Original Research

Superior labral anteroposterior (SLAP) tears are common labral injuries. They occur at the attachment of the long head of the biceps tendon on the superior glenoid and extend anterior and/or posterior to the biceps anchor. The mechanism of action for SLAP tears is traction on the superior labrum by the long head of the biceps tendon, resulting in “peeling” of the labrum off the glenoid. Such forces may result from repetitive overhead arm motion (pitching) or from direct trauma.

Clinical diagnosis is challenging with SLAP tears, as they often present with nonspecific shoulder pain and may not be associated with an acute injury. A further complication is that they are often associated with other shoulder pathology, such as rotator cuff tears.¹ As physical examination is typically nonspecific, proper diagnostic imaging is essential for diagnosis.

We prefer to assess potential SLAP tears with magnetic resonance arthrography (MRA).² Dilute (1:200) gadolinium contrast material (12-15 mL) is introduced into the glenohumeral joint under sonographic or fluoroscopic guidance. Capsular distention by dilute intra-articular contrast enables superior imaging resolution of the labroligamentous complex. We think the increase in diagnostic confidence enabled by direct arthrography outweighs the additional invasiveness and cost associated with MRA relative to noncontrast magnetic resonance imaging (MRI).

The MRA protocol differs from our routine noncontrast shoulder imaging. We perform a fat-saturated coronal oblique T1 sequence that maximizes the conspicuity of intra-articular contrast in the plane that optimally visualizes the superior labrum. Three planes of intermediate-weighted fast spin echo not only contrast the high-signal intra-articular fluid with the low-signal fibrocartilaginous labrum and the stratified intermediate signal of glenoid articular cartilage, but they also allow optimal assessment of the rotator cuff. In addition, we perform a fat-saturated coronal T2 sequence that highlights all fluid signal structures as well as edema.

SLAP tears appear on MRA as the insinuation of intra-articular contrast between the articular cartilage and the attachment of the superior labrum,³ within the substance of the labrum, or as detachment of the labrum from the glenoid rim⁴ (Figure 1). Findings can range from labral fraying to complete detachment with displacement. Tears can extend into other quadrants of the labrum, extend from a Bankart lesion, or involve the biceps tendon and/or the glenohumeral ligaments (Figures 2–4). Up to 10 types of SLAP tears have been described on arthroscopy. This classification scheme, however, is seldom helpful in the interpretation of SLAP tears on MRI. More important in guiding treatment is having a detailed description of the tear, including location, extent, and morphology, along with associated abnormalities.

Several findings can aid in the diagnosis of SLAP tears. Normal anatomical variants of the anterior-superior labrum do not extend posterior to the biceps anchor—an important finding for discerning normal morphologic variants from tears. Therefore, high signal within the posterior third of the superior labrum or extension of high signal laterally within the labrum and away from the glenoid suggests a SLAP tear.⁵ A paralabral cyst is almost always associated with a labral tear,¹ so signal abnormality of the superior labrum with a paralabral cyst suggests a SLAP tear (Figure 5).

MRA is not the only method for diagnosing SLAP tears. Standard 3-Tesla MRI had 83% sensitivity and 99% specificity for diagnosing SLAP tears in a recent study, though MRA had 98% sensitivity and 99% specificity—a statistically significant sensitivity difference.⁶ In another study, computed tomography arthrography (CTA) had 95% sensitivity and 88% specificity for diagnosing recurrent SLAP tears after surgery.⁷ CTA is associated with ionizing radiation and is limited in its assessment of other structures that may show concomitant abnormalities, such as the articular cartilage and the rotator cuff. Indirect MRA, wherein magnetic resonance sequences are obtained after intravenous injection of gadolinium contrast and exercise of the affected shoulder, had a high sensitivity of detection of labral tears of all types.⁸

MRA is most sensitive and specific for diagnosing SLAP tears; 3-Tesla MRI, indirect MRA, and CTA are useful alternative modalities for cases in which MRA cannot be performed.

Superior labral anteroposterior (SLAP) tears are common labral injuries. They occur at the attachment of the long head of the biceps tendon on the superior glenoid and extend anterior and/or posterior to the biceps anchor. The mechanism of action for SLAP tears is traction on the superior labrum by the long head of the biceps tendon, resulting in “peeling” of the labrum off the glenoid. Such forces may result from repetitive overhead arm motion (pitching) or from direct trauma.

Clinical diagnosis is challenging with SLAP tears, as they often present with nonspecific shoulder pain and may not be associated with an acute injury. A further complication is that they are often associated with other shoulder pathology, such as rotator cuff tears.¹ As physical examination is typically nonspecific, proper diagnostic imaging is essential for diagnosis.

We prefer to assess potential SLAP tears with magnetic resonance arthrography (MRA).² Dilute (1:200) gadolinium contrast material (12-15 mL) is introduced into the glenohumeral joint under sonographic or fluoroscopic guidance. Capsular distention by dilute intra-articular contrast enables superior imaging resolution of the labroligamentous complex. We think the increase in diagnostic confidence enabled by direct arthrography outweighs the additional invasiveness and cost associated with MRA relative to noncontrast magnetic resonance imaging (MRI).

The MRA protocol differs from our routine noncontrast shoulder imaging. We perform a fat-saturated coronal oblique T1 sequence that maximizes the conspicuity of intra-articular contrast in the plane that optimally visualizes the superior labrum. Three planes of intermediate-weighted fast spin echo not only contrast the high-signal intra-articular fluid with the low-signal fibrocartilaginous labrum and the stratified intermediate signal of glenoid articular cartilage, but they also allow optimal assessment of the rotator cuff. In addition, we perform a fat-saturated coronal T2 sequence that highlights all fluid signal structures as well as edema.

SLAP tears appear on MRA as the insinuation of intra-articular contrast between the articular cartilage and the attachment of the superior labrum,³ within the substance of the labrum, or as detachment of the labrum from the glenoid rim⁴ (Figure 1). Findings can range from labral fraying to complete detachment with displacement. Tears can extend into other quadrants of the labrum, extend from a Bankart lesion, or involve the biceps tendon and/or the glenohumeral ligaments (Figures 2–4). Up to 10 types of SLAP tears have been described on arthroscopy. This classification scheme, however, is seldom helpful in the interpretation of SLAP tears on MRI. More important in guiding treatment is having a detailed description of the tear, including location, extent, and morphology, along with associated abnormalities.

Several findings can aid in the diagnosis of SLAP tears. Normal anatomical variants of the anterior-superior labrum do not extend posterior to the biceps anchor—an important finding for discerning normal morphologic variants from tears. Therefore, high signal within the posterior third of the superior labrum or extension of high signal laterally within the labrum and away from the glenoid suggests a SLAP tear.⁵ A paralabral cyst is almost always associated with a labral tear,¹ so signal abnormality of the superior labrum with a paralabral cyst suggests a SLAP tear (Figure 5).

MRA is not the only method for diagnosing SLAP tears. Standard 3-Tesla MRI had 83% sensitivity and 99% specificity for diagnosing SLAP tears in a recent study, though MRA had 98% sensitivity and 99% specificity—a statistically significant sensitivity difference.⁶ In another study, computed tomography arthrography (CTA) had 95% sensitivity and 88% specificity for diagnosing recurrent SLAP tears after surgery.⁷ CTA is associated with ionizing radiation and is limited in its assessment of other structures that may show concomitant abnormalities, such as the articular cartilage and the rotator cuff. Indirect MRA, wherein magnetic resonance sequences are obtained after intravenous injection of gadolinium contrast and exercise of the affected shoulder, had a high sensitivity of detection of labral tears of all types.⁸

MRA is most sensitive and specific for diagnosing SLAP tears; 3-Tesla MRI, indirect MRA, and CTA are useful alternative modalities for cases in which MRA cannot be performed.

Superior labral anteroposterior (SLAP) tears are common labral injuries. They occur at the attachment of the long head of the biceps tendon on the superior glenoid and extend anterior and/or posterior to the biceps anchor. The mechanism of action for SLAP tears is traction on the superior labrum by the long head of the biceps tendon, resulting in “peeling” of the labrum off the glenoid. Such forces may result from repetitive overhead arm motion (pitching) or from direct trauma.

Clinical diagnosis is challenging with SLAP tears, as they often present with nonspecific shoulder pain and may not be associated with an acute injury. A further complication is that they are often associated with other shoulder pathology, such as rotator cuff tears.¹ As physical examination is typically nonspecific, proper diagnostic imaging is essential for diagnosis.

We prefer to assess potential SLAP tears with magnetic resonance arthrography (MRA).² Dilute (1:200) gadolinium contrast material (12-15 mL) is introduced into the glenohumeral joint under sonographic or fluoroscopic guidance. Capsular distention by dilute intra-articular contrast enables superior imaging resolution of the labroligamentous complex. We think the increase in diagnostic confidence enabled by direct arthrography outweighs the additional invasiveness and cost associated with MRA relative to noncontrast magnetic resonance imaging (MRI).

The MRA protocol differs from our routine noncontrast shoulder imaging. We perform a fat-saturated coronal oblique T1 sequence that maximizes the conspicuity of intra-articular contrast in the plane that optimally visualizes the superior labrum. Three planes of intermediate-weighted fast spin echo not only contrast the high-signal intra-articular fluid with the low-signal fibrocartilaginous labrum and the stratified intermediate signal of glenoid articular cartilage, but they also allow optimal assessment of the rotator cuff. In addition, we perform a fat-saturated coronal T2 sequence that highlights all fluid signal structures as well as edema.

SLAP tears appear on MRA as the insinuation of intra-articular contrast between the articular cartilage and the attachment of the superior labrum,³ within the substance of the labrum, or as detachment of the labrum from the glenoid rim⁴ (Figure 1). Findings can range from labral fraying to complete detachment with displacement. Tears can extend into other quadrants of the labrum, extend from a Bankart lesion, or involve the biceps tendon and/or the glenohumeral ligaments (Figures 2–4). Up to 10 types of SLAP tears have been described on arthroscopy. This classification scheme, however, is seldom helpful in the interpretation of SLAP tears on MRI. More important in guiding treatment is having a detailed description of the tear, including location, extent, and morphology, along with associated abnormalities.

Several findings can aid in the diagnosis of SLAP tears. Normal anatomical variants of the anterior-superior labrum do not extend posterior to the biceps anchor—an important finding for discerning normal morphologic variants from tears. Therefore, high signal within the posterior third of the superior labrum or extension of high signal laterally within the labrum and away from the glenoid suggests a SLAP tear.⁵ A paralabral cyst is almost always associated with a labral tear,¹ so signal abnormality of the superior labrum with a paralabral cyst suggests a SLAP tear (Figure 5).

MRA is not the only method for diagnosing SLAP tears. Standard 3-Tesla MRI had 83% sensitivity and 99% specificity for diagnosing SLAP tears in a recent study, though MRA had 98% sensitivity and 99% specificity—a statistically significant sensitivity difference.⁶ In another study, computed tomography arthrography (CTA) had 95% sensitivity and 88% specificity for diagnosing recurrent SLAP tears after surgery.⁷ CTA is associated with ionizing radiation and is limited in its assessment of other structures that may show concomitant abnormalities, such as the articular cartilage and the rotator cuff. Indirect MRA, wherein magnetic resonance sequences are obtained after intravenous injection of gadolinium contrast and exercise of the affected shoulder, had a high sensitivity of detection of labral tears of all types.⁸

MRA is most sensitive and specific for diagnosing SLAP tears; 3-Tesla MRI, indirect MRA, and CTA are useful alternative modalities for cases in which MRA cannot be performed.

