International Orthopaedics (SICOT) DOI 10.1007/s00264-015-2842-3
REVIEW ARTICLE
Unilateral versus bilateral pedicle screw fixation in short-segment lumbar spinal fusion: a meta-analysis of randomised controlled trials Zengfeng Xin 1 & Weixu Li 1
Received: 17 May 2015 / Accepted: 23 May 2015 # SICOT aisbl 2015
Abstract Objective We performed this meta-analysis of randomised controlled trials to compare the efficacy and safety of unilateral with bilateral fixation in short-segment lumbar spinal fusion. Methods Predefined terms were used to search electronic databases to identify relevant research. Randomised controlled trials (RCTs) published in English and Chinese during 1990– 2015 investigating efficacy and safety of unilateral and bilateral fixation in short-segment lumbar spinal fusion were included. Data of fusion rate, complications, visual analogue scale (VAS), Oswestry Disability Index (ODI), estimated blood loss (EBL) and length of hospital stay were extracted and analysed. Two reviewers independently searched information sources, selected eligible research, analysed data and evaluated risk of bias. Results Eleven RCTs comprising 756 participants were analysed. There was no significant difference in fusion rate, device-related complication, ODI, VAS and length of hospital stay between bilateral and unilateral groups. The unilateral group had the obvious advantage of reduced blood loss [mean difference (MD)−143.57, 95 % confidence interval (Cl) 206.61 to -80.54, P < 0.0001) and operation time (MD 52.72, 95 % Cl -73.58 to -31.87, P12 months of follow-up. For research that reported more than one VAS results, we extracted VAS for back pain. Binary outcome data was summarised by relative risk (RR) and 95 % confidence intervals (95 % Cl). As for continuous data, mean difference (MD) and 95 % Cl were summarised to improve clinical relevance. Standard deviation (SD) was abstracted from raw data. All calculations and data pooling were done using Review Manager 5.3. In each analysis, both fixedeffects and random-effects models were used; if there was no significant heterogeneity existed (I2 ≤60 %), we reported results using the fixed-effects model; otherwise, if the I2 ≥60 %, which meant significant heterogeneity existed, we reported results using the random-effects model. For adverse events, analysis of variance (ANOVA) was used; for continuous variables, Fisher exact test (for data ≤40) and for categorical data, chi-square test (for data >40) was used Significance was defined as a p 60 % of I2 . If analysis was defined as having heterogeneity, sources of heterogeneity were explored by examining the impact of the risk of bias and we tried to determine the reason. To strengthen our results, we also performed sensitivity analysis by assessing changes in the overall effect after
International Orthopaedics (SICOT) Table 1
Search process in Ovid MEDLINE database
Results
Search process
Key words
1
Unilateral: title, abstract, name of substance word, subject heading word Bilateral: title, original title, abstract, name of substance word, subject heading word or/1-2 Spine fusion: title, original title, abstract, name of substance word, subject heading word Spinal fusion: title, original title, abstract, name of substance word, subject heading word Lumbar fusion: title, original title, abstract, name of substance word, subject heading word PLIF: title, original title, abstract, name of substance word, subject heading word TLIF: title, original title, abstract, name of substance word, subject heading word PLF.mp: title, original title, abstract, name of substance word, subject heading word or/4–10 3 and 10
2 3 4 5 6 7 8 9 10 11
Study selection
TLIF transforminal lumbar interbody fusion, PLIF posterior lumbar interbody fusion
removing trials with a high risk of bias or that had some ambiguity as to whether they met inclusion criteria. Also, as mentioned above, each item was analysed using both fixed-effects and random-effects models. Comparison of the two results was used to test sensitivity. Fig. 1 Study selection flow diagram
Studies identified through database searching N=1336
