obliquus and 16 simple patellectomy) Inclusion criteria: skeletally mature people comminuted fracture of the patella with ≥ 5 fragments seen in x‐rays Exclusion criteria: concomitant fractures radiographic signs of osteoarthritis of the knee Age: mean 28.3 years patellectomy + advancement of vastus medialis obliquus group: mean 28.1 years simple patellectomy group: mean 28.4 years Gender: 16 men; 12 women Classification of injury: all comminuted fractures (no mention of open fractures)
Interventions	Timing of intervention: not reported Types of surgical interventions: patellectomy with vastus medialis obliquus advancement simple patellectomy Rehabilitation: same in both groups First postoperative day: partial weight‐bearing with immobiliser and isometric quadriceps exercises Fourth postoperative day: active assisted exercises and immobiliser discarded Flexion of the knee to 90° was achieved by 4 weeks after the operation
Outcomes	Length of follow‐up: mean 4.2 years; range 3 to 5.5 years Loss of follow‐up: 0 Primary outcomes Patient‐rated knee function scores – measured by score described by Levack 1985 (non‐validated) Anterior knee pain – assessed by a dichotomous endpoint Major adverse outcomes – patellar tendon subluxation (heterotopic ossification also assessed) Secondary outcomes: Observer‐rated measures of knee function – assessed using a non‐validated score: limitation of activity, loss of quadriceps strength, functional assessment using a non‐validated score described by Marshall 1977 Health‐related quality of life scores – not assessed Return, including time to return, to previous activities – not assessed Cosmetic appearance – not assessed
Notes	We contacted the authors to obtain further information; however, they did not reply. They described outcomes data at final follow‐up only (long‐term follow‐up).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Participants were randomly assigned by drawing lots.
Allocation concealment (selection bias)	Unclear risk	Allocation concealment was not reported.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Participants and personnel were probable not blinded.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Outcomes assessors were not blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	No loss to follow‐up.
Selective reporting (reporting bias)	High risk	Patient‐rated knee function scores and observer‐rated measures of knee function measured by a non‐validated score. Anterior knee pain assessed only by a dichotomous endpoint. Health‐related quality of life scores not assessed. No details of the trial protocol available.
Other bias	Unclear risk	Insufficient information provided, the authors reported only the following baseline characteristics: age and gender.

Hai‐feng 2015.

Study characteristics
Methods	Study design: randomised controlled trial Duration of study: January 2009 to January 2012 Protocol finalised before recruitment of participants: not reported Details of trial registration: not reported Funding sources: none known
Participants	Place of study: Department of Orthopedics and Trauma, 208 Hospital of PLA, Changchun, Jilin Province, China Number of participants assigned: 57 Number of participants assessed: 57 Inclusion criteria: any comminuted patellar fracture signed informed consent form Exclusion criteria: people with contraindications for surgery Age: mean 38.5; range 29 to 68 years Gender: 30 men; 27 women Classification of injury: not reported. All fractures were considered comminuted (it was classified by the cause of fracture: 35 cases of fall, 22 cases of car accidents)
Interventions	Timing of intervention: 1–4 days after trauma Types of surgical interventions Group 1: open reduction and adjustable patella claws (new metallic implant) plus absorbable suture fixation Group 2: open reduction and Kirschner wire tension band fixation Rehabilitation: did not use external fixation after surgery, the quadriceps function exercise began after the operation. After 2 days, continuous passive movement was used for passive rehabilitation training.
Outcomes	Length of final follow‐up: 2 years Loss of follow‐up: none lost to follow‐up Primary outcomes Patient‐rated knee function scores – HSS Knee function score Anterior knee pain – not assessed Major adverse outcomes – delayed wound healing, infection, loosening, hardware breakage, skin irritation and frictional articular surface pain Secondary outcomes: Observer‐rated measures of knee function – not assessed Health‐related quality of life scores – not assessed Return, including time to return, to previous activities – not assessed Cosmetic appearance – not assessed Other outcomes: incision length, comparison of surgery time, fracture healing (the criteria for fracture healing included local no tenderness, no longitudinal slap pain, no abnormal activity locally; x‐ray films showed blurred fracture lines, continuous osteophytes through the fracture line, etc.)
Notes	We contacted the authors to obtain further information; however, they did not reply.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Participants randomly assigned by random number table method.
Allocation concealment (selection bias)	Unclear risk	Allocation concealment not reported.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Participants and personnel not blinded.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Outcomes assessors not blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	No loss to follow‐up.
Selective reporting (reporting bias)	High risk	Anterior knee pain, observer‐rated measures of knee function, health‐related quality of life scores, return to previous activities, time to return to previous activities and cosmetic appearance were not assessed. Not reported if there was a trial protocol or a previous trial registration.
Other bias	High risk	The authors did not provide sufficient information about the surgeons and care providers, or sufficient details of rehabilitation after surgery. They did not report details relating to follow‐up (number of postsurgery medical consultation, timing of postsurgery medical consultation). There was a potential conflict of interest reflecting the lack of independent scrutiny as the trial tested a new device for treating fractures of the patella.

Juutilainen 1995.