As total health care costs reach almost $3 trillion per year—capturing more than 17% of the total US gross domestic product—payers are searching for more effective ways to limit health care spending.^1,2 One increasingly discussed plan is payment bundling.³ This one-lump-sum payment model arose as a result of rapid year-on-year increases in total reimbursements under the current, fee-for-service model. The Centers for Medicare & Medicaid Services hypothesized that a single all-inclusive payment for a procedure or set of services would incentivize improvements in patient-centered care and disincentivize cost-shifting behaviors.⁴ Bundled reimbursement is becoming increasingly common in orthopedic practice. With the recent introduction of the Bundled Payment for Care Improvement Initiative, several orthopedic practices around the United States are already actively engaged in creating models for bundled payment for common elective procedures and for associated services provided up to 90 days after surgery.^3,5

Bundled payment increases the burden on the provider to understand the cost of care provided during a care cycle. However, not only has the current system blinded physicians to the cost of care, but current antitrust legislation has made discussions of pricing with colleagues (so-called price collusion) illegal and subject to fines of up to $1 million per person and $100 million per organization,⁶ therefore limiting orthopedic physician involvement.

Given these legal constraints, instead of measuring direct costs of goods, we developed a “grocery list” approach in which direct comparisons are made of resources (goods and services) used and delivered during the entire 90-day cycle of care for patients who undergo anatomical total shoulder arthroplasty (TSA) or reverse shoulder arthroplasty (RSA). We used one surgeon’s practice experience as a model for measuring other orthopedic surgeons’ resource utilization, based on their electronic medical records (EMR) system data. By capturing the costs of the components of resource utilization rather than just the final cost of care, we can assess, compare, understand, endorse, and address these driving factors.

1. The significance of resource utilization

To maximize the efficiency of their practices, high-volume shoulder surgeons have introduced standardization to health care delivery.⁷ Identifying specific efficiencies makes uniform acceptance of beneficial practice patterns possible.

 To facilitate comparison of goods and services used during an episode of surgical care, Virani and colleagues^8,9 studied the costs of TSA and RSA and calculated the top 10 driving cost factors for these procedures (Figure 1). Their systematic analysis provided a framework for a common method of communication, allowing an orthopedic surgeon to gain a more complete understanding of the resources used during a particular operative procedure in his or her practice, and allowing several physicians to compare and contrast the resources collectively used for a single procedure, facilitating an understanding of different practice patterns within a local community. At a societal level, these data can be collected to help guide overall recommendations.

2. How we defined utilization

To define the resources used, we had to decide which procedure components cost the most. Virani and colleagues^8,9 found that the top 10 cost drivers accounted for 93.11% and 94.77% of the total cost of the TSA and RSA care cycles, respectively (Figure 1). For each cost driver, information on resources used (goods, services, overhead) was collected on 2 forms, the Hospital Utilization Form (7 hospital-based items) and the Clinical Utilization Form (3 non-hospital-based items). To make hospital data easy to compile, we piloted use of a “smart form” in the EpicCare EMR system to isolate and auto-populate specific data fields.

3. EMR data collection

With EMR becoming mandatory for all public and private health care providers starting in 2014, utilization data are now included in a single unified system. Working with our in-house information technology department, we developed an algorithm to populate this information in a separate, easy-to-follow hospital utilization form. This form can be adopted by other institutions. Although EpicCare EMR is used by 52% of hospitals and at our institution, the data points required to make the same measurements are generalizable and exist in other EMRs.

Smartlinks, a tool in this EMR, allows utilization data to be quickly retrieved from different locations in a medical record and allows a form to be electronically completed in seconds. Data can be retrieved for any patient in the EMR system, regardless of when that patient’s hospital stay occurred. We populated data from surgeries performed 2 years before the start of this project.

4. What we can learn from these data

Data from a pilot study of 25 patients who underwent primary anatomical TSA for osteoarthritis and 25 patients who underwent primary RSA for massive rotator cuff tear allowed us to generate graphical representations of a single surgeon’s practice patterns that most affected the cost of care. Time in holding, time in the operating room, time in the postanesthesia care unit, and percentage of patients receiving different medications were recorded for each procedure (Figures 2–11). The study did not capture the wide variances in practice patterns in shoulder arthroplasty, and therefore other surgeons’ resource utilization may differ from ours. However, replicating this methodology at other institutions will produce a more robust data set from which conclusions about resource utilization and, indirectly, cost of care can be made.

5. Future possibilities

By using existing EMR tools to better understand resource utilization, orthopedic surgeons can play a constructive role in the dialogue on health care costs and new reimbursement models. The data presented here are not meant to be interpreted as hard and fast numbers on global resource utilization, but instead we intend to establish a model for collecting data on resource utilization. Resource utilization begins the dialogue that allows orthopedic surgeons and specialty societies to speak a common language without discussing actual cost numbers, which is discouraged under antitrust regulation. The data presented will allow comparisons to be made between surgeons in all practice settings to highlight areas of inconsistency in order to further improve patient care. Although this work involved only 50 patients undergoing only 2 types of surgeries, the resource-capturing methodology can be expanded to include more procedures and orthopedic practices. As all hospitals are now required to have EMRs, the metrics tracked in this work can be found on any patient medical record and auto-populated using our open-source utilization forms. Starting this data collection at your hospital may require no more than a conversation with the informatics department, as the metrics can for the most part be populated into a database on surgeon request.

As orthopedic surgeons return to the economic health care discussion, this information could prove essential in helping the individual surgeon and the orthopedic community justify the cost of care as well as fully understand the cost drivers for musculoskeletal care.

Click here to read the commentary on this article by Peter D. McCann, MD

As total health care costs reach almost $3 trillion per year—capturing more than 17% of the total US gross domestic product—payers are searching for more effective ways to limit health care spending.^1,2 One increasingly discussed plan is payment bundling.³ This one-lump-sum payment model arose as a result of rapid year-on-year increases in total reimbursements under the current, fee-for-service model. The Centers for Medicare & Medicaid Services hypothesized that a single all-inclusive payment for a procedure or set of services would incentivize improvements in patient-centered care and disincentivize cost-shifting behaviors.⁴ Bundled reimbursement is becoming increasingly common in orthopedic practice. With the recent introduction of the Bundled Payment for Care Improvement Initiative, several orthopedic practices around the United States are already actively engaged in creating models for bundled payment for common elective procedures and for associated services provided up to 90 days after surgery.^3,5

Bundled payment increases the burden on the provider to understand the cost of care provided during a care cycle. However, not only has the current system blinded physicians to the cost of care, but current antitrust legislation has made discussions of pricing with colleagues (so-called price collusion) illegal and subject to fines of up to $1 million per person and $100 million per organization,⁶ therefore limiting orthopedic physician involvement.

Given these legal constraints, instead of measuring direct costs of goods, we developed a “grocery list” approach in which direct comparisons are made of resources (goods and services) used and delivered during the entire 90-day cycle of care for patients who undergo anatomical total shoulder arthroplasty (TSA) or reverse shoulder arthroplasty (RSA). We used one surgeon’s practice experience as a model for measuring other orthopedic surgeons’ resource utilization, based on their electronic medical records (EMR) system data. By capturing the costs of the components of resource utilization rather than just the final cost of care, we can assess, compare, understand, endorse, and address these driving factors.

1. The significance of resource utilization

To maximize the efficiency of their practices, high-volume shoulder surgeons have introduced standardization to health care delivery.⁷ Identifying specific efficiencies makes uniform acceptance of beneficial practice patterns possible.

 To facilitate comparison of goods and services used during an episode of surgical care, Virani and colleagues^8,9 studied the costs of TSA and RSA and calculated the top 10 driving cost factors for these procedures (Figure 1). Their systematic analysis provided a framework for a common method of communication, allowing an orthopedic surgeon to gain a more complete understanding of the resources used during a particular operative procedure in his or her practice, and allowing several physicians to compare and contrast the resources collectively used for a single procedure, facilitating an understanding of different practice patterns within a local community. At a societal level, these data can be collected to help guide overall recommendations.

2. How we defined utilization

To define the resources used, we had to decide which procedure components cost the most. Virani and colleagues^8,9 found that the top 10 cost drivers accounted for 93.11% and 94.77% of the total cost of the TSA and RSA care cycles, respectively (Figure 1). For each cost driver, information on resources used (goods, services, overhead) was collected on 2 forms, the Hospital Utilization Form (7 hospital-based items) and the Clinical Utilization Form (3 non-hospital-based items). To make hospital data easy to compile, we piloted use of a “smart form” in the EpicCare EMR system to isolate and auto-populate specific data fields.

3. EMR data collection

With EMR becoming mandatory for all public and private health care providers starting in 2014, utilization data are now included in a single unified system. Working with our in-house information technology department, we developed an algorithm to populate this information in a separate, easy-to-follow hospital utilization form. This form can be adopted by other institutions. Although EpicCare EMR is used by 52% of hospitals and at our institution, the data points required to make the same measurements are generalizable and exist in other EMRs.

Smartlinks, a tool in this EMR, allows utilization data to be quickly retrieved from different locations in a medical record and allows a form to be electronically completed in seconds. Data can be retrieved for any patient in the EMR system, regardless of when that patient’s hospital stay occurred. We populated data from surgeries performed 2 years before the start of this project.

4. What we can learn from these data

Data from a pilot study of 25 patients who underwent primary anatomical TSA for osteoarthritis and 25 patients who underwent primary RSA for massive rotator cuff tear allowed us to generate graphical representations of a single surgeon’s practice patterns that most affected the cost of care. Time in holding, time in the operating room, time in the postanesthesia care unit, and percentage of patients receiving different medications were recorded for each procedure (Figures 2–11). The study did not capture the wide variances in practice patterns in shoulder arthroplasty, and therefore other surgeons’ resource utilization may differ from ours. However, replicating this methodology at other institutions will produce a more robust data set from which conclusions about resource utilization and, indirectly, cost of care can be made.

5. Future possibilities

By using existing EMR tools to better understand resource utilization, orthopedic surgeons can play a constructive role in the dialogue on health care costs and new reimbursement models. The data presented here are not meant to be interpreted as hard and fast numbers on global resource utilization, but instead we intend to establish a model for collecting data on resource utilization. Resource utilization begins the dialogue that allows orthopedic surgeons and specialty societies to speak a common language without discussing actual cost numbers, which is discouraged under antitrust regulation. The data presented will allow comparisons to be made between surgeons in all practice settings to highlight areas of inconsistency in order to further improve patient care. Although this work involved only 50 patients undergoing only 2 types of surgeries, the resource-capturing methodology can be expanded to include more procedures and orthopedic practices. As all hospitals are now required to have EMRs, the metrics tracked in this work can be found on any patient medical record and auto-populated using our open-source utilization forms. Starting this data collection at your hospital may require no more than a conversation with the informatics department, as the metrics can for the most part be populated into a database on surgeon request.

As orthopedic surgeons return to the economic health care discussion, this information could prove essential in helping the individual surgeon and the orthopedic community justify the cost of care as well as fully understand the cost drivers for musculoskeletal care.

Click here to read the commentary on this article by Peter D. McCann, MD

As total health care costs reach almost $3 trillion per year—capturing more than 17% of the total US gross domestic product—payers are searching for more effective ways to limit health care spending.^1,2 One increasingly discussed plan is payment bundling.³ This one-lump-sum payment model arose as a result of rapid year-on-year increases in total reimbursements under the current, fee-for-service model. The Centers for Medicare & Medicaid Services hypothesized that a single all-inclusive payment for a procedure or set of services would incentivize improvements in patient-centered care and disincentivize cost-shifting behaviors.⁴ Bundled reimbursement is becoming increasingly common in orthopedic practice. With the recent introduction of the Bundled Payment for Care Improvement Initiative, several orthopedic practices around the United States are already actively engaged in creating models for bundled payment for common elective procedures and for associated services provided up to 90 days after surgery.^3,5

Bundled payment increases the burden on the provider to understand the cost of care provided during a care cycle. However, not only has the current system blinded physicians to the cost of care, but current antitrust legislation has made discussions of pricing with colleagues (so-called price collusion) illegal and subject to fines of up to $1 million per person and $100 million per organization,⁶ therefore limiting orthopedic physician involvement.

