Introduction
Trauma is the most common reason for wild birds to be presented to wildlife centers,1 and collisions with windows are among the leading causes of trauma. Window collisions often result in fractures or luxations of the bones of the pectoral girdle (coracoid, clavicle, and scapula).2 Birds with coracoid fractures or, to a certain extent, fractures of the other pectoral girdle bones will be unable to fly or will only be able to fly short distances without achieving appropriate lift.3 Making a clinical diagnosis of a coracoid fracture or a fracture of the other pectoral girdle bones can be difficult, as affected birds may only have a minor wing droop or no clinical signs at all.4 Thus, the diagnosis is usually made through a combination of palpation and diagnostic imaging (eg, radiography).5
Radiographic examination of the pectoral girdle in birds is challenging because of superimposition of the pectoral girdle bones on standard ventrodorsal (VD) and dorsoventral (DV) radiographic views. A recent study6 showed that the addition of a caudoventral-to-craniodorsal oblique view made at 45° to the frontal plane (Cd45V-CrD oblique view or H view) to the standard VD view could increase the chances of correctly identifying fractures of the pectoral girdle. However, that study was limited by a low number of patients (n = 24) and lack of postmortem examination in many cases, and birds included in that study consisted only of raptors ranging in size from an eastern screech owl (Megascops asio) to a bald eagle (Haliaeetus leucocephalus). In addition, the generator that was used allowed for redirection of the beam in horizontal and oblique angles, which may not be possible or practical in clinical practice. The objectives of the study reported here were to determine the prevalence of pectoral girdle fractures in wild passerines found dead following presumed window collision and evaluate the diagnostic accuracy of various radiographic views for diagnosis of pectoral girdle fractures in these birds.
Materials and Methods
Animals
Cadavers collected by one of the authors (TJO) as part of a study assessing the impact of window design on frequency of window collision (data not shown) were used in the study. Birds of the order Passeriformes found beneath windows at sites being monitored for that study were collected, labeled as to species and sex (if known), and stored at–20 °C in a freezer. Cadavers were frozen for up to 7 years. Cadavers were allowed to thaw in a refrigerator for 12 to 24 hours prior to diagnostic imaging and necropsy.
Diagnostic imaging
A portable x-ray generator (HF120/60HPPWV PowerPlusTM; MinXray Inc.) mounted on a custom stand was used to obtain radiographic images of each cadaver, with exposure variables of 50 kVp and 0.6 mAs used for all radiographic images. Images were stored in a picture-archiving and communication system (eFilm Workstation 4.0; Merge Healthcare). For all imaging, the plate was placed 91 cm away from the x-ray tube, as measured with a tape measure built into the machine. Prior to placement of the cadavers on the radiographic detector plate, the wings, legs, and neck were gently flexed and extended to increase flexibility and allow for proper positioning. All imaging was performed by a single individual (CLM).
Seven radiographic views (VD, DV, right lateral, and 4 oblique views) were taken of each cadaver (Figure 1). For both the VD and DV projections, the bird was placed on the plate with wings held in an extended position and the shoulder joints level with the horizontal line of the crosshair markings at the center of the plate. For both positions, the mandible was positioned as parallel with the plate as possible when securing the head. For the right lateral position, a 1-inch-thick foam wedge was placed between the wings at the level of the carpus or dorsum to hold the wings superimposed over one another. The legs were taped down overlapping one another as much as possible so they would also be superimposed. After each image was obtained, the operator made a brief, subjective interpretation of the bird's alignment and adjusted the bird and collected additional images as necessary to ensure appropriate symmetry.
Diagram of the setup used to obtain ventrodorsal (VD; A), dorsoventral (DV; B), right lateral (C), and 4 oblique radiographic views of a bird. For 2 of the oblique views, the bird was taped to the radiographic plate, which was propped up at a 45° angle with foam wedges while the x-ray tube was positioned vertically above it. The bird's head was placed at the top of the plate, and radiographic images were obtained with the bird in ventral recumbency, resulting in a craniodorsal-to-caudoventral oblique view made at 45° to the frontal plane (Cr45D-CdV oblique view; D), and in dorsal recumbency, resulting in a cranioventral-to-caudodorsal oblique view made at 45° to the frontal plane (Cr45V-CdD oblique view; E). For the remaining 2 oblique views, the bird was taped to the radiographic plate, which was laid flat on the table while the x-ray tube was positioned at a 45° angle. Radiographic images were obtained with the bird in dorsal recumbency, resulting in a caudoventral-to-craniodorsal oblique view made at 45° to the frontal plane (Cd45V-CrD oblique view; F), and in ventral recumbency, resulting in a caudodorsal-to-cranioventral oblique view made at 45° to the frontal plane (Cd45D-CrV oblique view; G).
