2.1 Literature search results
As shown in Figure 1, a total of 2 577 relevant studies (PubMed 1026, Embase 525, Web of Science 655, Cochrane Library 372) were retrieved in this study, the remaining 2243 studies were excluded after excluding duplicates, 210 articles were screened after reading the article name and abstract, and after reading the full text, 1 81 articles that did not meet the inclusion criteria were excluded, and finally, 29 articles[1, 4, 5, 14-39] were included, including ten randomised controlled trials, 19 Cohort studies. The basic information table of included studies is shown in Table 1.
Table 1.Basic information of the included studies
id
|
Study
|
Year
|
Intervention
|
Total
|
Age(/years)
|
Follow-up
|
1
|
Kirby, R. M.
|
1981
|
traction (plaster)/ intramedullary nail
|
13/13
|
12.7/14
|
16.4 months/ 13 months
|
2
|
Bar-On, E.
|
1997
|
external fixation / intramedullary nail
|
10/10
|
8.3 /9.3
|
14 months
|
3
|
Flynn, J. M.
|
2004
|
traction (plaster))/ intramedullary nail
|
35/48
|
8.7/10.2
|
3 years
|
4
|
Barlas, K.
|
2006
|
external fixation / intramedullary nail
|
20/20
|
8.2/9.2
|
——
|
5
|
Ramseier, L. E.
|
2007
|
external fixation / intramedullary nail
|
23/12
|
11.7/12.1
|
——
|
6
|
Park, K. C.
|
2012
|
plate/ intramedullary nail
|
23/22
|
13.6/14.2
|
20.9 months/ 22.4 months
|
7
|
Assaghir, Y.
|
2013
|
traction (plaster)/ intramedullary nail
|
52/52
|
4.1/4.9
|
41.9 months/ 40.3 months
|
8
|
G Rasool
|
2013
|
traction (plaster)/ intramedullary nail
|
51/49
|
8/9
|
2,4,12,24 weeks after discharge
|
9
|
Soleimanpour, J.
|
2013
|
traction (plaster)/ intramedullary nail
|
60/60
|
8.33/8.73
|
more than 1 year
|
10
|
I. Ahmad
|
2015
|
traction (plaster)/ intramedullary nail
|
30/30
|
4.62/4.43
|
——
|
11
|
Kousar, R.
|
2018
|
traction (plaster)/ intramedullary nail
|
30/36
|
7.2/8.1
|
6 months
|
12
|
Luo, Y.
|
2019
|
plate/ intramedullary nail
|
12/10
|
5.9
|
24 months
|
13
|
Brnjoš, K.
|
2020
|
traction (plaster))/ intramedullary nail
|
2878/1181
|
3.5/4.9
|
——
|
14
|
Gordon, J. E.
|
2020
|
traction (plaster)/ intramedullary nail
|
39/36
|
2.71/3.16
|
56.3 weeks /55.9 weeks
|
15
|
Li, J.
|
2020
|
external fixation / intramedullary nail
|
38/33
|
8.3/8.0
|
more than 2 years
|
16
|
Barnett, S. A.
|
2021
|
traction(plaster)/intramedullary nail
|
91/52
|
3/4.9
|
——
|
17
|
El-Adly, W.
|
2022
|
plate/ intramedullary nail
|
25/25
|
8.28/7.96
|
1 year
|
18
|
Hong, P.
|
2022
|
plate/ external fixation
|
36/35
|
8.2/8.3
|
more than 18 months
|
19
|
James, D.
|
2022
|
plate/ intramedullary nail
|
20/20
|
10.2/9.45
|
2 years
|
20
|
Yigit, S.
|
2020
|
plate/ intramedullary nail
|
28/32
|
11/9.7
|
29.8 months
|
21
|
Li, J.
|
2020
|
plate/ intramedullary nail
|
45/77
|
8.0/8.1
|
more than 2 years
|
22
|
Bajelidze, G.
|
2020
|
plate/ intramedullary nail
|
61/76
|
9
|
10.68 months
|
23
|
Wang, W.
