Surgeon-related Factors in the Surgical Treatment of Proximal Humerus Fractures

Abstract

Introduction:

In this study, we reviewed proximal humeral fractures surgically treated with open reduction and internal fixation (ORIF) to test the null hypothesis that there is no association between fracture characteristics and surgeon characteristics in outcomes after surgical management of displaced proximal humeral fractures with ORIF.

Methods:

A retrospective review of surgically treated proximal humeral fractures was done at a tertiary-level hospital. The data were organized into two categories: fracture characteristics and surgeon characteristics.

Results:

There were 314 proximal humeral fractures treated with ORIF. In total, there were 112 failures (36%). Bivariate Pearson correlations demonstrated that the number of proximal humerus fixation surgeries performed by an individual surgeon was associated with adequate calcar reduction (r = 0.995, P < 0.001) and greater tuberosity reduction (r = 0.994, P < 0.001). Years of experience was positively associated with adequate calcar reduction (r = 0.594, P = 0.012) and greater tuberosity reduction (r = 0.589, P = 0.013). Regression analysis of two surgeons versus the rest of the cohort demonstrated significantly lower failure rates (P = 0.001).

Discussion:

Complex proximal humeral fractures (3-part and 4-part) have better calcar reduction and tuberosity reduction when treated by surgeons with greater volume of proximal humeral fracture surgery. Among surgeons with high volume, there may be individuals who have the unique ability to provide markedly better results.

Proximal humeral fractures are the third most common osteoporotic fragility fracture, affecting approximately 250 per 100,000 people aged older than 65 years. Most proximal humeral fractures can be treated nonsurgically.^1,2 However, there are many studies which demonstrate that nonsurgical treatment of displaced proximal humeral fractures often results in less than satisfactory outcomes.^3,4

Studies have attempted to identify factors predictive of revision surgery rates in the surgical treatment of proximal humeral fractures.⁵ A study of 307 shoulders treated with angular stable osteosynthesis found that displaced fractures, specifically AO type C fractures, resulted in reduced head vascularity, and varus configuration, which all carried a poor prognosis.⁶ AO type C fractures carried the worst prognosis with a revision surgery rate of 38.1%. Age, local bone mineral density, and adequate restoration of medial cortical support have also been found to markedly influence failure rates.^7,8 Schnetzke et al⁹ demonstrated that satisfactory outcome of locked-plate fixation of proximal humeral fractures depends on the quality of reduction. In these studies, patient-related factors, fracture characteristics, and surgery-related factors were analyzed; however, surgeon-related factors were not analyzed.

Surgeon backgrounds vary in education, specialty, case volume, and years of experience. These factors suggest that surgeon characteristics should be considered when assessing the outcomes of those patients with proximal humeral fractures requiring surgery. The purpose of this study was to look at fracture characteristics and surgeon characteristics in the treatment of displaced proximal humeral fractures.

In this study, we reviewed all proximal humeral fractures surgically treated with open reduction and internal fixation (ORIF) to test the null hypothesis that there is no association between fracture characteristics and surgeon characteristics in outcomes after surgical management of displaced proximal humeral fractures with ORIF.

Methods

A retrospective review of all surgically treated proximal humeral fractures from January 1, 2000, to December 31, 2015, was done at a tertiary-level hospital. We queried an institutional database for patients using diagnosis codes associated with proximal humeral fractures and current procedural terminology codes associated with surgical treatment of proximal humeral fractures. We excluded patients aged younger than 18 years, pregnant patients, pathologic fractures, and patients with less than 12 weeks of radiographic follow-up.

We extracted longitudinal medical record and billing data for all patients in the sample across the entire study timeframe. Data included surgical notes associated with the proximal humeral surgery, clinical notes, and surgical history. The data were organized into two categories: fracture characteristics and surgeon characteristics. Fracture characteristics included the parts (2-part, 3-part, and 4-part)¹⁰ and angulation (varus and valgus); surgeon-related characteristics included volume of proximal humeral fracture surgery and years of experience.