Citation: Journal of the American Veterinary Medical Association 260, 6; 10.2460/javma.20.11.0642
In addition to the standard VD, DV, and right lateral radiographic views, 4 oblique views were obtained. Two of these consisted of the cadaver being taped to the radiographic plate, which was propped up on foam wedges on the table while the x-ray tube was positioned vertically above it. A protractor was used to ensure that the plate was at a 45° angle with the table. The bird's head was placed at the top of the plate to simulate clinical conditions, and radiographic images were obtained with the bird in both ventral recumbency, resulting in a craniodorsal-to-caudoventral oblique view made at 45° to the frontal plane (Cr45D-CdV oblique view), and dorsal recumbency, resulting in a cranioventral-to-caudodorsal oblique view made at 45° to the frontal plane (Cr45V-CdD oblique view).
The remaining 2 oblique views consisted of the cadaver being taped to the radiographic plate, which was laid flat on the table while the x-ray tube was positioned at a 45° angle with the plate, as measured with markers on the machine. With this setup, images were obtained with the bird in both ventral recumbency, resulting in a caudodorsal-to-cranioventral oblique view made at 45° to the frontal plane (Cd45D-CrV oblique view), and dorsal recumbency, resulting in a caudoventral-to-craniodorsal oblique view made at 45° to the frontal plane (Cd45V-CrD oblique view [H view]).
For all radiographic views, the bird was positioned at the center markings on the plate, and the crosshairs of the collimator were positioned over the bird's pectoral girdle. Radiographs were not evaluated at the time they were obtained and were only assessed to ensure proper alignment.
Necropsy
After collection of radiographs, a gross necropsy was performed on each bird by a single individual (CLM), with specific focus on the integrity of the coracoids, clavicles, and scapulae. The bird's weight was measured with a gram scale accurate to 0.01 g (Shenzhen Amier Technology Co). Species information based on an identification made at the time of cadaver collection was recorded, and a brief physical exam was performed on each bird. The wings, legs, and pectoral girdles were inspected and palpated to detect hematomas and obvious fractures. The mouth was inspected for signs of hemorrhage. Once this information was recorded, the pectoral girdle bones were dissected and examined for integrity. Each individual bone was recorded as fractured or intact.
Radiographic interpretation
Following completion of the necropsies, each radiographic image was assigned a random number with an online randomizer (www.randomizer.org). Each image was then assessed with an open medical image viewer (Horos version 2.3.0; Horos Project) in random order by a single individual (CLM) with all identifying features (including view, position, species, and original order) masked. Each bone of the pectoral girdle was classified as fractured or intact. Because right and left bones were not distinguishable on the lateral radiographic view, paired bones were interpreted together.
Statistical analysis
Frequency data were reported as counts and percentages. Sensitivity, specificity, positive predictive value, negative predictive value, and overall accuracy and their respective 95% CIs were calculated as measures of diagnostic accuracy with an online calculator (Diagnostic test evaluation calculator; MedCalc Software Ltd). Diagnostic accuracy of each radiographic view was calculated (1) for each specific bone and (2) for the pectoral girdle as a whole (ie, having at least 1 fracture of the pectoral girdle vs having no fractures of the pectoral girdle). Diagnostic accuracy for the pectoral girdle as a whole was calculated on the basis of the clinical concept that treatment would be similar for passerine birds with any fractures of the pectoral girdle, but likely different from treatment for birds with no fractures of the pectoral girdle. Diagnostic accuracy was calculated for each of the 7 individual radiographic views and for 8 combinations of radiograph views. Each combination included the lateral view, the VD or DV view, and 1 of the 4 oblique views.
Results
Population
One hundred five cadavers collected as window collision specimens were considered for inclusion in the study. One was excluded because it belonged to the order Piciformes, and another was excluded because it had been found alongside a road and was suspected to have died as a result of car impact trauma. The remaining 103 cadavers fit the inclusion criteria and were included in the study.
Median weight of birds included in the study was 14 g (range, 5 to 46 g; SD, 6.6 g). Lincoln's sparrows (Melospiza lincolnii) were the most frequently represented species (n = 25), followed by painted buntings (Passerina ciris; 13). The remaining cadavers consisted of clay-colored sparrows (Spizella pallida; n = 7), grasshopper sparrows (Ammodramus savannarum; 7), Swainson's thrushes (Catharus ustulatus; 7), and 26 other species represented by ≤ 4 individuals each (Supplementary Table S1).
