Proximal humeral fractures are one of the most common fractures, accounting for 4%-5% of total body fractures, and their incidence increases with age. Neer classification is the most widely used classification system, and it divides the proximal humeral fracture into four parts: humeral head, greater tuberosity, lesser tuberosity and humeral shaft. According to the fracture displacement (>1 cm) and angulation (>45°) , proximal humeral fracture is divided into 1 to 4-part. For non-displaced or slightly displaced proximal humeral fractures, conservative treatment can usually achieve satisfactory clinical outcomes. However, the treatment of complex proximal humeral fractures (Neer 3 or 4-part fractures and fracture-dislocations) is more difficult. Conservative treatment, locking plate and humeral head replacement are the most commonly used treatments in clinic, but there is still debate about their clinical efficacy. Conservative treatment mainly involves proper manual reduction and fixation by suspension or other means. It is theoretically difficult to perform satisfactory fracture reduction and reliable fixation, and may bring a relatively high rate of malunion and other complications. Neer et al, first proposed humeral head replacement for treatment of displaced 3 or 4-part fractures and fracture-dislocations, and reported higher patient satisfaction. However, humeral head replacement is more traumatic with complex surgical technique, long postoperative recovery time and more complications, which often leads to poor function. Later studies have reported poor functional results as well. Locking plate fixation is also a classic treatment for displaced proximal humeral fractures. Plate with angle-stabilized screws is used for fracture fixation as a whole. Theoretically, complications such as internal fixation failure and screw penetration can be significantly reduced. Currently, there have been many systematic reviews on the proximal humeral fractures, but the included literatures are mostly retrospective studies with low levels of evidence. Objective This study conducted a meta-analysis of currently published randomized controlled trials to obtain a higher level of evidence for clinical guidelines in the treatment of complex proximal humeral fractures in the elderly.

The randomized controlled trials that met the inclusive criteria and reported functional scores and complications were searched on Pubmed, EMBASE and Cochrane General Register of Controlled Trials. The evidence levels of literatures were evaluated for the extraction of functional scores and complications. Metaman 5.0 software was used for meta-analysis.

A total of 4 randomized controlled trials involving 215 patients were included in the study. Two of the literatures (105 patients) compared the overall and group Constant scores and complications between conservative treatment and humeral head replacement of Neer part-4 proximal humeral fractures. The meta-analysis showed a higher short-term overall Constant score (WMD =5.69, 95% CI: 0.34-11.03, P=0.04) of conservative treatment compared with humeral head replacement for Neer part-4 proximal humeral fractures. The VAS score (WMD =-2.75, 95% CI: -4.45- -1.04, P= 0.002) was significantly improved one year after humeral head replacement, but the there was no significant difference compared with conservative treatment (WMD =-2.46 , 95% CI: -8.88-3.96, P=0.45) . Another two literatures (110 patients) compared Constant scores and complications between conservative treatment and locking plate fixation of Neer part-3 and part-4 proximal humeral fractures. The meta-analysis showed no significant difference in the improvement of functional scores and complications between them. One of the studies showed that for Neer part-3 fractures, locking plate fixation may improve the overall functional score of the patient.

The functional score was higher for conservative treatment of Neer part-4 fractures in the short term. The humeral head replacement can improve long-term pain, but it cannot improve the overall functional score. Locking plate fixation may improve overall functional scores in patients with Neer part-3 fractures.