Study characteristics
Methods	Study design: prospective randomised trial Duration of study: not reported Protocol finalised before recruitment of participants: not reported Details of trial registration: not reported Funding sources: study was supported by grants from the Academy of Finland, the Foundation of Orthopaedic and Traumatology in Finland, and the Finnish Parliament
Participants	Place of study: Helsinki University Central Hospital, Helsinki, Finland Number of participants assigned: 10 (6 with biodegradable implant and 4 with metallic implant) Number of participants assessed: 9 (5 with biodegradable implant and 4 with metallic implant) Inclusion criteria: aged > 16 years fracture of the patella with 2 or 3 fragments seen in x‐rays Exclusion criteria: fractures with > 3 fragments > 14 days since the trauma open fractures people with mental illness or chronic alcoholism Age: mean 47.6; range 29 to 69 years biodegradable group: mean 44.4; range 34 to 62 years metallic group: mean 51.7; range 29 to 69 years Gender: 2 men; 7 women Classification of injury: all fractures were transverse or oblique with 2 or 3 fragments (closed fractures)
Interventions	Timing of intervention: biodegradable group: mean 2.0; range 0–7 days metallic group: mean 6.3; range 0–13 days Types of surgical interventions Biodegradable group: 2 self‐reinforced polyglycolide acid screws or plugs and a "special wire" made of self‐reinforced poly‐L‐lactide acid Metallic group: tension band with 2 K‐wires and 1 metallic cerclage wire Rehabilitation: same in both groups, a plaster cast used for 6 weeks, weight‐bearing allowed immediately postoperatively. Stitches removed after 2 weeks.
Outcomes	Length of follow‐up Biodegradable group: mean 1.4; range 0.3 to 2 years Metallic group: mean 1.6; range 1 to 2 years Participants were evaluated at 2, 4, 6 weeks, 3 and 6 months, 1 and 2 years postoperatively Loss of follow‐up: authors reported that 1 participant from the biodegradable group was excluded from the comparison because the implant failed after a new trauma Primary outcomes Patient‐rated knee function scores – not assessed Anterior knee pain – indirectly assessed using a non‐validated score Major adverse outcomes Secondary outcomes Observer‐rated measures of knee function – indirectly assessed using a non‐validated score Health‐related quality of life scores – not assessed Return, including time to return, to previous activities – not assessed Cosmetic appearance – not assessed
Notes	The authors analysed olecranon and patella fractures in the same study. All metallic implants were removed 1 year after the primary operation. We contacted the authors to obtain further information; however, they did not reply. The authors reported that they assessed outcomes at 2, 4, 6 weeks, 3 and 6 months, 1 and 2 years postoperatively; however, they only described outcome data at final follow‐up (long‐term follow‐up).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Method of generating the random sequence not reported.
Allocation concealment (selection bias)	Unclear risk	Whether and how allocation was concealed not reported.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Participants and personnel were probably not blinded.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Outcome assessors were not blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Only 1 participant was lost to follow‐up.
Selective reporting (reporting bias)	High risk	Patient‐rated knee function and health‐related quality of life scores were not assessed. Anterior knee pain and observer‐rated measures of knee function were indirectly assessed using a non‐validated score. No details of the trial protocol available.
Other bias	Unclear risk	Authors did not provide sufficient information about the surgeons and care providers or sufficient details of rehabilitation after surgery.

Lin 2015.

Study characteristics
Methods	Study design: randomised controlled trial Duration of study: June 2008 to June 2013 Protocol finalised before recruitment of participants: pilot study of 14 participants finalised first Details of trial registration: registration trial (Chinese Clinical Trial Register): current controlled trials ChiCTR‐PRCH‐14005017, registration date 14 June 2014. CNO – CN‐01169535 and a pilot of 14 participants was done to ensure feasibility. Funding sources: not reported
Participants	Place of study: Union Hospital of Tongji Medical College, Huazhong University of Science and Technology, China Number of participants assigned: 63 Number of participants assessed: 52 Inclusion criteria: closed transverse patellar frat cures (with displacement < 8 mm, and no comminution at fracture site) Exclusion criteria: open fractures or multiple fractures people who presented > 5 days after the injury C1.2 or C1.3 fractures transverse fractures with displacement > 8 mm according to computed tomography Age: mean 51.7; range 22 to 76 years percutaneous group: mean 50.0 (SD 16.3) years open surgery group: mean 52.5 (SD 17.4) years Gender: 28 men; 24 women Classification of injury: percutaneous group: mechanism (15 fall; 4 sports related trauma; 7 traffic accident) type (AO/OTA) transverse, middle (45C1.1): 15 (57.7%) transverse, proximal (45C1.21): 4 (15.4%) transverse, distal (45C1.3): 7 (27%) interfragmentary gap: mean 4.3 (SD 1.8) mm open surgery group: mechanism (14 fall; 6 sports related trauma; 6 traffic accident) type (AO/OTA) transverse, middle (45C1.1): 17 (65.4%) transverse, proximal (45C1.21): 3 (11.5%) transverse, distal (45C1.3): 6 (23%) interfragmentary gap: mean 4.7 (SD 1.7) mm
Interventions	Timing of intervention: operative time to injury: mean 3 days after the initial injury percutaneous group: mean 2.8 (SD 1.2) days open surgery group: mean 3.3 (SD 1.4) days Types of surgical interventions: CRCF ORTF of transverse patellar fractures Rehabilitation: external immobilisers were not used in any of the groups. Participants performed quadriceps‐femur contraction exercises soon after the operation. Unrestricted passive ROM was started postoperatively at the earliest time point, depending on the participant's pain tolerance. Active ROM was encouraged by 3 weeks postoperatively, and full weightbearing was started by 8 weeks
Outcomes	Length of final follow‐up: minimum 1 year Participants were evaluated at 15 days, and 1, 3, 6 and 12 months Loss of follow‐up: percutaneous group: 6 participants in 3 months open surgery group: 5 participants in 3 months Primary outcomes Patient‐rated knee function scores – Lysholm Knee Scoring Scale Anterior knee pain – by VAS score Major adverse outcomes – complications (postoperative displacement, infection, painful hardware, tension band loosening or migration of the hardware, reoperation rate) Secondary outcomes Observer‐rated measures of knee function – knee ROM, flexion, Health‐related quality of life scores – not assessed Return, including time to return, to previous activities – not assessed Cosmetic appearance – not assessed Hardware removal: indicated in cases of painful hardware, skin irritation, postoperative fracture displacement before healing Other outcomes: radiographic union
Notes	We contacted the authors to obtain further information; however, they did not reply.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Sequence generation performed by a computer random number generator.
Allocation concealment (selection bias)	Unclear risk	Whether and how allocation was concealed was not reported.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Participants and personnel were probably not blinded.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Outcome assessors were not blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	82.5% of participants completed the follow‐up. Missing outcome data balanced in numbers across intervention groups, with similar reasons for missing data across groups (only 11 were missed: 6 from the percutaneous group and 5 from the open group).
Selective reporting (reporting bias)	Unclear risk	Patient‐rated knee function (Lysholm) was assessed, but health‐related quality of life score, return to previous activities, time to return to previous activities and cosmetic appearance were not assessed. Anterior knee pain and observer‐rated measures of knee function were assessed using VAS and goniometry. The trial was registered in the Chinese Clinical Trial Register, and a pilot was performed before.
Other bias	Low risk	The authors provided sufficient information about the surgeons and care providers and showed sufficient details of rehabilitation after surgery. They reported several limitations of this study: the percutaneous surgery was indicated for displaced transverse patellar fractures (but not > 8 mm of displacement), this was a single centre and open‐label trial, and the sample size was small. These were not judged in this category.

