Pedicle screw fixation for traumatic fractures of the thoracic and lumbar spine
Abstract
Background
Spine fractures are common. The treatment of traumatic fractures of the thoracic and lumbar spine remains controversial but surgery involving pedicle screw fixation has become a popular option.
Objectives
To assess the effects (benefits and harms) of pedicle screw fixation for traumatic fractures of the thoracic and lumbar spine.
Search methods
We searched the Cochrane Bone, Joint and Muscle Trauma Group Specialised Register (March 2011), the Cochrane Central Register of Controlled Trials (CENTRAL; The Cochrane Library, 2011 Issue 1), MEDLINE (1948 to March 2011), EMBASE (1980 to 2011 Week 11), the Chinese Biomedical Database (CBM Database) (1978 to March 2011), the WHO International Clinical Trials Registry Platform (March 2011), reference lists of articles and conference proceedings.
Selection criteria
Randomised controlled trials (RCTs) and quasi‐randomised controlled trials comparing pedicle screw fixation and other methods of surgical treatment, or different methods of pedicle screw fixation, for treating traumatic fractures of the thoracic and lumbar spine.
Data collection and analysis
Three review authors independently performed study selection, risk of bias assessment and data extraction. Limited meta‐analysis was performed.
Main results
Pedicle screw fixation versus other methods of surgery that do not involve pedicle screw fixation was not looked at in any of the identified trials. Studies that were identified investigated different methods of pedicle fixation.
Five randomised and three quasi‐randomised controlled trials were included. All were at high or unclear risk of various biases, including selection, performance and detection bias. A total of 448 patients with thoracic and lumbar spine fractures were included in the review. Participants were restricted to individuals without neurological impairment in five trials. The mean ages of study populations of the eight trials ranged from 33 to 41 years, and participants had generally experienced traumatic injury. Mean follow‐up for trial participants in the eight trials ranged from 28 to 72 months.
Five comparisons were tested.
Two trials compared short‐segment instrumentation versus long‐segment instrumentation. These studies found no significant differences between the two groups in self‐reported function and quality of life at final follow‐up. Aside from one participant, who sustained partial neurological deterioration that was resolved by further surgery (group not known), no neurological deterioration was noted in these trials.
One trial comparing short‐segment instrumentation with transpedicular bone grafting versus short‐segment fixation alone found no significant difference between the two groups related to patient‐perceived function and pain at final follow‐up. All participants had normal findings on neurological examination at final follow‐up.
Two trials compared posterior instrumentation with fracture level screw incorporation ('including' group) versus posterior instrumentation alone ('bridging' group). Investigators reported no differences between the two groups in patient‐reported function, quality of life, or pain at final follow‐up. One trial confirmed that all participants had normal findings on neurological examination at final follow‐up.
One trial comparing monosegmental pedicle screw instrumentation versus short‐segment pedicle instrumentation found no significant differences between the two groups in Oswestry Disability Index results or in pain scores at final follow‐up. No neurological deterioration was reported.
Three trials compared posterior instrumentation with fusion versus posterior instrumentation without fusion. Researchers found no differences between the two groups in function and quality of life or pain. No participants showed a decline in neurological status in any of the three trials, and no significant difference was reported between groups in the numbers whose status had improved at final follow‐up. Two trials stated that patients in the fusion group frequently had donor site pain. Other reported complications included deep vein thrombosis and superficial infection.
Authors' conclusions
This review included only eight small trials and five different comparisons of methods of pedicle fixation in various participants while looking at a variety of outcomes at different time points. Overall, evidence is insufficient to inform the selection of different methods of pedicle screw fixation or the combined use of fusion. However, in the absence of robust evidence to support fusion, it is important to factor the risk of long‐term donor site pain related to bone harvesting into the decision of whether to use this intervention. Further research involving high‐quality randomised trials is needed.
Plain language summary
Pedicle screw fixation methods for traumatic fractures of the thoracic and lumbar spine
Thoracic and lumbar spine fractures are the most common injuries of the spine. An exaggerated curvature (kyphosis) at the end of treatment may predispose to later back pain and a poor functional outcome. If the nerve root or spinal cord is damaged, partial or complete loss of sensory and motor function in the legs, and urinary and faecal incontinence may result. Treatment depends on the individual characteristics of the fracture, with options including bed rest alone, closed reduction of the fracture and functional bracing, and surgery involving open reduction and internal fixation of the fracture. Surgery frequently involves posterior pedicle screw fixation, where typically screws are placed in the 'pedicle' parts of the vertebrae (bones of the spine) adjacent to the damaged vertebrae and connected by rods to hold the bones in place and stabilise the fracture while it heals. This review examined the evidence for the different types of pedicle screw fixation and for additional support such as fusion, where bone graft (usually taken from bone near the hip region of the patient) or substitute is added to the spine. The latter aims to reduce movement of the injured segment and any associated pain.
The review authors found eight trials that included a total of 448 patients with thoracic and lumbar spine fractures. These trials were small and were at risk of bias that could have affected their findings. Five comparisons of different methods of pedicle screw fixation were tested by the included trials. For each comparison, no differences were found in function, activities of daily living, or pain in the two treatment groups. Aside from one person, who required further surgery to treat a temporary decline in neurological status, no report described any trial participant who showed a permanent decline in neurological status. Notably, two of the three trials testing fusion reported that a quarter to two thirds of participants had long‐term pain at the donor site where the bone graft had been taken.
The review authors concluded that evidence is insufficient to inform on the selection of different methods of pedicle screw fixation or on the use of fusion with pedicle screw fixation.
Authors' conclusions
Background
Description of the condition
Spine fractures are common injuries in today's society (Bensch 2004; Cassar‐Pullicino 2002; Jelly 2000; Trivedi 2002). In a Canadian study including 2063 patients with spinal fracture, Hu 1996 found an annual incidence of spinal fracture of 64 per 100,000 people. Of 904 patients admitted to hospital in Hu 1996, 30% had a thoracic fracture and 43% had an injury to the lumbosacral spine. Vives 2008 reported that thoracic spinal injuries accounted for 19% and lumbar spinal injuries for 37% of 135 pedestrians who had sustained spinal injuries after being struck by a motor vehicle.
Among young people, thoracic and lumbar spine fractures are usually caused by high‐energy accidents such as falls or motor vehicle accidents, whereas in elderly people, osteoporosis is the dominant cause (Robertson 2002). In people younger than 60 years of age, and the incidence of spine fractures is twice as high in men as in women (Jansson 2010).
Description of the intervention
Thoracic and lumbar spine injuries range from injuries that can be managed conservatively with bracing, including most compression fractures, to highly unstable injuries often associated with neurological compromise. Indications for surgery are based on the presence of neurological compromise, deformity, and instability. Denis 1983 put forward a three‐column conception of the spine. He described the importance of the middle column of the vertebra and expressed the need for more aggressive intervention, that is, surgery, if two of the three columns of the spine are disrupted. Vaccaro 2005 proposed the Thoracolumbar Injury Severity Score to determine which people may benefit from operative intervention based on the mechanism of injury, the neurological status of the patient, and the integrity of the posterior ligamentous complex.
Surgical management of thoracic and lumbar spine fractures initially involved open reduction of the deformity followed by fixation of the fractured segments by two plates bolted to the spinous processes or lamina above and below the level of injury. More modern techniques of internal fixation now include posterior pedicle screw fixation and/or direct anterior column decompression and reconstruction using bone or synthetic grafts with or without internal fixation.
