A cross-sectional survey comparing management of proximal humerus fractures across a multinational upper limb specialist surgeon panel

Abstract

Background

Proximal humerus fractures (PHFs) are the most common traumatic bony injury to the shoulder. There is lack of consensus in the optimal method of management for fractures with varying morphologies/patterns. This study assesses treatment choices for 3- or 4-part PHFs across surgeons from a multinational panel, focusing on operative versus non-operative strategies and whether inter-surgeon agreement exists within and between countries.

Methods

A prospective cross-sectional survey was completed by a cohort of consultant/attending-level shoulder surgeons from the United Kingdom, United States, India, and Australia. Treatment decisions for 20 individual cases were assessed with standardization of clinical data and imaging. The case range included patients with a breadth of demographics for age (20-80 years) and varying comorbidities including a comprehensive range of fracture morphologies for 3/4-part Neer PHFs with increasing severity, fragmentation, tuberosity displacement, and head split or dislocations. Statistical analysis was performed using a mixed-effects logistic regression model and Krippendorff's alpha. Statistical significance was set at P < .05.

Results

Seventy-five surgeons worldwide responded to the study. There was surgeon representation from Asia (India, 26/75, 34.7%), Europe (UK, 24/75, 32.0%), North America (US, 18/75, 24.0%), and Oceania (Australia, 7/75, 9.3%). Thirty-nine (39/75, 52%) surgeons had over 10 years of experience in treating PHFs. International agreement was low between surgeons (K-A 0.238) regarding overall decision-making. After adjustment for surgeon country and patient demographics, patient age (<65 vs. ≥ 65 years) was not independently associated with the likelihood of recommending operative management (estimate 0.34, P = .67). There was a wide variation among surgeons, with 51.7% of UK-based, 72.3% of Indian, 57.8% of American, and 69.3% of Australian surgeons preferring operative management across the 20 cases. Likelihood for operative decision-making was more pronounced in Australia (P = .024) and India (P < .001), predominantly private healthcare systems, compared to the United Kingdom. There was no significant difference between the United States and the United Kingdom (P = .89). The commonest operative modality in all countries was open reduction and internal fixation. In all countries, reverse total shoulder arthroplasty was preferred to hemiarthroplasty. Intra-medullary nailing was the least recommended modality in the United Kingdom, while hemiarthroplasty was the least recommended in all others.

Conclusion

This study provides an insight into the international variation among surgeons where healthcare characteristics and patient factors contribute to the decision-making processes involved in the management of PHFs. Further research is required to develop standardized, evidence-based guidelines on the management of PHFs.

Proximal humerus fractures (PHFs) account for 5-6% of all fractures.² Fracture incidence exhibits a bimodal age distribution with the highest incidence in patients above the age of 65 years following low-energy injury mechanisms, while PHFs in the younger population usually result from high-energy trauma.⁵

The Neer classification system stratifies PHFs according to fracture anatomy. The proximal humerus is divided into “parts,” which broadly reflects fracture complexity.⁷ Generally, undisplaced 1- or 2-part fractures can be managed non-operatively in a collar and cuff; however, no consensus exists for the management of multi-fragmentary or “complex” fractures. Hertel et al delineated characteristics of complex fractures which predict intraoperative ischemia of the humeral head and may influence operative decision-making.⁴ With respect to displaced humeral neck fractures, the PROPHER trial⁹ found no significant difference between the clinical effectiveness of non-operative and operative management; however, data from other Level 1 studies have contested these findings.¹⁰

Contrasting ideologies have been reflected in studies assessing surgeon opinion on management of PHFs with no consensus on 3- or 4-part fractures with regard to conservative management, operative fixation, or arthroplasty.¹⁰ Patient demographics and surgical fitness are important tools in formulating management in trauma and orthopedic surgery. While some studies have assessed the role of demographic information in influencing decision-making between trauma and arthroplasty surgeons1, 2, 3, 4, 5, 6, 7, 8; there is no literature assessing the management of 3- or 4-part fractures across an international panel of surgeons in a variety of healthcare settings.