Luna‐Pizarro 2006.

Study characteristics
Methods	Study design: randomised controlled trial Duration of study: March 2000 to May 2002 Protocol finalised before recruitment of participants: protocol referred to in the trial report Details of trial registration: not reported Funding sources: none known
Participants	Place of study: tertiary trauma centre in Mexico City, Mexico Number of participants assigned: 53 (27 with percutaneous patellar osteosynthesis system and 26 with open surgery) Number of participants assessed: 45 (23 with percutaneous patellar osteosynthesis system and 22 with open surgery) Inclusion criteria: aged > 16 years clinical and radiographic evidence of patella fracture with > 3 mm displacement acute fracture (clinical course < 48 hours) patient informed consent Exclusion criteria: comminuted fractures chronic degenerative joint disease uncompensated diabetes previous knee surgical intervention open fractures poly‐fractures cranioencephalic trauma severe osteopenia, peripheral neural damage alcoholism and drug abuse Age: overall mean 47; range 16 to74 years percutaneous group: mean 51 (SD 14.8) years open surgery group: mean 44 (SD 18.2) years Gender: 30 men; 23 women Classification of injury: all fractures classified according to Orthopaedic Trauma Association fracture classification (closed fractures). 2 types described: complete articular, disrupted extensor mechanism, transverse and middle; and complete articular, disrupted extensor mechanism, transverse distal
Interventions	Timing of intervention: up to 48 hours after injury Types of surgical interventions Percutaneous patellar osteosynthesis using a locally designed device. A haemostatic clamp was introduced to reduce the fracture. After reduction, 2 K‐wires were placed percutaneously through the sliding guides from proximal to distal. Then a 1.2‐mm wire was introduced through the inferior‐lateral portal toward the inferior‐medial portal and, after repositioning the knee, reintroduced through the inferior‐medial portal toward the superior‐lateral portal. The "figure‐of‐eight" configuration was completed by reintroducing the wire through the inferior‐lateral portal towards the superior‐medial portal under the skin and over the patella. Then the wire was reintroduced through the superior‐medial portal towards the superior‐lateral portal under the K‐wires and closed with a subcutaneous twist tight knot. Open surgery (customary technique with K‐wire and metallic tension band technique) Rehabilitation: same in both groups, participants initiated physiotherapy and rehabilitation 12 hours after surgery, isomeric and isotonic contractions of the quadriceps for 30 minutes, 4 times a day. Dressing was removed at the third postoperative day, no immobilisation was used.
Outcomes	Length of follow‐up: 2 years Participants were evaluated at 4 and 8 weeks and at 12 and 24 months after the surgery Loss of follow‐up: percutaneous group: 4 participants in 24 months open surgery group: 4 participants in 24 months Primary outcomes Patient‐rated knee function scores – not assessed Anterior knee pain – by VAS at 4 and 8 weeks using a scale from 0 (no pain) to 10 (the most intense pain) Major adverse outcomes – infection, displaced fragment, complications/problems with hardware, hardware removal Secondary outcomes Observer‐rated measures of knee function – knee ROM and Knee Society Clinical Rating System Health‐related quality of life scores – not assessed Return, including time to return, to previous activities – not assessed Cosmetic appearance – not assessed
Notes	We tried unsuccessfully to contact the authors to obtain further information. Different surgeons performed the 2 operations: 2 performed the percutaneous operation and 4 the open operation.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Method of generating the random sequence was not reported.
Allocation concealment (selection bias)	Unclear risk	Sealed envelopes were used. No details of sufficient safeguards to ensure concealment.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Participants and personnel were not blinded
Blinding of outcome assessment (detection bias) All outcomes	High risk	Outcome assessors were not blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Missing outcome data balanced in numbers across intervention groups, with similar reasons for missing data across groups (8 participants were missing in the evaluation at 24 months).
Selective reporting (reporting bias)	High risk	Patient‐rated knee function and health‐related quality of life scores were not assessed. No details of the trial protocol available.
Other bias	High risk	There was a potential conflict of interest reflecting the lack of independent scrutiny as the trial tested a new percutaneous technique using a locally designed device.