Posterior pedicle screw fixation is widely used in clinical practice. The pedicle screw is generally placed through the pedicle of the vertebral arch － the segment between the transverse process and the vertebral body － into the vertebral body without breaching the bony pedicle wall. This trajectory provides three‐column fixation, which is biomechanically desirable. Pedicle screw fixation can be divided into several categories based on the lengths of the fixed segments and of pedicle screws in the injured vertebra, as well as on transpedicular grafting and posterolateral fusion:
-
Short‐segment pedicle screw instrumentation (SSPI) is performed with pedicle screws inserted bilaterally into two vertebrae, one above and one below the fractured vertebrae, with longitudinal rods connecting the tail ends of the pedicle screws. There are four screws and two rods in all (Dick 1985).
-
Long‐segment pedicle screw instrumentation (LSPI) is performed with pedicle screws inserted bilaterally into four vertebrae, two above and two below the fractured vertebrae, with longitudinal rods connecting the tail ends of the pedicle screws. There are eight screws and two rods in all (Tezeren 2005a).
-
Monosegmental pedicle screw instrumentation (MSPI) is performed with pedicle screws inserted bilaterally at the level of the fracture and one level adjacent, either superior or inferior depending on the location of the intact endplate (Liu 2009).
If necessary, SSPI and LSPI can be combined with one additional pedicle screw, which is inserted higher than the other screws into the intact pedicle of the injured segment for additional fixation (Wang 2006a). SSPI and LSPI can also be combined with transpedicular grafting directly through a posterolateral approach. The depressed endplate is elevated through a transpedicular approach (De Boeck 1999).
Some studies have reported that autogenous bone was grafted onto the decorticated laminae of the fixed segments to fuse the relevant segments. Autogenous bone graft can be combined with all of the techniques above (Wang 2006a).
How the intervention might work
According to Denis's three‐column conception of the spine, which divides the bony and ligamentous components of the spinal column into three columns, the middle column fracture in the posterior wall of the vertebral body significantly affects spinal stability and may lead to spinal cord injuries. Pedicle screw fixation allows immediate stable fixation, as the screws traverse all three columns. This technique has three main advantages over other internal spinal fixation constructs: the ability to provide three‐column fixation, the ability to facilitate the instrumentation of short segments, and the ability to maintain anatomical alignment of the spinal column, or the best possible alignment if anatomical alignment is not possible. However, this technique also has its disadvantages. Because of its close proximity to the spinal canal and surrounding vessels, misplacement of the pedicle screw can lead to disastrous complications.
Why it is important to do this review
Many people have permanent functional impairment and pain after thoracic and lumbar spine fractures. Especially when associated with spinal cord injury, long‐term reductions in quality of life and ability to work are frequent (Robertson 2002). Although a large number of publications have described various surgical techniques for the reduction and fixation of spinal fractures, no general consensus has been reached with regard to the optimal treatment (Trivedi 2002). This lack of consensus includes pedicle screw fixation and the various methods of pedicle screw fixation. This points to the need for a systematic review of the evidence to inform clinical decisions pertaining to pedicle screw fixation for traumatic fractures of the thoracic and lumbar spine.
Objectives
To assess the effects (benefits and harms) of pedicle screw fixation for traumatic fractures of the thoracic and lumbar spine.
Our two main comparisons were between:
-
Pedicle screw fixation and other methods of surgical treatment (Hook‐Rod fixation; direct anterior column decompression and reconstruction using bone or synthetic grafts with or without internal fixation); and
-
Different methods of pedicle screw fixation.
Methods
Criteria for considering studies for this review
Types of studies
Randomised controlled trials (RCTs) and quasi‐randomised (methods of allocating participants to a treatment that were not strictly random, e.g. by date of birth, hospital record number, alternation) controlled trials comparing the treatment of traumatic fractures of the thoracic and lumbar spine performed using various pedicle screw fixation techniques will be included.
Types of participants
People with fractures of the thoracic and lumbar spine. Ideally, the time from trauma to treatment was less than or equal to three weeks. However, we accepted trials where some people were treated after three weeks, provided the proportion of late presenters is less than 20%.
Types of interventions
We set up two main comparisons.
1. Pedicle screw fixation versus other methods of surgery that do not involve pedicle screw fixation.
Other surgical treatments for traumatic fractures of the thoracic and lumbar spine include Hook‐Rod fixation, direct anterior column decompression, and reconstruction using bone or synthetic grafts with or without internal fixation.
2. Different methods of pedicle fixation, as listed below:
-
Short‐segment pedicle screw instrumentation.
-
Long‐segment pedicle screw instrumentation.
-
Short‐segment pedicle screw instrumentation with transpedicular grafting.
-
Short‐segment pedicle screw instrumentation with pedicle screw in the injured vertebra.
-
Monosegment pedicle screw instrumentation.
Last, we compared posterolateral fusion with pedicle screw fixation versus pedicle screw fixation alone.
Types of outcome measures
Primary outcomes
-
Self‐reported function and quality of life.
-
Improvement or deterioration in neurological status (e.g. classic Frankel score).
-
Pain (e.g. Denis pain scale, pain assessed by visual analogue scale (VAS)).
Secondary outcomes
-
Postoperative complications:
-
Deep venous thrombosis.
-
Pulmonary embolism.
-
Superficial infection.
-
Deep infection.
-
Malposition of the implant.
-
Implant failure.
-
Miscellaneous complications.
-
-
Radiological evaluation
-
The Cobb angle.
-
Vertebral body translation percentage.
-
The anterior vertebral body compression percentage.
-
The sagittal‐to‐transverse canal diameter ratio, the canal total cross‐sectional area (measured or calculated), and the percent canal occlusion.
-
-
Intraoperative blood loss, transfusion
-
Duration of surgery
-
Employment (e.g. Denis work scale for previously employed people)
-
Length of hospital stay
Search methods for identification of studies
Electronic searches
We searched the Cochrane Bone, Joint and Muscle Trauma Group Specialised Register (March 2011), the Cochrane Central Register of Controlled Trials (CENTRAL; The Cochrane Library, 2011 Issue 1), MEDLINE (1948 to March Week 2 2011), EMBASE (1980 to 2011 Week 11) and the Chinese Biomedical Database (CBM Database) (1978 to March 2011). No language restrictions were applied.
The MEDLINE subject‐specific search was combined with the sensitivity‐maximizing version of the Cochrane Highly Sensitive Search Strategy for identifying randomised trials (Lefebvre 2009). The EMBASE subject‐specific search was combined with the Scottish Intercollegiate Guidelines Network search filter for RCTs. Details of the search strategies for The Cochrane Library, MEDLINE, and EMBASE are shown in Appendix 1.
Ongoing and unpublished trials were identified by searching the WHO International Clinical Trials Registry Platform (March 2011).
Searching other resources
-
We inspected all citations within included and excluded studies for additional relevant trials.
-
We wrote to the corresponding authors of included trials for additional information.
-
Manufacturers of relevant products were contacted to identify any unpublished studies.
-
We checked the conference proceedings of the first China International Congress of Lumbar Spine, the 2011 Chinese Medical Association Trauma Conference, and the first Congress of Trauma of Yangtze River Delta.
Data collection and analysis
Selection of studies
Three review authors (LMC, JJW and RZ) independently screened search results for potentially eligible trials. We obtained the full reports for these and for any study for which there was doubt or dispute about its eligibility. Three review authors (LMC, JJW and RZ) performed study selection. Any disagreements were resolved by discussion. If necessary, study authors were contacted for clarification of their methods to inform study selection.
Data extraction and management
Three review authors (LMC, JJW and RZ) independently extracted data from the included studies using a piloted form. Any disagreement was discussed and decisions documented, and, if necessary, study authors were contacted for clarification.
Assessment of risk of bias in included studies
Three review authors (LMC, JJW and RZ) independently assessed risk of bias using the tool described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). We assessed risk of bias in relation to sequence generation, allocation concealment, blinding (participants and personnel), blinding of outcome assessment, completeness of outcome data, selective reporting, and other biases. The bias from major imbalances in key baseline characteristics (age, gender, neurological deficit, and timing of presentation) and performance bias resulting from differences in surgeon experience were also considered. Where necessary, study authors were contacted for clarification. Any disagreements between review authors were resolved by discussion.