This study aims to assess treatment choices for 3- and 4-part PHFs across surgeons from a multinational panel, focusing on operative versus non-operative strategies, preferred surgical procedures, and if inter-surgeon agreement exists within a particular country.

Materials and methods

A prospective cross-sectional survey was completed by consultant upper limb surgeons in the United Kingdom, United States, India, and Australia using the Jisc online survey platform (Bristol, United Kingdom).

Survey candidates

Cohorts of consultant (attending surgeons) trauma and orthopedic surgeons were chosen from a range of healthcare settings. To ensure generalizability of data and a range of responses, the authors recruited clinicians from government-funded institutes, private hospitals, or institutions with hybrid funding.

The clinicians were contacted through established networks and professional connections between hospital centers and researchers, and invitations were posted on social media (Instagram, X/Twitter, LinkedIn) to widen recruitment to the survey. Survey response rate is difficult to assess, given the posting of the survey link on social media platforms. Once recruited, surgeon demographic data including grade, years of experience, and subspecialty (trauma surgery, upper limb subspecialist, or arthroplasty surgery) were collected.

Identifying appropriate cases

Twenty individual cases of PHFs with varying Neer/AO foundation/orthopaedic trauma association classifications and complexity were identified by the authors (A.P., A.S., M.B., H.S.) at a large trauma unit in the United Kingdom. The inclusion criteria for cases were the following:

Age between 20 and 80 years old.

Three- or 4-part PHFs according to Neer's classification and/or fracture-dislocations.

All cases could be managed with conservative and operative management (both fixation and arthroplasty).

Pathologic fractures, deemed “simple” (1- or 2-part fractures), open fractures, or fractures with associated vascular injury were excluded.

The full survey (without clinical images) can be seen in Appendix 1; the clinical images for cases can be seen in Appendix 2.

Data pertaining to patient demographics including age, gender, comorbidities, and occupational status were collected from electronic records. Radiographs and cross-sectional imaging, including pertinent computed tomography image slices and 3-dimensional reconstructions, were obtained from the Picture Archiving and Communication System and included for review. A consultant anesthetist reviewed each of the 20 cases and provided an American College of Anaesthetics, physical status classification system score. The cases were anonymized and presented in a standardized manner. The key features of the 20 cases are summarized in Table I. An initial pilot was released to a focus group for feedback to ensure peer-reviewed standardization of the survey questions and imaging (Appendix 1).

Table I.

Summary of the key features of the 20 included cases.

Case	Age	ASA	Neer classification	Dominant arm affected?	Pertinent medical and social background
1	71	3	4-part	Yes	Diabetes mellitus, osteoarthritis, hypercholesterolemia, Abbreviated Mental Test Score 5/10
2	65	2	3-part	No	Cervical spondylosis, alcohol abuse.
3	70	3	3-part	No	Alcohol excess, hypercholesterolemia
4	48	2	4-part	Yes	Anxiety depressive disorder, fibromyalgia
5	29	1	3-part	No	Attention deficit hyperactivity disorder
6	59	2	3-part	No	Parkinson's disease
7	84	3	3-part	Yes	Previous lymphoma with radiotherapy to the neck, premorbid poor shoulder function. Main carer for disabled son.
8	61	3	3-part	No	Uncontrolled type 1 diabetes mellitus.
9	74	4	3-part	Yes	Atrial fibrillation (on anticoagulation), congestive cardial failure, stroke with visual and cognitive deficit (weakness on injured side), postural hypotension, absence seizures.
10	71	3	3-part	Yes	Ischemia heart disease, myocardial infarction, open abdominal aortic aneurysm repair. Current smoker.
11	46	2	3-part	Yes	Hypertension, early menopause.
12	62	3	3-part	No	Type 2 diabetes mellitus, hypertension, hypothyroidism, vertigo, BMI 40.
13	63	2	4-part	Yes	Recurrent falls secondary to epilepsy.
14	69	1	3-part	Yes	Fit and well. Keen tennis player.
15	63	2	3-part	No	Hypertension, high BMI, type 2 diabetes mellitus
16	60	2	3-part	Yes	Asthma
17	65	2	3-part	Yes	Hypothyroidism, gastroesophageal reflux, Parkinson's disease.
18	68	1	3-part	Yes	Fit and well.
19	44	2	3-part	Yes	BMI 35.
20	60	3	3-part	No	Type 2 diabetes mellitus, hypertension, recent stroke (1 week prior to fracture)

