Literature search
A total of 608 potentially relevant studies were identified. After removing 211 duplicates, we screened 397 papers. By reading the title and abstract, 355 papers were excluded according to the inclusion and exclusion criteria. A total of 42 studies were assessed by reading the full text; eventually, 16 studies involving 1050 patients (464 patients in the MIPO group and 586 patients in the ORIF group) were included in the meta-analysis [9, 10, 15–28]. The flow diagram of the included studies is shown in Figure 1. The characteristics of the included studies are listed in Table 1.
Table 1 here
Methodological quality
The methodological quality of the RCTs [9, 27] was assessed by the Cochrane Handbook, the assessment results are summarized in Figure 2. The quality index scores of the non-RCTs [10, 15–26, 28] were 14–20. The assessment results are summarized in Table 2.
Table 2 here
Results of the meta-analysis
Blood loss
Nine studies involving 610 patients reported blood loss [10, 15–18, 21, 22, 25, 27]. Heterogeneity tests indicated high heterogeneity (P < 0.00001; I2 = 98%); a random-effects model was used. The result showed lesser blood loss in the MIPO group than that in the ORIF group (MD = –115.26; 95% CI: –167.48 to –63.03; P < 0.0001; Figure 3).
Operation time
Thirteen studies [9, 10, 15–19, 21, 22, 25–28], with 921 patients, mentioned operation time, and the heterogeneity test indicated significant heterogeneity (P < 0.00001; I2 = 96%); thus, a random-effects model was adopted. The results showed shorter operation time in the MIPO group than that in the ORIF group (MD = –20.71; 95% CI: –30.21 to –11.22; P < 0.0001; Figure 4).
Radiation time
Three articles [23, 26, 28], with 189 patients, stated radiation time, but used different units of time; therefore, the SMD was adopted. A random-effects model was used, with obvious heterogeneity (P < 0.00001; I2 = 98%). The duration of radiation exposure in the MIPO group was longer than that in the ORIF group (MD = 4.36; 95% CI: 1.21 to 7.51; P = 0.007; Figure 5).
Postoperative pain
The visual analogue scale (VAS) was used to evaluate postoperative pain, and seven studies [15, 18, 19, 24–27], with 486 patients, reported the VAS score. A random-effects model was used, with obvious heterogeneity (P < 0.0001; I2 = 79%). The meta-analysis showed a significantly lower VAS score in the MIPO group than in the ORIF group (MD = –0.54; 95% CI: –1.04 to –0.04; P = 0.04; Figure 6).
Union time
Ten studies [9, 15–19, 21, 23, 27, 28], with 608 patients, indicated postoperative union time, but used different units of time; therefore, the SMD was adopted. Heterogeneity tests indicated that significant heterogeneity (P = 0.0003; I2 = 71%); thus, a random-effects model was adopted. The meta-analysis showed a shorter time to union in the MIPO group than in the ORIF group (SMD = –0.38; 95% CI: –0.70 to –0.06; P = 0.02; Figure 7).
Functional outcomes
The constant score of Neer type II fractures were provided in four studies [9, 15, 23, 27, 28]. A fixed-effects model was used (P = 0.16; I2 = 39%), and analysis showed significantly higher score in the MIPO group than in the ORIF group (MD = 2.24; 95% CI: 0.82 to 3.65; P = 0.02; Figure 8).
The constant score of Neer type III fractures were stated in four studies [9, 15, 24, 27]. A fixed-effects model was used (P = 0.42; I2 = 0%), and analysis showed significantly higher score in the MIPO group than in the ORIF group (MD = 1.95; 95% CI: 0.98 to 2.92; P < 0.001; Figure 8).
Meta-analysis was not performed for the constant score of Neer type IV PHFs because it was reported by only one study.
Axillary nerve injury
Four studies [17, 22, 23, 25], with 305 patients, reported axillary nerve injury. A fixed-effects model was used (P = 0.95; I2 = 0%), and results showed a significantly higher rate of axillary nerve injury in the MIPO group than in the ORIF group (OR = 4.88; 95% CI: 1.03 to 23.25; P = 0.05).
Complications
Thirteen studies reported complications. A fixed-effects model was used (P = 0.88; I2 = 0%), and pooled results showed no significant difference in total complication rate between the two groups (OR = 0.74; 95% CI: 0.51 to 1.07; P = 0.11; Figure 9). However, the MIPO group had a significantly higher rate of axillary nerve injury that the ORIF group (OR = 4.88; 95% CI: 1.03 to 23.25; P = 0.05; I2 = 0%). The pooled results of the following complications showed no significant difference between the two groups (Table 3): avascular necrosis, impingement, screw perforation, implant loosening, delayed union or nonunion, limited abduction, and varus.
Table 3 here
Publication bias
Funnel plots of the total complication rate showed no significant visual evidence of publication bias (Figure 9).