Measures of treatment effect
We calculated risk ratios (RRs) with 95% confidence intervals (CIs) for dichotomous outcomes, and mean differences with 95% CIs for continuous outcomes.
Unit of analysis issues
We considered that the unit of randomisation was likely to be the individual patient. If trials with a cluster‐randomised design had been included, we planned to include their data in a meta‐analysis only if it was possible to make an appropriate correction for clustering. If that was not possible, we planned to report their results in the text only. We were alert to other unit of analysis issues such as those related to multiple observations for the same outcome.
Dealing with missing data
Where necessary, we attempted to contact trial authors to request missing data. Where possible, we performed intention‐to‐treat analyses, which included all people randomly assigned. However, where drop‐outs were identified, the actual denominators of participants contributing data at the relevant outcome assessment were used and sensitivity analyses used to explore the potential effects of the missing data. We were alert to the potential mislabelling or non‐identification of standard errors and standard deviations. Unless missing standard deviations could be derived from confidence intervals or standard errors, we stipulated that we would not assume values for the purpose of presenting these in the analyses.
Assessment of heterogeneity
We considered all included studies initially without viewing comparison data to judge clinical heterogeneity. We assessed heterogeneity by visual inspection of the forest plot (analysis) along with consideration of the Chi² test for heterogeneity and the I² statistic (Higgins 2003). The I² statistic provided an estimate of the percentage of inconsistency thought to be due to chance. We interpreted the I² results in the approximate ranges presented in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2009).
Assessment of reporting biases
Reporting biases arise when the reporting and dissemination of research findings are influenced by the nature and direction of results. We were aware that funnel plots may be useful in investigating reporting biases but were of limited power to detect small‐study effects (Egger 1997). We would have used funnel plots only in cases where more than 10 studies were included in a meta‐analysis.
Data synthesis
Where considered appropriate, results of comparable groups of trials were pooled. For dichotomous variables, RRs and 95% CIs were calculated. For continuous variables, mean differences (MDs) and 95% CIs were calculated. Where data were derived from disparate outcome measures, we planned to calculate standardised mean differences (SMDs) instead. The fixed‐effect model was used where clearly statistically homogeneous results were provided. Otherwise, the random‐effects model was used in keeping with our expectations of substantial clinical and methodological heterogeneity, which in turn could generate substantial statistical heterogeneity.
Subgroup analysis and investigation of heterogeneity
We aimed to perform subgroup analyses to explore effect size differences in relation to the neurological deficit (presence versus absence), the timing of presentation, and the type of fracture. To test whether the subgroups are statistically significantly different from one another, we planned to inspect the overlap of CIs and to perform the test for subgroup differences available in RevMan.
Sensitivity analysis
Where appropriate, we performed sensitivity analyses to examine various aspects of trial and review methodology, including best and worst case scenario analyses where dichotomous data for primary outcomes were missing.
Results
Description of studies
Results of the search
For this search (completed March 2011), we screened a total of 361 records from the following databases: the Cochrane Bone, Joint and Muscle Trauma Group Specialised Register (20 records); Cochrane Central Register of Controlled Trials (20), MEDLINE (105), EMBASE (104), the Chinese Biomedical Database (4) and the World Health Organization (WHO) International Clinical Trials Registry Platform (106). We also identified two potentially eligible studies from other conference proceedings.
After initial screening of titles and abstracts, 15 papers were selected for possible inclusion. Review of the full texts led us to include eight trials, which consisted of five randomised trials and three quasi‐randomised trials, and to exclude seven other studies for reasons given in the Characteristics of excluded studies.
Included studies
Five randomised controlled trials (Dai 2009;Farrokhi 2010;Guven 2009;Wang 2006;Wei 2010) and three quasi‐randomised controlled trials (Alanay 2001;Tezeren 2005;Tezeren 2009) were included. All of them were published in full reports. A total of 448 participants were included in this review. The unit of randomisation was the individual participant. Details of the individual trials can be found in the Characteristics of included studies.
Setting
The trials were conducted in mainland China (2 trials), Iran (1 trial), Turkey (4 trials), and Taiwan (1 trial). All were single‐centred except Farrokhi 2010, which involved three hospital sites. Of the five trials providing details of the timing of trial recruitment, Wang 2006 started the earliest in 1996 and Wei 2010 started the latest in 2003.
Participants
All studies included adult participants only. The youngest participant was 15 years old (Tezeren 2009) and the oldest was 75 years (Farrokhi 2010). The mean ages of study populations of the eight trials ranged from 33 to 41 years. No information on gender was provided in Alanay 2001. The percentage of male participants in the other studies ranged from 64% (Guven 2009) to 83% (Tezeren 2005). Most fractures resulted from falls from a height. Other causes included vehicle accidents, direct trauma, participation in sports, and building collapse.
In six trials, indications for surgery included more than 50% loss of vertebral body height, kyphosis progressing by 20% or more, or more than 50% of canal involvement (Alanay 2001; Farrokhi 2010;Guven 2009;Tezeren 2005; Tezeren 2009; Wei 2010). Dai 2009 included patients with a single‐level burst fracture involving the thoracolumbar spine and a load‐sharing score of ≤ 6. Wang 2006 included patients who met the following inclusion criteria: neurologically intact spine with a kyphotic angle ≥ 20°, decreased vertebral body height of 50% or a canal compromise of 50%, incomplete neurological deficit with a canal compromise of 50%, complete neurological deficit, and multilevel spinal injury or multiple traumas. Participants were restricted to individuals without neurological impairment in five trials (Alanay 2001; Guven 2009; Tezeren 2005; Tezeren 2009;Wei 2010).
Interventions
Pedicle screw fixation versus other methods of surgery that do not involve pedicle screw fixation
None of the completed and published studies identified in the detailed searches fulfilled the inclusion criteria of this comparison.
Different methods of pedicle fixation
Seven trials had two interventions groups, whereas Guven 2009 had four intervention groups: short‐segment instrumentation alone; long‐segment instrumentation alone; short‐segment instrumentation with fracture level screw incorporation; and long‐segment instrumentation with fracture level screw incorporation. Guven 2009 appears in two of the five comparisons listed below:
-
Short‐segment instrumentation versus long‐segment instrumentation was compared in two trials (Guven 2009;Tezeren 2005) involving 90 participants.
-
Short‐segment instrumentation with transpedicular grafting versus short‐segment instrumentation alone was compared in one trial (Alanay 2001), involving 20 participants.
-
Posterior instrumentation with fracture level screw incorporation versus posterior instrumentation alone was compared in two trials (Farrokhi 2010;Guven 2009), involving 152 participants.
-
Monosegment pedicle screw instrumentation versus short‐segment pedicle screw instrumentation was compared in one trial (Wei 2010), involving 85 participants.
-
Posterior instrumentation with fusion versus posterior instrumentation alone was compared in three trials (Dai 2009;Tezeren 2009; Wang 2006), involving 173 participants.
Outcomes
Mean follow‐ups for trial participants in seven trials ranged from 28 to 50 months; a minimum five‐year follow‐up was reported in the remaining trial (Dai 2009).
Primary outcomes
Self‐reported function and quality of life
Methods used to assess self‐reported function and quality of life were stated in all trials. A Likert questionnaire (Prolo 1986) that included five items each for function and pain was used in Alanay 2001 and Guven 2009. The 36‐Item Short Form Health Survey (SF‐36) (Ware 1992) was used in Dai 2009. The Low Back Outcome Score (LBOS) devised by Greenough and Fraser (Greenough 1992) was used in Tezeren 2005, Tezeren 2009, Wang 2006, and Wei 2010. Wei 2010 also used the Oswestry Disability Index (Fairbank 1980). In Farrokhi 2010, functional quality of life was measured by the Denis Work Scale (Denis 1984).