|
2019
|
plate/ intramedullary nail
|
60/60
|
6.55/10.36
|
5 months
|
24
|
Bajelidze, G.
|
2019
|
traction(plaster)/intramedullary nail
|
44/49
|
3.5
|
12 months
|
25
|
Andreacchio, A.
|
2016
|
external fixation /intramedullary nail
|
15/23
|
5.9/5.6
|
67 months /60.4 months
|
26
|
Shemshaki, H. R.
|
2011
|
traction (plaster)/ intramedullary nail
|
23/23
|
6.5 /7.1
|
24 weeks
|
27
|
Mehdinasab, S. A.
|
2008
|
traction (plaster))/ intramedullary nail
|
30/36
|
7.2/8.1
|
6 months
|
28
|
Caglar, O.
|
2006
|
plate/ intramedullary nail
|
22/18
|
8/8
|
——
|
29
|
Song, H. R.
|
2004
|
traction (plaster)/ intramedullary nail
|
24/27
|
6/11
|
59.2 months
|
2.2 Evaluation of methodological quality of the included studies
According to the quality evaluation of the included studies, in 10 randomised controlled trials, referring to the Cochrane risk of bias tool, it is necessary to perform quality evaluation analysis and risk of bias evaluation on the included studies from six aspects including selection bias, performance bias, detection bias, attrition bias, reporting bias and other bias, and this study was analysed using the Review Manager 5.4.
Bias was evaluated for the included studies, ten studies specified the generation method of reasonable random sequence, two studies set allocation concealment so that patients and researchers could not predict the grouping results, a total of one study reported the blind setting for investigators and subjects, that is, both investigators and subjects did not know the interventions received during the trial, one study blinded outcomes, that is, the investigators who evaluated the study results did not know the interventions received by the subjects when collection outcomes, the outcome data reported in nine studies were complete and did not affect the outcome evaluation, a total of ten studies stated the study results and reported the expected outcomes, there was no risk of bias from other sources in all studies, so other bias options in all studies were low risk.
The summary graph of risk of bias for quality evaluation is shown in Figure 2, where green indicates low risk of bias, yellow indicates unclear risk of bias, and red indicates high risk of bias.
Included cohort studies were assessed for quality using the N OS scale. A total of eight items, with a full score of 9 points, the higher the score, the smaller the possible risk of bias. From the table, it can be concluded that the overall reliability of cohort studies is high in 7 studies with a total score of more than 7 points and in 14 studies with a total score of more than 6 points.
2.3 Results of network meta-analysis
2.3.1 Surgical duration
Figure 3 shows the result evaluation system for the surgical duration, of which Figure 3 (a) is the evidence network diagram for operation time, which shows that this outcome measure involves four treatment measures, including plate, traction (plaster), external fixation and intramedullary nailing and involving a total of 13 studies [4, 23, 25-33, 35, 38] . The global ring inconsistency test was performed for the studies involved (P = 0.775 > 0.05), and the local inconsistency test was performed using the node-splitting method, as shown in Table 2.a, the results were not significant, and the consistency model could be used for analysis. Figure 3 (b) is a comparative plot of the surface under the cumulative ranking curve (SUCRA) for the surgical duration of the four interventions. According to this figure, when randomised controlled trials and cohort studies were included at the same time, the ASES score of the four interventions was ranked in order of traction (plaster) (SUCRA = 97.0), external fixation (SUCRA = 58.1), intramedullary nailing (SUCRA = 44.4) and plate (SUCRA = 0.4). Table 2.b and Figure 3 (c) are the league table and the intuitive forest plot of the table for the comparison between every two interventions. It can be seen that traction (plaster), external fixation, and intramedullary nailing were superior to plate and had significant statistical differences in the evaluation of the surgical duration, and the others had no significant statistical differences. The funnel plot is shown in Figure 3 (d), and no significant publication bias was observed.