ASA, American College of Anaesthetics, physical status classification system; BMI, body mass index.

Clinicians were initially presented with imaging and the age, in years, of the patient in each case. Following case review, clinicians were invited to offer a treatment option which included: non-operative management, fixation (open reduction and internal fixation [ORIF] or intramedullary fixation), or shoulder arthroplasty. Shoulder arthroplasty was further divided into hemiarthroplasty, anatomical total shoulder arthroplasty, or reverse total shoulder arthroplasty. Furthermore, clinicians were provided with comorbidities and hand dominance for each case and asked to select a treatment option in view of this.

Statistical analysis

Demographic data were reported descriptively. Statistical analysis to assess inter-surgeon agreement on treatment was performed using Krippendorff's alpha. Treatment decisions were compared both nationally and internationally. Values < 0.6 were significant for poor agreement between observers and therefore unreliable in drawing conclusions.

Responses for the management of each case were distinguished as operative (encompassing ORIF and arthroplasty) or non-operative. For the purposes of statistical analysis, patient age was dichotomized a priori into 2 clinically relevant categories (<65 and ≥ 65 years).

To evaluate whether operative decision-making differed according to patient age and surgeon country, a mixed-effects logistic regression model was used with operative versus non-operative management as the binary outcome. Patient age group and surgeon country were included as fixed effects. Random intercepts were specified for both surgeon and case to account for repeated assessments by individual surgeons and multiple ratings per case, thereby adjusting for between-surgeon variability in treatment preference and unmeasured case-specific characteristics such as fracture complexity. Other factors such as hospital system (private/public), years of experience, and training background have not been examined as part of this work.

Model estimates are presented as log-odds with corresponding OR where appropriate. The independent effect of patient age was assessed after adjustment for surgeon and case effects. Differences in operative decision-making between countries were evaluated relative to the United Kingdom, which was specified as the reference category. To assess whether the influence of age differed between countries, an additional model including an age-by-country interaction term was fitted and compared with the main effects model using a likelihood ratio test. Statistical significance was defined as P < .05. All analyses were performed using R (R Foundation for Statistical Computing, Vienna, Austria) with the lme4 package.

Results

Responder demographics

Seventy-five surgeons responded to the international survey. There was surgeon representation from India, 26/75, 34.7%; the United Kingdom, 24/75, 32.0%; the United States, 18/75, 24.0%; and Australia, 7/75, 9.3%. Thirty-nine (39/75, 52%) surgeons had over 10 years of experience in treating PHFs. Thirty-eight (38/75, 50.7%) surgeons predominantly performed arthroplasty in their shoulder subspecialty practice. Of 24, 23 (95.8%) of UK-based surgeons were predominantly based in government-funded institutes (National Health Service Trusts), while 82.4% (42/51) of non-UK-based surgeons practiced predominantly in the private sectors. The surgeon/responder demographic summary is reported in Table II.

Table II.

Summary demographics of all surgeons—UK, India, US, and Australian.