Neurological status
Five trials included only neurologically intact participants (Alanay 2001; Guven 2009; Tezeren 2005; Tezeren 2009; Wei 2010). The neurological status of participants in the other three trials (Dai 2009; Farrokhi 2010; Wang 2006) was assessed with the Frankel Scale (Frankel 1969). Dai 2009 also used the motor score of the American Spinal Injury Association (Maynard 1997).
Pain
Back pain and graft site pain were quantified separately with the use of a visual analogue scale (range 0 to 10 cm) (Million 1982) in two trials (Dai 2009;Farrokhi 2010). Separate pain data were unavailable for the other six trials that used composite outcome measures, which included an assessment of pain.
Secondary outcomes
Postoperative complications
Postoperative complications were disclosed in all trials and consisted of medical complications or surgical complications, which included deep venous thrombosis, superficial infection, deep infection, neurological deterioration, implant failure, and miscellaneous complications.
Radiographic outcomes
Radiographic measurement parameters consisted of sagittal Index, local kyphosis, anterior body height compression, midsagittal spinal canal diameter at the injury level, posterior vertebral body height ratio, lumbar (T12‐S1) lordosis, and limitation of motion.
Perioperative outcomes (intraoperative blood loss and transfusion, duration of surgery)
All eight trials reported intraoperative blood loss and average operation time.
Employment (e.g. Denis Work Scale for previously employed people)
In three trials, employment status data were collected as part of a composite measure of function and activity: Alanay 2001 and Guven 2009 used the Likert questionnaire, and Farrokhi 2010 used the Denis Work Scale. However, none of these studies reported employment data.
Length of hospital stay
With the exception of Tezeren 2005 and Wei 2010, included trials reported on length of hospital stay.
Excluded studies
Seven studies were excluded for reasons given in the Characteristics of excluded studies.
Risk of bias in included studies
In the following sections, we report on items related to randomisation, allocation concealment, blinding (participants and personnel), blinding of outcome assessment, incomplete outcome data, selective reporting, and care programme comparability. Details for individual trials are presented in the Characteristics of included studies, and graphical representations of the risk of bias judgements can be seen in Figure 1 and Figure 2.
Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
Allocation
Five trials were at low risk of selection bias reflecting adequate sequence generation, resulting from using computer‐based randomisation (Dai 2009; Farrokhi 2010; Guven 2009) or a random‐number generator (Wei 2010) or from tossing a coin (Wang 2006). However, none of these provided sufficient details to confirm that there was allocation concealment. All three quasi‐randomised trials (Alanay 2001; Tezeren 2009; Tezeren 2005) for which allocation was based on admission sequence were judged at high risk of selection bias.
Blinding
Blinding of the surgeons performing the operations is not possible for these trials, but the effect on bias is unclear. Alanay 2001, in which investigators confirmed in an email that there was blinded assessment, was judged at low risk of detection bias. Although three other trials (Dai 2009, Guven 2009, Wei 2010) reported independent assessors, none confirmed assessor blinding or the use of safeguards to ensure blinding.
Incomplete outcome data
All trials were judged at unclear risk of attrition bias. Data on operative time, blood loss, and hospitalisation were reported for only 17 of 20 participants in Alanay 2001. At seven‐year follow‐up, losses were 22/37 in the fusion group and 22/36 in the non‐fusion group in Dai 2009. Losses to follow‐up were not described in the other trials.
Selective reporting
All trials were judged at unclear risk of reporting bias given the lack of prospective trial registration and protocols.
Other potential sources of bias
Generally information concerning the equivalence of the so‐called 'care programmes' was insufficient, as was information needed to judge participants' compliance with rehabilitation programmes.
Effects of interventions
The five comparisons are reported in turn below.
Comparison 1: short‐segment instrumentation versus long‐segment instrumentation
Short‐segment (SS) instrumentation versus long‐segment (LS) instrumentation was assessed in 90 participants with thoracolumbar burst fracture by Guven 2009 and Tezeren 2005. Thirty‐six participants in Guven 2009 treated with SS or LS were treated with fracture level screw incorporation at the same time (SS+/ LS+). In Tezeren 2005, all fractures were type B according to the Denis Classification. The participants in Guven 2009 had type A and B fractures according to the Denis (three‐column) Classification with posterior longitudinal ligament injury. Participants were followed up for an average of 50 months in Guven 2009 and 29 months in Tezeren 2005.
Primary outcomes
Self‐reported function and quality of life, and pain
Without presenting data, Guven 2009 reported that overall function and pain scores, based on results from a patient‐completed Likert questionnaire, showed good clinical outcomes for all groups with no significant differences between group differences at final follow‐up (reported P = 0.426). Tezeren 2005 found no significant differences between the two groups in LBOS at final follow‐up (MD ‐2.00, 95% Cl ‐5.27 to 1.27; seeAnalysis 1.1).
Neurological status
Both trials included only neurologically intact participants. Guven 2009 reported that no neurological complications. Tezeren 2005 reported that one participant had a "sustained partial neurological deficit (Frankel C) in the early postoperative period due to epidural hematoma", which resolved in a few days after a posterior decompression was performed. The intervention group for this patient was not identified.
Secondary outcomes
Postoperative complications
Guven 2009 reported two deep venous thromboses in the SS+ group and one superficial infection in the SS group. All fusions healed well without the need for revision surgery.
Tezeren 2005 reported that no implants were removed, including in one SS group participant who had a "pedicle screw dislodgement". One trial participant incurred a partial neurological deficit (see above), and another a superficial infection treated by debridement and antibiotics. Tezeren 2005 did not report the treatment group for either patient.
Radiographic outcomes (anterior body height compression, local kyphosis angle)
At final follow‐up, the LS group showed better clinical outcomes in correction of anterior body height compression (MD 5.03%, 95% Cl 2.12% to 7.94%) and local kyphosis angle (MD 2.98°, 95% Cl 0.96° to 5.01°; seeAnalysis 1.2) than were seen in the SS group. Although the results from the two comparisons in Guven 2009 and those of Tezeren 2005 favoured long‐segment instrumentation, these results are significantly heterogeneous (I² = 96%; I² = 80%), and so the pooled results are provided for illustrative purposes only.
Perioperative outcomes
Short‐segment surgery took statistically significantly less time to perform (MD ‐9.92 min, 95% Cl ‐19.31 to ‐0.54 min; seeAnalysis 1.3). No differences were noted between the two groups in intraoperative blood loss (MD ‐10.28 ml, 95% Cl ‐65.93 to 45.37 ml; seeAnalysis 1.3). Given the heterogeneous data, random‐effects results are presented for both outcomes. However, fixed‐effect results are similar, as are those noted when the data from Tezeren 2005 are excluded.
Employment
Neither trial presented data on employment.
Length of hospital stay
Guven 2009 found that the SS groups stayed in hospital on average one day less (MD ‐1.28 days, 95% Cl ‐2.16 to ‐0.40 days; seeAnalysis 1.4). Tezeren 2005 did not report on length of hospital stay.
Comparison 2: short‐segment instrumentation with transpedicular grafting versus short‐segment fixation alone
Short‐segment instrumentation with transpedicular grafting (TPG) versus short‐segment fixation alone (NTPG) was assessed in 20 participants with burst fractures between T11 and L3 without neurological impairment by Alanay 2001. Participants had type A, B, and E fractures according to the Denis (three‐column) Classification.
Because of large discrepancies between reported P values that indicated no statistically significant effect (P > 0.05) and those derived from inputting of continuous outcome measures data into the analyses, we treated the reported standard deviations as if these were standard errors and adjusted accordingly.