Table 2.a Local inconsistency test results of point method
Side
|
Direct
|
Indirect
|
Difference
|
|
Tau
|
|
Coef
|
Std.Err
|
Coef
|
Std.Err
|
Coef
|
Std.Err
|
P>|z|
|
|
A C
|
-22.40
|
20.28
|
-31.53
|
15.96
|
9.13
|
25.80
|
0.72
|
20.14
|
A D
|
-22.95
|
7.03
|
-13.44
|
24.68
|
-9.50
|
25.66
|
0.71
|
20.13
|
B D *
|
45.6
|
23.52
|
-56.95
|
4936.02
|
102.55
|
4936.09
|
0.98
|
19.21
|
C D
|
8.701866
|
14.34
|
-0.66
|
21.44
|
9.36
|
25.79
|
0.72
|
20.14
|
Table 2.b Operation time trapezoidal table
Plate
|
0.00
(0.00,0.00)
|
0.00 (0.00,0.01)
|
0.00 (0.00,0.00)
|
2.94e+29
(5.07e+08,5.07e+08)
|
Traction (plaster)
|
1.91e+17 (0.00,0.00)
|
6.35e+19 (0.00,0.61)
|
1.53e+12
(99.21,2.37e+22)
|
0.00
(90414.41,90414.41)
|
External fixation
|
331.71 (0.00,1.61e+12)
|
4.63e+09
(14295.42,1.50e+15)
|
0.00
(1.64,1.64)
|
0.00 (0.00,1.46e+07)
|
Intramedullary nailing
|
2.3.2 Length of hospital stay
Figure 4 shows the result evaluation system for the length of hospital stay, of which Figure 4 (a) is the evidence network diagram for the length of hospital stay, which shows that this outcome measure involves four treatment measures including plate, traction (plaster), external fixation and intramedullary nailing and involving a total of 17 studies[5, 17, 18, 20, 23-34, 36]. The global ring inconsistency test was performed for the studies involved (P = 0.300 > 0.05), and the local inconsistency test was performed using the node-splitting method, as shown in Table 3.a, the results were not significant, and the consistency model could be used for analysis. Figure 4 (b) is a comparative plot of the surface under the cumulative ranking curve (SUCRA) for the length of hospital stay of the four interventions. The figure shows that when randomised controlled trials and cohort studies were included at the same time, the ASES scores of the four interventions were ranked in order of external fixation (SUCRA = 68.7), intramedullary nailing (SUCRA = 67.2), plate (SUCRA = 37.9), and traction (plaster) (SUCRA = 26.1). Table 3.b and Figure 4 (c) are the league table and the intuitive forest plot of the table for the comparison between every two interventions. Although the ranking of each treatment was different in the evaluation of the length of hospital stay according to the figure, there was no significant difference in the outcome measures. The funnel plot is shown in Figure 4 (d), and no significant publication bias was observed.
Table 3.a Local inconsistency test results of point method
Side
|
Direct
|
Indirect
|
Difference
|
|
Tau
|
|
Coef
|
Std.Err
|
Coef
|
Std.Err
|
Coef
|
Std.Err
|
P>|z|
|
|
A C
|
1
|
8.44
|
-6.12
|
7.33
|
7.12
|
11.18
|
0.52
|
8.44
|
A D
|
-2.66
|
3.19
|
2.87
|
8.82
|
-5.53
|
9.38
|
0.56
|
8.36
|
B D *
|
-3.23
|
3.14
|
-4.29
|
35.34
|
1.07
|
35.47
|
0.98
|
8.26
|
C D
|
3.69
|
6.72
|
-3.65
|
8.95
|
7.34
|
11.19
|
0.51
|
8.44
|
Table 3.b Length of hospital stay trapezoidal table
Plate
|
3.36 (0.00,14401.66)
|
0.05 (0.00,1975.79)
|
0.13 (0.00,43.76)
|
0.30 (0.00,1272.77)
|
Traction (plaster)
|
0.01 (0.00,2257.49)
|
0.04 (0.00,17.80)
|
20.62 (0.00,840167.38)
|
69.37 (0.00,1.09e+07)
|
External fixation
|
2.73 (0.00,81760.78)
|
7.55 (0.02,2492.64)
|
25.39 (0.06,11472.38)
|
0.37 (0.00,10952.24)
|
Intramedullary nail
|
2.3.3 Blood loss