Years experience		Practice type			Sector		Setting
<10 yr	>10 yr	Arthroplasty	Trauma	Mixed	Public	Private	Town	Large city	Unknown
48% (36/75)	52% (39/75)	50.7% (38/75)	21.3% (16/75)	28% (21/75)	42.7% (32/75)	57.3% (43/75)	16% (12/75)	52% (39/75)	59% (24/75)

After review of patient demographics and imaging, 51.7% of UK-based, 72.3% of Indian, 57.8% of American, and 69.3% of Australian surgeons opted for operative management across the 20 cases. The commonest operative modality in all countries was ORIF. In all countries, reverse total shoulder arthroplasty was preferred to hemiarthroplasty. Intra-medullary nailing was the least recommended modality in the United Kingdom (1.7%, 8/480 outcome decisions), while hemiarthroplasty was the least recommended in all others. International agreement was low between surgeons (K-A 0.238) regarding overall decision-making. The breakdown of demographics for individual countries (the United Kingdom, India, the United States, and Australia) can be found in Table III.

Table III.

Summary demographics of UK, Indian, US, and Australian surgeons.

Country	Years experience		Practice type			Sector		Setting
	<10 yr	>10 yr	Arthroplasty	Trauma	Mixed	Public	Private	Town	Large city
United Kingdom	45.8% (11/24)	54.2% (13/24)	37.5% (9/24)	-	62.5% (15/24)	95.8% (23/24)	4.2% (1/24)	-	-
India	23.1% (6/26)	76.9% (20/26)	42.3% (11/26)	42.3% (11/26)	15.4% (4/26)	15.4% (4/26)	84.6% (22/26)	30.8% (8/26)	69.2% (18/26)
United States	22.2% (4/18)	77.8% (14/18)	83.3% (15/18)	5.6% (1/18)	11.1% (2/18)	0% (0/18)	100% (18/18)	16.7% (3/18)	83.3% (15/18)
Australia	28.6% (2/7)	71.4% (5/7)	42.9% (3/7)	57.1% (4/7)	-	71.4% (5/7)	28.6% (2/7)	14.3% (1/7)	85.7% (6/7)

Effect of comorbidities, age, and country preference on management

In the mixed-effects logistic regression model including patient age group and surgeon country as fixed effects and random intercepts for surgeon and case, considerable heterogeneity was observed between cases (standard deviation, 1.74) and surgeons (standard deviation, 1.21), with greater variability attributable to demographic factors. After adjustment for surgeon and case effects, patient age category (<65 vs. ≥ 65 years) was not independently associated with the likelihood of recommending operative management (estimate 0.34, P = .67).

In contrast, operative decision-making differed significantly by country. Using the United Kingdom as the reference category, surgeons practicing in Australia were significantly more likely to recommend operative management (estimate 1.32, odds ratio [OR] = 3.8, P = .024), as were surgeons practicing in India (estimate 1.54, OR = 4.7, P < .001). No significant difference was observed between surgeons practicing in the United States and those in the United Kingdom (estimate 0.06, P = .89). An additional model including an interaction term between age group and country did not demonstrate a significant improvement in model fit compared with the main effects model (likelihood ratio test P = .30), indicating that the influence of patient age on operative decision-making did not differ significantly between countries.

UK-based surgeons

Twenty-four UK-based surgeons responded to the survey. Of 24, 13 (54.2%) had >10 years of practice in treating PHFs. Of 24, 15 (62.5%) had a mixed practice between trauma and arthroplasty. Most surgeons advised surgical intervention when provided with imaging only (63% versus 37%). ORIF predominated (41.0%), followed by arthroplasty (18.8%) and intramedullary nailing (3.2%) with imaging only. Once patient demographic data were provided, the rate of non-operative management increased from 37.0% to 45.1%. A low level of agreement was noted in the overall decision-making among UK-based surgeons; K-alpha agreement score = 0.291. Figure 1 outlines individual treatment decisions with imaging only and after provision of demographic details for UK-based surgeons.

Figure 1.

Individual treatment decisions with imaging only and after provision of demographic details by UK-based surgeons. ORIF, open reduction and internal fixation; rTSA, reverse total shoulder arthroplasty; IM, intra-medullary.

