Evaluation of open reduction and internal fixation of proximal humerus fractures using a locking plate with smooth pegs – A retrospective study

Abstract

Background

Optimal treatment of displaced proximal humeral fractures is controversial. This retrospective study aims to identify complications and clinical outcomes using a locking plate with smooth pegs instead of screws (S3 plate).

Method

Eighty-two patients with displaced proximal humeral fracture classified with 2–4 fragments (Neer’s classification) treated with open reduction and internal fixation (ORIF) with S3 plate were studied retrospectively. Clinical outcome according to constant score; Single Shoulder Value; Disabilities of Arm, Shoulder and Hand; and European Quality of life-5 dimensions and complication rate defined radiologically including peg penetration, avascular necrosis, and loss of reduction was assessed minimum 2.5 years after surgery.

Results

A total of 11 peg penetrations were identified (13.6%). Avascular necrosis was seen in 8.5% (n = 7). Mean constant score at follow-up was 64.4 with a relative constant score of 87% (standard deviation 18%) compared to the contralateral uninjured side. The mean Disabilities of Arm, Shoulder and Hand score was 12.7 and mean European Quality of life-5 dimensions score 0.83. The mean Single Shoulder Value was 78.3. No cases of deep infection were seen.

Conclusions

Fixation with S3 plate shows a proper osteosynthesis and the functional outcome is good. Symptomatic peg penetrations are rare and the incidence is lower compared to what has been reported with locked screws.

Introduction

Proximal humeral fractures are the third most commonly observed osteoporotic fracture types in elderly patients.¹ The incidence of proximal humerus fractures has been growing steadily, and it has been suggested that the rate might triple over the next three decades.² The fracture type is most common among patients more than 60 years of age and women are overrepresented.³ The majority of the fractures are treated conservatively.⁴ Within the surgically treated fractures, a variety of treatment options is available, such as intramedullary nails, hemiarthroplasty, and reversed shoulder replacement. Open reduction and internal fixation (ORIF) using locking plates are commonly used, but have in several studies shown a high rate of complications such as screw perforation, avascular necrosis (AVN), loss of reduction, and subacromial impingement.^5–7 The optimal surgical treatment method of displaced proximal humeral fractures is controversial. The replacement of screws with smooth pegs in the locking plate was performed to reduce screw perforation and to provide optimum subchondral support in osteoporotic bone. There has been skepticism to pegs with a concern of their ability to achieve a stable fracture fixation. The present study was conducted to retrospectively evaluate the clinical functional outcome and complication rate among patients treated with open reduction and internal fixation with locking plate with smooth pegs (S3 proximal humerus plate, at present Zimmer Biomet, Warsaw, Indiana, USA), Figure 1.

Figure 1.

(a) Anterior–posterior X-rays of a proximal humerus fracture, (b) postoperative X-ray after open reduction and internal fixation with S3 plate, and (c) four years postoperatively.

Patients and methods

One hundred and thirty-five consecutive cases with patients operated with S3 plate in our Institution due to displaced proximal humeral fracture from 1 January 2011 until 18 February 2013 were identified. The patients were contacted by letter, telephone calls, or SMS and offered to participate in this study.

Protocol for clinical evaluation

Complications were identified by the medical history record, by evaluation of follow-up X-rays, and by a form completed by the patients at the physiotherapist visit.

Patients’ shoulder function was evaluated by a physiotherapist to assess individual constant scores of the injured and the contralateral shoulder. The strength in 90° abduction was measured with a spring balance (Isobex 4.0, Medical Device solutions AG, Bern, Switzerland). Painful test of strength or inability to reach 90° abduction resulted in a strength score of 0 according to recommendations.⁸ In addition to constant score, the following questionnaires were used: Disabilities of Arm, Shoulder and Hand (DASH), Single Shoulder Value (SSV), and European Quality of life-5 dimensions (EQ-5D). SSV was evaluated by an oral question during the physiotherapist visit, and patients were asked to define the function of their operated shoulder on a scale from 1 to 100. EQ-5D was calculated with a British tariff, according to Swedish practice.⁹

Radiographic follow-up

Follow-up radiography with frontal, axial, and subscapular views and clinical evaluation by a physiotherapist was performed minimum 2.5 years after surgery.

The acute X-ray and CT scan when available were reviewed and the fractures were classified according to Neer.¹⁰ Postoperative X-ray and X-ray at follow-up were interpreted by two orthopedic surgeons independently. Definition of AVN was assessed radiographically and considered present when the humeral head had collapsed and had lost its spherical contour (Figure 2). A peg penetration was diagnosed when the tip of the peg was seen above the cortical level of an intact caput humeri on the follow-up X-ray.