Figure 5 shows the result evaluation system of blood loss, of which Figure 5 (a) is the evidence network diagram of blood loss, which shows that this outcome measure involves four treatment measures, including plate, traction (plaster), external fixation and intramedullary nailing and involving a total of eight studies[4, 23, 25, 27, 29-31, 33]. The global inconsistency test was performed for the involved studies (P = 0.921 > 0.05), and the local inconsistency test was performed using the node-splitting method, as shown in Table 4.a, the results were not significant, and the consistency model could be used for analysis. Figure 5 (b) is a comparative plot of the surface under the cumulative ranking curve (SUCRA) for blood loss of the four interventions. According to the figure, when randomised controlled trials and cohort studies were included at the same time, the ASES scores of the three interventions were ranked in order of intramedullary nailing (SUCRA=73.3), external fixation (SUCRA=72.2), and plate (SUCRA=4.5). Table 4.b and Figure 5 (c) are the league table and the intuitive forest plot of the table for the comparison between every two interventions. It can be seen that intramedullary nailing was superior to plate and had a significant statistical difference in the evaluation of blood loss, and the others had no significant statistical difference.
Table 4.a Inconsistency test results of point method
Side
|
Direct
|
Indirect
|
Difference
|
|
Tau
|
|
Coef
|
Std.Err
|
Coef
|
Std.Err
|
Coef
|
Std.Err
|
P>|z|
|
|
A C*
|
-58.69
|
16.99
|
3.48
|
160.80
|
-62.17
|
161.70
|
0.70
|
43.61
|
B C *
|
6.50
|
43.63
|
-117.84
|
320.44
|
124.34
|
323.40
|
0.70
|
43.61
|
Table 4.b Blood loss trapezoidal table
Plate
|
0.00
(0.00,1.17e+12)
|
0.00
(0.00,0.00)
|
1.12e+27
(0.00,0.00)
|
External fixation
|
70.89
(0.00,0.00)
|
1.58e+25 (7.66e+10,7.66e+10)
|
0.01 (5.48e+34,5.48e+34)
|
Intramedullary nail
|
2.3.4 Flynn Score
Figure 6 shows the result evaluation system of the Flynn score, of which Figure 6 (a) is the evidence network diagram for the Flynn score, which shows that this outcome measure involves four treatment measures, including plate, traction (plaster), external fixation and intramedullary nailing and involving a total of 13 studies[4, 16, 18, 19, 25, 27, 28, 30, 32-34, 37, 39]. The global ring inconsistency test was performed for the involved studies (P = 0.748 > 0.05), and the local inconsistency test was performed using the node-splitting method. As shown in Table 5.a, the results were not significant, the consistency model could be used for analysis. Figure 6 (b) is a comparative plot of the surface under the cumulative ranking curve (SUCRA) for the Flynn score of the four interventions. According to the figure, when randomised controlled trials and cohort studies were included at the same time, the ASES scores of the four interventions were ranked in order of intramedullary nail (SUCRA=77.2), plate (SUCRA=62.7), external fixation (SUCRA=36.4) and traction (plaster) (SUCRA = 23.8). Table 5.b and Figure 6 (c) are the league table and the intuitive forest plot of the table for the comparison between every two interventions. Although the ranking of each treatment was different in the evaluation of the Flynn score according to the figure, there was no significant difference. The funnel plot is shown in Figure 6 (d), and no significant publication bias was observed.