Necropsy findings
At necropsy, 14 of the 103 (13.6%) birds had at least 1 coracoidal fracture, 18 (17.5%) had at least 1 clavicular fracture, and 41 (39.8%) had at least 1 scapular fracture (Supplementary Table S2). Overall, 56 of the 103 (54.4%) birds had at least 1 fracture of the pectoral girdle. Most birds (n = 46) had unilateral fractures, but 10 birds had bilateral fractures. The right coracoid, scapula, and clavicle were affected in 10, 25, and 9 birds, respectively, whereas the left coracoid, scapula, and clavicle were affected in 6, 21, and 10 birds, respectively.
Diagnostic accuracy for individual bone fractures
Percentages of true positive, true negative, false positive, and false negative results when each individual radiographic view was used to identify fractures of each bone of the pectoral girdle were calculated (Figure 2;Supplementary Table S3). Some views had > 20% false negative results, including the VD view when used to diagnose fractures of the right scapula, the DV view when used to diagnose fractures of both scapulae, the lateral view when used to diagnose fractures of both scapulae, and the Cd45D-CrV view when used to diagnose fractures of the right scapula. Some views had > 5% false positive results, including the DV view when used to diagnose fractures of the left clavicle and the Cd45V-CrD view when used to diagnose fractures of the left scapula.
Bar plots representing percentages of true positive (black), true negative (white), false positive (dark gray), and false negative (light gray) results when 7 individual radiographic views were used to diagnose fractures of the scapula (A), clavicle (B), and coracoid (C) in 103 wild passerines that presumptively died as a result of window collisions. Lat = Lateral. See Figure 1 for remainder of key.
Citation: Journal of the American Veterinary Medical Association 260, 6; 10.2460/javma.20.11.0642
Diagnostic accuracy for pectoral girdle fractures
Results of examination of each radiographic view were cross tabulated with necropsy findings (Table 1), and numbers of fractures identified with each view were summarized (Supplementary Table S2). When individual radiographic views were used to classify birds as having or not having a pectoral girdle fracture, sensitivity ranged from 21.3% to 51.1%, specificity ranged from 85.7% to 100.0%, and overall accuracy ranged from 63.1% to 72.8% (Table 2). When combinations of radiographic views were used to classify birds as having or not having a pectoral girdle fracture, sensitivity ranged from 51.1% to 66.0%, specificity ranged from 76.8% to 96.4%, and overall accuracy ranged from 68.0% to 79.6% (Table 3).
Cross-tabulation of results of examination of 7 radiographic views for fractures of the pectoral girdle versus necropsy findings (gold standard) for 103 wild passerines that presumptively died as a result of window collisions.
| Radiographic projection | Necropsy | |
|---|---|---|
| Fracture present | Fracture absent | |
| VD | ||
| Fracture detected | 17 (16.5) | 0 (0.0) |
| Fracture not detected | 30 (29.1) | 56 (54.4) |
| DV | ||
| Fracture detected | 21 (20.4) | 5 (4.8) |
| Fracture not detected | 26 (25.2) | 51 (49.5) |
| Right lateral | ||
| Fracture detected | 10 (9.7) | 1 (1.0) |
| Fracture not detected | 37 (35.9) | 55 (53.4) |
| Cr45V-CdD | ||
| Fracture detected | 21 (20.4) | 2 (1.9) |
| Fracture not detected | 26 (25.2) | 54 (52.4) |
| Cr45D-CdV | ||
| Fracture detected | 20 (19.4) | 5 (4.8) |
| Fracture not detected | 27 (26.2) | 51 (49.5) |
| Cd45V-CrD | ||
| Fracture detected | 24 (23.3) | 8 (7.8) |
| Fracture not detected | 23 (22.3) | 48 (46.6) |
| Cd45D-CrV | ||
| Fracture detected | 18 (17.5) | 3 (2.9) |
| Fracture not detected | 29 (28.2) | 53 (51.5) |
Data are given as number (%). At necropsy, 56 of the 103 (54.4%) birds had at least 1 fracture of the pectoral girdle, and 47 (45.6%) did not have any fractures of the pectoral girdle.