Given these legal constraints, instead of measuring direct costs of goods, we developed a “grocery list” approach in which direct comparisons are made of resources (goods and services) used and delivered during the entire 90-day cycle of care for patients who undergo anatomical total shoulder arthroplasty (TSA) or reverse shoulder arthroplasty (RSA). We used one surgeon’s practice experience as a model for measuring other orthopedic surgeons’ resource utilization, based on their electronic medical records (EMR) system data. By capturing the costs of the components of resource utilization rather than just the final cost of care, we can assess, compare, understand, endorse, and address these driving factors.

1. The significance of resource utilization

To maximize the efficiency of their practices, high-volume shoulder surgeons have introduced standardization to health care delivery.⁷ Identifying specific efficiencies makes uniform acceptance of beneficial practice patterns possible.

 To facilitate comparison of goods and services used during an episode of surgical care, Virani and colleagues^8,9 studied the costs of TSA and RSA and calculated the top 10 driving cost factors for these procedures (Figure 1). Their systematic analysis provided a framework for a common method of communication, allowing an orthopedic surgeon to gain a more complete understanding of the resources used during a particular operative procedure in his or her practice, and allowing several physicians to compare and contrast the resources collectively used for a single procedure, facilitating an understanding of different practice patterns within a local community. At a societal level, these data can be collected to help guide overall recommendations.

2. How we defined utilization

To define the resources used, we had to decide which procedure components cost the most. Virani and colleagues^8,9 found that the top 10 cost drivers accounted for 93.11% and 94.77% of the total cost of the TSA and RSA care cycles, respectively (Figure 1). For each cost driver, information on resources used (goods, services, overhead) was collected on 2 forms, the Hospital Utilization Form (7 hospital-based items) and the Clinical Utilization Form (3 non-hospital-based items). To make hospital data easy to compile, we piloted use of a “smart form” in the EpicCare EMR system to isolate and auto-populate specific data fields.

3. EMR data collection

With EMR becoming mandatory for all public and private health care providers starting in 2014, utilization data are now included in a single unified system. Working with our in-house information technology department, we developed an algorithm to populate this information in a separate, easy-to-follow hospital utilization form. This form can be adopted by other institutions. Although EpicCare EMR is used by 52% of hospitals and at our institution, the data points required to make the same measurements are generalizable and exist in other EMRs.

Smartlinks, a tool in this EMR, allows utilization data to be quickly retrieved from different locations in a medical record and allows a form to be electronically completed in seconds. Data can be retrieved for any patient in the EMR system, regardless of when that patient’s hospital stay occurred. We populated data from surgeries performed 2 years before the start of this project.

4. What we can learn from these data

Data from a pilot study of 25 patients who underwent primary anatomical TSA for osteoarthritis and 25 patients who underwent primary RSA for massive rotator cuff tear allowed us to generate graphical representations of a single surgeon’s practice patterns that most affected the cost of care. Time in holding, time in the operating room, time in the postanesthesia care unit, and percentage of patients receiving different medications were recorded for each procedure (Figures 2–11). The study did not capture the wide variances in practice patterns in shoulder arthroplasty, and therefore other surgeons’ resource utilization may differ from ours. However, replicating this methodology at other institutions will produce a more robust data set from which conclusions about resource utilization and, indirectly, cost of care can be made.

5. Future possibilities

By using existing EMR tools to better understand resource utilization, orthopedic surgeons can play a constructive role in the dialogue on health care costs and new reimbursement models. The data presented here are not meant to be interpreted as hard and fast numbers on global resource utilization, but instead we intend to establish a model for collecting data on resource utilization. Resource utilization begins the dialogue that allows orthopedic surgeons and specialty societies to speak a common language without discussing actual cost numbers, which is discouraged under antitrust regulation. The data presented will allow comparisons to be made between surgeons in all practice settings to highlight areas of inconsistency in order to further improve patient care. Although this work involved only 50 patients undergoing only 2 types of surgeries, the resource-capturing methodology can be expanded to include more procedures and orthopedic practices. As all hospitals are now required to have EMRs, the metrics tracked in this work can be found on any patient medical record and auto-populated using our open-source utilization forms. Starting this data collection at your hospital may require no more than a conversation with the informatics department, as the metrics can for the most part be populated into a database on surgeon request.

As orthopedic surgeons return to the economic health care discussion, this information could prove essential in helping the individual surgeon and the orthopedic community justify the cost of care as well as fully understand the cost drivers for musculoskeletal care.

Click here to read the commentary on this article by Peter D. McCann, MD

Proximal humerus fractures are exceedingly common and account for almost 5% of all fractures. As osteoporosis is a risk factor for these fractures, their incidence rises with patient age.¹

In 1970, Neer² described these type of fractures and classified them as having 2, 3, or 4 parts based on the amount of angulation and displacement of the humeral head and the greater and lesser tuberosities with respect to the shaft.

Three- and 4-part proximal humerus fractures can be treated either nonoperatively, or surgically with closed reduction and percutaneous fixation, intramedullary fixation, open reduction and internal fixation (ORIF), or arthroplasty. There remains controversy over the best treatment, but a key component of any surgical treatment is anatomical reduction, stable fixation, and then healing of the tuberosities. A current common form of treatment is augmentation with an allograft fibula placed in the medullary canal. Although not formally reported, anecdotal evidence demonstrates that revision to arthroplasty is very difficult in the setting of an ingrown graft in the medullary canal of the humerus.

In this article, we present a novel technique of using allograft femoral head to reconstruct the metaphysis in ORIF of comminuted proximal humerus fractures.

Technique

Presented in Figure 1 are preoperative images of a representative displaced 4-part proximal humerus fracture treated surgically using the technique described here. General anesthesia is used. After intubation on the operating table, the patient is placed in the beach-chair position with about 75° of hip flexion. All bony prominences are padded, and the head and trunk are well secured. A pneumatic arm positioner is used to alleviate the need for an assistant to manipulate the arm. An image intensifier is used before preparing to verify that appropriate images of the proximal humerus can be obtained. Once adequate images are confirmed, the floor can be marked at the position of the fluoroscopic unit’s wheels to allow easy reproduction of images once the arm is prepared and draped. The intensifier is then removed from the field, the shoulder is prepared and draped in usual fashion, and prophylactic antibiotics are administered.

A deltopectoral incision is used, and sharp dissection is made through the subcutaneous tissue to raise full-thickness subcutaneous flaps on each side. The deltopectoral interval is sharply dissected while protecting the cephalic vein. Subdeltoid adhesions are then released. Palpation of the axillary nerve in the quadrilateral space to identify its location is helpful to avoid injury during the procedure.

The fracture is then identified, and No. 5 permanent suture is placed through the posterior and superior rotator cuff and through the subscapularis insertion (Figure 2). The tuberosities are freed from the humeral head sharply. A blunt elevator is then used to gently elevate the humeral head upward, with care taken to avoid comminuting the metaphyseal bone while levering. Reduction is achieved by manipulating the sutures and levering the head with the elevator while placing the arm in extension and posterior translation. Fluoroscopic images are used to verify correct anatomical alignment. Generally, the metaphysis demonstrates comminution and impaction, with poor bone quality necessitating use of bone graft.

A frozen allograft femoral head is then obtained and split into 2 equal pieces using a saw (Figures 3–5). One piece is fashioned with a saw and a burr into a trapezoid such that the proximal portion is wider, and the distal, tapered portion is sized to fit the canal. The broad, proximal portion of the graft will serve as a pedestal to reduce the head to the shaft. Measuring the internal diameter of the humeral canal can be useful in estimating the necessary dimensions of the distal portion of the allograft. The graft often needs several small adjustments that necessitate attempting to place it in the intramedullary canal and then trimming as necessary to ensure proper fit distally within the shaft. For this reason, it is beneficial to perform the graft preparation near the surgical field. Once completed, the distal portion is then impacted into the humeral canal (Figure 6). Because of this impaction, there is no possibility for subsidence or pistoning of the graft within the canal, which can occur with a fibular graft. The humeral head is reduced onto the shaft with the already placed sutures; this is achieved by abducting the shoulder. The image intensifier is then used to confirm appropriate alignment and positioning of the fragments, making sure that both neck–shaft angle and medial calcar alignment have been restored (Figures 7, 8).

An appropriately sized proximal humerus plate is then selected based on the location of the fracture line. We have used standard lateral proximal humerus locking plates as well as laterality-specific anterolateral proximal humerus plates and found that both are suitable for incorporation of the screws through the graft and into the head. The plate is positioned on the humerus, and a guide pin is placed by hand through the proximal-most hole so that the appropriate height of the plate can be verified on fluoroscopy. The first screw is then a nonlocking bicortical screw placed through the oval hole in the shaft of the plate to allow further fine manipulation of the plate more proximally or distally as needed. The final height is confirmed, and the screw is firmly tightened (Figure 9). The locking-screw guide is fixed to the proximal portion of the plate, and 2 locking screws are then placed into the head. The arm is then rotated to an anteroposterior view by placing the arm in external rotation and neutral flexion and is then abducted and internally rotated to recreate a lateral view to perform final verification of the position of the plate on orthogonal images. If the surgeon is satisfied with the position of the plate, another nonlocking screw is placed distally, and then the proximal holes are used to place locking screws as needed. If the surgeon is not satisfied, the 2 proximal screws can be removed and the plate repositioned.

After each screw is placed, fluoroscopy is used to ensure there has been no breach of the articular surface. The number of proximal screws placed depends on fracture configuration and surgeon preference.

The sutures through the rotator cuff are then fixed to the plate, securing the tuberosities. Final intraoperative radiographs are used to confirm reduction, alignment, and final position of hardware (Figure 10). After copious irrigation, a surgical drain is placed as needed, and the wound is closed in layered fashion. Three years after surgery, follow-up examination revealed no radiographic change in alignment, no necrosis, and no varus collapse (Figure 11), and the patient was pain-free during activities.

Discussion

Surgical treatment of comminuted proximal humerus fractures usually consists of some type of plate fixation with screw fixation of the shaft, screws or smooth pegs to support the chondral surfaces, and screw fixation or suture cerclage of the tuberosities.

Fixed-angle locking-plate-and-screw constructs increased the biomechanical stability and pullout strength of proximal humerus plates.^3,4 Nevertheless, avascular necrosis, malunion, and nonunion are still known complications of proximal humerus fractures, especially those with comminution, with up to 14% of patients still experiencing loss of fixation.⁵

For this reason, several authors have proposed using allograft bone and/or augmentation with calcium-containing cement to supplement fixation and provide an endosteal form of support for the head and tuberosities to decrease the risk for varus collapse. Osteobiologics (eg, calcium phosphate or sulfate cement) have been shown to decrease the risk for loss of reduction of proximal humerus fractures and decrease the risk for intra-articular screw penetration.^6,7 Many calcium phosphate cements are commercially available. Cost and availability are 2 reasons that these supplements are not more widely used. Cancellous chips have also been used to aid in the reduction of proximal humerus fractures.⁸ No randomized study has been conducted to show a clinical advantage of this technique, though retrospective studies have shown that it is not as advantageous as using calcium phosphate cement with respect to loss of reduction or screw penetration.⁶ Certainly, cancellous chips are easily available in most hospitals and are less expensive than some alternatives. A recent review of these techniques in osteoporotic proximal humerus fractures found no clear indication for using one of these supplements over another.⁹

However, some fracture patterns require a structural graft to reduce the tuberosities and head component. Although described more than 30 years ago as a treatment for nonunions with an intramedullary “peg” of iliac crest graft,¹⁰ the graft most commonly reported today is allograft fibula.^11-15 This technique consists of preparing the humeral shaft and often the fractured head segment with reaming to create a channel to receive the graft. Even with use of a small fibula, it is often time-consuming to use a saw, rasp, or burr to size the fibular segment to fit the medullary canal of the humerus. Once in place, the graft provides a strut on which the head fragment can be reduced and around which the tuberosities can be reduced. Although this technique is successful clinically and is biomechanically superior to plate-only constructs,^16,17 concerns remain.