The preliminary search yielded 1331 results, which were excluded primarily by title and abstract. Nineteen trials were reviewed in full text; 11 were included in the meta-analysis. Eight trials were excluded because of study design or no desirable outcomes reported. Study characteristics Characteristics of the 11 trials [2–4, 6–9, 15–18] aresummarised in Table 2. Studies and paticipants The 11 prospective randomised controlled trials selected for analysis comprised 756 participants, with 367 in experimental groups and 389 in control groups. Two trials reported on posterior lumbar interbody fusion (PLIF) [15], one on posterolateral fusion of the lumbar spine (PLF) [7] and eight were minimally invasive, performing transforminal lumbar interbody fusion (TLIF). One study [17] was written in Chinese and the other ten in English. In all trials, the experimental group had unilateral pedicle screw fixation and a comparison group in which bilateral pedicle screws were used. Five studies [4, 8, 9, 16, 18] performed one-level fusions, and six [2, 3, 6, 7, 15, 17] performed short-segment fusion (1 or 2 levels). Nine trials evaluated fusion rate, which was our primary outcome, all trials reported complications, six reported VAS and five reported ODI
Additional studies identified through other sources N=2
Studies after duplicates removed N=1331
Studies assessed by full-text N=19
Studies included N=118-18
Excluded non-clinical or non-comparative studies N=1312
Excluded non-RCTs or studies without desirable outcomes reported N=8
24 in Exp group 23 in Con group 40 in Exp group 41 in Con group
Aoki 2012 [8]
56 in Exp group 52 in Con group
37 in Exp group 43 in Con group
16 in Exp group 20 in Con group
33 in Exp group 35 in Con group
26 in Exp group 27 in Con group
50 in Exp group 52 in Con group
20 in Exp group 20 in Con group
19 in Exp group 20 in Con group
Xie 2012 [11]
Xue 2012 [12]
Dahdaleh 2013 [13]
Zhang 2013 [14]
Choi 2013 [15]
Lin 2012 [17]
Feng 2011 [16]
Dong 2014 [18] Exp group: PLIF, uni-PS+cage Con group: PLIF, bi-PS+cage
Fusion rate, VAS, ODI, JOA, EBL, OPT, LOS
VAS, ODI, JOA, EBL, OPT, LOS, Cost,
Fusion rate, VAS, ODI, EBL, OPT, LOS
Exp group: TLIF, uni-PS+cage Con group: TLIF, bi-PS+cage Exp group: TLIF, uni-PS+cage Con group: TLIF, bi-PS+cage
Fusion rate, EBL, OPT
Fusion rate, VAS, ODI, SF-36, EBL, OPT, LOS
Fusion rate, VAS, ODI, SF-36, EBL, LOS
Fusion rate, VAS, ODI mProlo, cost, EBL, OPT, LOS
Fusion rate, SF-36, JOA, EBL, OPT, LOS
Cage migration
Fusion rate, VAS, JOA, EBL, OPT Fusion rate SF-36, EBL, OPT, LOS
Main outcomes
Exp group: MIS-TLIF, uni-PS+cage Con group: MIS-TLIF, bi-PS+ cage
Exp group: TLIF, uni-PS+cage Con group: TLIF, bi-PS+cage
Exp group: TLIF, uni-PS+cage Con group: TLIF, bi-PS+cage
Exp group: TLIF, uni-PS+cage Con group: TLIF, bi-PS+cage
Exp group: PLIF, uni-PS+cage Con group: PLIF, bi-PS+cage
Exp group: TLIF, uni-PS+cage Con group: TLIF, bi-PS+cage
Exp group: TLIF, uni-PS+cage Con group: TLIF, bi-PS+cage Exp group: PLF, uni-PS Con group: PLF, bi-PS
Interventions 24 months Loss to follow-up: 6 % at 24th month 36 months Loss to follow-up: 0 at 36th month 6 months Loss to follow-up: 12 % at 6th month 36 months Loss to follow-up: 0 % at 36th month 24 months Loss to follow-up: 0 at 24th month 12 months Loss to follow-up: 12 % at 12th month 24 months Loss to follow-up: 0 at 24th month 24 months Loss to follow-up: 2 % at 24th month 12–21 months Loss to follow-up: 0 at 21st month 3 months Loss to follow-up: 0 at 3rd month 28–43 months Loss to follow-up: 0 at 36th month
Follow-up
None
None
None
None
National Natural Science Foundation of China
None
None
National Natural Science Foundation of China
None
None
None
Sponsor
a
Each group recalculated after excluding loss to follow-up
pedicle screw, bi-PS bilateral pedicle screw, SF-36 36-Item Short Form Healthy Survey, ODI Oswestry Disability Index, VAS visual analogue scale, JOA Japanese Orthopedic Association, EBL estimated blood loss, LOS length of hospital stay, OPT operation time, mProlo modified Prolo scores
Exp experimental, Con controls, PLF posterolateral fusion of lumbar spine, TLIF transforaminal lumbar interbody fusion, MIS-TLIF minimally invasive transforminal lumbar interbody fusion, PLIF posterior lumbar interbody fusion, uni-PS unilateral
46 in Exp group 56 in Con group
Duncan 2013 [10]
Fernández-Fairen 2007 [9]
Participantsa
Characteristics of the 11 randomised controlled trials in the meta-analysis
Study
Table 2
International Orthopaedics (SICOT)
International Orthopaedics (SICOT)
Device-related complications
results. Other outcomes, such as blood loss, operation time and hospital stay, were reported by most trials.