Cr45D-CdV = Craniodorsal-to-caudoventral oblique view made at 45° to the frontal plane. Cd45D-CrV = Caudodorsal-to-cranioventral oblique view made at 45° to the frontal plane. Cr45V-CdD = Cranioventral-to-caudodorsal oblique view made at 45° to the frontal plane. Cd45V-CrD = Caudoventral-to-craniodorsal oblique view made at 45° to the frontal plane. DV = Dorsoventral. VD = Ventrodorsal.
Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and overall accuracy of 7 individual radiographic views used to diagnose fractures of the pectoral girdle in 103 wild passerines that presumptively died as a result of window collisions.
| Radiographic view | Sensitivity | Specificity | PPV | NPV | Accuracy |
|---|---|---|---|---|---|
| VD | 36.2 (22.7–51.5) | 100.0 (93.6–100.0) | 100.0 (NC) | 65.1 (60.1–69.8) | 70.9 (61.1–79.4) |
| DV | 44.7 (30.2–59.9) | 91.1 (80.4–97.0) | 80.8 (63.2–91.1) | 66.2 (60.0–72.0) | 69.9 (60.1–78.5) |
| Right lateral | 21.3 (10.7–35.7) | 98.2 (90.4–99.9) | 90.9 (57.0–98.7) | 59.8 (56.1–63.4) | 63.1 (53.0–72.4) |
| Cr45V-CdD | 44.7 (30.1–59.9) | 96.4 (87.7–99.6) | 91.3 (72.2–97.7) | 67.5 (61.5–73.0) | 72.8 (63.2–81.1) |
| Cr45D-CdV | 42.6 (28.3–57.8) | 91.1 (80.4–97.0) | 80.0 (61.9–90.8) | 65.4 (59.3–71.0) | 68.9 (59.1–77.7) |
| Cd45V-CrD | 51.1 (36.1–65.9) | 85.7 (73.8–93.6) | 75.0 (59.8–85.8) | 67.6 (60.5–74.0) | 69.9 (60.1–78.6) |
| Cd45D-CrV | 38.3 (24.5–53.6) | 94.6 (85.1–98.9) | 85.7 (65.3–95.0) | 64.3 (59.1–69.8) | 68.9 (59.1–77.7) |
Sensitivity, specificity, PPV, NPV, and overall accuracy of 8 combinations of radiographic views used to diagnose fractures of the pectoral girdle in 103 wild passerines that presumptively died as a result of window collisions.
| Combination of views | Sensitivity | Specificity | PPV | NPV | Accuracy |
|---|---|---|---|---|---|
| Lateral, VD, and Cr45V-CdD | 59.6 (44.3–73.6) | 96.4 (87.7–99.6) | 93.3 (77.9–98.2) | 74.0 (66.7–80.1) | 79.6 (70.5–86.9) |
| Lateral, VD, and Cr45D-CdV | 53.2 (38.1–67.9) | 91.1 (80.4–97.0) | 83.3 (67.5–92.3) | 69.9 (62.8–76.1) | 73.8 (64.2–82.0) |
| Lateral, VD, and Cd45V-CrD | 57.5 (42.2–71.7) | 85.7 (73.8–93.6) | 77.1 (62.9–87.0) | 70.6 (62.9–77.3) | 72.8 (63.2–81.1) |
| Lateral, VD, and Cd45D-CrV | 53.2 (38.1–67.9) | 94.6 (85.1–98.9) | 89.3 (72.9–96.3) | 70.7 (63.8–76.7) | 75.7 (66.3–83.6) |
| Lateral, DV, and Cr45V-CdD | 61.7 (46.4–75.5) | 87.5 (75.9–94.8) | 80.6 (66.7–90.0) | 73.1 (65.1–80.0) | 75.7 (66.3–83.6) |
| Lateral, DV, and Cr45D-CdV | 51.1 (36.1–65.9) | 82.1 (69.6–91.1) | 70.6 (56.2–81.8) | 66.7 (59.3–73.3) | 68.0 (58.0–76.8) |
| Lateral, DV, and Cd45V-CrD | 66.0 (50.7–79.1) | 76.8 (63.6–87.0) | 70.5 (58.7–80.0) | 72.9 (63.8–80.4) | 71.8 (62.1–80.3) |
| Lateral, DV, and Cd45D-CrV | 55.3 (40.1–69.8) | 85.7 (73.8–93.6) | 76.5 (62.0–86.6) | 69.6 (62.0–76.2) | 71.8 (62.1–80.3) |
Discussion
Results of the present study suggested that radiography had limited diagnostic accuracy for identifying fractures of the pectoral girdle in passerine birds suspected to have died as a result of window collisions. In particular, radiographic diagnoses had only moderate agreement with necropsy diagnoses, with the overall accuracy of radiography ranging from 63.1% to 79.6%. Both individual radiographic views and combinations of views resulted in several false negative but few false positive results when classifying birds as having or not having a fracture of the pectoral girdle. The oblique views evaluated in the present study were not associated with conspicuous improvements in diagnostic accuracy when compared with the accuracy of standard VD and DV views or with each other. Additionally, the method of acquiring oblique images (ie, whether the machine or radiographic plate was angled) did not have an impact on the diagnostic accuracy of those images.