Data were gathered in an encrypted, password-protected drive from the medical center's electronic medical record system, and HIPAA compliance was maintained throughout. Human Subjects Institutional Review Board approval was obtained.

Fracture Characteristics

For each patient, two fellowship-trained orthopaedic surgeons reviewed all preoperative, intraoperative, and postoperative radiographs. The initial injury radiographs were used to describe the fracture pattern. The intraoperative and immediate postoperative images were assessed for alignment, reduction, type of fixation, and quality of fixation.

Failure was defined as a gross loss of alignment from initial fixation, hardware failure, glenohumeral dislocation, osteonecrosis, and superior migration of the humeral head.¹¹ Since most of the alignment loss occurs within 12 weeks of fixation,¹² all included cases had at least 12 weeks of radiographic follow-up. For all discrepancies between the two surgeons, those cases were reviewed and a consensus decision was made.

Surgeon Characteristics

For each surgeon that performed surgical treatment of proximal humeral fractures during this study period, the volume of proximal humeral fracture surgery and years of experience were collected.

Statistical Analysis

Data were analyzed using IBM SPSS version 24 statistical software. Normality of distribution was calculated for continuous variables using Kolmogorov-Smirnov and Shapiro-Wilk tests and by examining skewness and kurtosis. Analyses were conducted using bivariate Pearson correlations and multiple regression analyses. Results were considered significant at P < 0.05, with confidence intervals set at 95%.

Results

During the study period, there were 545 patients treated surgically for proximal humeral fractures, performed by 50 surgeons. Eighteen of these surgeons were attending surgeons, and 32 surgeons were fellows with attending privileges. On average, the surgeons performed 1.9 proximal humeral fracture surgeries annually and had 4.4 years of experience (Table 1: surgeon).

Table 1.

Surgeon Volume and Experience

	PHF (cases/yr)	Experience (yr)
Mean	1.9	4.4
Median	1.0	1.0
Mode	1.0	1.0
SD	1.6	5.7
Minimum	0.2	1.0
Maximum	8.3	16.0

PHF = proximal humeral fracture surgery

Of this cohort, there were 314 proximal humeral fractures treated with ORIF. Fifty-one patients (16%) had fractures displaced in varus and 98 patients (31%) in valgus. Eighty-eight (28%) 2-part fractures, 118 (38%) 3-part fractures, and 106 (34%) 4-part fractures were observed. Twenty-eight patients (9%) with fracture dislocations and 8 patients (3%) with head split fractures were observed (Table 2: descriptive statistics). An adequate reduction was considered less than 2 mm of displacement.

Table 2.

Fracture Characteristics

Anatomical neck fracture	7 (2%)
Surgical neck fracture	310 (99%)
Lesser tuberosity fracture	110 (35%)
Greater tuberosity fracture	81 (26%)
2-part	88 (28%)
3-part	118 (38%)
4-part	106 (34%)
Valgus fracture pattern	98 (31%)
Varus fracture pattern	51 (16%)

Twenty-three patients underwent revision surgery (7%). Seventy-nine (25%) has a loss of alignment and 48 patients developed osteonecrosis (15%). In total, there were 112 failures (36%).

All Surgeons

Two hundred seventy-nine patients (89%) were treated by attending physicians, and 35 patients (11%) were treated by fellows with attending privileges. When looking at all 50 surgeons (18 attendings and 32 fellows), regression analysis demonstrated that failure of fixation was inversely correlated with the number of proximal humerus fixation surgeries performed by an individual surgeon in our cohort (P = 0.015). Surgeons who perform low volumes of proximal humeral fracture treatment with ORIF have higher failure rates in 3-part fractures and 4-part fractures (Table 3: regression).

Table 3.