Mao 2013.

Study characteristics
Methods	Study design: randomised controlled trial Duration of study: April 2008 to June 2010 Protocol was published before recruitment of participants: not reported Details of trial registration: not reported Funding sources: none known
Participants	Place of study: Changhai Hospital, Shanghai, China Number of participants assigned: 40 (20 with percutaneous patellar osteosynthesis and 20 with open surgery) Number of participants assessed: 39 (20 with percutaneous patellar osteosynthesis and 19 with open surgery) Inclusion criteria: aged 18 to 65 years acute fracture (clinical course < 48 hours) clinical and radiographic evidence of a transverse patella fracture with > 3 mm displacement willingness to participate in study after providing informed consent Exclusion criteria: open fracture comminuted or multiple fractures multiple trauma fracture associated with pre‐existing osteoarthritis of the knee previous surgical intervention of the knee peripheral neural damage uncompensated diabetes serious osteoporosis Age: overall mean 41.8; range 22 to 65 years percutaneous group: mean 40.2; range 26 to 65 years open surgery group: mean 43.5; range 22 to 59 years Gender: 25 men; 15 women Classification of injury: all fractures were transverse patella fracture with > 3 mm displacement (closed fractures)
Interventions	Timing of intervention: up to 48 hours after injury Types of surgical interventions Percutaneous patellar osteosynthesis using the CPS (combination of interfragmentary screws molten to a tension band wire). 1 end of the system is a partially threaded 4.0 mm cancellous screw, the other end of the system is a guide needle which can be passed through bone tunnels, the material connecting the 2 ends is a cable with a special braiding structure. Open surgery (customary technique with K‐wire and metallic tension band technique) Rehabilitation: same in both groups, an elastic bandage was used for 2 days after surgery to reduce swelling and haematoma of the knee; no immobilisation was used; passive exercise was initiated on the first day after surgery; participants used a continuous passive motion machine, and active flexion exercises of the knee in a prone position were encouraged; active extension exercises of the knee were allowed 3 weeks after surgery, and full weight‐bearing was allowed after the fracture was radiographically healed
Outcomes	Length of follow‐up: 2 years Participants were evaluated at 1, 3, 6, 12 and 24 months after the surgery Loss of follow‐up: 1 participant from the open surgery group was excluded from the comparison at 24 months because they sustained a broken tibia as a result of a traffic accident 16 months after surgery Primary outcomes Patient‐rated knee function scores – not assessed Anterior knee pain – by VAS at 1, 3 and 6 months using a scale from 0 (no pain) to 10 (the most intense pain) Major adverse outcomes – delayed wound healing, displaced fragment, complications/problems with hardware, hardware removal Secondary outcomes Observer‐rated measures of knee function – knee ROM and Böstman score (Böstman 1981) Health‐related quality of life scores – not assessed Return, including time to return, to previous activities – not assessed Cosmetic appearance – not assessed
Notes	Journal supplements available with individual participant data for baseline characteristics and results. Both operations were conducted by same 2 surgeons using the same surgical protocol. We found that Yong‐liang 2015 was another report of Mao 2013, published in a Chinese journal, which reported results for 80 participants. However, the new report did not acknowledge Mao 2013 as a report of the trial and there was a surprisingly high degree of similarity on both the baseline data (e.g. identical means age and gender ratio) and the results (e.g. operating time, Böstman scores at immediate follow‐up) that was unexplained. We decided not to include data from Yong‐liang 2015 to prevent double‐counting.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Sequence generation performed by a computer random number generator.
Allocation concealment (selection bias)	Low risk	Sequentially numbered, opaque, sealed envelopes.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Participants and personnel were not blinded.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Outcome assessors were not blinded. However, there was some indication of data collection by clinicians other than the 2 surgeons.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Only 1 participant was lost to follow‐up.
Selective reporting (reporting bias)	High risk	Patient‐rated knee function and health‐related quality of life scores were not assessed. No details of the trial protocol available.
Other bias	High risk	There was a potential conflict of interest reflecting the lack of independent scrutiny as the trial tested a new percutaneous technique, separately reported by the authors, using a new device.

Shao 2019.

