2.1 Study Selection And Characteristics
The literature search identified 216 studies for consideration.Following removal of duplicates and a screening of abstracts, only 122 studies were eligible for inclusion. An examination of these full-text manuscripts was conducted, and only 12 studies met all of the eligibility criteria to be included in the analysis[4-16], among which 627 cases were in the experimental group and 656 cases in the control group, a total of 1283 cases.There are 7 literatures that mention the random grouping method: 1 pieces of numerical randomization [5], 4 pieces of hospital randomization [8–10,12], and 2 piece of envelope randomization [4,7].Two references mentioned the blind method, none mentioned the report hiding, while four mentioned the reasons and the number of people lost to visit.Jadad score: 3 article 5 points [4,5,8], 4 article 4 points [7,9, 10, 12,13], 3 article 2 points [2, 11], the rest article are 2 points.
Figure 1 PRISMA flow chart depicting selection of articles
2.2 Meta Analysis Index Results
2.2.1 Comparison of operative time
Operation time of two groups was mentioned in 11 studies [4–11, 13–15].The heterogeneity test in the literature showed that P < 0.00001, I2 = 99%, and the random effect model was used for analysis.The results showed that the proximal femur anti-rotation intramedullary nailing group had shorter operation time (MD=-17.02, 95%Cl: -29.64,-4.77, P < 0༎01, in Fig. 1). It indicates that the operation time of proximal femur anti-rotation intramedullary nail is shorter for the treatment of senile unstable intertrochanteric fracture.
2.2.2 Comparison Of Intraoperative Blood Loss
There were 9 studies on intraoperative blood loss [4–6, 8–11, 13, 15].The heterogeneity test in the literature showed that P < 0.000 01, I2 = 99%, and the random effect model was used for analysis.The results showed that the intraoperative blood loss in the proximal femur anti-rotation intramedullary nail group was less than that in the hemihip replacement group ((MD=-216.91, 95%CI: -269.84, -163.99, P < 0.000 1, in Fig. 2).This suggests that the proximal femur anti-rotation intramedullary nail group can reduce intraoperative blood loss compared with the treatment of semi-hip replacement for senile unstable intertrochanteric fractures.
Figure 1. Comparison of operative time between PFNA and hemiarthroplasty
Figure 2. Comparison of intraoperative blood loss between PFNA and hemiarthroplasty
2.2.3 Comparison Of The Time Of Weight Bearing
5 studies reported postoperative loading time [4,5,8,10,15].The heterogeneity test in the literature showed that P < 0.000 01, I2 = 99%, and the random effect model was used for analysis.The results showed that the time of weight bearing of proximal femur anti-rotation intramedullary nail was longer than that of hemihip replacement (MD = 4.8, 95%CI: 1.31, 8.29, P < 0.01, in Fig. 3).It is suggested that the treatment of unstable intertrochanteric fracture with proximal femur anti-rotation intramedullary nail can be compared with that of semi-hip replacement.
2.2.4 Comparison Of Hospitalization Time
Hospital stay was mentioned in 9 studies [4,7–11, 13–15].The heterogeneity test in the literature showed that P < 0.000 01, I2 = 99%, and the random effect model was used for analysis.The results showed tha Hospital stay of proximal femur anti-rotation intramedullary nail was shorter than that of hemihip replacemen (MD=-1.72, 95%CI:-3.25, -0.19, P < 0.05, in Fig. 4).
2.2.5 Comparison Of Harris Hip Score
Harris Hip Score was mentioned in 9 studies [5–8,11–13, 15].The heterogeneity test in the literature showed that P = 0.007, I2=62%, and the random effect model was used for analysis. Postoperative Harris function score was higher in the PFNA group, the difference was statistically significant (MD = 4.78, 95༅Cl: 2.59, 6.97, P < 0༎01, in Fig. 5).
2.2.6 Comparison Of Harris Hip Score Grade
5 studies reported Harris Hip Score grade [5–8,11–13, 15].The heterogeneity test in the lite rature showed that P = 0.41, I2 = 0%, using fixed effect model analysis. The postoperative Harris score was better in the PFNA group (OR = 1.54, 95%Cl: 1.09, 2.16, P < 0༎05, in Fig. 6).
Figure 3. Comparison of weight bearing days between PFNA and hemiarthroplasty
Figure 4. Comparison of hospital stay between PFNA and hemiarthroplasty
Figure 5. Comparison of Harris Hip Score between PFNA and hemiarthroplasty
Figure 6. Comparison of Harris grade between PFNA and hemiarthroplasty
2.2.7 Comparison of postoperative complications
5 studies reported postoperative complications [4,7,8,910,11].The heterogeneity test in the literature showed that P = 0.41, I2 = 0%, using fixed effect model analysis. There was no difference in postoperative medical complications between the two groups, and no statistical significance(P = 0.14)༈OR = 0.64, 95% CI: 0.35ཞ1.16, P = 0.14, in Fig. 7).
2.2.8 Comparison Of Postoperative Implantation Complications
Postoperative implantation complications was mentioned in 6 studies [4,5,8,10,13,15]. The heterogeneity test in the literature showed that P = 0.27, I2=21%, using fixed effect model analysis.. There was no difference in postoperative medical complications between the two groups, and no statistical significance(P = 0.78)༈OR = 0.93, 95༅ CI: 0.57ཞ1.52, P = 0.78, in Fig. 8).
Figure 7. Comparison of postoperative complications between PFNA and hemiarthroplasty
Figure 8. Comparison of postoperative implantation complications between PFNA and hemiarthroplasty
2.2.9 Comparison of Mortality rates
Mortality rates was mentioned in 10 studies [4–6,8–11,13–15]. The heterogeneity test in the literature showed that P = 0.97, I2=0%, using fixed effect model analysis. Postoperative mortality was lower in the PFNA group (OR = 0.50, 95% CI: 0.36 ~ 0.70, P < 0.0001, in Fig. 9)
2.2.10 Comparison of infection rates
Infection rates was mentioned in 7 studies [4–9]. The heterogeneity test in the literature showed that P = 0.75, I2=0%, using fixed effect model analysis. There was no difference in postoperative Infection rates between the two groups(OR = 0.50, 95༅ CI: 0.36ཞ0.70, P > 0༎05, in Fig. 10). At the same time, we also compared the on postoperative bedsores (4–6), urinary tract infections (4–6, 13), and venous thrombosis (4–6, 8), the differences between the two groups were not statistically significant (P > 0.05).
2.2.11 Comparison Of Re-operation Rates
Re-operation rates was mentioned in 9 studies [5–11, 13, 15]. The heterogeneity test in the studies showed that P = 0.41, I2=2%, using fixed effect model analysis. There was no difference in postoperative Infection rates between the two groups(OR = 1.33, 95༅ CI: 0.82ཞ2.20, P > 0༎05, in Fig. 11).
Figure 9.Forest plot of comparison of Mortality rates
Figure 10.Forest plot of comparison of infection rates
Figure 11 Forest plot of comparison of re-operation rates
2.2 Risk of Bias
The risk of bias in the studies we included was very low. The first two authors independently read each article on randomization, blinding, and outcome parameters, and then used RevMan 5.3 software to generate the risk of bias. (in Figure 12).
Fig. 12 Risk of bias graph and Risk of bias summary