Predictors of Failure and Fracture Reduction

Fracture Type	Dependent Variable (Outcome)	Independent Variable (Predictor)	B	Sig (P < 0.05)	CI (95%)
All	Failure	PHF	0.005	0.115	0.999-1.012
		EXP	−0.023	0.352	0.930-1.026
	Calcar reduction	PHF	−0.001	0.721	0.993-1.005
		EXP	−0.109	0.373	0.974-1.073
	Tuberosity reduction	PHF	−0.005	0.131	0.988-1.002
		EXP	0.025	0.351	0.973-1.079
2-part	Failure	PHF	−0.006	0.405	0.982-1.007
		EXP	−0.005	0.917	0.901-1.098
	Calcar reduction	PHF	0.004	0.572	0.991-1.017
		EXP	−0.013	0.805	0.893-1.092
	Tuberosity reduction	PHF	−0.004	0.609	0.981-1.011
		EXP	0.090	0.216	0.949-1.260
3&4 part	Failure	PHF	0.009	0.015	1.002-1.017
		EXP	−0.033	0.253	0.913-1.024
	Calcar reduction	PHF	−0.003	0.370	0.989-1.004
		EXP	0.032	0.260	0.976-1.093
	Tuberosity reduction	PHF	−0.004	0.406	0.987-1.005
		EXP	0.018	0.623	0.948-1.094
Varus	Failure	PHF	−0.011	0.252	0.969-1.008
		EXP	0.084	0.279	0.934-1.268
	Calcar reduction	PHF	0.000	0.992	0.981-1.020
		EXP	0.030	0.703	0.882-1.204
	Tuberosity reduction	PHF	−0.014	0.178	0.966-1.007
		EXP	0.100	0.254	0.931-1.312
Valgus	Failure	PHF	0.007	0.189	0.996-1.018
		EXP	−0.025	0.590	0.890-1.069
	Calcar reduction	PHF	−0.002	0.686	0.987-1.009
		EXP	0.050	0.286	0.959-1.154
	Tuberosity reduction	PHF	−0.003	0.669	0.985-1.010
		EXP	0.064	0.228	0.961-1.183

EXP = years of experience, PHF = proximal humeral fracture volume

Attendings

Among the 18 attending surgeons, the failure rate was 35% (97/279). Bivariate Pearson correlations of the 18 attending surgeons demonstrated that the number of proximal humerus fixation surgeries performed by an individual surgeon was associated with adequate calcar reduction (r = 0.995, P < 0.001) and adequate greater tuberosity reduction (r = 0.994, P < 0.001). Years of experience was positively associated with adequate calcar reduction (r = 0.594, P = 0.012) and adequate greater tuberosity reduction (r = 0.589, P = 0.013). Higher volume of proximal humeral fracture treatment with ORIF was associated with higher failure rates (r = 0.932, P ≤ 0.001).

High-volume Surgeons

Seven surgeons performed 81% (253/314) of all ORIF of proximal humeral fractures. Among these seven surgeons, the failure rate was 34% (range: 16.3% to 55.6%) (Table 4: high-volume). Two particular surgeons demonstrated lower failure rates (16.3% and 29.2%), and regression analysis of these two surgeons versus the rest of the cohort demonstrated significantly lower failure rates (P = 0.001) (Table 5: high-volume surgeon—regression).

Table 4.

High-volume Surgeons' Failure Rates

Surgeon#	Volume (n)	Failures	Failure Rate
7	22	10	45%
8	30	9	30%
24	49	8	16%
32	84	35	42%
35	35	13	37%
47	24	7	29%
48	9	5	56%
Total	253	87	34%

Table 5.

Exceptional Surgeons (#24 and #47) Failure Rate Versus Cohort

Surgeon#	24	47	24 & 47	All other surgeons
Volume (n)	49	24	73	180
Failures	8	7	15	72
Failure rate	16%	29%	21%	40%
P-value	0.003	0.110	0.001

Discussion

Surgical treatment of displaced proximal humeral fractures yields better patient-reported outcomes; however, up to 30% of proximal humeral fractures treated surgically will undergo additional surgery, suggesting that proximal humerus ORIF is highly technical and dependent on experience.^{13,14,15,16,17} Previous studies have demonstrated that factors associated with loss of alignment after proximal humeral ORIF include patient age, bone quality, varus angulation, lack of medial cortical support, diabetes mellitus, and the length of metaphyseal head extension.^{12,13,18,19,20,21,22,23} However, there is a paucity of literature looking at surgeon-related factors in the treatment of proximal humeral fractures.