One such concern is keeping this graft in routine supply at most hospitals. Supply and pricing from vendors can differ significantly between hospitals, and a surgeon may need to request grafts in advance, which makes their use nonviable in a trauma case. Certain grafts are often kept in routine supply based on their overall utilization. At our institution, allograft femoral heads meet this criterion and are routinely stocked.

Of more importance are the ramifications of these procedures for future revision surgeries. The need for arthroplasty revision is common after ORIF of a proximal humerus fracture.¹⁸

Arthroplasty revision is an already challenging procedure that becomes more complex with the need to remove 6 to 8 cm of ingrown endosteal bone from a shell of outer osteoporotic cortical bone. Our experience with these complex revisions provided the impetus to search for an alternate graft type that still provides a strut for reducing the head and tuberosities but limits the amount of endosteal bone that would need to be removed in arthroplasty revision in order to place a stemmed component into the humeral canal.

Some currently available arthroplasty fracture systems modify the previous anatomy of the stem to provide a more anatomical platform to reduce the tuberosities to a broader metaphyseal construct that incorporates bone grafting to assist with healing.

Because of these concerns and factors, we adapted our technique to create an individual-specific pedestal with allograft femoral head that can be anatomically matched to each patient. This provides a strut to reduce the head and tuberosity fragments but still limits the amount of allograft bone needed to seat into the existing canal. The geometry of the allograft can also be customized to the fracture, with most 3- and 4-part fractures needing a trapezoidal strut that resembles the metaphyseal portion of a fracture-specific shoulder arthroplasty implant.

We have used this technique for comminuted 3- and 4-part fractures of the proximal humerus in 14 cases with at least 2-year follow-up and in several more cases that have not reached 2-year follow-up. All cases have gone on to radiographic union; none have had to be revised either with revision ORIF or to an arthroplasty. Formal measurements of final postoperative range of motion have not been tabulated in all cases, as some cases have been lost to follow-up after radiographic union was achieved. Medium- and long-term results are not yet available, but no short-term complications have been noted.

Disadvantages of this technique are that, while an individualized graft is created, proper shaping still takes time, and a moderate amount of the femoral head is not used. However, we have found that, if a graft is inadvertently undersized, there is still ample femoral head remaining to create another sized graft. Other disadvantages are the added cost and the (rare) risk of disease transmission, which come with use of any allograft, but the technique is used instead of another type of allograft, so these disadvantages are largely equivalent. At our hospital, differences in cost and availability between femoral head or fibular allografts are negligible.

This procedure, which is easily performed in a short amount of time, allows a stable base of bone graft to be used as an aid in the anatomical reduction of proximal humerus fractures, without the need for reaming and preparation of the medullary canal and without further increasing the difficulty associated with a future revision procedure.

Proximal humerus fractures are exceedingly common and account for almost 5% of all fractures. As osteoporosis is a risk factor for these fractures, their incidence rises with patient age.¹

In 1970, Neer² described these type of fractures and classified them as having 2, 3, or 4 parts based on the amount of angulation and displacement of the humeral head and the greater and lesser tuberosities with respect to the shaft.

Three- and 4-part proximal humerus fractures can be treated either nonoperatively, or surgically with closed reduction and percutaneous fixation, intramedullary fixation, open reduction and internal fixation (ORIF), or arthroplasty. There remains controversy over the best treatment, but a key component of any surgical treatment is anatomical reduction, stable fixation, and then healing of the tuberosities. A current common form of treatment is augmentation with an allograft fibula placed in the medullary canal. Although not formally reported, anecdotal evidence demonstrates that revision to arthroplasty is very difficult in the setting of an ingrown graft in the medullary canal of the humerus.

In this article, we present a novel technique of using allograft femoral head to reconstruct the metaphysis in ORIF of comminuted proximal humerus fractures.

Technique

Presented in Figure 1 are preoperative images of a representative displaced 4-part proximal humerus fracture treated surgically using the technique described here. General anesthesia is used. After intubation on the operating table, the patient is placed in the beach-chair position with about 75° of hip flexion. All bony prominences are padded, and the head and trunk are well secured. A pneumatic arm positioner is used to alleviate the need for an assistant to manipulate the arm. An image intensifier is used before preparing to verify that appropriate images of the proximal humerus can be obtained. Once adequate images are confirmed, the floor can be marked at the position of the fluoroscopic unit’s wheels to allow easy reproduction of images once the arm is prepared and draped. The intensifier is then removed from the field, the shoulder is prepared and draped in usual fashion, and prophylactic antibiotics are administered.

A deltopectoral incision is used, and sharp dissection is made through the subcutaneous tissue to raise full-thickness subcutaneous flaps on each side. The deltopectoral interval is sharply dissected while protecting the cephalic vein. Subdeltoid adhesions are then released. Palpation of the axillary nerve in the quadrilateral space to identify its location is helpful to avoid injury during the procedure.

The fracture is then identified, and No. 5 permanent suture is placed through the posterior and superior rotator cuff and through the subscapularis insertion (Figure 2). The tuberosities are freed from the humeral head sharply. A blunt elevator is then used to gently elevate the humeral head upward, with care taken to avoid comminuting the metaphyseal bone while levering. Reduction is achieved by manipulating the sutures and levering the head with the elevator while placing the arm in extension and posterior translation. Fluoroscopic images are used to verify correct anatomical alignment. Generally, the metaphysis demonstrates comminution and impaction, with poor bone quality necessitating use of bone graft.

A frozen allograft femoral head is then obtained and split into 2 equal pieces using a saw (Figures 3–5). One piece is fashioned with a saw and a burr into a trapezoid such that the proximal portion is wider, and the distal, tapered portion is sized to fit the canal. The broad, proximal portion of the graft will serve as a pedestal to reduce the head to the shaft. Measuring the internal diameter of the humeral canal can be useful in estimating the necessary dimensions of the distal portion of the allograft. The graft often needs several small adjustments that necessitate attempting to place it in the intramedullary canal and then trimming as necessary to ensure proper fit distally within the shaft. For this reason, it is beneficial to perform the graft preparation near the surgical field. Once completed, the distal portion is then impacted into the humeral canal (Figure 6). Because of this impaction, there is no possibility for subsidence or pistoning of the graft within the canal, which can occur with a fibular graft. The humeral head is reduced onto the shaft with the already placed sutures; this is achieved by abducting the shoulder. The image intensifier is then used to confirm appropriate alignment and positioning of the fragments, making sure that both neck–shaft angle and medial calcar alignment have been restored (Figures 7, 8).

An appropriately sized proximal humerus plate is then selected based on the location of the fracture line. We have used standard lateral proximal humerus locking plates as well as laterality-specific anterolateral proximal humerus plates and found that both are suitable for incorporation of the screws through the graft and into the head. The plate is positioned on the humerus, and a guide pin is placed by hand through the proximal-most hole so that the appropriate height of the plate can be verified on fluoroscopy. The first screw is then a nonlocking bicortical screw placed through the oval hole in the shaft of the plate to allow further fine manipulation of the plate more proximally or distally as needed. The final height is confirmed, and the screw is firmly tightened (Figure 9). The locking-screw guide is fixed to the proximal portion of the plate, and 2 locking screws are then placed into the head. The arm is then rotated to an anteroposterior view by placing the arm in external rotation and neutral flexion and is then abducted and internally rotated to recreate a lateral view to perform final verification of the position of the plate on orthogonal images. If the surgeon is satisfied with the position of the plate, another nonlocking screw is placed distally, and then the proximal holes are used to place locking screws as needed. If the surgeon is not satisfied, the 2 proximal screws can be removed and the plate repositioned.

After each screw is placed, fluoroscopy is used to ensure there has been no breach of the articular surface. The number of proximal screws placed depends on fracture configuration and surgeon preference.

The sutures through the rotator cuff are then fixed to the plate, securing the tuberosities. Final intraoperative radiographs are used to confirm reduction, alignment, and final position of hardware (Figure 10). After copious irrigation, a surgical drain is placed as needed, and the wound is closed in layered fashion. Three years after surgery, follow-up examination revealed no radiographic change in alignment, no necrosis, and no varus collapse (Figure 11), and the patient was pain-free during activities.

Discussion

Surgical treatment of comminuted proximal humerus fractures usually consists of some type of plate fixation with screw fixation of the shaft, screws or smooth pegs to support the chondral surfaces, and screw fixation or suture cerclage of the tuberosities.

Fixed-angle locking-plate-and-screw constructs increased the biomechanical stability and pullout strength of proximal humerus plates.^3,4 Nevertheless, avascular necrosis, malunion, and nonunion are still known complications of proximal humerus fractures, especially those with comminution, with up to 14% of patients still experiencing loss of fixation.⁵

For this reason, several authors have proposed using allograft bone and/or augmentation with calcium-containing cement to supplement fixation and provide an endosteal form of support for the head and tuberosities to decrease the risk for varus collapse. Osteobiologics (eg, calcium phosphate or sulfate cement) have been shown to decrease the risk for loss of reduction of proximal humerus fractures and decrease the risk for intra-articular screw penetration.^6,7 Many calcium phosphate cements are commercially available. Cost and availability are 2 reasons that these supplements are not more widely used. Cancellous chips have also been used to aid in the reduction of proximal humerus fractures.⁸ No randomized study has been conducted to show a clinical advantage of this technique, though retrospective studies have shown that it is not as advantageous as using calcium phosphate cement with respect to loss of reduction or screw penetration.⁶ Certainly, cancellous chips are easily available in most hospitals and are less expensive than some alternatives. A recent review of these techniques in osteoporotic proximal humerus fractures found no clear indication for using one of these supplements over another.⁹

However, some fracture patterns require a structural graft to reduce the tuberosities and head component. Although described more than 30 years ago as a treatment for nonunions with an intramedullary “peg” of iliac crest graft,¹⁰ the graft most commonly reported today is allograft fibula.^11-15 This technique consists of preparing the humeral shaft and often the fractured head segment with reaming to create a channel to receive the graft. Even with use of a small fibula, it is often time-consuming to use a saw, rasp, or burr to size the fibular segment to fit the medullary canal of the humerus. Once in place, the graft provides a strut on which the head fragment can be reduced and around which the tuberosities can be reduced. Although this technique is successful clinically and is biomechanically superior to plate-only constructs,^16,17 concerns remain.

One such concern is keeping this graft in routine supply at most hospitals. Supply and pricing from vendors can differ significantly between hospitals, and a surgeon may need to request grafts in advance, which makes their use nonviable in a trauma case. Certain grafts are often kept in routine supply based on their overall utilization. At our institution, allograft femoral heads meet this criterion and are routinely stocked.

Of more importance are the ramifications of these procedures for future revision surgeries. The need for arthroplasty revision is common after ORIF of a proximal humerus fracture.¹⁸

Arthroplasty revision is an already challenging procedure that becomes more complex with the need to remove 6 to 8 cm of ingrown endosteal bone from a shell of outer osteoporotic cortical bone. Our experience with these complex revisions provided the impetus to search for an alternate graft type that still provides a strut for reducing the head and tuberosities but limits the amount of endosteal bone that would need to be removed in arthroplasty revision in order to place a stemmed component into the humeral canal.

Some currently available arthroplasty fracture systems modify the previous anatomy of the stem to provide a more anatomical platform to reduce the tuberosities to a broader metaphyseal construct that incorporates bone grafting to assist with healing.

Because of these concerns and factors, we adapted our technique to create an individual-specific pedestal with allograft femoral head that can be anatomically matched to each patient. This provides a strut to reduce the head and tuberosity fragments but still limits the amount of allograft bone needed to seat into the existing canal. The geometry of the allograft can also be customized to the fracture, with most 3- and 4-part fractures needing a trapezoidal strut that resembles the metaphyseal portion of a fracture-specific shoulder arthroplasty implant.