Study characteristics
Methods	Study design: randomised prospective controlled trial (randomly assigned by a coin flip) Duration of study: 18 March 2009 to 17 March 2012 Protocol finalised before recruitment of participants: not reported. It was approved by the ethics review committee of the Second Military Medical university Details of trial registration: not reported Funding sources: not reported
Participants	Place of study: Department of Orthopedics, Changai Hospital, The Second Military Medical University, Shanghai 200433, PR China. Number of participants assigned: 38 (21 minimally invasive surgical technique and 17 open surgery technique) Number of participants assessed: 38 (21 minimally invasive surgical technique and 17 open surgery technique) Inclusion criteria: clinical and x‐ray evidence of transverse patella displacements > 3 mm sign informed consent Exclusion criteria: open fracture lower limb fracture on the same side chronic degenerative joint disease previous knee surgical intervention peripheral neural damage osteoporosis defined as bone mineral density value ≥ 2.5 SDs below peak bone mass (T score < –2.5) measured by dual energy x‐ray absorptiometry Age: overall range 19 to 55 years minimally invasive surgical group: mean 42.2 (SD 12.4) open surgery group: mean 40.3 (SD 10.5) Gender: 25 men; 13 women Side of injury: left knee 18; right knee 20 Time lag between injury and operation: minimally invasive surgical group: 33.6 (SD 18.4) hours open surgery group: 38.3 (SD 20.2) hours Classification of injury: all fractures were considered transverse and closed (classified by mechanism of injury: 29 cases of fall, 7 cases of traffic accident, 2 cases of sports related)
Interventions	Timing of intervention: TLIO (time lag between injury and operation): range 15 to 59 hours Types of surgical interventions minimally invasive surgical group used closed reduction and percutaneous fixation with the CPS open surgery used open reduction and internal fixation with the CPS Both interventions used: The CPS The Zimmer Cable‐Ready CPS (Zimmer Biomet Holdings, Inc., Warsaw, IN, USA) is a new type of internal fixation system, which consists of a cable and a pin. One end of the system is the pin, a partially threaded 4.0 mm cancellous screw, which can provide compression between fragments and prevent the cancellous screw from backing out. The other end of the system is a guide needle, which can easily be passed through bone tunnels. The material connecting the 2 ends is a cable with a special braiding structure. It is braided from 19 wires with each wire containing 7 monofilament wires with a diameter of 0.005 mm. All surgeries were performed by the same surgeon. Spinal anaesthesia or combined spinal‐epidural anaesthesia was adopted for the surgery. The participant was positioned in the supine position, with a sterile soft cushion used to support the injured knee, leaving the joint in a 20° flexed position, allowing the knee to perform flexion‐extension movements. Rehabilitation: for all participants, an elastic bandage was used for 2 days after surgery to reduce swelling of the knee and the likelihood of a haematoma. No immobilisation was recommended. Passive exercise was initiated 1 day after the surgery by a continuous passive motion machine. 1‐h sessions, starting from 0 to 60°, increasing 15° per day until 90° was achieved. This training lasted for 3–5 days. Active flexion exercises of the knee joint in prone position were encouraged. The patients began to get out of bed 3 days after surgery and were encouraged to start bearing weight on the leg that was operated on during level walking. Active extension exercises of the knee joint were allowed 3 weeks after surgery, and full weightbearing was allowed after the fracture was healed following a radiographic assessment.
Outcomes	Length of follow‐up: clinical and radiographical follow‐up were performed at 4 and 8 weeks, and 12 months postoperatively Loss of follow‐up: 0 lost to follow‐up Primary outcomes Patient‐rated knee function scores – not assessed Anterior knee pain – at 4 and 8 weeks, pain was measured using a VAS with scores between 0 (no pain) and 10 (the most intense pain) Major adverse outcomes: reported on defect of cartilage surface, arthritic changes, breaking or backing out of pin, or cable rupture detected through radiological examinations Secondary outcomes Observer‐rated measures of knee function – clinical grading scale – the Böstman score (Böstman 1981). The total score was stratified as follows: < 20, unsatisfactory; 20 to 27, good; 28 to 30, excellent. Active flexion and extension of the knee joint were measured in degrees by goniometry. Health‐related quality of life scores – not assessed Return, including time to return, to previous activities – not assessed Cosmetic appearance – not assessed Other outcomes: radiographical follow‐up performed at 4 and 8 weeks, and 12 months postoperatively. All the x‐ray films were read independently by 2 well‐trained orthopaedic surgeons. Mean operating time was measured
Notes	We contacted the authors to obtain further information; however, they did not reply.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Sequence generation performed by a coin flip.
Allocation concealment (selection bias)	Unclear risk	Whether and how allocation was concealed was not reported.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Participants and personnel were probably not blinded.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Outcome assessors were not blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	No loss to follow‐up.
Selective reporting (reporting bias)	High risk	Patient‐rated knee function, health‐related quality of life scores, return to previous activities, time to return to previous activities and cosmetic appearance were not assessed. No details of the trial protocol are available.
Other bias	High risk	There was a potential conflict of interest reflecting the lack of independent scrutiny as the trial tested a new percutaneous technique, separately reported by the authors, using a new device: the cable pin.

Tian 2015.