The design of the pectoral girdle in birds is one of the most important anatomic features facilitating flight. The coracoids serve as supportive struts for the wings and help prevent collapse of the thoracic cavity during downward wing strokes.4 The clavicles serve as a strut between the bird's shoulder joints and can act as elastic springs to assist in flight and respiration.7 The scapulae provide support for the thoracic cavity during flight and are angled with the coracoids to allow more force to lift the wings.7 Together, these 3 bones articulate to form the triosseal canal, through which runs the tendon of the supracoracoideus muscle, which is the principal muscle used to produce the upstroke during flight.7 The humerus also articulates with the bones of the pectoral girdle to form the shoulder joint. Fractures to these bones are common and have a substantial impact on the flight capacity of birds.
The present study found a higher incidence (56/103 [54.4%]) of pectoral girdle fractures than did the study by Visser et al6 (9/24 [37.5%]), which investigated raptors that were the victims of suspected trauma. Consequently, the present study also reported a higher incidence of each type of fracture, with the incidence of scapular fractures being more than double the findings in the previous study. However, the cadavers used in the present study were presumptive natural fatalities; therefore, the authors cannot rule out the possibility that some of the fractures were the result of postmortem trauma or were unrelated to window collision.
One previous study8 has evaluated the frequencies and types of injuries in birds of various species that had collided with windows or towers, with a focus on head and neck injuries. That study found that nearly every bird (98% to 99%) had evidence of subdermal intracranial hemorrhage and suggested that hemorrhage within and around the brain was the likely cause of death in most collision fatalities. All the birds examined in this previous study were evaluated superficially and found to have no injuries beyond those sustained to the head or neck. Unlike the previous study, we found more than half the subjects had at least 1 pectoral girdle fracture. Many of these fractures would likely not have been evident on an external exam, which serves to emphasize the need for diagnostic testing, such as radiographic imaging, to obtain an accurate diagnosis for these patients.
A study by Cousins et al2 investigating window collisions and other trauma in New Zealand pigeons, also known as kererū (Hemiphaga novaeseelandiae), found a higher incidence of pectoral girdle fractures (25/40 [62.5%]) than the present study. That study also found that 24 of 40 (60%) birds that died as a result of window collisions had fractures of the coracoids, and 20 (50%) had fractures of the clavicles. The authors hypothesized that the high mass of those birds (mean, 570 g) would lead to an increased strike force and, therefore, a higher propensity for fractures.2 The body shape of Columbiformes leads to a higher density with respect to their length when compared with raptors, which may explain why the study by Cousins et al2 found higher incidences of fracture than the study by Visser et al,6 despite examining birds of a similar weight class.
Overall, the present study found that 45° oblique views, independent of recumbency or generator orientation, had similar accuracy in identifying fractures of the pectoral girdle when compared with the standard VD and DV views. The confidence intervals for sensitivity and specificity of each view had moderate overlap with those of the other views.
Additionally, whether the bird was raised to a 45° angle or the x-ray tube was slanted at a 45° angle did not have a substantial effect on diagnostic accuracy. This suggested that angling the patient on a foam wedge could be used to obtain an oblique view similar to the H view described by Visser et al6 without requiring a moveable x-ray generator. Because both methods could be used to obtain CrV-CdD and CdV-CrD oblique views, the images acquired by each method should be nearly identical. The primary differences would arise on the basis of which side of the bird was against the plate (ie, dorsal vs ventral recumbency), because objects in closer proximity to the radiographic plate have decreased scatter and decreased magnification artifact, resulting in a sharper image. Given the small body size of the birds used in the present study, artifacts due to scatter and magnification were likely negligible, and recumbency of the patient likely had minimal impact on image quality. However, in larger patients, it may be desirable to preferentially angle the patient or the x-ray machine depending on which aspect of the pectoral girdle is more of a concern. For example, it may be preferable to select an oblique image that allows for the patient to remain in dorsal recumbency if the scapulae are the primary concern.