Table 5.a Inconsistency test results of point method
Side
|
Direct
|
Indirect
|
Difference
|
|
Tau
|
|
Coef
|
Std.Err
|
Coef
|
Std.Err
|
Coef
|
Std.Err
|
P>|z|
|
|
A C
|
-0.03
|
2.51
|
-1.17
|
1.87
|
1.14
|
3.13
|
0.71
|
1.51
|
A D
|
0.17
|
0.90
|
1.31
|
3.00
|
-1.14
|
3.13
|
0.72
|
1.51
|
B D *
|
1.21
|
0.82
|
0.11
|
93.88
|
1.10
|
93.88
|
0.99
|
1.45
|
C D
|
1.34
|
1.633
|
0.20
|
2.67
|
1.14
|
3.13
|
0.72
|
1.51
|
Table 5.b Flynn Score trapezoidal table
Plate
|
0.38
(0.04,3.82)
|
0.46
(0.03,8.36)
|
1.29
(0.25,6.81)
|
2.61 (0.26,25.94)
|
Traction (plaster)
|
1.20 (0.05,27.68)
|
3.37 (0.67,16.85)
|
2.17 (0.12,39.40)
|
0.83 (0.04,19.21)
|
External fixation
|
2.81 (0.19,41.34)
|
0.77
(0.15,4.08)
|
0.30
(0.06,1.48)
|
0.36
(0.02,5.25)
|
Intramedullary nail
|
2.3.5 Adverse events
Figure 8 shows the result evaluation system of adverse events, of which Figure 7 (a) is the evidence network diagram of adverse events, which shows that this outcome measure involves four treatment measures, including plate, traction (plaster), external fixation and intramedullary nailing and involving a total of 24 studies[1, 4, 5, 14-22, 24-26, 28, 29, 31, 33-38]. The global inconsistency test was performed for the involved studies (P = 0.1504 > 0.05), and the local inconsistency test was performed using the node-splitting method. As shown in Table 6.a, the results are not significant, the consistency model can be used for analysis. Figure 7 (b) is a comparative plot of the surface under the cumulative ranking curve (SUCRA) for adverse events of the four interventions. According to the figure, when randomised controlled trials and cohort studies were included at the same time, the ASES scores of the four interventions were ranked in order of plate (SUCRA = 95.0), intramedullary nail (SUCRA = 71.5), traction (cast) (SUCRA = 31.5) and external fixation (SUCRA = 2.0). Table 6.b and Figure 7 (c) are the league table and the intuitive forest plot of the table for the comparison between every two interventions. It can be seen that intramedullary nailing was more effective than external fixation and traction (plaster) and had significant statistical differences in the evaluation of adverse events, plate was superior to intramedullary nailing and external fixation and had significant statistical differences, and the others had no significant statistical differences. The funnel plot is shown in Figure 7 (d), and no significant publication bias was observed.s
Table 6.a Local inconsistency test results of point method
Side
|
Direct
|
Indirect
|
Difference
|
|
Tau
|
|
Coef
|
Std.Err
|
Coef
|
Std.Err
|
Coef
|
Std.Err
|
P>|z|
|
|
A C
|
0.75
|
1.48
|
3.19
|
0.83
|
-2.44
|
1.70
|
0.15
|
0. 80
|
A D
|
0.89
|
0.59
|
-1.54
|
1.59
|
2.44
|
1.69
|
0.15
|
0.80
|
B D *
|
-1.04
|
0.30
|
1.26
|
44.86
|
-2.30
|
44.86
|
0.96
|
0.81
|
C D
|
-2.30
|
0.58
|
0.14
|
1.60
|
-2.44
|
1.70
|
0.15
|
0.80
|
Table 6.b Adverse events trapezoidal table
Plate
|
5.16
(1.50,17.81)
|
13.65
(3.30,56.52)
|
1.82
(0.61,5.40)
|
0.19
(0.06,0.67)
|
Traction (plaster)
|
2.64
(0.78,8.97)
|
0.35
(0.20,0.63)
|
0.07
(0.02,0.30)
|
0.38
(0.11,1.28)
|
External fixation
|
0.13
(0.05,0.39)
|
0.55
(0.19,1.63)
|
2.84
(1.58,5.09)
|
7.50
(2.57,21.93)
|
Intramedullary nail
|