Study characteristics
Methods	Study design: randomised controlled trial Duration of study: February 2008 to December 2011 Protocol finalised before recruitment of participants: not reported Details of trial registration: not reported Funding sources: none known
Participants	Place of study: Beijing Chao‐Yang Hospital Affiliated to Capital Medical University, Beijing, China Number of participants assigned: 73 (39 TBW group, 34 CPS, open) Number of participants assessed: 73 (39 TBW group, 34 CPS, open) Inclusion criteria: people with transverse patella fractures (The Orthopaedic Trauma Association classification 34C1) with an articular incongruity (step‐off) > 2 mm and the separated distance > 3 mm Exclusion criteria: skeletal immaturity (aged < 18 years) ASA score > IV bilateral fractures, refractures, pathological fractures and fractures with a deep infection Age: TBW group: mean 44.53; range 21 to 72 years CPS group: mean 46.15; range 23 to 73 years Gender: 27 men; 46 women Classification of injury: all fractures were transverse patella fractures (The Orthopaedic Trauma Association classification 34C1) with an articular incongruity (step‐off) > 2 mm and separated distance > 3 mm
Interventions	Timing of intervention: time from injury to surgery range 2–7 days Types of surgical interventions: CPS (a combination of interfragmentary screws molten to a tension band wire), which was manufactured by Zimmer. Open reduction and the Zimmer Cable‐Ready CPS (Zimmer, Warsaw, IN) is a design consisting of a partially threaded 4.0 mm cancellous lag screw connected to a stainless‐steel multifilament cable. The lag screw is 35–60 mm in length, and the cable (diameter 1.3 mm, length 448 mm) is braided from 19 wires, each containing 7 monofilament wires. TBW: open reduction and modified TBW technique not detailed in the study, just described that the technique was performed according to a previously described method (Weber 1980). Rehabilitation: standardised postoperative rehabilitation protocol was performed. An elastic bandage was used for 2 days after surgery to reduce haematomas of the knee and deep vein thrombosis. All participants were encouraged to practice isometric knee extension and straight leg elevation. Controlled passive ROM was started 3 days postoperatively with an initial ROM 0–40° of flexion and increasing by 20° to 30° weekly until a full ROM was achieved, within 4 weeks. Partial weight‐bearing was allowed 2 weeks after operation, and full weight‐bearing was allowed after the fracture was radiographically healed.
Outcomes	Length of final follow‐up: all participants were followed up for 12 to 29 months. All participants underwent regular radiological examinations (2, 4, 6, 8, 12 and 16 weeks) until union was confirmed, radiological union defined as the loss of the fracture line and presence of bony trabecular continuity. Postoperative knee function evaluated clinically in all participants using the modified HSS Knee score 1 year after surgery. Loss of follow‐up: no loss reported Primary outcomes Patient‐rated knee function scores – postoperative knee function evaluated clinically in all participants using the modified HSS Knee score at 1 year after surgery Anterior knee pain – not assessed Major adverse outcomes – incision and deep infection, complications/problems with hardware Secondary outcomes Observer‐rated measures of knee function – not assessed Health‐related quality of life scores – not assessed Return, including time to return, to previous activities – not assessed Cosmetic appearance – not assessed Other outcomes: incision length, comparison of surgical time, intraoperative blood loss (with pneumatic tourniquet), fracture healing in radiological examinations (until union was confirmed, radiological union defined as the loss of the fracture line and presence of bony trabecular continuity).
Notes	We tried to contact the authors to obtain further information; however, they did not reply.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	The sequence generation was performed by a computer random number generator (SPSS software, version 19.0, Chicago, IL).
Allocation concealment (selection bias)	Unclear risk	Whether and how allocation was concealed was not reported.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Participants and personnel were probably not blinded.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Outcome assessors were not blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	No loss to follow‐up.
Selective reporting (reporting bias)	High risk	Anterior knee pain, health‐related quality of life scores, observer‐rated measures of knee function, return to previous activities, time to return to previous activities and cosmetic appearance were not assessed. No details of the trial protocol are available.
Other bias	High risk	There was a potential conflict of interest reflecting the lack of independent scrutiny as the trial tested a new open technique. The authors did not provide sufficient information about the surgeons and care providers, and about the tension band technique (modified TBW)

Wei 2019.

Study characteristics
Methods	Study design: quasi‐randomised controlled trial (according to the admission date) Duration of study: January 2015 to June 2017 Protocol finalised before recruitment of participants: not reported Details of trial registration: not reported Funding sources: none known
Participants	Place of study: Beijing, China Number of participants assigned: 56 (28 in 3‐dimensional strapping reduction group, 28 in towel clamp reduction group) Number of participants assessed: 56 (28 in 3‐dimensional strapping reduction group, 28 in towel clamp reduction group) Inclusion criteria: people with patella fractures informed consent Exclusion criteria: pathological fractures osteoarthritis rheumatoid arthritis other fractures Age: 3‐dimensional strapping reduction group: mean 39.6; range 20 to 65 years towel clamp reduction group: mean 37.8; range 21 to 66 years Gender: 29 men; 27 women Classification of injury: 3‐dimensional strapping reduction group: 19 cases of fall and 7 cases of traffic accidents and 2 cases of other causes; 12 cases on left patella and 16 cases on right. All closed fractures towel clamp reduction group: 18 cases of fall and 10 cases of traffic accidents; 11 cases on left patella and 17 cases on right. All closed fractures
Interventions	Timing of intervention: 1–9 days after injury Types of surgical interventions: Both groups of surgery were performed by the same group of physicians 3‐dimensional strapping reduction group: participant was placed in supine position after spine anaesthesia. The anterior patella incision was used, the blood clot and decidual tissue at the fracture end were cleaned, then the fracture line was observed and counted. The surgeons used 1‐0 Ai Sibang suture (number of sutures = number of major fracture lines + 1) from the fracture end through the patella joint to the soft tissue around, restored the patella fracture under direction vision, and tightened the sutures 1 by 1 on the outer surface of the patella. After the C‐arm X‐ray machine confirmed that the patellar surface was satisfactory, the knee was bent 70°, and the tension band cable was used to finally fix the banding suture. For comminuted fractures, a large free fracture block could be fixed with Herbert nails for final fixation. Towel clamp reduction group: participant was placed in supine position after spine anaesthesia. The anterior patella incision was used, the blood clot and decidual tissue at the fracture end were cleaned, then the fracture line was observed and counted. After clearing the soft tissue of the fracture end, the patella fracture was fixed by the towel clamp. If the reduction of the fracture was not satisfactory (with the evaluation of C‐arm X‐ray machine), the towel clamp was loosened and then adjusted until the joint surface was flat. For comminuted fractures, first used the Herbert nail to fix the larger free fracture block, the towel clamp was reset, and the tension band cable was finally fixed. Rehabilitation: quadriceps function exercise began immediately after surgery. The continuous passive motion passively fixed and extended the knee reaching 90° within 2 weeks, and normal passive flexion and extension activities were restored within 4 weeks. Within 2 weeks, the trial participants could go to weight‐bearing walking and subsequently resume the active flexion and extension activities within 8 weeks.
Outcomes	Length of final follow‐up: mean 12.5; range 11 to 15 months Loss of follow‐up: none lost to follow‐up (implied) Primary outcomes Patient‐rated knee function scores – postoperative knee function was evaluated clinically in all participants using the modified HSS knee score Anterior knee pain – not assessed Major adverse outcomes – infection, bone non‐union and broken nails Secondary outcomes Observer‐rated measures of knee function – not assessed Health‐related quality of life scores – not assessed Return, including time to return, to previous activities – not assessed Cosmetic appearance – not assessed Other outcomes: operation time, intraoperative fluoroscopy time, fracture healing time
Notes	We tried to contact the authors to obtain further information; however, they did not reply. It was found Wei 2017a an interim publication of Wei 2019 , published in a Chinese journal that reported results for 32 participants (Wei 2019 reported results for 56 participants in another Chinese journal, and did not report or mention a previous interim publication). So, only the results from Wei 2019 were included in this review because it is the most up‐to‐date report of the same trial.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Method of generating the random sequence not described, it was only described as allocated according to the admission date (we considered this study a quasi‐randomised controlled trial).
Allocation concealment (selection bias)	High risk	Whether and how allocation was concealed was not reported. The description of the method of allocation is unhelpful but almost certainly the non‐probability consecutive sampling rules out randomisation.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Participants and personnel were probably not blinded.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Outcome assessors were not blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	No loss to follow‐up.
Selective reporting (reporting bias)	High risk	Anterior knee pain, health‐related quality of life scores, return to previous activities, time to return to previous activities and cosmetic appearance were not assessed. Not reported if there was a trial protocol or a previous trial registration.
Other bias	High risk	The authors did not provide sufficient information about the surgeons and care providers, or sufficient details of rehabilitation after surgery. They did not report details in follow‐up (number of postsurgery medical consultation, timing of postsurgery medical consultation for follow‐up). There was a potential conflict of interest reflecting the lack of independent scrutiny as the trial tested a new reduction technique for treating fractures of the patella.