Primary outcomes
Self‐reported function and quality of life, and pain
Participant‐perceived function and pain at the latest follow‐up (32 months) were assessed by having participants complete a 10‐item Likert questionnaire. Scores showed 'good' clinical outcomes for both groups with no significant difference between the two groups (MD ‐0.50, 95% Cl ‐1.37 to 0.37; seeAnalysis 2.1).
Neurological status
No participants with prior impairment in neurological function were included in Alanay 2001, and all participants had normal findings on a neurological examination performed at final follow‐up.
Secondary outcomes
Postoperative complications
Screw breakage occurred in one participant in each group. The treatment group was not identified for the two other reported complications: deep vein thrombosis and superficial wound infection.
Radiographic outcomes (local kyphosis, sagittal index, anterior body height compression)
No differences were noted between the two groups in correction loss in local kyphosis between immediate postoperative and final follow‐up (MD 0.70°, 95% Cl ‐4.84° to 6.24°); in the sagittal index at final follow‐up (MD 0.10°, 95% Cl ‐4.64° to 4.84°); or in anterior body height compression at final follow‐up (MD ‐4.30%, 95% Cl ‐11.00% to 2.40%; seeAnalysis 2.2). A correction loss of greater than 10° was seen in five participants in the transpedicular grafting (TPG) group and in four of those in the non‐transpedicular grafting (NTPG) group (RR 1.25, 95% CI 0.46 to 3.33; seeAnalysis 2.3); one of the participants in each group had screw breakage.
Perioperative outcomes
No significant differences were noted between the two groups in intraoperative blood loss (MD ‐12.00 ml, 95% Cl ‐219.92 to 195.92 ml) or in duration of surgery (MD 16.00 min, 95% Cl ‐12.94 to 44.94 min; seeAnalysis 2.4).
Employment
This was not reported.
Length of hospital stay
No statistically significant differences were noted between the two groups in length of hospital stay (MD ‐4.00 days, 95% Cl ‐10.20 to 2.20 days; seeAnalysis 2.5).
Comparison 3: posterior instrumentation with fracture level screw incorporation versus posterior instrumentation alone
Posterior pedicular fixation including the fracture level ("including group") versus posterior pedicular fixation excluding the fracture level ("bridging group") was assessed in 152 consecutive participants with fractures between T10 and L3 by Farrokhi 2010 and Guven 2009. In Farrokhi 2010, all participants received short‐segment instrumentation. In Guven 2009, 36 participants were treated with short‐segment instrumentation with or without fracture level screw incorporation, and the other 36 were treated with long‐segment instrumentation with or without fracture level screw incorporation. Participants in Farrokhi 2010 had type A, B, and C fractures according to the Magerl Classification. Those in Guven 2009 had type A and B fractures according to the Denis (three‐column) Classification with posterior longitudinal ligament injury. Participants were followed up for an average of 37 months in Farrokhi 2010 and 50 months in Guven 2009.
Primary outcomes
Self‐reported function and quality of life
Neither trial reported data that could be presented in the analyses. Farrokhi 2010 reported "no significant difference between the two groups" in Denis Work Scale results at final follow‐up (reported P = 0.08). Guven 2009 reported that overall function and pain scores, based on results from a patient‐completed Likert questionnaire, showed good clinical outcomes for all groups with no significant between‐group differences at final follow‐up (reported P = 0.426).
Neurological status
Although 20 participants had neurological deficits (thus Frankel scores other than E) in Farrokhi 2010 at trial entry, no data were presented on neurological status at final follow‐up. Guven 2009, which included only neurologically intact participants, reported no neurological complications at final follow‐up.
Pain
In Farrokhi 2010, participants in the 'including group' had lower pain scores at follow‐up, but the difference between the two groups was of marginal significance, both statistically and clinically (MD ‐0.70, 95% Cl ‐1.40 to 0.00; seeAnalysis 3.1). Separate pain data were not available for Guven 2009.
Secondary outcomes
Postoperative complications
No differences between the 'including group' and the 'bridging group' in incidence of postsurgical infection were reported (seeAnalysis 3.2). Guven 2009 reported two deep venous thromboses in the 'including group'. All fusions healed well without the need for revision surgery in Guven 2009. Farrokhi 2010 found higher implant failure, mainly rod displacement or breakage, in the bridging group (2/38 vs 9/42; RR 0.25; 95% CI 0.06 to 1.07; seeAnalysis 3.2).
Radiographic outcomes (sagittal plane kyphosis, limitation of motion, anterior body height compression)
Data from individual comparisons showed better radiological outcomes for the 'including group' but were significantly heterogeneous. Given the very high inconsistency indexes (I² = 74% for kyphosis angle and 96% for anterior body height compression), data were pooled using the random‐effects model for illustrative purposes. At final follow‐up, the 'including group' showed statistically significantly better clinical outcomes in the correction of local kyphosis angle (MD ‐3.04°, 95% Cl ‐5.73° to ‐0.35°) and in anterior body height compression (MD ‐3.86%, 95% Cl ‐7.29% to ‐0.43%) compared with the 'bridging group' (seeAnalysis 3.3). Farrokhi 2010, however, found that the difference between the two groups in limitation of motion at final follow‐up was not statistically significant (MD ‐5.00°, 95% Cl ‐11.14° to 1.14°; seeAnalysis 3.3).
Perioperative outcomes
No significant differences were noted between the two groups in intraoperative blood loss (MD 2.80 ml, 95% Cl ‐24.49 to 30.10 ml) or in duration of surgery (MD ‐2.43 min, 95% Cl ‐7.42 to 2.56 min; seeAnalysis 3.4).
Employment
Neither trial provided data on employment.
Length of hospital stay
Pooled results from the two trials showed no significant differences in length of hospital stay between the two groups (MD 0.91 days, 95% Cl ‐0.67 to 2.48 days; seeAnalysis 3.5).
Comparison 4: monosegment pedicle screw instrumentation (MSPI) versus short‐segment pedicle screw instrumentation (SSPI)
Wei 2010 investigated this comparison in 85 patients with single‐level closed thoracolumbar burst fractures (AO‐ASIF type A3; complete burst fractures such as type A3.3 were excluded) involving T11 to L2 without neurological impairment. Participants were followed up for an average of 27.8 months.
Self‐reported function and quality of life
Self‐reported function and quality of life were assessed using the LBOS and the Oswestry Disability Index (ODI) at last follow‐up. Data provided in the trial report showed a statistically significant difference (P < 0.0001) in favour of MSPI in the final LBOS scores (MD 14.70, 95% Cl 10.76 to 18.64; seeAnalysis 4.1). However, this was not reported to be statistically significant in the trial report (P = 0.07); this lack of statistical significance would also be the case if standard errors rather than standard deviations had been reported (seeAnalysis 4.1). No significant differences were noted between the two treatment groups in ODI results (MD ‐3.50, 95% Cl ‐8.09 to 1.09; seeAnalysis 4.2).
Neurological status
Wei 2010 reported "no neurological deteriorations after surgery".
Pain
No significant difference was reported between the two treatment groups in visual analogue pain scores at final follow‐up (MD ‐0.20, 95% Cl ‐0.48 to 0.08; seeAnalysis 4.3); data were provided via email by the trial author.
Secondary outcomes
Postoperative complications
Wei 2010 reported that one patient in the MSPI group who had removed his hyperextension braces prematurely was found to have screw dislodgement. This patient had severe back pain and underwent an addition operation to remove the loosened screws along with long‐segment fixation to correct the progressed kyphosis. Reported complications included four urinary infections and one superficial infection, all of which were treated by antibiotics. Treatment groups were not identified for these complications, nor for the participant who broke a screw after a fall down stairs after 18 months' follow‐up (the implants were left in situ given that the participant had no symptoms).