Seven of 11 studies reported device-related complications, which included cage migration, screw malposition, root irritation and screw loosening. There was no statistically significant device-related complication rate between bilateral and unilateral groups (Fig. 3).
Risk of bias Summary assessment of the risk of bias for each study is detailed in Table 3. Because it was hard to blind therapists and patients, blinding was applied to the radiologists who evaluated the evidence of fusion. For one study [6], which did not evaluate fusion rate, we blinded investigators who evaluated clinical outcomes. All eligible trials were evaluated as having a low or unknown risk of bias.
Clinical outcome We evaluated clinical outcome by performing a metaanalysis of ODI and VAS results. Six studies reported VAS and five reported ODI. One study [16] was excluded for short follow-up (3 months); another [17] had a 12- to 21-month follow-up; however, VAS and ODI data were collected 1 month after operation, so we also excluded it. Pooled data from the remaining four studies documented no significant difference in ODI or VAS (Fig. 4).
Meta-analysis There were no significant differences between bilateral and unilateral groups for patient age, gender, diagnosis, disc location and number of fusion levels.
Fusion rate Surgery and hospitalisation We defined fusion as bilateral formation of bridging trabecular bone connecting vertebrae and the absence of motion of fused segments (
REVIEW ARTICLE
Unilateral versus bilateral pedicle screw fixation in short-segment lumbar spinal fusion: a meta-analysis of randomised controlled trials Zengfeng Xin 1 & Weixu Li 1
Received: 17 May 2015 / Accepted: 23 May 2015 # SICOT aisbl 2015
Abstract Objective We performed this meta-analysis of randomised controlled trials to compare the efficacy and safety of unilateral with bilateral fixation in short-segment lumbar spinal fusion. Methods Predefined terms were used to search electronic databases to identify relevant research. Randomised controlled trials (RCTs) published in English and Chinese during 1990– 2015 investigating efficacy and safety of unilateral and bilateral fixation in short-segment lumbar spinal fusion were included. Data of fusion rate, complications, visual analogue scale (VAS), Oswestry Disability Index (ODI), estimated blood loss (EBL) and length of hospital stay were extracted and analysed. Two reviewers independently searched information sources, selected eligible research, analysed data and evaluated risk of bias. Results Eleven RCTs comprising 756 participants were analysed. There was no significant difference in fusion rate, device-related complication, ODI, VAS and length of hospital stay between bilateral and unilateral groups. The unilateral group had the obvious advantage of reduced blood loss [mean difference (MD)−143.57, 95 % confidence interval (Cl) 206.61 to -80.54, P < 0.0001) and operation time (MD 52.72, 95 % Cl -73.58 to -31.87, P12 months of follow-up. For research that reported more than one VAS results, we extracted VAS for back pain. Binary outcome data was summarised by relative risk (RR) and 95 % confidence intervals (95 % Cl). As for continuous data, mean difference (MD) and 95 % Cl were summarised to improve clinical relevance. Standard deviation (SD) was abstracted from raw data. All calculations and data pooling were done using Review Manager 5.3. In each analysis, both fixedeffects and random-effects models were used; if there was no significant heterogeneity existed (I2 ≤60 %), we reported results using the fixed-effects model; otherwise, if the I2 ≥60 %, which meant significant heterogeneity existed, we reported results using the random-effects model. For adverse events, analysis of variance (ANOVA) was used; for continuous variables, Fisher exact test (for data ≤40) and for categorical data, chi-square test (for data >40) was used Significance was defined as a p 60 % of I2 . If analysis was defined as having heterogeneity, sources of heterogeneity were explored by examining the impact of the risk of bias and we tried to determine the reason. To strengthen our results, we also performed sensitivity analysis by assessing changes in the overall effect after