Figure 2.

AVN with loss of the spherical contour of the humeral head.

Surgical technique

Patients were positioned in beach chair position under general anesthesia and in cases applicable with regional anesthesia with ultrasound-guided interscalene nerve block. A single-dose antibiotics, Cloxacillin 2 g, was given 30 min before surgery. Tranexamic acid, 10 mg/kg bodyweight was given preoperatively. A standard deltopectoral approach was used. Acceptable reduction was confirmed intraoperatively with fluoroscopy before the plate was applied. After plate fixation, the final result was documented either peroperatively or with conventional X-ray postoperatively.

Data analysis

A database in Microsoft Excel (Microsoft Corporation, Washington, DC) was set up. Statistical analyses were performed by IBM SPSS Statistics (IBM Analytics, New York) and Stata (StataCorp, College Station, Texas).

Statistical calculations

The risk of experiencing at least one complication and the risk of experiencing peg penetration were estimated as the proportion of the number of incident cases and the total number of patients in respective category with an associated exact binominal 95% confidence interval.¹¹

In the analyses where the incidence between different age categories was compared, a risk ratio between patients < = 70 years and above 70 years was calculated. The difference in risk between these two groups was tested with Fisher’s exact test. In addition, a logistic regression was used to analyze the differences between all four age categories, where the age group < = 50 years was used as a reference. The result from this analysis was presented with odds ratio and a p-value from a Wald test.

In the analyses where different fracture types were compared, Fisher’s exact test was used to note any difference between the groups. A logistic regression was used to compare the three groups, where the two-fragment group was used as the reference. The result from this analysis was presented with odds ratio and a p-value from a Wald test.

Results

Patient demographics

Twenty-one of the 135 identified patients were excluded due to death or them having moved from the Stockholm area, thus 114 patients were available for the study. Eighty-two patients participated with a physiotherapist visit and completed evaluation forms and 80 of these performed follow-up X-ray at the time for the study, which was minimum 2.5 years after surgery. Thus, a follow-up rate of 72% (n = 82) was achieved. There were 60 women and 22 men with a mean age of 61.8 (SD 13.9; range 13–85) years. Immediate postoperative X-ray had been performed in all cases except one. One patient did not complete the questionnaires DASH and EQ-5D.

In addition to the 82 patients included in the study, 12 patients completed questionnaires of DASH, EQ-5D, and SSV but did not complete follow-up with X-ray and physiotherapist visit. The results of these 12 patients will be added to the group of 82 patients and reported as a group. With these patients included, the total number of patients participating in the study was 94 (82%). Ninety-three patients in total completed EQ-5D and DASH.

Preoperative planning and treatment details

Beside standard X-ray, CT scan was performed in 23 cases (28%) before final decision of surgery. The mean delay between accident and ORIF was seven (SD 5.6; range 0–26) days. Five different surgeons were involved in the treatment of the patients included in the study. Each operated on an average of 16 (median 17; range 5–23) patients. Osteosutures or sutures through rotator cuff tendon fixed to the plate in order to further stabilize the fracture were performed in 51 (62%) cases. Patients attended assisted training by physiotherapist during a mean period of five (SD 6; range 0–36) months postoperatively.

No open fractures were reported. Thirteen patients (15%) were smokers at the time of surgery. Fifty-five sustained a low-energy trauma defined by falling in the same level. Twenty-seven patients sustained a moderate- or high-energy trauma defined by falling from height (horse or stairs), traffic accident, or accident while skiing or skating. The fractures were classified according to Neer: 39% two-part fractures, 44% three-part fractures, and 17% four-part fractures (Table 1). The most common fracture type was three-part fractures with displacement of the greater tuberosity (n = 36). Twenty-one (26%) of the fractures had a preoperative varus malalignment, 32 (39%) valgus malalignment, and 16 (20%) neutral alignment.

Table 1.

Fracture classification.

	No. of patients
Neer’s classification
Two part	32
Three part	36
Four part	14
Further fracture description
Fracture dislocation	5
Head splitting	1
Impaction fracture	1
Two-part fractures completely displaced	5

Twenty-three patients reported problems with the contralateral arm and/or shoulder. Problems reported were previous fracture or arthrosis of shoulder, elbow or wrist, impingement, congenital paresthesia, and tularemia infection. In 37 patients (45%), the dominant arm was treated.