AO: Arbeitssgemeinschaft fur Osteosynthesefragen; ASA: American Society of Anesthesiologists; CPS: cable pin system; CRCF: closed reduction and percutaneous cannulated screw fixation; HSS: Hospital for Special Surgery; ORTF: open reduction and internal fixation with tension band wiring; OTA: Orthopaedic Trauma Association; ROM: range of motion; SD: standard deviation; TBW: tension band wiring; VAS: visual analogue scale.

Characteristics of excluded studies [ordered by study ID]

Study	Reason for exclusion
Alayan 2018	Design of study not relevant: experimental cadaveric study.
Chang 2011	Design of study not relevant: case series.
Depeng 2019	Design of study not relevant: retrospective cohort study.
Gosal 2001	Design of study not relevant: not a randomised or quasi‐randomised controlled trial – participants were allocated by surgeon preference.
Heusinkveld 2013b	Design of study not relevant: systematic review.
Hoshino 2013	Design of study not relevant: retrospective cohort study.
Jiang 2018	Design of study not relevant: case‐control study.
Kyung 2017	Design of study not relevant: retrospective cohort study.
Lee 2014	Design of study not relevant: retrospective cohort study.
Luna Pizarro 2008	Design of study not relevant: case series.
Mao 2012	Design of study not relevant: case series.
Massoud 2017	Design of study not relevant: prospective cohort study
Matthews 2018	Design of study not relevant: systematic review and case report.
Meng 2019	Design of study not relevant: retrospective cohort study.
Ren 2018	Design of study not relevant: systematic review (with only retrospective cohort studies).
Tang 2013	Design of study not relevant: case series.
Tang 2018	Design of study not relevant: case series.
Wang 2016	Design of study not relevant: not a randomised or quasi‐randomised controlled trial – participants were allocated by surgeon preference.
Xu 2013	Design of study not relevant: not a randomised or quasi‐randomised controlled trial – the participants could choose the intervention.
Zderic 2017	Design of study not relevant: experimental cadaveric study.
Zha 2017	Design of study not relevant: systematic review.
Zhang 2018	Design of study not relevant: systematic review.

Characteristics of ongoing studies [ordered by study ID]

ChiCTR1800016105.

Study name	A randomised controlled trial of modified tension band reducing flexion restriction of knee joint due to the rotation of K‐wire in the treatment of patellar fracture
Methods	Study design: randomised controlled trial Random sequence generation: not reported Allocation concealment: not reported Blinding: not reported
Participants	Men and women with displaced patellar fracture (separation > 3 mm or the step height of fracture > 2 mm) Location: China Target number of participants: 148 (74 modified tension band group, 74 traditional tension band) Inclusion criteria: not reported Exclusion criteria: not reported
Interventions	Modified tension band in the treatment of patellar fracture Surgical intervention (tested): modified tension band Surgical intervention (control): traditional tension band
Outcomes	Primary outcome Range of knee joint flexion Secondary outcomes: not reported Timing of outcomes measurement: not reported
Starting date	Main ID: ChiCTR1800016105 Date of registration: 11 May 2018 Last updated: not reported Date of first enrolment: not reported Status: not reported Estimated study completion date: not reported
Contact information	Name: Liangyuan Address: not reported Telephone: not reported Email: Wenliangyuan1964@126.com Affiliation: not reported
Notes	We contacted the authors to obtain further information; however, they did not reply.

NCT03445819.