India-based surgeons

Seventy-seven percent (20/26) of surgeons had a minimum of 10 years' experience in managing PHFs. Of 26, 18 (69.2%) of surgeons practiced in large cities and the majority (22/26, 84.6%) worked predominantly in the private sector. Most surgeons opted for surgical management when provided imaging only (81.7% versus 18.3%), particularly ORIF (57%) followed by arthroplasty (13.7%) and intramedullary nailing (11%). This rate of recommended operative management reduced once demographic data were made available (72.4% versus 81.7%). A low level of agreement was noted between India-based surgeons; K-alpha agreement score = 0.235. Figure 2 outlines individual treatment decisions with imaging only and after provision of demographic details for India-based surgeons.

Figure 2.

Individual treatment decisions with imaging only and after provision of demographic details by India-based surgeons. ORIF, open reduction and internal fixation; rTSA, reverse total shoulder arthroplasty; IM, intra-medullary.

US-based surgeons

Eighteen US-based surgeons responded to the survey. Of 18, 4 (22.2%) had greater than 10 years of practice in treating PHF. Of 18, 15 (83.3%) were predominantly arthroplasty surgeons, with only 5.6% (1/18) specializing solely in trauma. Of 18, 15 (83.3%) were based in a large city, and all (100%) surgeons were based in the private sector. Of 18, 2 (11.1%) claimed vested interests in industry. Most US surgeons opted for surgical management with imaging only (59.1%), which reduced to 54.6% once patient demographic data were made available (Fig. 3). ORIF was the most common operative modality (36.2%), followed by arthroplasty (18.1%) and intramedullary nailing (4.7%). The K-alpha agreement between US-based surgeons was K-alpha = 0.282.

Figure 3.

Individual treatment decisions with imaging only and after provision of demographic details by US-based surgeons. ORIF, open reduction and internal fixation; rTSA, reverse total shoulder arthroplasty; IM, intra-medullary.

Australian surgeons

Of 7, 2 (28.6%) had greater than 10 years of practice in treating PHF. Of 7, 4 (57.1%) were predominantly arthroplasty surgeons, with 42.9% (3/7) specializing in trauma only. Of 7, 6 (85.7%) were based in a large city. Of 7, 5 (71.4%) predominantly worked in the public sector. One claimed a vested interest in industry. Most Australian surgeons opted for surgical management with imaging only (74.5%), which reduced to 68.1% once patient demographic data were made available (Fig. 4). Open reduction internal fixation was the most common operative modality (45.4%) followed by arthroplasty (24.1%) and intramedullary nailing (5.0%). K-alpha agreement score = 0.266.

Figure 4.

Individual treatment decisions with imaging only and after provision of demographic details by Australian surgeons. ORIF, open reduction and internal fixation; rTSA, reverse total shoulder arthroplasty.

Discussion

The management of complex PHFs remains a subject of significant debate. Despite advancements in surgical techniques and implant design, there continues to be no consensus on the optimal strategy, particularly in elderly or medically complex patients. This ongoing debate reflects the multifactorial nature of decision-making in PHF management, where patient characteristics, patient preference, surgeon preference, and differing healthcare systems all exert influence on the treatment patients receive.

Operative management, particularly ORIF, was the most employed treatment modality among respondents in this study. This aligns with current trends observed in trauma surgery, where anatomic restoration and early mobilization are prioritized. However, our findings suggest a significant shift toward non-operative management when age, comorbidities, and cognitive impairment are taken into consideration. This is in keeping with the growing awareness of the potential complications associated with surgery in frail individuals and is consistent with current literature supporting conservative management in these cases and the idea that treatment of proximal humeral fractures (more than many other fractures) is dependent on patient factors and not just the characteristics of the fracture.¹¹

However, when controlling for case and surgeon characteristics, age alone did not independently define operative decision-making. This suggests that surgical decision-making may be guided by fracture-specific features such as fracture pattern/degree of comminution rather than age itself. This challenges pre-existing notions that age alone is a significant determinant of treatment choice in PHFs.