We have used this technique for comminuted 3- and 4-part fractures of the proximal humerus in 14 cases with at least 2-year follow-up and in several more cases that have not reached 2-year follow-up. All cases have gone on to radiographic union; none have had to be revised either with revision ORIF or to an arthroplasty. Formal measurements of final postoperative range of motion have not been tabulated in all cases, as some cases have been lost to follow-up after radiographic union was achieved. Medium- and long-term results are not yet available, but no short-term complications have been noted.

Disadvantages of this technique are that, while an individualized graft is created, proper shaping still takes time, and a moderate amount of the femoral head is not used. However, we have found that, if a graft is inadvertently undersized, there is still ample femoral head remaining to create another sized graft. Other disadvantages are the added cost and the (rare) risk of disease transmission, which come with use of any allograft, but the technique is used instead of another type of allograft, so these disadvantages are largely equivalent. At our hospital, differences in cost and availability between femoral head or fibular allografts are negligible.

This procedure, which is easily performed in a short amount of time, allows a stable base of bone graft to be used as an aid in the anatomical reduction of proximal humerus fractures, without the need for reaming and preparation of the medullary canal and without further increasing the difficulty associated with a future revision procedure.

Proximal humerus fractures are exceedingly common and account for almost 5% of all fractures. As osteoporosis is a risk factor for these fractures, their incidence rises with patient age.¹

In 1970, Neer² described these type of fractures and classified them as having 2, 3, or 4 parts based on the amount of angulation and displacement of the humeral head and the greater and lesser tuberosities with respect to the shaft.

Three- and 4-part proximal humerus fractures can be treated either nonoperatively, or surgically with closed reduction and percutaneous fixation, intramedullary fixation, open reduction and internal fixation (ORIF), or arthroplasty. There remains controversy over the best treatment, but a key component of any surgical treatment is anatomical reduction, stable fixation, and then healing of the tuberosities. A current common form of treatment is augmentation with an allograft fibula placed in the medullary canal. Although not formally reported, anecdotal evidence demonstrates that revision to arthroplasty is very difficult in the setting of an ingrown graft in the medullary canal of the humerus.

In this article, we present a novel technique of using allograft femoral head to reconstruct the metaphysis in ORIF of comminuted proximal humerus fractures.

Technique

Presented in Figure 1 are preoperative images of a representative displaced 4-part proximal humerus fracture treated surgically using the technique described here. General anesthesia is used. After intubation on the operating table, the patient is placed in the beach-chair position with about 75° of hip flexion. All bony prominences are padded, and the head and trunk are well secured. A pneumatic arm positioner is used to alleviate the need for an assistant to manipulate the arm. An image intensifier is used before preparing to verify that appropriate images of the proximal humerus can be obtained. Once adequate images are confirmed, the floor can be marked at the position of the fluoroscopic unit’s wheels to allow easy reproduction of images once the arm is prepared and draped. The intensifier is then removed from the field, the shoulder is prepared and draped in usual fashion, and prophylactic antibiotics are administered.

A deltopectoral incision is used, and sharp dissection is made through the subcutaneous tissue to raise full-thickness subcutaneous flaps on each side. The deltopectoral interval is sharply dissected while protecting the cephalic vein. Subdeltoid adhesions are then released. Palpation of the axillary nerve in the quadrilateral space to identify its location is helpful to avoid injury during the procedure.

The fracture is then identified, and No. 5 permanent suture is placed through the posterior and superior rotator cuff and through the subscapularis insertion (Figure 2). The tuberosities are freed from the humeral head sharply. A blunt elevator is then used to gently elevate the humeral head upward, with care taken to avoid comminuting the metaphyseal bone while levering. Reduction is achieved by manipulating the sutures and levering the head with the elevator while placing the arm in extension and posterior translation. Fluoroscopic images are used to verify correct anatomical alignment. Generally, the metaphysis demonstrates comminution and impaction, with poor bone quality necessitating use of bone graft.

A frozen allograft femoral head is then obtained and split into 2 equal pieces using a saw (Figures 3–5). One piece is fashioned with a saw and a burr into a trapezoid such that the proximal portion is wider, and the distal, tapered portion is sized to fit the canal. The broad, proximal portion of the graft will serve as a pedestal to reduce the head to the shaft. Measuring the internal diameter of the humeral canal can be useful in estimating the necessary dimensions of the distal portion of the allograft. The graft often needs several small adjustments that necessitate attempting to place it in the intramedullary canal and then trimming as necessary to ensure proper fit distally within the shaft. For this reason, it is beneficial to perform the graft preparation near the surgical field. Once completed, the distal portion is then impacted into the humeral canal (Figure 6). Because of this impaction, there is no possibility for subsidence or pistoning of the graft within the canal, which can occur with a fibular graft. The humeral head is reduced onto the shaft with the already placed sutures; this is achieved by abducting the shoulder. The image intensifier is then used to confirm appropriate alignment and positioning of the fragments, making sure that both neck–shaft angle and medial calcar alignment have been restored (Figures 7, 8).

An appropriately sized proximal humerus plate is then selected based on the location of the fracture line. We have used standard lateral proximal humerus locking plates as well as laterality-specific anterolateral proximal humerus plates and found that both are suitable for incorporation of the screws through the graft and into the head. The plate is positioned on the humerus, and a guide pin is placed by hand through the proximal-most hole so that the appropriate height of the plate can be verified on fluoroscopy. The first screw is then a nonlocking bicortical screw placed through the oval hole in the shaft of the plate to allow further fine manipulation of the plate more proximally or distally as needed. The final height is confirmed, and the screw is firmly tightened (Figure 9). The locking-screw guide is fixed to the proximal portion of the plate, and 2 locking screws are then placed into the head. The arm is then rotated to an anteroposterior view by placing the arm in external rotation and neutral flexion and is then abducted and internally rotated to recreate a lateral view to perform final verification of the position of the plate on orthogonal images. If the surgeon is satisfied with the position of the plate, another nonlocking screw is placed distally, and then the proximal holes are used to place locking screws as needed. If the surgeon is not satisfied, the 2 proximal screws can be removed and the plate repositioned.

After each screw is placed, fluoroscopy is used to ensure there has been no breach of the articular surface. The number of proximal screws placed depends on fracture configuration and surgeon preference.

The sutures through the rotator cuff are then fixed to the plate, securing the tuberosities. Final intraoperative radiographs are used to confirm reduction, alignment, and final position of hardware (Figure 10). After copious irrigation, a surgical drain is placed as needed, and the wound is closed in layered fashion. Three years after surgery, follow-up examination revealed no radiographic change in alignment, no necrosis, and no varus collapse (Figure 11), and the patient was pain-free during activities.

Discussion

Surgical treatment of comminuted proximal humerus fractures usually consists of some type of plate fixation with screw fixation of the shaft, screws or smooth pegs to support the chondral surfaces, and screw fixation or suture cerclage of the tuberosities.

Fixed-angle locking-plate-and-screw constructs increased the biomechanical stability and pullout strength of proximal humerus plates.^3,4 Nevertheless, avascular necrosis, malunion, and nonunion are still known complications of proximal humerus fractures, especially those with comminution, with up to 14% of patients still experiencing loss of fixation.⁵

For this reason, several authors have proposed using allograft bone and/or augmentation with calcium-containing cement to supplement fixation and provide an endosteal form of support for the head and tuberosities to decrease the risk for varus collapse. Osteobiologics (eg, calcium phosphate or sulfate cement) have been shown to decrease the risk for loss of reduction of proximal humerus fractures and decrease the risk for intra-articular screw penetration.^6,7 Many calcium phosphate cements are commercially available. Cost and availability are 2 reasons that these supplements are not more widely used. Cancellous chips have also been used to aid in the reduction of proximal humerus fractures.⁸ No randomized study has been conducted to show a clinical advantage of this technique, though retrospective studies have shown that it is not as advantageous as using calcium phosphate cement with respect to loss of reduction or screw penetration.⁶ Certainly, cancellous chips are easily available in most hospitals and are less expensive than some alternatives. A recent review of these techniques in osteoporotic proximal humerus fractures found no clear indication for using one of these supplements over another.⁹

However, some fracture patterns require a structural graft to reduce the tuberosities and head component. Although described more than 30 years ago as a treatment for nonunions with an intramedullary “peg” of iliac crest graft,¹⁰ the graft most commonly reported today is allograft fibula.^11-15 This technique consists of preparing the humeral shaft and often the fractured head segment with reaming to create a channel to receive the graft. Even with use of a small fibula, it is often time-consuming to use a saw, rasp, or burr to size the fibular segment to fit the medullary canal of the humerus. Once in place, the graft provides a strut on which the head fragment can be reduced and around which the tuberosities can be reduced. Although this technique is successful clinically and is biomechanically superior to plate-only constructs,^16,17 concerns remain.

One such concern is keeping this graft in routine supply at most hospitals. Supply and pricing from vendors can differ significantly between hospitals, and a surgeon may need to request grafts in advance, which makes their use nonviable in a trauma case. Certain grafts are often kept in routine supply based on their overall utilization. At our institution, allograft femoral heads meet this criterion and are routinely stocked.

Of more importance are the ramifications of these procedures for future revision surgeries. The need for arthroplasty revision is common after ORIF of a proximal humerus fracture.¹⁸

Arthroplasty revision is an already challenging procedure that becomes more complex with the need to remove 6 to 8 cm of ingrown endosteal bone from a shell of outer osteoporotic cortical bone. Our experience with these complex revisions provided the impetus to search for an alternate graft type that still provides a strut for reducing the head and tuberosities but limits the amount of endosteal bone that would need to be removed in arthroplasty revision in order to place a stemmed component into the humeral canal.

Some currently available arthroplasty fracture systems modify the previous anatomy of the stem to provide a more anatomical platform to reduce the tuberosities to a broader metaphyseal construct that incorporates bone grafting to assist with healing.

Because of these concerns and factors, we adapted our technique to create an individual-specific pedestal with allograft femoral head that can be anatomically matched to each patient. This provides a strut to reduce the head and tuberosity fragments but still limits the amount of allograft bone needed to seat into the existing canal. The geometry of the allograft can also be customized to the fracture, with most 3- and 4-part fractures needing a trapezoidal strut that resembles the metaphyseal portion of a fracture-specific shoulder arthroplasty implant.

We have used this technique for comminuted 3- and 4-part fractures of the proximal humerus in 14 cases with at least 2-year follow-up and in several more cases that have not reached 2-year follow-up. All cases have gone on to radiographic union; none have had to be revised either with revision ORIF or to an arthroplasty. Formal measurements of final postoperative range of motion have not been tabulated in all cases, as some cases have been lost to follow-up after radiographic union was achieved. Medium- and long-term results are not yet available, but no short-term complications have been noted.

Disadvantages of this technique are that, while an individualized graft is created, proper shaping still takes time, and a moderate amount of the femoral head is not used. However, we have found that, if a graft is inadvertently undersized, there is still ample femoral head remaining to create another sized graft. Other disadvantages are the added cost and the (rare) risk of disease transmission, which come with use of any allograft, but the technique is used instead of another type of allograft, so these disadvantages are largely equivalent. At our hospital, differences in cost and availability between femoral head or fibular allografts are negligible.

This procedure, which is easily performed in a short amount of time, allows a stable base of bone graft to be used as an aid in the anatomical reduction of proximal humerus fractures, without the need for reaming and preparation of the medullary canal and without further increasing the difficulty associated with a future revision procedure.

In children, pelvic fractures are uncommon, with an incidence ranging from 1% to 4.6% of all pediatric fractures,^1-4 and acetabular fractures make up only 0.8% to 15% of pelvic fractures.^1,3,5,6 Acetabular fractures are so uncommon in children partly because of the cartilaginous nature of the immature acetabulum. The increased cartilage volume relative to adults provides greater capacity for energy absorption, resulting in greater elastic and plastic deformation before fracture occurrence. More force is therefore required to cause a fracture, and associated visceral injuries, head injuries, and long-bone fractures are common.^3,7,8

The impact of acetabular fractures on adolescents warrants special attention because any resulting disability will affect them during their most productive years. Both avascular necrosis (AVN) and degenerative arthritis are particularly devastating complications in this age group. Complications such as premature physeal closure^9-15 are unique to adolescents, and there is little information available on how injury in older children affects growth in this area.