Radiographic outcomes (sagittal index, anterior body height compression)
At final follow‐up, the sagittal index was statistically significantly lower in the MSPI group (MD 2.30°, 95% Cl 0.79° to 3.81°; seeAnalysis 4.4). No statistically significant difference was noted between the two groups in anterior body height compression (MD 2.30%, 95% Cl ‐2.39% to 6.99%; seeAnalysis 4.4).
Perioperative outcomes
The MSPI group had less blood loss (MD ‐303.00 ml, 95% Cl ‐347.85 to ‐258.15 ml) and a shorter duration of surgery (MD ‐51.00 min, 95% Cl ‐62.49 to ‐39.51 min) than the SSPI group (seeAnalysis 4.5). Both results were statistically significant.
Employment
Wei 2010 did not report this outcome.
Length of hospital stay
Wei 2010 did not report this outcome.
Comparison 5: posterior instrumentation with fusion versus posterior instrumentation alone
Three trials (Dai 2009; Tezeren 2009; Wang 2006) involving 115 participants with thoracolumbar burst fractures compared fusion with no fusion (non‐fusion). Participants in Dai 2009 and Wang 2006 were treated by short‐segment fixation, and participants in Tezeren 2009 were treated by long‐segment instrumentation. Dai 2009 included patients with a single‐level Denis type B burst fracture involving the thoracolumbar spine and a load‐sharing score of ≤ 6. Participants in Tezeren 2009 and Wang 2006 had type A, B, and C fractures according to the Denis (three‐column) Classification. Study inclusion in Tezeren 2009 was limited to neurologically intact participants. Participants were followed up for a minimum of five years in Dai 2009, and for an average of 34.6 months in Tezeren 2009 and 41 months in Wang 2006.
Primary outcomes
Self‐reported function and quality of life (36‐Item Short Form Health Survey, Low Back Outcome Score)
Dai 2009, which reported separate SF‐36 physical component summary (PCS) and mental component summary (MCS) scores, found no difference between the two treatment groups at final follow‐up (PCS: MD ‐1.50, 95% Cl ‐4.02 to 1.02; MCS, MD ‐1.20, 95% Cl ‐3.49 to 1.09; seeAnalysis 5.1). A similar finding applied to pooled LBOS data from Tezeren 2009 and Wang 2006 (MD ‐0.60, 95% Cl ‐3.27 to 2.06; seeAnalysis 5.1).
Neurological status
At final follow‐up, no participants showed a decline in neurological status in any of the three trials.
Based on the Frankel scale, 25 participants in Dai 2009 and 16 participants in Wang 2006 had a neurological deficit (grades A to D) at recruitment. At final follow‐up, most had recovered. No significant differences were reported between the groups in numbers who had recovered by one Frankel grade or more (RR 0.89, 95% Cl 0.70 to 1.11) nor in numbers with no neurological deficit (RR 1.0, 95% CI 0.91 to 1.10; seeAnalysis 5.2).
Pain
In Dai 2009, no significant differences were observed between groups in visual analogue scale scores at final follow‐up (MD ‐0.10, 95% Cl ‐0.74 to 0.54; seeAnalysis 5.3). Separate pain data were unavailable for Tezeren 2009 and Wang 2006.
Secondary outcomes
Postoperative complications
No significant differences were reported between the fusion group and the non‐fusion group in implant failure (RR 1.42, 95% Cl 0.43 to 4.68). However, two thirds of participants in the fusion group of Dai 2009 and nearly one quarter of those in Wang 2006 had long‐term donor site pain (RR 31.67, 95% Cl 4.48 to 224.11; seeAnalysis 5.4).
Radiographic outcomes (anterior body height compression, sagittal alignment)
Results for anterior body height compression at final follow‐up favoured the fusion group in Tezeren 2009 but the non‐fusion group in Wang 2006. Pooling, for illustrative purposes, of these heterogeneous data showed no difference between the two groups in anterior body height compression (MD 1.50%, 95% Cl ‐8.99% to 11.98%; seeAnalysis 5.5).
Pooled data from Dai 2009 and Wang 2006 showed no significant differences between groups in the local kyphosis angle (MD ‐0.36°, 95% Cl ‐0.99° to 0.28°; seeAnalysis 5.5).
Tezeren 2009 found no differences between groups in the sagittal index (MD ‐0.50°, 95% Cl ‐1.64 to 0.64°; seeAnalysis 5.5).
Perioperative outcomes
Pooled results, using the random‐effects model because of significant heterogeneity, showed that the fusion group had significantly greater intraoperative blood loss ((MD 189.49 ml, 95% CI 86.54 to 292.45 ml) and longer duration of surgery (MD 64.29 min; 95% CI 42.94 to 85.64 min; seeAnalysis 5.6). Similar findings in favour of the non‐fusion group were found when the fixed‐effect model was used.
Employment
The employment of participants was not mentioned in any of the three studies.
Length of hospital stay
Pooled data from the three trials showed no significant differences between the fusion group and the non‐fusion group in length of hospital stay (MD ‐0.88 days, 95% Cl ‐2.18 to 0.42 days; seeAnalysis 5.7).
Discussion
Summary of main results
Eight randomised controlled trials involved a total of 448 participants were included and divided into five comparison groups. The main results were as follows:
Pedicle screw fixation versus other methods of surgery that do not involve pedicle screw fixation
None of the completed and published studies identified in the detailed searches fulfilled the inclusion criteria for this comparison.
Different methods of pedicle fixation
Two trials (Guven 2009;Tezeren 2005) compared short‐segment instrumentation versus long‐segment instrumentation. Investigators found no significant differences between the two groups in self‐reported function and quality of life at final follow‐up. Only patients without neurological impairment were included in these trials. One patient, whose group was not identified, in Tezeren 2005 incurred a postoperative partial neurological deficit, which was resolved after further surgery.
One trial (Alanay 2001) compared short‐segment instrumentation with transpedicular grafting (TPG) versus short‐segment fixation alone (NTPG). Alanay 2001 found no significant differences between the two groups in patient‐perceived function and pain at final follow‐up. All participants had normal findings on neurological examination at baseline and at final follow‐up in Alanay 2001.
Two trials (Farrokhi 2010;Guven 2009) compared posterior instrumentation with fracture level screw incorporation (including group) versus posterior instrumentation alone (bridging group). Researchers reported no differences between the two groups in patient‐reported function and quality of life. Farrokhi 2010 also found no difference between the two groups in postoperative pain scores. Guven 2009, which recruited only neurologically intact patients, reported that all participants had normal findings on neurological examination at final follow‐up. Farrokhi 2010 did not report postoperative neurological status.
Wei 2010 compared monosegmental pedicle screw instrumentation (MSPI) versus short‐segment pedicle screw instrumentation (SSPI). When disregarding the inconsistent LBOS data, Wei 2010 found no significant differences between the two groups in the OSI results or in pain scores at final follow‐up. No neurological deterioration was reported.
Three trials (Dai 2009;Tezeren 2009; Wang 2006) compared posterior instrumentation fusion with fusion versus without fusion. Investigators found no differences between two groups in function and quality of life or spinal pain. No participants showed a decline in neurological status in any of the three trials, and no significant difference was noted between groups in the numbers whose status had improved at final follow‐up. However, many participants in the fusion groups of both trials had long‐term donor site pain.
Overall completeness and applicability of evidence
We included in this review eight randomised controlled trials, which tested five separate comparisons. Overall, the evidence is not robust because of the risk of bias and the small sample sizes. Limitations in the data were compounded by the use of different rating scales of self‐reported function and quality of life and, in particular, by the lack of data in Guven 2009, which contributed to two comparisons. The range of follow‐up times in some trials is also a matter of concern (e.g. 6 to 84 months in Farrokhi 2010). Given that most participants were of working age, the lack of data on employment is a matter of special concern.