Geographic variation in treatment preferences was also remarkable. Respondents from the United Kingdom and United States were significantly more likely to recommend conservative treatment than their colleagues in India and Australia. This divergence may be partially explained by differences in healthcare systems, particularly the government-funded models in the UK, where resource allocation and cost-effectiveness may influence clinical decisions. Variations in patient expectations, surgical training, and even medico-legal considerations may also play into these geographic differences. Additionally, demographic characteristics of the surgeon cohorts could have contributed; UK and US surgeons were generally more experienced with a higher rate of conservative management, while Indian-based surgeons were more likely to favor operative treatment, particularly intramedullary nailing. This raises important questions regarding the role of surgeon skillset, kit availability, and patient-related expectations in influencing operative preferences in different regions. Cultural factors, personal beliefs or inclinations, and training backgrounds likely contribute as well, emphasizing the complexity of global treatment variability.³ Similarly, no significant differences were noted between patient age and country, indicating that age as a factor does not differ significantly between countries. This suggests that international differences in treatment strategy are not driven by age thresholds but by broader practice patterns and system-level factors.

Arthroplasty usage presented another point of interest. While overall rates of arthroplasty were similar across most countries, Australia showed a relatively higher proportion. This appears to reflect the subspecialty distribution of respondents; within the Australian cohort, there were more arthroplasty-focused surgeons, who predictably preferred arthroplasty over certain fixation strategies. However, it should be noted that ORIF still remained the most popular operative choice in this group of surgeons. This observation supports the broader trend that subspecialty training, and familiarity with a specific procedure, influences surgical decision-making.

When analyzing the entire dataset, a consistent pattern emerged: surgeons trained in arthroplasty were more likely to recommend joint replacement as a primary treatment modality, particularly in complex fracture patterns and in older patients. This supports the notion that a surgeon's familiarity with a specific procedure often guides clinical decisions, potentially introducing bias into treatment algorithms.^1,6

Key strengths of this paper include the usage of standardized cases reviewed by an international panel of upper limb surgeons with the application of a cross-classified mixed-effects model that accounts for repeated surgeon assessments and multiple ratings per case. This approach provides a robust assessment of decision-making behavior while minimizing confounding from surgeon- or case-specific factors.

The limitations of this study must also be acknowledged. The overall sample size was limited, and inter-case agreement among respondents was low, highlighting the subjectivity and variability in clinical decision-making. Additionally, the geographic and subspecialty imbalances within the respondent pool may limit generalizability. Furthermore, whilst cases were based upon real patients, a high number had complex medical histories, which may have disqualified them from consideration of surgical management where a less extensive medical history may not have done. This lack of consensus reinforces the need for clearer guidelines and higher-quality comparative studies to inform best practices. Nonetheless, these findings offer valuable insight into current practice patterns and underscore the need for more standardized, evidence-based guidelines.

Conclusion

This study provides a unique insight into the geographic variance, surgeon characteristics, and patient factors that contribute to the decision-making processes involved in the management of PHFs. There is poor inter-rater agreement between surgeons both within a country and between countries highlighting the complexity of PHF management and the range of acceptable surgical options that exist. However, the absence of an independent age effect, alongside marked international variation, highlights the limitations of using chronological age alone as a decision threshold in clinical guidelines. These findings support a more individualized approach to treatment selection that prioritizes fracture characteristics and patient-specific functional considerations over age-based cutoffs. Further work, which would be of interest, would be to repeat the study again with the same survey respondents and the same clinical cases to assess intra-observer reliability and assess if the differences between countries and healthcare systems remain.

Disclaimers:

Funding: No funding was disclosed by the authors.

Conflicts of interest: The authors, their immediate families, and any research foundations with which they are affiliated have not received any financial payments or other benefits from any commercial entity related to the subject of this article.