Surgical volume has been found to be a surgeon-related factor that affects patient outcomes. In the total joint literature, there has been a direct relationship between surgeon volume and hospital volume in patient outcomes after a total hip and total knee arthroplasty.^24,25 Surgeons with greater surgical volume have demonstrated decreased surgical times, resulting in decreased infection rates and blood loss. The literature in shoulder arthroplasty has echoed the same findings. As expected, a direct correlation exists between surgical volume and patient outcomes in shoulder surgery—higher-volume surgeons have lower complication rates.^26,27 Low-volume surgeons had increased surgical complications, length of stay, surgical time and surgical cost.²⁸

Helfen et al²⁹ looked at a surgeon's shoulder surgery volume and its relationship with 2-part proximal humeral fracture surgery outcomes. This study found that senior shoulder surgeons who performed more than 50 shoulder surgeries a year had lower rates of revision surgery when treating 2-part proximal humeral fractures.

In our study, surgeons who perform low volumes of proximal humeral fracture treatment with ORIF have higher failure rates in 3-part fractures and 4-part fractures. Among attending surgeons, those who perform higher volumes of proximal humeral fracture treatment with ORIF have higher rates of adequate calcar and tuberosity reduction. Calcar and greater tuberosity reduction has been shown to affect surgical outcomes.^18,20,30 However, despite demonstrating improved rates of calcar reduction and greater tuberosity reduction, attending surgeons had a markedly higher failure rate. It is possible that simpler fracture patterns were likely treated by surgeons of less experience (fellows with attending privileges) and fractures of greater complexity were likely referred to the attending surgeons of greater volume.

The results of our study demonstrate that surgical volume is surgeon-related factors, which leads to better reduction of the calcar and greater tuberosity in complex proximal humeral fractures (3-part and 4-part fractures); however, volume is not the only surgeon-related factor. Overall, failure rates were high in all categories of surgeons, which is very likely a reflection of the difficulty of the procedure and need for alternative methods of treatment.

Among the high-volume surgeons, two particular surgeons demonstrated markedly lower failure rates, but they did not have the greatest volume. The findings of this study could not identify a surgeon-related factor that improved the outcomes with ORIF for these two particular surgeons. It is possible that these surgeons are more familiar with arthroplasty and therefore opted for reverse total shoulder arthroplasty or hemiarthroplasty more frequently, while other surgeons opted to perform ORIF for more complex fractures. It is possible that failure rates are higher with ORIF with certain surgeons because either (1) they are not as familiar with arthroplasty or (2) they philosophically opt for ORIF more readily. Since there are no identifiable surgeon-related factor related to this improved patient outcome, after patient outcomes can assist in identify surgeons who can best treat proximal humeral fractures.

There are some limitations to this study. It is a retrospective review of radiographs and charts; therefore, the clinical results are assumed from the radiographs. In addition, this study did not look specifically into patient-related factors, which are related to surgical outcomes, such as osteoporosis and age. The study was conducted at a level 1 trauma center and serves as a referral center; therefore, many of the fractures who presented were more complex than that might be treated at a community hospital. As such, these complex fracture patterns will have higher failure rates. One of the strengths of this study was the long study span (15 years); however, a limitation of the long study span is the change in indications for surgery, evolving literature on the management of these fractures and technological advances in fixation implants and techniques.

Conclusion

Complex proximal humeral fractures have better calcar reduction and tuberosity reduction when treated by surgeons with greater volume of proximal humeral fracture surgery. Among surgeons with high volume, there may be individuals who have the unique ability to markedly provide better results; these individuals should be identified by after patient outcomes.