There have been very few studies of the outcomes of these injuries in children. Mostly, there have been case reports and small series primarily dealing with nonoperative management of acetabular fractures in adolescents.^{3,10,11,16-20} By contrast, operative treatment of acetabular fractures in adults has been well described, and outcomes widely reported. As a result, much of our knowledge about managing these injuries is extrapolated from the adult literature. Although treatment of acetabular fractures in adults has evolved substantially, treatment of these injuries in adolescents remains primarily nonoperative. We conducted a study to evaluate outcomes of treatment of adolescent acetabular fractures.

Patients and Methods

After obtaining institutional review board approval for this study, we retrospectively reviewed the cases of all adolescent patients admitted with a diagnosis of acetabular fracture to 2 academic institutions between 1991 and 2003. Thirty-eight patients (28 males, 10 females) were identified. Mean age at time of injury was 15 years (range, 11-18 years). Mean follow-up was 3.2 years (range, 5-180 months).

Data on fracture types, treatment methods, associated injuries, complications, union rates, pain, and return to normal activities were collected. Acetabular fractures were classified according to the system of Letournel and Judet.²¹ There were 20 elementary and 18 associated fractures.

Of the 38 patients, 30 sustained high-energy trauma in motor vehicle accidents (25) or in falls from significant heights (5). The other 8 patients injured themselves playing sports (4 had severe traumatic brain injury, 2 had labial wounds, and 2 had injuries involving the abdominal viscera). Twelve patients had associated pelvic ring injuries, 18 had femoral head dislocations, 2 had femoral head fractures, and 13 had evidence of impaction injury to the femoral head articular cartilage. Twelve patients had marginal impaction of the acetabular wall. Fifteen patients had open triradiate physes at time of injury (Table 1).

Thirty-seven of the 38 patients were treated with open reduction and internal fixation (ORIF) by an experienced orthopedic trauma surgeon; 1 patient with a stable posterior wall fracture was treated nonoperatively. Surgical indications were articular displacement of more than 1 mm, hip joint instability, irreducible hip dislocation, and intra-articular fracture fragments. In the 37 surgically treated cases, the approaches used were Kocher-Langenbeck (22), ilioinguinal (8), combined Kocher-Langenbeck/ilioinguinal (5), and triradiate (2).

Immediate postoperative radiographs were evaluated by 3 orthopedic surgeons blinded to the patients’ clinical outcomes. Displacement was evaluated on anteroposterior (AP) and Judet views of the pelvis, as described by Matta,²² and reductions were classified as anatomical (0-1 mm of displacement), imperfect (>1 to 3 mm), poor (>3 mm), or surgical secondary congruence (Table 2).

Results

Thirty-seven patients underwent acetabular fracture ORIF. Immediate postoperative radiographs showed 30 anatomical reductions and 7 imperfect reductions. One patient had surgical secondary congruence and developed AVN of the hip. We could not identify an association between the quality of the reduction and the outcome with respect to pain or return to activity. However, no patient had a poor reduction. An illustrative case is presented in Figures 1 to 4.

All acetabular fractures united within 4.5 months (range, 3.0-8.0 months) after the index procedure. Early postoperative complications included 3 cases of meralgia paresthetica and 13 cases of abductor weakness. Meralgia paresthetica resolved spontaneously in all 3 patients. Of the 13 patients with abductor weakness, 11 improved with physical therapy, 1 was limited by the head injury, and 1 subsequently underwent hip fusion. One patient had a deep vein thrombosis (DVT) that was identified before surgery and managed with warfarin.

Other complications included 1 case of deep infection of the surgical wound. This infection presented 4 months after surgery and was treated with débridement, hardware removal, and a 3-month course of antibiotics. Two patients who sustained hip dislocations at time of injury developed AVN of the femoral head. Both developed osteoarthritis, and 1 underwent hip fusion. Eight patients developed heterotopic ossification on the side of the acetabular fracture; 4 of them underwent surgical excision. Four patients required a separate operation for hardware removal. Four patients with triradiate cartilage involvement went on to premature closure. No patient had any leg-length discrepancy or dysplasia at time of follow-up.

Thirty-four of the 38 patients returned to their regular activities. For these patients, mean time to return to full activity was 7.0 months (range, 3-30 months); there was no difference in mean time to return to full activity between skeletally mature and skeletally immature patients (6.6 vs 7.4 months; P = .57). Of the other 4 patients, 1 had permanent cognitive and physical disability with an ataxic gait as a result of a traumatic brain injury, 2 were limited by AVN (1 underwent hip fusion), and 1 was limited by an ipsilateral knee injury.

Of the 38 patients, 29 were pain-free; 6 had occasional, intermittent mild pain that did not limit their activities; and 3 had severe, activity-limiting pain. Of the 6 patients with mild pain, 2 had femoral impaction injuries, and 4 had marginal impaction injuries. Of the 3 patients with severe pain, 2 developed femoral head AVN, and 1 had multiple ipsilateral extremity injuries involving the femur, knee, and tibia.

Discussion

The traditional treatment for acetabular fractures in children has been nonoperative,^8,10 and there are few specific treatment guidelines.¹³ Recent recommendations are nonoperative treatment for minimally displaced fractures (<1 mm) and acetabular fracture ORIF for fractures displaced more than 2 mm.¹¹ No clear consensus exists on management for fractures displaced 1 to 2 mm. Few studies have investigated the outcomes of operative management of these fractures in the pediatric or adolescent population.

In our series of adolescent acetabular fractures, we examined unions, complications, and return to activity. Of 38 patients with acetabular fractures, 37 were treated with ORIF. Anatomical reduction was achieved in the majority of patients. Posterior wall fractures were by far the most common fracture type, which is consistent with previous reports.^10,11 All acetabular fractures united, and most patients were pain-free at latest follow-up. There was a low incidence of major complications in our patient population. One major complication was a DVT in a 14-year-old boy who was in a motor vehicle accident and sustained a T-type fracture of the right acetabulum with contralateral femoral shaft and ankle fractures. The DVT was in the right internal iliac and common femoral veins and was diagnosed on magnetic resonance venography. The patient was treated with warfarin for 3 months without incident.

Two patients developed AVN of the femoral head. One of these patients was an 11-year-old girl who was in a motor vehicle accident and sustained a T-type fracture with marginal impaction of the posterior wall, posterior hip dislocation, and a pelvic ring injury. She was treated with ORIF through combined Kocher-Langenbeck/ilioinguinal approaches. By 4 months after surgery, the acetabular fracture was united. Nine months after surgery, she still had pain (activity-limiting) and a 35° flexion contracture of the hip, and she was ambulating with a cane. The diagnosis was AVN of the hip. The patient underwent hip fusion 1 year after surgery.

The second patient with femoral head AVN was a 12-year-old boy who fell while skiing and sustained a fracture of the posterior wall and a hip dislocation with impaction of the femoral head. Initial treatment at an outside institution consisted of open reduction of the hip and excision of a “loose body” from the joint. Eight weeks after surgery, the patient continued to have pain and was referred to our institution. A second operation was performed. Findings included a defect involving 40% of the posterior wall, and signs that the posterior wall had been excised during the initial operation. The patient eventually developed AVN of the hip. This patient was also diagnosed with a deep wound infection 4 months after surgery. He presented with pain and a fluid collection around the hip. The infection was not confirmed through fluid culture, and, as he eventually developed AVN of the hip, his symptoms may have been the result of chondrolysis or AVN rather than infection.

There were no cases of nonunion or malunion, leg-length discrepancy, or permanent sciatic nerve palsy. Although there were a few cases of premature closure of the triradiate cartilage, no acetabular dysplasia was seen at latest follow-up, likely because of the relative maturity of our pediatric group (age range, 11-18 years). Age at time of injury is thought to be the most important factor influencing growth and development of the acetabulum.^9,13 In addition, previous studies have demonstrated a tendency toward acetabular fractures in patients with mature triradiate cartilage—versus pelvic ring injuries in patients with immature triradiate cartilage.^8,11 This may also account for the older age of our study group.

Minor complications (eg, meralgia paresthetica) resolved spontaneously. The most common complications were abductor weakness and heterotopic ossification. In only 4 cases was a secondary procedure for excision of the heterotopic bone required. Abductor weakness, more commonly associated with a Kocher-Langenbeck approach to the hip, resolved with therapy in almost all cases. Only 4 of our patients required removal of hardware from the acetabulum.

Although the majority of acetabular fractures resulted from high-energy trauma, 8 cases were sports-related. Six of these involved posterior wall fractures, suggesting the injury resulted from a fall on flexed knee and hip. This was not known to be a common mechanism of injury in this age group.^3,7 An additional concern was how to size the posterior wall fragment when not ossified. At one center, preoperative magnetic resonance imaging (MRI) was effectively used to size the osteochondral posterior wall fragment as standard protocol for patients with posterior wall fractures in this age group—resulting in better decisions regarding the need for ORIF. At the other institution, preoperative MRI was not performed routinely for this subset of patients.

Thirty-four of our 38 patients returned to their normal activities. Of the other 4 patients, 1 was permanently disabled secondary to traumatic brain injury, 1 had other ipsilateral extremity injuries that limited his mobility, and 2 developed AVN of the femoral head. Both patients with AVN had hip dislocations. Four of the 6 patients who were symptomatic during activity sustained impaction injuries of the femoral head or posterior wall. This suggests that poorer outcomes may be associated with dislocation or with articular injuries—similar to what has been reported in the adult literature.

This study had several limitations. First, it was a retrospective case series, so there was no control group for comparison. Second, the relatively short follow-up did not allow evaluation of the incidence of degenerative arthritis secondary to articular injury, the symptoms of which may develop 1 to 2 decades after injury.¹³ This phenomenon was well described by Letournel and Judet²¹ in the adult population, and there is no reason to presume the adolescent population is any different. Third, our sample was small and unlikely to represent a uniform sampling of the general pediatric population. Fourth, it was not possible to draw detailed conclusions about the outcome of ORIF for a particular type of acetabular fracture. Fifth, we did not see as many of the associated visceral injuries that are so prevalent in the literature. This may reflect improvement in safety specifications for automobiles, or our group may not have had the most severe or high-energy injuries. Here our population sample may have skewed our results, leading to better than expected outcomes.

One last study limitation, a major one, was the age of our population, 11 to 18 years, which makes it difficult to extrapolate results to the entire pediatric population. On one hand, a more immature skeleton has a higher chance of remodeling and is more forgiving of deformities and small amounts of displacement. On the other hand, injury and premature triradiate cartilage fusion in a younger patient can lead to significant deformity and acetabular dysplasia.⁹ Whether ORIF of these fractures would alter the outcome of an injury to the triradiate cartilage is yet to be determined.

Conclusion

In agreement with earlier studies,^10,11,15,18 the good outcomes in our series correlated with congruence of reduction. Outcome predictors such as dislocation, femoral head injury, and marginal impaction are similar to those described in the adult literature. Although our study did not have a nonoperative group for comparison, the favorable outcomes of ORIF of acetabular fractures suggest that a more aggressive approach to treatment should be considered. Given the added benefits of early, pain-free mobilization, we think that only stable, undisplaced fractures (<1 mm) should be managed nonoperatively. In the adolescent population, we recommend ORIF for optimal management of unstable acetabular fractures, fractures with any hip subluxation, and fractures displaced more than 1 mm.

In children, pelvic fractures are uncommon, with an incidence ranging from 1% to 4.6% of all pediatric fractures,^1-4 and acetabular fractures make up only 0.8% to 15% of pelvic fractures.^1,3,5,6 Acetabular fractures are so uncommon in children partly because of the cartilaginous nature of the immature acetabulum. The increased cartilage volume relative to adults provides greater capacity for energy absorption, resulting in greater elastic and plastic deformation before fracture occurrence. More force is therefore required to cause a fracture, and associated visceral injuries, head injuries, and long-bone fractures are common.^3,7,8

The impact of acetabular fractures on adolescents warrants special attention because any resulting disability will affect them during their most productive years. Both avascular necrosis (AVN) and degenerative arthritis are particularly devastating complications in this age group. Complications such as premature physeal closure^9-15 are unique to adolescents, and there is little information available on how injury in older children affects growth in this area.

