A Scoping Review of Biomechanical Testing for Proximal Humerus Fracture Implants

David Cruickshank; Kelly A. Lefaivre; Herman Johal; Norma J. MacIntyre; Sheila A. Sprague; Taryn Scott; Pierre Guy; Peter A. Cripton; Michael McKee; Mohit Bhandari; Gerard P. Slobogean


BMC Musculoskelet Disord. 2015;16(175) 

In This Article


Literature Review

The initial literature search of the PHF literature, which included clinical and basic science studies, resulted in identification of 5,406 titles. 2,540 were found to be duplicates, seven were book titles, and two were retracted studies; these were all excluded. An additional 1,459 titles were removed because they did not meet the eligibility criteria. After review, 1,051 proximal humerus fracture studies were included in our database. From our eligible PHF database, 94 were identified as basic science or biomechanical papers. For the purpose of this study, we excluded an additional 16 non-English language publications, seven basic science articles, three finite element analysis studies, and one review article (Fig. 1). Therefore, 67 proximal humerus biomechanical published studies were included in the current analysis (Additional file 1 http://www.biomedcentral.com/1471-2474/16/175/additional).

Figure 1.

Search and screening flow chart

Study Characteristics

The majority of the included publications originated from Europe (48 %) to North America (39 %), comprising 87 % of the total studies. Few biomechanical studies have been published from Asia (4 %), South/Central America (3 %), to the Middle East (3 %) (Fig. 2). The earliest included study identified dates back to 1988 with nothing published again until 1993. Since that time, however, there has been an exponential increase in biomechanical publications, with 13 studies published in 2012 alone (Fig. 3).

Figure 2.

Location of research

Figure 3.

Frequency of studies published per year

Specimen Characteristics

The sample size was reported in every study and the average sample size was found to be 27 ± 28.9 specimens with a range of 5 to 150 specimens (Table 1). The most commonly tested specimen was cadaver bones (87 %) followed by saw bones (7 %), animal bones (4 %), and wood (1 %). Of the cadaver studies, the obtained cadavers were frozen in 45 studies (75 %), embalmed in 12 (20 %), and fresh in 3 (5 %). Of the 58 studies that used cadaver specimens, only 33 (57 %) included information on the age of the cadavers used. In the studies that reported age of the cadaver, the average age was found to be 73.3 ± 8.5 years, with a range of 32 to 101 years of age. Only 45 studies (67 %) undertook some form of pre-testing investigations including plain radiographs (32 studies), bone mineral density testing (31 studies), and CT scans (12 studies) (Table 2).

Sixty studies tested proximal humerus fracture implants in a specific simulated fracture pattern. The most common fracture simulated was a two-part proximal humerus fracture in 41 studies (68 %), followed by a three-part fracture in 14 studies (23 %), and a four-part fracture in five studies (8 %) (Table 1). Of the two-part fracture simulations, 37 involved the surgical neck, three involved the anatomic neck, and one study described making a two-part fracture model, but the location of the osteotomy was not stated. None of the included studies examined fractures of the greater tuberosity.

The most common method of specimen preparation was to create the fracture using a saw, followed by reduction and fixation with the specified construct. Often, in order to simulate medial comminution, a section of bone would be removed and a gap created. This modification of the specimen ensured that reduction and alignment was maintained solely by the implant in the absence of a medial cortical support. This was performed in 35 studies. When compared to fracture type, medial comminution was simulated in 27 (66 %) of the two-part fracture studies, in seven (50 %) of the three-part fracture studies and in none of the four-part fracture studies. One study specified that a gap osteotomy was performed but did not specify the location or the fracture pattern.

Implants Evaluated

The most frequently tested implant was a fixed angle locking plate, which was tested in 35 studies (Fig. 4). Intramedullary devices, including intramedullary nails, were tested in 17 studies, followed by non-locking plates in 13 studies, and blade plates in eight studies. Interestingly, arthroplasty implants were only tested in five studies and only included hemi-arthroplasty implants. We did not identify any studies that focused on biomechanical testing of total shoulder implants or reverse total shoulder implants for the treatment of proximal humerus fractures. An overview of the implants evaluated in each study is found in Additional file 2 http://www.biomedcentral.com/1471-2474/16/175/additional.

Figure 4.

Frequency of implant testing

Construct Testing

The apparatus and testing procedure of the constructs was found to be highly heterogeneous between studies and many different testing platforms, configurations, and devices were described in the included studies. Despite this heterogeneity, the majority of the studies tested their constructs under similar biomechanical themes, which has allowed us to summarize them. Specifically, the most commonly tested force was torque (34 studies), followed by axial load (30 studies), and cantilever bending, usually in varus or valgus (23 studies) (Table 2).

The testing parameters including the magnitude of the force (20 studies), how the force was applied (64 studies), and the loading mode (54 studies); all testing parameters varied significantly between studies (Table 2). Cyclic loading was utilized in 33 studies, load to failure was used in 28 studies, and compounding cyclic load to failure was used in six studies. In the studies that used cyclic loading to test the construct, the number of cycles varied from 5 to 1,000,000; the most commonly used number of cycles was 1000 (seven studies). Many studies (34 studies) used a combination of testing modes, for example a construct would be put through cyclic loading to a set number of cycles and then undergo a load to failure test.

Supplementary fixation methods were infrequently evaluated in the biomechanics literature. Sutures were used to augment fixation in three studies; two of the studies tested hemi-arthroplasty implants and one study tested locking plates. Tension band wiring, either on its own or as an augment to a construct, was used in four studies. Bone grafting with structural grafts was tested in three studies and two studies examined the use of cement as an augment.