Study name	A prospective randomized trial of non‐operative versus operative management of patella fractures in the elderly
Methods	Study design: parallel randomised controlled trial Random sequence generation: not reported Allocation concealment: not reported Blinding: data analysts will be blinded to the study treatment. Wherever possible, outcome assessors will be blinded to the study treatment.
Participants	Aged ≥ 65 years (older adult) with isolated, closed, fracture of the patella displaced by ≥ 5 mm Location: multicentre study; London Health Sciences Centre, St Michael's Hospital and Sunnybrook Health Sciences Centre, Ontario, Canada Target number of participants: 84 Inclusion criteria Men or women aged ≥ 65 years and ambulatory prior to injury (with or without walking aids) Isolated, closed, fracture of the patella displaced by ≥ 5 mm (displacement will be determined by measuring the widest displacement on any x‐ray view with the knee in full extension) Participant scores 3 to 6 on the Clinical Frailty Scale. This score corresponds to a low‐demand patient who is ambulatory and functionally independent Within 14 days of injury, the participant can perform a straight leg raise with < 30° of extensor lag Bilateral patella (participant is randomised once, both patella are treated the same) Willing and able to sign consent (in‐person or via substitute decision maker), follow the protocol and attend follow‐up visits Able to read and understand English (or a qualified interpreter is available) Exclusion criteria Associated knee injury (tibia or distal femur fractures) or other lower limb injuries that would interfere with rehabilitation or outcome Neurovascular injuries at the level of the knee requiring surgery Pathological fractures Periprosthetic fractures, or other knee surgery which would contraindicate inclusion in the study (e.g. previous patella ORIF or previous surgery involving patella such as patella tendon or quads repair) Medical contraindication to surgery Likely problems, in the judgement of the investigators, with maintaining follow‐up (i.e. people with no fixed address, intellectually challenged people without adequate support, etc.)
Interventions	Surgical treatment (intervention group) Typeofsurgical intervention: ORIF of the patellar fracture will be performed using screws, wires, pins or plate fixation at the discretion of the treating surgeon. The trial is designed using a pragmatic method to allow participating surgeons from the multiple participating sites to perform fixation as per the standard of care at their institution. Postoperative care will include standard‐of‐care antibiotics and deep vein thrombosis prophylaxis, both prescribed at the discretion of the attending surgeon. Conservative treatment (control group) Type of conservative intervention: identical treatment to the operative group, minus the surgery. Participants will be weight‐bearing as tolerated immediately in a removable knee immobiliser, with progressive range‐of‐motion exercises begun at 2 weeks following randomisation.
Outcomes	Primary outcomes Knee injury and Osteoarthritis Outcome Score at 1 year. A self‐administered questionnaire that assesses 5 outcomes: pain, symptoms, activities of daily living, sport and recreation function, and knee‐related quality of life. A normalised score (100 indicating no symptoms and 0 indicating extreme symptoms) is calculated for each subscale. Secondary outcomes Visual analogue pain scale (0: 'no pain at all' to 10: 'pain as bad as it could be') Range of motion: flexion and extension Timing of outcomes measurement: up to 24 months
Starting date	Main ID: NCT03445819 Date of registration: 13 February 2018 Last updated: 13 December 2018 Date of first enrolment: 26 February 2018 Status: recruiting Estimated study completion date: 31 December 2022
Contact information	Contact: Milena Vicente, vicentem@smh.ca Name: Milena Vicente Address: St Michael's Hospital, Toronto, Ontario, Canada Telephone: RN 416‐864‐6060 ext 2608 Email: vicentem@smh.ca Affiliation: St Michael's Hospital, Toronto, Ontario, Canada
Notes	We contacted the authors to obtain further information; however, they did not reply.

ORIF: open reduction and internal fixation.

Differences between protocol and review

Differences from protocol in the second version of the review (2021)

Most of the changes to the methods for the 2021 update reflect the uptake of new methodology and reporting as described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). These include more explicit reporting of data analysis and collection, assessment of the quality of the evidence using GRADE, production of a 'Summary of findings' table and implementation, with some restructuring, of subsection headings such as those in 'Background', 'Data collection', 'Analysis' and 'Discussion'.

In Measures of treatment effect, we removed our stated intention to report the number needed to treat for an additional beneficial outcome (NNTB) and the number needed to treat for an additional harmful outcome (NNTH). We did not do this and instead, where appropriate, phased absolute effects based on the approach taken in 'Summary of findings' tables.

We selected the percutaneous patellar osteosynthesis versus open surgery for treating fractures of the patella in adults as the main comparison, for presentation in a 'Summary of findings' table because this is the most important comparison tested by the included trials; it also featured the largest number of trials. We also selected the seven outcomes for presentation in any 'Summary of findings' table; see end of Data collection and analysis.

Differences from protocol in the first version of the review (2015)

We added "hardware removal" as a secondary outcome.

Contributions of authors

Author contributions for this review update are as follows.

• Lead and guarantor: JS

• Obtaining trial reports: JS

• Study selection: JS, ML and MJ

• Data extraction: JS, JB and FM

• Data entry into Review Manager 5: JS and FM

• Contacting authors of eligible trials to inform choices for study inclusion and to request additional information: JS and JC

• Conducting the analysis: JS, ML and FM

• Interpretation of the results: JS, MJ and JC

• Compiling the first draft of the review and responding to feedback from editors: JS and ML

• Drafting the final version of the review: JS, ML and JC

• All authors commented on and approved the final version of the review

Contributions of editorial base

Helen Handoll: edited the review update; advised on methodology and review content; and approved the final version for publication.

Joanne Elliott: co‐ordinated the editorial process; advised on content; and edited the review update.

Maria Clarke: ran search updates and edited the Search Methods section.

Sources of support

Internal sources

• Department of Orthopaedics, Universidade Federal de Sao Paulo, Brazil

• Hospital Israelita Albert Einstein, Brazil

External sources

• National Institute for Health Research (NIHR), UK

  via Cochrane Infrastructure funding to the Cochrane Bone, Joint and Muscle Trauma Group, UK

Declarations of interest

JS: none.

ML: none.

MJ: none.

FM: none.

JB: none.

New search for studies and content updated (conclusions changed)