Four trials did not report on the timing of trial recruitment (Alanay 2001;Tezeren 2005;Tezeren 2009;Wang 2006). Participants were restricted to individuals without neurological impairment in five trials (Alanay 2001; Guven 2009; Tezeren 2005; Tezeren 2009; Wei 2010). Participants in the eight trials had no uniformity of fracture classification. The fractures in Wei 2010 were AO type A3 fractures. Most of the fractures in Tezeren 2005, Guven 2009, Dai 2009, Tezeren 2009, Wang 2006, and Alanay 2001 were types A and B according to the Denis Classification. Farrokhi 2010 included participants with type A, B, and C fractures according to the Magerl Classification. However, similar fractures were included by the trials in each comparison.
Quality of the evidence
Trial size was an important consideration, and most of the included trials were unlikely to have been sufficiently powered to detect between‐group differences for a range of outcome measures, should they exist. A strong possibility of biased results resulted from methodological weaknesses in several trials. In particular, the three quasi‐randomised trials are at high risk of selection bias. Some aspects of trial methodology, notably concealment of allocation, were always possible, but others, such as blinding, represented more of a challenge for these trials. In particular, blinding of the surgeons who performed the operations was not possible. Incomplete outcome data were also a potential source of bias. Overall, the quality of the evidence is poor.
Potential biases in the review process
This review was conducted in accordance with criteria and methods set out in a published protocol. We believe that our search strategy was comprehensive. It included handsearching of conference proceedings and searches for ongoing and recently completed trials. However, it was possible that we may have missed some potentially eligible trials, including any that have been published more recently.
Although every step of the process was conducted independently by two review authors, various choices had arisen in the compilation of analyses of this review. Generally, results at final follow‐up rather than change from baseline have been presented. This could result in a disparity between the results presented here for individual trials and their trial reports.
Although we sent requests to authors for missing data, we were only partially successful. In some cases, we discovered that the trial data were no longer available.
Agreements and disagreements with other studies or reviews
We identified one relevant systematic review on the management of traumatic thoracic and lumbar spine fractures that included comparisons of different methods of pedicle screw fixation (Verlaan 2004). Verlaan 2004 included not only randomised controlled trials but also retrospective and prospective case series. Five surgical categories were recognised in Verlaan 2004. These included posterior short‐segment (PS), posterior long‐segment (PL), reports on both posterior short‐ and long‐segment (PSL), anterior (A), and anterior combined with posterior (AP) techniques. A total of 132 papers, most of which were retrospective case series, were included, representing 5748 participants. Verlaan 2004 concluded that none of the five techniques used was able to maintain the corrected kyphosis angle, that functional outcome after surgery seems to be better than is generally believed, and that complications are relatively rare.
Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
Comparison 1 Short‐segment instrumentation versus long‐segment instrumentation, Outcome 1 Low Back Outcome Score (0 to 75: best outcome) at average 29.6 months follow‐up.
Comparison 1 Short‐segment instrumentation versus long‐segment instrumentation, Outcome 2 Radiological outcomes at final follow‐up.
Comparison 1 Short‐segment instrumentation versus long‐segment instrumentation, Outcome 3 Peri‐operative outcomes.
Comparison 1 Short‐segment instrumentation versus long‐segment instrumentation, Outcome 4 Length of hospital stay (days).
Comparison 2 Short‐segment instrumentation with transpedicular grafting (TPG) versus short‐segment fixation alone (NTPG), Outcome 1 Self‐reported function and pain (Likert scale: 0 to 10: best) at 32 months.
Comparison 2 Short‐segment instrumentation with transpedicular grafting (TPG) versus short‐segment fixation alone (NTPG), Outcome 2 Radiological outcomes at 32 months follow‐up.
Comparison 2 Short‐segment instrumentation with transpedicular grafting (TPG) versus short‐segment fixation alone (NTPG), Outcome 3 Kyphosis correction loss > 10 degrees.
Comparison 2 Short‐segment instrumentation with transpedicular grafting (TPG) versus short‐segment fixation alone (NTPG), Outcome 4 Peri‐operative outcomes.
Comparison 2 Short‐segment instrumentation with transpedicular grafting (TPG) versus short‐segment fixation alone (NTPG), Outcome 5 Length of hospital stay (days).
Comparison 3 Posterior pedicular fixation with fracture level screw incorporation (including) versus without incorporation (bridging), Outcome 1 Pain (VAS scale: 0 to 10: worst) at 37 months follow.
Comparison 3 Posterior pedicular fixation with fracture level screw incorporation (including) versus without incorporation (bridging), Outcome 2 Postoperative complications.
Comparison 3 Posterior pedicular fixation with fracture level screw incorporation (including) versus without incorporation (bridging), Outcome 3 Radiological outcomes at final follow‐up.
Comparison 3 Posterior pedicular fixation with fracture level screw incorporation (including) versus without incorporation (bridging), Outcome 4 Peri‐operative outcomes.
Comparison 3 Posterior pedicular fixation with fracture level screw incorporation (including) versus without incorporation (bridging), Outcome 5 Length of hospital stay (days).
Comparison 4 Monosegmental transpedicular fixation (MSPI) versus short‐segment pedicle instrumentation (SSPI), Outcome 1 Low Back Outcome Score (0 to 75: best outcome) at average 27.8 months follow‐up.
Comparison 4 Monosegmental transpedicular fixation (MSPI) versus short‐segment pedicle instrumentation (SSPI), Outcome 2 Oswestry Disability Index (0 to 100: worst disability) at average 27.8 months follow‐up.
Comparison 4 Monosegmental transpedicular fixation (MSPI) versus short‐segment pedicle instrumentation (SSPI), Outcome 3 Pain (VAS scale: 0 to 10: worst) at 27.8 months follow up.
Comparison 4 Monosegmental transpedicular fixation (MSPI) versus short‐segment pedicle instrumentation (SSPI), Outcome 4 Radiological outcomes at 27.8 months follow‐up.
Comparison 4 Monosegmental transpedicular fixation (MSPI) versus short‐segment pedicle instrumentation (SSPI), Outcome 5 Peri‐operative outcomes.
Comparison 5 Fusion versus non‐fusion, Outcome 1 Self‐reported function and quality of life scores at final follow‐up.
Comparison 5 Fusion versus non‐fusion, Outcome 2 Neurological status at final follow‐up.
Comparison 5 Fusion versus non‐fusion, Outcome 3 Pain (VAS scale: 0 to 10: worst) at 72 months follow up.
Comparison 5 Fusion versus non‐fusion, Outcome 4 Postoperative complications.
Comparison 5 Fusion versus non‐fusion, Outcome 5 Radiographic outcomes at final follow‐up.
Comparison 5 Fusion versus non‐fusion, Outcome 6 Peri‐operative outcomes.
Comparison 5 Fusion versus non‐fusion, Outcome 7 Length of hospital stay (days).