International Orthopaedics (SICOT) Table 1
Search process in Ovid MEDLINE database
Results
Search process
Key words
1
Unilateral: title, abstract, name of substance word, subject heading word Bilateral: title, original title, abstract, name of substance word, subject heading word or/1-2 Spine fusion: title, original title, abstract, name of substance word, subject heading word Spinal fusion: title, original title, abstract, name of substance word, subject heading word Lumbar fusion: title, original title, abstract, name of substance word, subject heading word PLIF: title, original title, abstract, name of substance word, subject heading word TLIF: title, original title, abstract, name of substance word, subject heading word PLF.mp: title, original title, abstract, name of substance word, subject heading word or/4–10 3 and 10
2 3 4 5 6 7 8 9 10 11
Study selection
TLIF transforminal lumbar interbody fusion, PLIF posterior lumbar interbody fusion
removing trials with a high risk of bias or that had some ambiguity as to whether they met inclusion criteria. Also, as mentioned above, each item was analysed using both fixed-effects and random-effects models. Comparison of the two results was used to test sensitivity. Fig. 1 Study selection flow diagram
Studies identified through database searching N=1336
The preliminary search yielded 1331 results, which were excluded primarily by title and abstract. Nineteen trials were reviewed in full text; 11 were included in the meta-analysis. Eight trials were excluded because of study design or no desirable outcomes reported. Study characteristics Characteristics of the 11 trials [2–4, 6–9, 15–18] aresummarised in Table 2. Studies and paticipants The 11 prospective randomised controlled trials selected for analysis comprised 756 participants, with 367 in experimental groups and 389 in control groups. Two trials reported on posterior lumbar interbody fusion (PLIF) [15], one on posterolateral fusion of the lumbar spine (PLF) [7] and eight were minimally invasive, performing transforminal lumbar interbody fusion (TLIF). One study [17] was written in Chinese and the other ten in English. In all trials, the experimental group had unilateral pedicle screw fixation and a comparison group in which bilateral pedicle screws were used. Five studies [4, 8, 9, 16, 18] performed one-level fusions, and six [2, 3, 6, 7, 15, 17] performed short-segment fusion (1 or 2 levels). Nine trials evaluated fusion rate, which was our primary outcome, all trials reported complications, six reported VAS and five reported ODI
Additional studies identified through other sources N=2
Studies after duplicates removed N=1331
Studies assessed by full-text N=19
Studies included N=118-18
Excluded non-clinical or non-comparative studies N=1312
Excluded non-RCTs or studies without desirable outcomes reported N=8
24 in Exp group 23 in Con group 40 in Exp group 41 in Con group
Aoki 2012 [8]
56 in Exp group 52 in Con group
37 in Exp group 43 in Con group
16 in Exp group 20 in Con group
33 in Exp group 35 in Con group
26 in Exp group 27 in Con group
50 in Exp group 52 in Con group
20 in Exp group 20 in Con group
19 in Exp group 20 in Con group
Xie 2012 [11]
Xue 2012 [12]
Dahdaleh 2013 [13]
Zhang 2013 [14]
Choi 2013 [15]
Lin 2012 [17]
Feng 2011 [16]
Dong 2014 [18] Exp group: PLIF, uni-PS+cage Con group: PLIF, bi-PS+cage
Fusion rate, VAS, ODI, JOA, EBL, OPT, LOS
VAS, ODI, JOA, EBL, OPT, LOS, Cost,
Fusion rate, VAS, ODI, EBL, OPT, LOS
Exp group: TLIF, uni-PS+cage Con group: TLIF, bi-PS+cage Exp group: TLIF, uni-PS+cage Con group: TLIF, bi-PS+cage