Clinical results

The mean constant score at the follow-up visit was 64.4 (SD 13.5; range 17.6–81.8). Among patients reporting no problems with the contralateral side (n = 59), a mean relative constant score of 87% (SD 18%; median 93%; range 22–122%) was seen. The patient with constant score 17.6 and a relative constant score of 22% suffered from a four-part dislocation fracture and axillary nerve injury. Patients with at least one complication presented with a mean constant score, on operated shoulder, of 53.8 (SD 16.4; range 17.6–75.8) and patients with no complication 68.5 (SD 9.5; range 44.1–81.8).

The mean DASH score was 12.7 (SD 13.6; median 7.5; range 0–58.3, n = 81).

The mean EQ-5D score was 0.8 (SD 0.2; range 0.2–1). The mean score of health (EQ-5D VAS) presented in EQ-5D was 78.3 (SD 16.7; range 31–100). Mean SSV score was 78.3 (SD 21.5; range 20–100).

If the group of 12 patients participating only with DASH, EQ-5D, and SSV is added to the study group, no statistical difference was seen in the clinical results (n = 93): The mean DASH score was 13.3 (SD 14.78; median 7.9; range 0–64.2). The mean EQ-5D score was 0.8 (SD 0.2; range 0.1–1). The EQ-5D VAS presented was 77.5 (SD 18.0; range 30–100).

SSV was reported by 94 patients in total. The mean SSV score was 76.6 (SD 22.6; range 9–100).

Radiological results

Peg penetrations were identified in 11 cases (13.4%, n = 81). One of the two patients not performing follow-up X-ray had a peg penetration and had already been reoperated with plate removal and is hence included in the 11 cases mentioned above. Primary peg penetration was seen on postoperative X-ray in three cases. Secondary peg penetration was identified in eight patients (9.9%, n = 81). Reoperation with removal or change of pegs or removal of pegs and plate was performed in seven patients. In three cases, a plate removal was performed. Pegs were extracted or changed to shorter peg in four cases of whom one patient had an additional acromioplasty performed.

The highest risk of suffering from a peg penetration was seen in the age group 61–70 years, with a tendency to more than three times higher risk compared to patients 50 years old or younger. A tendency of a lower risk in patients over 70 years of age compared to the age class 61–70 was seen (Table 2). Patients with three fragments fracture showed the highest risk of peg penetrations (16.7%; CI 6.4–32.8%) whereas the patients with fractures of four fragment showed a tendency of slightly lesser risk of peg penetrations (14.3%; CI 6.4–32.8) (Table 3). These differences are not statistically significant.

Table 2.

Peg penetration per age class.

Age	N	N peg penetrations	Risk (%)	Exact 95% CI	Risk ratio (P)	Odds ratio (P)
<=50	13	1	7.7	0.2–36		Ref
51–60	15	1	6.7	0.2–31.9		0.86 (0.92)
61–70	33	7	21.2	9.0–38.9	0.68 (0.72)	3.23 (0.30)
>70	20	2	10.0	1.2–31.7		1.33 (0.82)

N = number of patients at risk.

Table 3.

Peg penetrations per fracture type according to Neer’s classification.

Fracture type	N	N peg penetrations	Risk (%)	Exact 95% CI	Fisher P	Odds ratio (P)
2	31	3	9.7	2.0–25.8		Ref
3	36	6	16.7	6.4–32.8		1.87 (0.41)
4	14	2	14.3	1.8–42.8	0.69	1.56 (0.65)

N = number of patients at risk.

Three (27%) of the peg penetrations were seen in the valgus malaligned fractures and five (45%) were identified in the group of varus malaligned fractures. One of the total number of 11 peg penetrations was seen among the neutral aligned fractures and two in the completely displaced fractures.

One (1.2%) case with primary loss of reduction was identified in a four-part fracture dislocation. No plate breakage or plate pullout were seen.

The AVN was identified in seven cases (8.8%, n = 80). Three patients were operated with removal of the S3 plate, and one of them underwent further surgery with hemiarthroplasty. One patient who presented with AVN on follow-up X-ray at 13 months was revised to reversed shoulder replacement 18 months after primary ORIF.

Associated complications

No infections leading to reoperation were seen. In one case, the wound inspection was described as irritated skin, and patient was prescribed oral antibiotics (Clindamycin). Culture was negative and CRP negative. Due to penetration of two pegs in this patient, reoperation with removal and insertion of shorter pegs was performed two months postoperatively. No signs of infections were reported during surgery. This patient with superficial wound infection reported a constant score relative to contralateral healthy shoulder of 83% and a SSV of 70 at follow-up. No deep infections were seen in any patient.