Notably, several radiographic views in the present study had ≥ 20% false negative results, meaning 1 in 5 patients would incorrectly be declared free of fractures. This serves to emphasize the importance of taking multiple views of every patient and considering all data available when making a diagnosis of pectoral girdle fractures. Clinically, missing a pectoral girdle fracture could result in the release of a bird unable to fly effectively in the wild, which could be a fatal error for that patient. Alternatively, false positive results (sometimes ≥ 5%) could result in unnecessary hospitalization and bandaging of healthy patients, leading to increased cost and patient stress. However, specificity was generally high, indicating that if a fracture was suspected on the basis of radiographic images, it was likely that a fracture was truly present.
Combinations of multiple radiographic views had high specificity and low sensitivity in the present study. Overall, the combination with the highest accuracy was the combination of right lateral, VD, and Cr45V-CdD oblique views. This combination had higher accuracy, sensitivity, and negative predictive value than any individual view. The specificity and positive predictive value were higher than values for any other combination but were not higher than values for each of the individual views. These results should be interpreted with caution, considering that a formal statistical comparison of the diagnostic accuracy of the various views was not performed to avoid potential type I statistical errors and that the confidence intervals for overall accuracy of the various views mostly overlapped.
There were several limitations to the present study that must be addressed. One limitation was that the cadavers used in this study were collected as presumptive window collision fatalities, but other potential causes of death and postmortem trauma could not be excluded. In addition, any window collision victims that survived the initial impact and either died elsewhere or survived could not be included. Additionally, even though necropsy was used as the gold standard in the present study, it is possible that some fractures could have been missed, leading to some radiographic diagnoses being incorrectly classified as false positive results. Another limitation was that data collection, radiographic imaging, and radiographic image interpretation were all performed by a single observer. This observer was a veterinary student who had received training in avian radiology and gross necropsy to help ensure consistent technique. However, it is possible that use of a more experienced observer or multiple observers would have provided different results. Future research efforts could focus on other orders of birds, greater size variability, and the impact of various observers on data collection and interpretation.
Supplementary Materials
Supplementary materials are posted online at the journal website: avmajournals.avma.org
Acknowledgments
This study was funded by the Morris Animal Foundation, the Oklahoma State University Summer Research Scholars Program, and the Dr. Kristie Plunkett Fund for Exotic Animal Medicine. Additional support for Dr. O’Connell came from the USDA National Institute of Food and Agriculture McIntire Stennis project (project No. 1023033). Avian cadavers were collected and curated under Oklahoma Department of Wildlife Conservation Scientific Collection Permits 6623, 6961, and 7390.
The authors thank Ms. Laura Siegel for the drawings used in Figure 1.
References
- 1. ↑
Molina-López RA, Casal J, Darwich L. Causes of morbidity in wild raptor populations admitted at a wildlife rehabilitation centre in Spain from 1995–2007: a long term retrospective study. PLoS One. 2011;6(9):e24603. doi: 10.1371/journal.pone.0024603
- 2. ↑
Cousins RA, Battley PF, Gartrell BD, Powlesland RG. Impact injuries and probability of survival in a large semiurban endemic pigeon in New Zealand, Hemiphaga novaeseelandiae. J Wildl Dis. 2012;48(3):567–574.
- 3. ↑
Davidson JR, Mitchell MA, Ramirez S. Plate fixation of a coracoid fracture in a bald eagle (Haliaeetus leucocephalus). J Avian Med Surg. 2005;19(4):303–308.
- 5. ↑
Scheelings TF. Coracoid fractures in wild birds: a comparison of surgical repair versus conservative treatment. J Avian Med Surg. 2014;28(4):304–308.
- 6. ↑
Visser M, Hespel A-M, de Swarte M, Bellah JR. Use of a caudoventral-craniodorsal oblique radiographic view made at 45° to the frontal plane to evaluate the pectoral girdle in raptors. J Am Vet Med Assoc. 2015;247(9):1037–1041.
- 7. ↑
Beaufrère H. A review of biomechanic and aerodynamic considerations of the avian thoracic limb. J Avian Med Surg. 2009;23(3):173–185.
- 8. ↑
Veltri CJ, Klem D Jr. Comparison of fatal bird injuries from collisions with towers and windows. J Field Ornithol. 2005;76(2):127–133.