Fusion rate, EBL, OPT
Fusion rate, VAS, ODI, SF-36, EBL, OPT, LOS
Fusion rate, VAS, ODI, SF-36, EBL, LOS
Fusion rate, VAS, ODI mProlo, cost, EBL, OPT, LOS
Fusion rate, SF-36, JOA, EBL, OPT, LOS
Cage migration
Fusion rate, VAS, JOA, EBL, OPT Fusion rate SF-36, EBL, OPT, LOS
Main outcomes
Exp group: MIS-TLIF, uni-PS+cage Con group: MIS-TLIF, bi-PS+ cage
Exp group: TLIF, uni-PS+cage Con group: TLIF, bi-PS+cage
Exp group: TLIF, uni-PS+cage Con group: TLIF, bi-PS+cage
Exp group: TLIF, uni-PS+cage Con group: TLIF, bi-PS+cage
Exp group: PLIF, uni-PS+cage Con group: PLIF, bi-PS+cage
Exp group: TLIF, uni-PS+cage Con group: TLIF, bi-PS+cage
Exp group: TLIF, uni-PS+cage Con group: TLIF, bi-PS+cage Exp group: PLF, uni-PS Con group: PLF, bi-PS
Interventions 24 months Loss to follow-up: 6 % at 24th month 36 months Loss to follow-up: 0 at 36th month 6 months Loss to follow-up: 12 % at 6th month 36 months Loss to follow-up: 0 % at 36th month 24 months Loss to follow-up: 0 at 24th month 12 months Loss to follow-up: 12 % at 12th month 24 months Loss to follow-up: 0 at 24th month 24 months Loss to follow-up: 2 % at 24th month 12–21 months Loss to follow-up: 0 at 21st month 3 months Loss to follow-up: 0 at 3rd month 28–43 months Loss to follow-up: 0 at 36th month
Follow-up
None
None
None
None
National Natural Science Foundation of China
None
None
National Natural Science Foundation of China
None
None
None
Sponsor
a
Each group recalculated after excluding loss to follow-up
pedicle screw, bi-PS bilateral pedicle screw, SF-36 36-Item Short Form Healthy Survey, ODI Oswestry Disability Index, VAS visual analogue scale, JOA Japanese Orthopedic Association, EBL estimated blood loss, LOS length of hospital stay, OPT operation time, mProlo modified Prolo scores
Exp experimental, Con controls, PLF posterolateral fusion of lumbar spine, TLIF transforaminal lumbar interbody fusion, MIS-TLIF minimally invasive transforminal lumbar interbody fusion, PLIF posterior lumbar interbody fusion, uni-PS unilateral
46 in Exp group 56 in Con group
Duncan 2013 [10]
Fernández-Fairen 2007 [9]
Participantsa
Characteristics of the 11 randomised controlled trials in the meta-analysis
Study
Table 2
International Orthopaedics (SICOT)
International Orthopaedics (SICOT)
Device-related complications
results. Other outcomes, such as blood loss, operation time and hospital stay, were reported by most trials.
Seven of 11 studies reported device-related complications, which included cage migration, screw malposition, root irritation and screw loosening. There was no statistically significant device-related complication rate between bilateral and unilateral groups (Fig. 3).
Risk of bias Summary assessment of the risk of bias for each study is detailed in Table 3. Because it was hard to blind therapists and patients, blinding was applied to the radiologists who evaluated the evidence of fusion. For one study [6], which did not evaluate fusion rate, we blinded investigators who evaluated clinical outcomes. All eligible trials were evaluated as having a low or unknown risk of bias.
Clinical outcome We evaluated clinical outcome by performing a metaanalysis of ODI and VAS results. Six studies reported VAS and five reported ODI. One study [16] was excluded for short follow-up (3 months); another [17] had a 12- to 21-month follow-up; however, VAS and ODI data were collected 1 month after operation, so we also excluded it. Pooled data from the remaining four studies documented no significant difference in ODI or VAS (Fig. 4).
Meta-analysis There were no significant differences between bilateral and unilateral groups for patient age, gender, diagnosis, disc location and number of fusion levels.
Fusion rate Surgery and hospitalisation We defined fusion as bilateral formation of bridging trabecular bone connecting vertebrae and the absence of motion of fused segments (