There have been very few studies of the outcomes of these injuries in children. Mostly, there have been case reports and small series primarily dealing with nonoperative management of acetabular fractures in adolescents.^{3,10,11,16-20} By contrast, operative treatment of acetabular fractures in adults has been well described, and outcomes widely reported. As a result, much of our knowledge about managing these injuries is extrapolated from the adult literature. Although treatment of acetabular fractures in adults has evolved substantially, treatment of these injuries in adolescents remains primarily nonoperative. We conducted a study to evaluate outcomes of treatment of adolescent acetabular fractures.

Patients and Methods

After obtaining institutional review board approval for this study, we retrospectively reviewed the cases of all adolescent patients admitted with a diagnosis of acetabular fracture to 2 academic institutions between 1991 and 2003. Thirty-eight patients (28 males, 10 females) were identified. Mean age at time of injury was 15 years (range, 11-18 years). Mean follow-up was 3.2 years (range, 5-180 months).

Data on fracture types, treatment methods, associated injuries, complications, union rates, pain, and return to normal activities were collected. Acetabular fractures were classified according to the system of Letournel and Judet.²¹ There were 20 elementary and 18 associated fractures.

Of the 38 patients, 30 sustained high-energy trauma in motor vehicle accidents (25) or in falls from significant heights (5). The other 8 patients injured themselves playing sports (4 had severe traumatic brain injury, 2 had labial wounds, and 2 had injuries involving the abdominal viscera). Twelve patients had associated pelvic ring injuries, 18 had femoral head dislocations, 2 had femoral head fractures, and 13 had evidence of impaction injury to the femoral head articular cartilage. Twelve patients had marginal impaction of the acetabular wall. Fifteen patients had open triradiate physes at time of injury (Table 1).

Thirty-seven of the 38 patients were treated with open reduction and internal fixation (ORIF) by an experienced orthopedic trauma surgeon; 1 patient with a stable posterior wall fracture was treated nonoperatively. Surgical indications were articular displacement of more than 1 mm, hip joint instability, irreducible hip dislocation, and intra-articular fracture fragments. In the 37 surgically treated cases, the approaches used were Kocher-Langenbeck (22), ilioinguinal (8), combined Kocher-Langenbeck/ilioinguinal (5), and triradiate (2).

Immediate postoperative radiographs were evaluated by 3 orthopedic surgeons blinded to the patients’ clinical outcomes. Displacement was evaluated on anteroposterior (AP) and Judet views of the pelvis, as described by Matta,²² and reductions were classified as anatomical (0-1 mm of displacement), imperfect (>1 to 3 mm), poor (>3 mm), or surgical secondary congruence (Table 2).

Results

Thirty-seven patients underwent acetabular fracture ORIF. Immediate postoperative radiographs showed 30 anatomical reductions and 7 imperfect reductions. One patient had surgical secondary congruence and developed AVN of the hip. We could not identify an association between the quality of the reduction and the outcome with respect to pain or return to activity. However, no patient had a poor reduction. An illustrative case is presented in Figures 1 to 4.

All acetabular fractures united within 4.5 months (range, 3.0-8.0 months) after the index procedure. Early postoperative complications included 3 cases of meralgia paresthetica and 13 cases of abductor weakness. Meralgia paresthetica resolved spontaneously in all 3 patients. Of the 13 patients with abductor weakness, 11 improved with physical therapy, 1 was limited by the head injury, and 1 subsequently underwent hip fusion. One patient had a deep vein thrombosis (DVT) that was identified before surgery and managed with warfarin.

Other complications included 1 case of deep infection of the surgical wound. This infection presented 4 months after surgery and was treated with débridement, hardware removal, and a 3-month course of antibiotics. Two patients who sustained hip dislocations at time of injury developed AVN of the femoral head. Both developed osteoarthritis, and 1 underwent hip fusion. Eight patients developed heterotopic ossification on the side of the acetabular fracture; 4 of them underwent surgical excision. Four patients required a separate operation for hardware removal. Four patients with triradiate cartilage involvement went on to premature closure. No patient had any leg-length discrepancy or dysplasia at time of follow-up.

Thirty-four of the 38 patients returned to their regular activities. For these patients, mean time to return to full activity was 7.0 months (range, 3-30 months); there was no difference in mean time to return to full activity between skeletally mature and skeletally immature patients (6.6 vs 7.4 months; P = .57). Of the other 4 patients, 1 had permanent cognitive and physical disability with an ataxic gait as a result of a traumatic brain injury, 2 were limited by AVN (1 underwent hip fusion), and 1 was limited by an ipsilateral knee injury.

Of the 38 patients, 29 were pain-free; 6 had occasional, intermittent mild pain that did not limit their activities; and 3 had severe, activity-limiting pain. Of the 6 patients with mild pain, 2 had femoral impaction injuries, and 4 had marginal impaction injuries. Of the 3 patients with severe pain, 2 developed femoral head AVN, and 1 had multiple ipsilateral extremity injuries involving the femur, knee, and tibia.

Discussion

The traditional treatment for acetabular fractures in children has been nonoperative,^8,10 and there are few specific treatment guidelines.¹³ Recent recommendations are nonoperative treatment for minimally displaced fractures (<1 mm) and acetabular fracture ORIF for fractures displaced more than 2 mm.¹¹ No clear consensus exists on management for fractures displaced 1 to 2 mm. Few studies have investigated the outcomes of operative management of these fractures in the pediatric or adolescent population.

In our series of adolescent acetabular fractures, we examined unions, complications, and return to activity. Of 38 patients with acetabular fractures, 37 were treated with ORIF. Anatomical reduction was achieved in the majority of patients. Posterior wall fractures were by far the most common fracture type, which is consistent with previous reports.^10,11 All acetabular fractures united, and most patients were pain-free at latest follow-up. There was a low incidence of major complications in our patient population. One major complication was a DVT in a 14-year-old boy who was in a motor vehicle accident and sustained a T-type fracture of the right acetabulum with contralateral femoral shaft and ankle fractures. The DVT was in the right internal iliac and common femoral veins and was diagnosed on magnetic resonance venography. The patient was treated with warfarin for 3 months without incident.

Two patients developed AVN of the femoral head. One of these patients was an 11-year-old girl who was in a motor vehicle accident and sustained a T-type fracture with marginal impaction of the posterior wall, posterior hip dislocation, and a pelvic ring injury. She was treated with ORIF through combined Kocher-Langenbeck/ilioinguinal approaches. By 4 months after surgery, the acetabular fracture was united. Nine months after surgery, she still had pain (activity-limiting) and a 35° flexion contracture of the hip, and she was ambulating with a cane. The diagnosis was AVN of the hip. The patient underwent hip fusion 1 year after surgery.

The second patient with femoral head AVN was a 12-year-old boy who fell while skiing and sustained a fracture of the posterior wall and a hip dislocation with impaction of the femoral head. Initial treatment at an outside institution consisted of open reduction of the hip and excision of a “loose body” from the joint. Eight weeks after surgery, the patient continued to have pain and was referred to our institution. A second operation was performed. Findings included a defect involving 40% of the posterior wall, and signs that the posterior wall had been excised during the initial operation. The patient eventually developed AVN of the hip. This patient was also diagnosed with a deep wound infection 4 months after surgery. He presented with pain and a fluid collection around the hip. The infection was not confirmed through fluid culture, and, as he eventually developed AVN of the hip, his symptoms may have been the result of chondrolysis or AVN rather than infection.

There were no cases of nonunion or malunion, leg-length discrepancy, or permanent sciatic nerve palsy. Although there were a few cases of premature closure of the triradiate cartilage, no acetabular dysplasia was seen at latest follow-up, likely because of the relative maturity of our pediatric group (age range, 11-18 years). Age at time of injury is thought to be the most important factor influencing growth and development of the acetabulum.^9,13 In addition, previous studies have demonstrated a tendency toward acetabular fractures in patients with mature triradiate cartilage—versus pelvic ring injuries in patients with immature triradiate cartilage.^8,11 This may also account for the older age of our study group.

Minor complications (eg, meralgia paresthetica) resolved spontaneously. The most common complications were abductor weakness and heterotopic ossification. In only 4 cases was a secondary procedure for excision of the heterotopic bone required. Abductor weakness, more commonly associated with a Kocher-Langenbeck approach to the hip, resolved with therapy in almost all cases. Only 4 of our patients required removal of hardware from the acetabulum.

Although the majority of acetabular fractures resulted from high-energy trauma, 8 cases were sports-related. Six of these involved posterior wall fractures, suggesting the injury resulted from a fall on flexed knee and hip. This was not known to be a common mechanism of injury in this age group.^3,7 An additional concern was how to size the posterior wall fragment when not ossified. At one center, preoperative magnetic resonance imaging (MRI) was effectively used to size the osteochondral posterior wall fragment as standard protocol for patients with posterior wall fractures in this age group—resulting in better decisions regarding the need for ORIF. At the other institution, preoperative MRI was not performed routinely for this subset of patients.

Thirty-four of our 38 patients returned to their normal activities. Of the other 4 patients, 1 was permanently disabled secondary to traumatic brain injury, 1 had other ipsilateral extremity injuries that limited his mobility, and 2 developed AVN of the femoral head. Both patients with AVN had hip dislocations. Four of the 6 patients who were symptomatic during activity sustained impaction injuries of the femoral head or posterior wall. This suggests that poorer outcomes may be associated with dislocation or with articular injuries—similar to what has been reported in the adult literature.

This study had several limitations. First, it was a retrospective case series, so there was no control group for comparison. Second, the relatively short follow-up did not allow evaluation of the incidence of degenerative arthritis secondary to articular injury, the symptoms of which may develop 1 to 2 decades after injury.¹³ This phenomenon was well described by Letournel and Judet²¹ in the adult population, and there is no reason to presume the adolescent population is any different. Third, our sample was small and unlikely to represent a uniform sampling of the general pediatric population. Fourth, it was not possible to draw detailed conclusions about the outcome of ORIF for a particular type of acetabular fracture. Fifth, we did not see as many of the associated visceral injuries that are so prevalent in the literature. This may reflect improvement in safety specifications for automobiles, or our group may not have had the most severe or high-energy injuries. Here our population sample may have skewed our results, leading to better than expected outcomes.

One last study limitation, a major one, was the age of our population, 11 to 18 years, which makes it difficult to extrapolate results to the entire pediatric population. On one hand, a more immature skeleton has a higher chance of remodeling and is more forgiving of deformities and small amounts of displacement. On the other hand, injury and premature triradiate cartilage fusion in a younger patient can lead to significant deformity and acetabular dysplasia.⁹ Whether ORIF of these fractures would alter the outcome of an injury to the triradiate cartilage is yet to be determined.

Conclusion

In agreement with earlier studies,^10,11,15,18 the good outcomes in our series correlated with congruence of reduction. Outcome predictors such as dislocation, femoral head injury, and marginal impaction are similar to those described in the adult literature. Although our study did not have a nonoperative group for comparison, the favorable outcomes of ORIF of acetabular fractures suggest that a more aggressive approach to treatment should be considered. Given the added benefits of early, pain-free mobilization, we think that only stable, undisplaced fractures (<1 mm) should be managed nonoperatively. In the adolescent population, we recommend ORIF for optimal management of unstable acetabular fractures, fractures with any hip subluxation, and fractures displaced more than 1 mm.