Axillary nerve injury diagnosed with EMG and ENEG was seen in two patients. One patient suffered from a four-part split head fracture and an injury of the anterior part of the axillary nerve. One patient suffered from a four-part fracture dislocation and plexus injury. In both cases, the axillary nerve function was not adequately documented before surgery at the time of admission. The cause of the nerve injury cannot be defined, and these nerve injuries are therefore included as potential neurologic complications.

The reoperation rate was 19.5% (n = 16).

Four patients presented with pain and stiffness of the shoulder, but no visible cause on X-ray. Reoperation with removal of plate and screws and mobilization during anesthesia were performed. Two of the patients improved after surgery, presenting at follow-up with a relative constant score compared to the contralateral uninjured arm of 87 and 95%, respectively.

The risk of obtaining a complication in age group 61–70 years is 35.3% (95% CI: 19.7–53.5). The incidence of complication is defined as the number of patients suffering from at least one of the complications listed in Table 4. A tendency of a lower risk of obtaining a complication in the age group more than 70 years compared to all other age groups is seen; 15% (95% CI: 3.2–37.9) (Table 5). There was a tendency of higher risk of complications among three- and four-part fractures classified according to Neer (Table 6).

Table 4.

Detailed listing of complications.

Detailed listing of complications	Quantity	Reoperationa
Infection
Deep	0
Superficial	1	0
Impingement or undefined cause of pain leading to reoperation	4	4
Neurologic complication	2	0
Early loss of reduction	1	1
Peg penetration
Primary	3	1
Secondary	8	6
Avascular necrosis	7	4
Hematoma	1	1b
Total number	27	17

^aReoperations were in some cases associated with more than one complication, and indicated here is the cause that most likely led to reoperation. Two patients underwent > 1 reoperations.

^bPostoperative hematoma after reoperation with removal of osteosynthesis due to AVN.

>1 complications are in some cases reported in the same patient.

Table 5.

Incidence of complication per age class.

Age	N	N complications	Risk (%)	Exact 95% CI	Risk ratio (P)	Odds ratio (P)
<=50	13	3	23.1	5.0–53.8		Ref
51–60	15	5	33.3	11.8–61.6		1.67 (0.55)
61–70	34	12	35.3	19.7–53.5	0.47 (0.16)	1.82 (0.42)
>70	20	3	15.0	3.2–37.9		0.59 (0.067)

N = number of patients at risk.

Table 6.

Complications per fracture type according to Neer’s classification.

Fracture type	N	N complications	Risk (%)	Exact 95% CI	Fisher P	Odds ratio (P)
2	31	3	9.7	2.0–25.8		Ref
3	36	6	16.7	6.4–32.8		1.87 (0.41)
4	14	2	14.3	1.8–42.8	0.69	1.56 (0.65)

N = number of patients at risk.

Discussion

The optimal treatment of proximal humerus fractures with ORIF remains controversial. The functional outcome in this study with smooth pegs is comparable with results using conventional locking plates with screws.

The mean constant score at follow-up was 64.4, which is lower than many studies of ORIF by proximal humerus fractures.^5,6,12 A relative constant score compared with the contralateral uninjured shoulder was calculated to be 87%, which is comparable to the results at one-year follow-up in the multicenter study by Brunner et al.⁵ using the PHILOS plate, where a relative constant score in relation to the contralateral uninjured shoulder was 87% (SD 16.6%; median 92%; 5–95% percentile 53–103%). The impact of various parameters on the strength score in constant score measurements, such as torso position and shoulder abduction is discussed in other studies.¹³ Strength measurement represents one-fourth of the total constant score and can hence have a big impact on the total score. In our unit, we apply strict rules on strength, which could explain the lower mean constant score compared to other studies. A relative constant score in relation to age and gender can be calculated but is not done in this study.⁸

The general population in the United States would score 10.1 on the DASH score according to a large general population survey by the AAOS.¹⁴ The mean DASH score in our study was 12.7 which is comparable to the multicenter study on PHILOS plates by Brunner et al.,⁵ which showed a mean DASH of 16 (SD 21; median 5.8; range 0–85) one year postoperatively.

Patients report to be satisfied with the function of the injured shoulder, with a mean SSV score of 78.3 of a maximum score of 100.