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Low Back Outcome Score (0 to 75: best outcome) at average 29.6 months follow‐up Show forest plot | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 2 Radiological outcomes at final follow‐up Show forest plot | 2 | Mean Difference (IV, Random, 95% CI) | Subtotals only | |
| 2.1 Anterior body height compression (%) | 2 | 90 | Mean Difference (IV, Random, 95% CI) | 5.03 [2.12, 7.94] |
| 2.2 Local kyphosis angle (degrees) | 2 | 90 | Mean Difference (IV, Random, 95% CI) | 2.98 [0.96, 5.01] |
| 3 Peri‐operative outcomes Show forest plot | 2 | Mean Difference (IV, Random, 95% CI) | Subtotals only | |
| 3.1 Intraoperative blood loss (ml) | 2 | 90 | Mean Difference (IV, Random, 95% CI) | ‐10.28 [‐65.93, 45.37] |
| 3.2 Duration of surgery (min) | 2 | 90 | Mean Difference (IV, Random, 95% CI) | ‐9.92 [‐19.31, ‐0.54] |
| 4 Length of hospital stay (days) Show forest plot | 1 | 72 | Mean Difference (IV, Fixed, 95% CI) | ‐1.28 [‐2.16, ‐0.40] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Self‐reported function and pain (Likert scale: 0 to 10: best) at 32 months Show forest plot | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 2 Radiological outcomes at 32 months follow‐up Show forest plot | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 2.1 Correction loss in local kyphosis | 1 | Mean Difference (IV, Fixed, 95% CI) | 0.0 [0.0, 0.0] | |
| 2.2 Sagittal index (degrees) | 1 | Mean Difference (IV, Fixed, 95% CI) | 0.0 [0.0, 0.0] | |
| 2.3 Anterior body height compression % | 1 | Mean Difference (IV, Fixed, 95% CI) | 0.0 [0.0, 0.0] | |
| 3 Kyphosis correction loss > 10 degrees Show forest plot | 1 | Risk Ratio (M‐H, Fixed, 95% CI) | Totals not selected | |
| 4 Peri‐operative outcomes Show forest plot | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 4.1 Intraoperative blood loss (ml) | 1 | Mean Difference (IV, Fixed, 95% CI) | 0.0 [0.0, 0.0] | |
| 4.2 Duration of surgery (min) | 1 | Mean Difference (IV, Fixed, 95% CI) | 0.0 [0.0, 0.0] | |
| 5 Length of hospital stay (days) Show forest plot | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Pain (VAS scale: 0 to 10: worst) at 37 months follow Show forest plot | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 2 Postoperative complications Show forest plot | 2 | Risk Ratio (M‐H, Fixed, 95% CI) | Subtotals only | |
| 2.1 Infection | 2 | 152 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.49 [0.12, 2.08] |
| 2.2 Deep vein thrombosis | 2 | 152 | Risk Ratio (M‐H, Fixed, 95% CI) | 5.0 [0.26, 97.37] |
| 2.3 Implant failure | 2 | 152 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.25 [0.06, 1.07] |
| 3 Radiological outcomes at final follow‐up Show forest plot | 2 | Mean Difference (IV, Random, 95% CI) | Subtotals only | |
| 3.1 Local kyphosis angle (degrees) | 2 | 152 | Mean Difference (IV, Random, 95% CI) | ‐3.04 [‐5.73, ‐0.35] |
| 3.2 Anterior body height compression (%) | 1 | 72 | Mean Difference (IV, Random, 95% CI) | ‐3.86 [‐7.29, ‐0.43] |
| 3.3 Limitation of motion (degrees) | 1 | 80 | Mean Difference (IV, Random, 95% CI) | ‐5.0 [‐11.14, 1.14] |
| 4 Peri‐operative outcomes Show forest plot | 2 | Mean Difference (IV, Fixed, 95% CI) | Subtotals only | |
| 4.1 Intraoperative blood loss (ml) | 2 | 152 | Mean Difference (IV, Fixed, 95% CI) | 2.80 [‐24.49, 30.10] |
| 4.2 Duration of surgery (min) | 2 | 152 | Mean Difference (IV, Fixed, 95% CI) | ‐2.43 [‐7.42, 2.56] |
| 5 Length of hospital stay (days) Show forest plot | 2 | 152 | Mean Difference (IV, Random, 95% CI) | 0.91 [‐0.67, 2.48] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Low Back Outcome Score (0 to 75: best outcome) at average 27.8 months follow‐up Show forest plot | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 1.1 Assuming reports SDs were actually SDs | 1 | Mean Difference (IV, Fixed, 95% CI) | 0.0 [0.0, 0.0] | |
| 1.2 Assuming reported SDs were SEs | 1 | Mean Difference (IV, Fixed, 95% CI) | 0.0 [0.0, 0.0] | |
| 2 Oswestry Disability Index (0 to 100: worst disability) at average 27.8 months follow‐up Show forest plot | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 3 Pain (VAS scale: 0 to 10: worst) at 27.8 months follow up Show forest plot | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 4 Radiological outcomes at 27.8 months follow‐up Show forest plot | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 4.1 Sagittal Index (°) | 1 | Mean Difference (IV, Fixed, 95% CI) | 0.0 [0.0, 0.0] | |
| 4.2 Anterior body height compression (%) | 1 | Mean Difference (IV, Fixed, 95% CI) | 0.0 [0.0, 0.0] | |
| 5 Peri‐operative outcomes Show forest plot | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 5.1 Intraoperative blood loss (ml) | 1 | Mean Difference (IV, Fixed, 95% CI) | 0.0 [0.0, 0.0] | |
| 5.2 Duration of surgery (min) | 1 | Mean Difference (IV, Fixed, 95% CI) | 0.0 [0.0, 0.0] | |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Self‐reported function and quality of life scores at final follow‐up Show forest plot | 3 | Mean Difference (IV, Fixed, 95% CI) | Subtotals only | |
| 1.1 Physical component summary score (SF‐36) (0 to 100: best outcome) | 1 | 73 | Mean Difference (IV, Fixed, 95% CI) | ‐1.5 [‐4.02, 1.02] |
| 1.2 Mental component summary score (SF‐36) (0 to 100: best outcome) | 1 | 73 | Mean Difference (IV, Fixed, 95% CI) | ‐1.20 [‐3.49, 1.09] |
| 1.3 Low Back Outcome Score (0 to 75: best outcome) | 2 | 100 | Mean Difference (IV, Fixed, 95% CI) | ‐0.60 [‐3.27, 2.06] |
| 2 Neurological status at final follow‐up Show forest plot | 2 | Risk Ratio (M‐H, Fixed, 95% CI) | Subtotals only | |
| 2.1 Neurological recovery of 1+ Frankel grade | 2 | 41 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.89 [0.70, 1.11] |
| 2.2 Frankel grade E (no neurological deficit) | 2 | 133 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.00 [0.91, 1.10] |
| 3 Pain (VAS scale: 0 to 10: worst) at 72 months follow up Show forest plot | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 4 Postoperative complications Show forest plot | 3 | Risk Ratio (M‐H, Fixed, 95% CI) | Subtotals only | |
| 4.1 Implant failure | 3 | 173 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.42 [0.43, 4.68] |
| 4.2 Donor site pain | 2 | 131 | Risk Ratio (M‐H, Fixed, 95% CI) | 31.67 [4.48, 224.11] |
| 5 Radiographic outcomes at final follow‐up Show forest plot | 3 | Mean Difference (IV, Random, 95% CI) | Subtotals only | |
| 5.1 Anterior body height compression (%) | 2 | 100 | Mean Difference (IV, Random, 95% CI) | 1.50 [‐8.99, 11.98] |
| 5.2 Local kyphosis angle (degrees) | 2 | 131 | Mean Difference (IV, Random, 95% CI) | ‐0.36 [‐0.99, 0.28] |
| 5.3 Sagittal Index (degrees) | 1 | 42 | Mean Difference (IV, Random, 95% CI) | ‐0.5 [‐1.64, 0.64] |
| 6 Peri‐operative outcomes Show forest plot | 3 | Mean Difference (IV, Random, 95% CI) | Subtotals only | |
| 6.1 Intraoperative blood loss (ml) | 3 | 173 | Mean Difference (IV, Random, 95% CI) | 189.49 [86.54, 292.45] |
| 6.2 Duration of surgery (min) | 3 | 173 | Mean Difference (IV, Random, 95% CI) | 64.29 [42.94, 85.64] |
| 7 Length of hospital stay (days) Show forest plot | 3 | 173 | Mean Difference (IV, Fixed, 95% CI) | ‐0.88 [‐2.18, 0.42] |