In children, pelvic fractures are uncommon, with an incidence ranging from 1% to 4.6% of all pediatric fractures,^1-4 and acetabular fractures make up only 0.8% to 15% of pelvic fractures.^1,3,5,6 Acetabular fractures are so uncommon in children partly because of the cartilaginous nature of the immature acetabulum. The increased cartilage volume relative to adults provides greater capacity for energy absorption, resulting in greater elastic and plastic deformation before fracture occurrence. More force is therefore required to cause a fracture, and associated visceral injuries, head injuries, and long-bone fractures are common.^3,7,8

The impact of acetabular fractures on adolescents warrants special attention because any resulting disability will affect them during their most productive years. Both avascular necrosis (AVN) and degenerative arthritis are particularly devastating complications in this age group. Complications such as premature physeal closure^9-15 are unique to adolescents, and there is little information available on how injury in older children affects growth in this area.

There have been very few studies of the outcomes of these injuries in children. Mostly, there have been case reports and small series primarily dealing with nonoperative management of acetabular fractures in adolescents.^{3,10,11,16-20} By contrast, operative treatment of acetabular fractures in adults has been well described, and outcomes widely reported. As a result, much of our knowledge about managing these injuries is extrapolated from the adult literature. Although treatment of acetabular fractures in adults has evolved substantially, treatment of these injuries in adolescents remains primarily nonoperative. We conducted a study to evaluate outcomes of treatment of adolescent acetabular fractures.

Patients and Methods

After obtaining institutional review board approval for this study, we retrospectively reviewed the cases of all adolescent patients admitted with a diagnosis of acetabular fracture to 2 academic institutions between 1991 and 2003. Thirty-eight patients (28 males, 10 females) were identified. Mean age at time of injury was 15 years (range, 11-18 years). Mean follow-up was 3.2 years (range, 5-180 months).

Data on fracture types, treatment methods, associated injuries, complications, union rates, pain, and return to normal activities were collected. Acetabular fractures were classified according to the system of Letournel and Judet.²¹ There were 20 elementary and 18 associated fractures.

Of the 38 patients, 30 sustained high-energy trauma in motor vehicle accidents (25) or in falls from significant heights (5). The other 8 patients injured themselves playing sports (4 had severe traumatic brain injury, 2 had labial wounds, and 2 had injuries involving the abdominal viscera). Twelve patients had associated pelvic ring injuries, 18 had femoral head dislocations, 2 had femoral head fractures, and 13 had evidence of impaction injury to the femoral head articular cartilage. Twelve patients had marginal impaction of the acetabular wall. Fifteen patients had open triradiate physes at time of injury (Table 1).

Thirty-seven of the 38 patients were treated with open reduction and internal fixation (ORIF) by an experienced orthopedic trauma surgeon; 1 patient with a stable posterior wall fracture was treated nonoperatively. Surgical indications were articular displacement of more than 1 mm, hip joint instability, irreducible hip dislocation, and intra-articular fracture fragments. In the 37 surgically treated cases, the approaches used were Kocher-Langenbeck (22), ilioinguinal (8), combined Kocher-Langenbeck/ilioinguinal (5), and triradiate (2).

Immediate postoperative radiographs were evaluated by 3 orthopedic surgeons blinded to the patients’ clinical outcomes. Displacement was evaluated on anteroposterior (AP) and Judet views of the pelvis, as described by Matta,²² and reductions were classified as anatomical (0-1 mm of displacement), imperfect (>1 to 3 mm), poor (>3 mm), or surgical secondary congruence (Table 2).

Results

Thirty-seven patients underwent acetabular fracture ORIF. Immediate postoperative radiographs showed 30 anatomical reductions and 7 imperfect reductions. One patient had surgical secondary congruence and developed AVN of the hip. We could not identify an association between the quality of the reduction and the outcome with respect to pain or return to activity. However, no patient had a poor reduction. An illustrative case is presented in Figures 1 to 4.

All acetabular fractures united within 4.5 months (range, 3.0-8.0 months) after the index procedure. Early postoperative complications included 3 cases of meralgia paresthetica and 13 cases of abductor weakness. Meralgia paresthetica resolved spontaneously in all 3 patients. Of the 13 patients with abductor weakness, 11 improved with physical therapy, 1 was limited by the head injury, and 1 subsequently underwent hip fusion. One patient had a deep vein thrombosis (DVT) that was identified before surgery and managed with warfarin.

Other complications included 1 case of deep infection of the surgical wound. This infection presented 4 months after surgery and was treated with débridement, hardware removal, and a 3-month course of antibiotics. Two patients who sustained hip dislocations at time of injury developed AVN of the femoral head. Both developed osteoarthritis, and 1 underwent hip fusion. Eight patients developed heterotopic ossification on the side of the acetabular fracture; 4 of them underwent surgical excision. Four patients required a separate operation for hardware removal. Four patients with triradiate cartilage involvement went on to premature closure. No patient had any leg-length discrepancy or dysplasia at time of follow-up.

Thirty-four of the 38 patients returned to their regular activities. For these patients, mean time to return to full activity was 7.0 months (range, 3-30 months); there was no difference in mean time to return to full activity between skeletally mature and skeletally immature patients (6.6 vs 7.4 months; P = .57). Of the other 4 patients, 1 had permanent cognitive and physical disability with an ataxic gait as a result of a traumatic brain injury, 2 were limited by AVN (1 underwent hip fusion), and 1 was limited by an ipsilateral knee injury.

Of the 38 patients, 29 were pain-free; 6 had occasional, intermittent mild pain that did not limit their activities; and 3 had severe, activity-limiting pain. Of the 6 patients with mild pain, 2 had femoral impaction injuries, and 4 had marginal impaction injuries. Of the 3 patients with severe pain, 2 developed femoral head AVN, and 1 had multiple ipsilateral extremity injuries involving the femur, knee, and tibia.

Discussion

The traditional treatment for acetabular fractures in children has been nonoperative,^8,10 and there are few specific treatment guidelines.¹³ Recent recommendations are nonoperative treatment for minimally displaced fractures (<1 mm) and acetabular fracture ORIF for fractures displaced more than 2 mm.¹¹ No clear consensus exists on management for fractures displaced 1 to 2 mm. Few studies have investigated the outcomes of operative management of these fractures in the pediatric or adolescent population.

In our series of adolescent acetabular fractures, we examined unions, complications, and return to activity. Of 38 patients with acetabular fractures, 37 were treated with ORIF. Anatomical reduction was achieved in the majority of patients. Posterior wall fractures were by far the most common fracture type, which is consistent with previous reports.^10,11 All acetabular fractures united, and most patients were pain-free at latest follow-up. There was a low incidence of major complications in our patient population. One major complication was a DVT in a 14-year-old boy who was in a motor vehicle accident and sustained a T-type fracture of the right acetabulum with contralateral femoral shaft and ankle fractures. The DVT was in the right internal iliac and common femoral veins and was diagnosed on magnetic resonance venography. The patient was treated with warfarin for 3 months without incident.

Two patients developed AVN of the femoral head. One of these patients was an 11-year-old girl who was in a motor vehicle accident and sustained a T-type fracture with marginal impaction of the posterior wall, posterior hip dislocation, and a pelvic ring injury. She was treated with ORIF through combined Kocher-Langenbeck/ilioinguinal approaches. By 4 months after surgery, the acetabular fracture was united. Nine months after surgery, she still had pain (activity-limiting) and a 35° flexion contracture of the hip, and she was ambulating with a cane. The diagnosis was AVN of the hip. The patient underwent hip fusion 1 year after surgery.

The second patient with femoral head AVN was a 12-year-old boy who fell while skiing and sustained a fracture of the posterior wall and a hip dislocation with impaction of the femoral head. Initial treatment at an outside institution consisted of open reduction of the hip and excision of a “loose body” from the joint. Eight weeks after surgery, the patient continued to have pain and was referred to our institution. A second operation was performed. Findings included a defect involving 40% of the posterior wall, and signs that the posterior wall had been excised during the initial operation. The patient eventually developed AVN of the hip. This patient was also diagnosed with a deep wound infection 4 months after surgery. He presented with pain and a fluid collection around the hip. The infection was not confirmed through fluid culture, and, as he eventually developed AVN of the hip, his symptoms may have been the result of chondrolysis or AVN rather than infection.

There were no cases of nonunion or malunion, leg-length discrepancy, or permanent sciatic nerve palsy. Although there were a few cases of premature closure of the triradiate cartilage, no acetabular dysplasia was seen at latest follow-up, likely because of the relative maturity of our pediatric group (age range, 11-18 years). Age at time of injury is thought to be the most important factor influencing growth and development of the acetabulum.^9,13 In addition, previous studies have demonstrated a tendency toward acetabular fractures in patients with mature triradiate cartilage—versus pelvic ring injuries in patients with immature triradiate cartilage.^8,11 This may also account for the older age of our study group.

Minor complications (eg, meralgia paresthetica) resolved spontaneously. The most common complications were abductor weakness and heterotopic ossification. In only 4 cases was a secondary procedure for excision of the heterotopic bone required. Abductor weakness, more commonly associated with a Kocher-Langenbeck approach to the hip, resolved with therapy in almost all cases. Only 4 of our patients required removal of hardware from the acetabulum.

Although the majority of acetabular fractures resulted from high-energy trauma, 8 cases were sports-related. Six of these involved posterior wall fractures, suggesting the injury resulted from a fall on flexed knee and hip. This was not known to be a common mechanism of injury in this age group.^3,7 An additional concern was how to size the posterior wall fragment when not ossified. At one center, preoperative magnetic resonance imaging (MRI) was effectively used to size the osteochondral posterior wall fragment as standard protocol for patients with posterior wall fractures in this age group—resulting in better decisions regarding the need for ORIF. At the other institution, preoperative MRI was not performed routinely for this subset of patients.

Thirty-four of our 38 patients returned to their normal activities. Of the other 4 patients, 1 was permanently disabled secondary to traumatic brain injury, 1 had other ipsilateral extremity injuries that limited his mobility, and 2 developed AVN of the femoral head. Both patients with AVN had hip dislocations. Four of the 6 patients who were symptomatic during activity sustained impaction injuries of the femoral head or posterior wall. This suggests that poorer outcomes may be associated with dislocation or with articular injuries—similar to what has been reported in the adult literature.

This study had several limitations. First, it was a retrospective case series, so there was no control group for comparison. Second, the relatively short follow-up did not allow evaluation of the incidence of degenerative arthritis secondary to articular injury, the symptoms of which may develop 1 to 2 decades after injury.¹³ This phenomenon was well described by Letournel and Judet²¹ in the adult population, and there is no reason to presume the adolescent population is any different. Third, our sample was small and unlikely to represent a uniform sampling of the general pediatric population. Fourth, it was not possible to draw detailed conclusions about the outcome of ORIF for a particular type of acetabular fracture. Fifth, we did not see as many of the associated visceral injuries that are so prevalent in the literature. This may reflect improvement in safety specifications for automobiles, or our group may not have had the most severe or high-energy injuries. Here our population sample may have skewed our results, leading to better than expected outcomes.

One last study limitation, a major one, was the age of our population, 11 to 18 years, which makes it difficult to extrapolate results to the entire pediatric population. On one hand, a more immature skeleton has a higher chance of remodeling and is more forgiving of deformities and small amounts of displacement. On the other hand, injury and premature triradiate cartilage fusion in a younger patient can lead to significant deformity and acetabular dysplasia.⁹ Whether ORIF of these fractures would alter the outcome of an injury to the triradiate cartilage is yet to be determined.

Conclusion

In agreement with earlier studies,^10,11,15,18 the good outcomes in our series correlated with congruence of reduction. Outcome predictors such as dislocation, femoral head injury, and marginal impaction are similar to those described in the adult literature. Although our study did not have a nonoperative group for comparison, the favorable outcomes of ORIF of acetabular fractures suggest that a more aggressive approach to treatment should be considered. Given the added benefits of early, pain-free mobilization, we think that only stable, undisplaced fractures (<1 mm) should be managed nonoperatively. In the adolescent population, we recommend ORIF for optimal management of unstable acetabular fractures, fractures with any hip subluxation, and fractures displaced more than 1 mm.