The results of the study population including patients participating only with EQ-5D, DASH, and SSV show that these patients do not affect the results in a significant way.

Less penetration using smooth pegs instead of screws was seen. In a multicenter study by Südkamp et al., intraoperative screw perforation of the humeral head was reported in 21 (14%) of 155 patients and a secondary screw perforation rate of 7% operated with locking proximal humerus plate. The overall rate of screw cutout was 21%.⁶ In a study of Zhu et al.¹⁵ screw penetrations when using locking proximal humerus plate and PHILOS plate was seen in five (19%) of 21 cases within three postoperative months. Brunner et al. found in a prospective multicenter analysis primary or secondary screw perforation in 35 (22%) of 158 proximal humeral fractures who underwent ORIF with PHILOS plate. Olerud et al.¹⁶ found a 14% rate of screw penetrations when using the PHILOS plate in two-part fractures. In our study, primary peg penetrations were reported in three (3.7%) of 82 patients and secondary peg penetrations in eight (9.9%) of 81 patients. Total peg penetrations were seen in 11 (13.6%) of 81 patients. In three of the 11 patients, the peg penetrations seen are not interfering with the glenoid. These patients did not undergo reoperation and presented at follow-up with a relative constant score of 84–101% compared to the contralateral uninjured side. These patients present what can be defined as asymptomatic peg penetrations. This leaves the complication rate of symptomatic peg penetrations to eight (9.9%) of 81 patients.

In a retrospective study of fracture fixation using the S3 plate in displaced three- and four-part fractures, a complication rate of 3.7% was reported.¹⁷ The patients were followed up by evaluation form and review of medical records and operation notes. Follow-up X-ray at the time of the study was not performed, although the study patients had undergone routine follow-up X-ray where signs of healing were criteria for discharge. The cutoff of one-year follow-up and absence of X-ray at the time of the study could to some extent explain the lower complication rate compared to our study. The mean age in the study was 54.1 years compared to 61.8 years in our study. The incidence of osteoporosis increases with age, and hence the risk of complication, and the younger study population is a possible reason for the lower complication rate reported. The low complication rate, however, states that the pegs give an adequate and not inferior fracture fixation compared to locked screws. A biomechanical study shows that the biomechanical characteristics of the S3 plate compared with conventional locking plate with threaded screws are superior for bending and performed similarly for torsion.¹⁸ A retrospective study comparing the S3 plate to the proximal humeral locking plate showed a tendency of lower complication rate using the S3 plate, but no significant difference was found between the two groups.¹⁹

Brunner et al. showed an increased risk of complication in the age group more than 70 years; 16.1% risk of experiencing implant-related complications and 32.1% for non-implant-related complications. The risks in the age group 61–70 are presented as 6.5 and 32.6%, respectively.⁵ In our study, the risk of experiencing a complication, both implant related and non-implant related, was 15.0% in the age group more than 70 years compared to 35.3% among patients 61–70 years old. Six patients (30%) in the age group older than 70 years sustained a two-part fracture, 10 patients (50%) a three-part fracture, and four patients (20%) a four-part fracture. Twelve patients had a preoperative valgus malalignment; two patients a varus malalignment; and three patients a neutral alignment, two were two-part fractures completely displaced and one was a fracture dislocation. The treatment algorithm used in our unit recommends reversed shoulder replacement in patients older than 75 years. The majority of varus malaligned fractures and complex three- to four-part fractures were also selected for reversed shoulder replacement. Due to a study of predicting factors of failure, age, local bone mineral density, anatomic reduction, and restoration of the medial cortical support were found to have a significant influence on the failure rate.²⁰ The risk of failure increases with the number of risk factors. We suggest that the patients in the age group > 70 years undergoing ORIF at our unit are better selected regarding risk factors, and hence less likely to sustain a complication.

Conclusions

The S3 plate is effective and gives a stable fixation for proximal humeral fractures. The complication rate tends to be lower than in conventional locking plates with screws. The open reduction and internal fixation of proximal humeral fractures in the oldest population has been associated with high complication rates, but with better selection of patients undergoing ORIF in this age group, the complication rate could be decreased. The results support the use of smooth pegs in the humeral head instead of screws.

However, this study is a retrospective study and there is a need for a randomized controlled trial comparing the S3 plating system with conventional locking plates with screws.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Ethical Review and Patient Consent

Trial registered with Stockholm Ethical Review Board. File number: EPN 2015/1491-31/1. www.epn.se