Accepted Manuscript Title: Location of Civilian Ballistic Femoral Fracture Indicates Likelihood of Arterial Injury Author: Leah Gitajn Paul Perdue John Hardcastle Robert V. O’Toole PII: DOI: Reference:
S0020-1383(14)00258-7 http://dx.doi.org/doi:10.1016/j.injury.2014.05.020 JINJ 5750
To appear in:
Injury, Int. J. Care Injured
Received date: Revised date: Accepted date:
7-10-2013 2-5-2014 18-5-2014
Please cite this article as: Gitajn L, Perdue P, Hardcastle J, O’Toole RV, Location of Civilian Ballistic Femoral Fracture Indicates Likelihood of Arterial Injury, Injury (2014), http://dx.doi.org/10.1016/j.injury.2014.05.020 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
1 2 3
Location of Civilian Ballistic Femoral Fracture Indicates Likelihood of Arterial Injury
Leah Gitajn, BS, Paul Perdue, BS, John Hardcastle, MD, and Robert V. O’Toole, MD
cr
5
ip t
4
6
From R Adams Cowley Shock Trauma Center, Department of Orthopaedics, University of Maryland School of Medicine, Baltimore, MD, United States
us
7 8
an
9 10
M
11 12
d
13
te
14
16 17 18 19
Ac ce p
15
Corresponding author:
20
Robert V. O’Toole, MD
21
R Adams Cowley Shock Trauma Center
22
Department of Orthopaedics
23
University of Maryland School of Medicine
24
22 S. Greene Street, Suite T3R62
25
Baltimore, MD 21201.
26
Tel.: 410.328.6292; fax: 410.328.2893.
27
E-mail address:
[email protected] 28 1 Page 1 of 30
ABSTRACT
29
Background: We evaluated whether the location of a ballistic femoral fracture helps predict the
30
presence of arterial injury. We hypothesized that fractures located in the distal third of the femur
31
are associated with a higher rate of arterial injury.
32
Methods: We conducted a retrospective review of electronic medical records at our level I
33
trauma centre and found 133 consecutive patients with femoral fractures from civilian gunshots
34
from 2002 through 2007, 14 of whom sustained arterial injury. Fracture extent was measured
35
with computerized viewing software and recorded with a standard technique, calculating
36
proximal, distal, and central locations of the fracture as a function of overall length of the bone.
37
Analyses were conducted with Student’s t, Chi-squared, and Fisher’s exact tests.
38
Results: The location of any fracture line in the distal third of the femur was associated with
39
increased risk of arterial injury (P < 0.05). The odds ratio for the presence of arterial injury when
40
the proximal fracture line was in the distal third of the femur was 5.63 (95% confidence interval,
41
1.7−18.6; P < 0.05) and when the distal fracture line was in the distal third of the femur was 6.72
42
(95% confidence interval, 1.78−25.44; P < 0.05).
43
Conclusions: A fracture line in the distal third of the femur after ballistic injury is six times
44
more likely to be associated with arterial injury and warrants careful evaluation. Our data show
45
that fracture location can help alert clinicians to possible arterial injury after ballistic femoral
46
fracture.
Ac ce p
te
d
M
an
us
cr
ip t
28
2 Page 2 of 30
47
48
Introduction
Penetrating extremity trauma is commonly encountered at busy trauma centres, and diagnosing all injuries in a timely manner is critical. Of particular importance is the evaluation of
50
vascular integrity after penetrating trauma. Rapid diagnosis of arterial injury is critical to prevent
51
complications of prolonged limb ischemia, including compartment syndrome, arteriovenous
52
fistula, ischemic contracture, and limb amputation.1,2
us
cr
ip t
49
Controversy exists regarding the appropriate diagnostic tools indicated in the setting of
54
penetrating extremity trauma. Historically, all penetrating extremity trauma underwent surgical
55
exploration. More recently, angiography has been used to diagnose vascular injury in all cases of
56
extremity trauma.3−6 However, the necessity for angiography in all penetrating trauma cases has
57
become controversial, and some data suggest that “hard” signs of vascular injury revealed by
58
physical examination detect those vascular injuries that require immediate surgical repair
59
rendering further imaging studies unnecessary. Hard signs include active hemorrhage, distal
60
ischemia, and bruit or thrill over the wound.7
M
d
te
Ac ce p
61
an
53
Fractures that occur as a result of penetrating trauma deserve particular attention.
62
Ballistic fractures are associated with a twofold higher rate of total vascular injury (both those
63
that present with hard signs during physical examination and those without hard signs).8,9
64
Additionally, the need for surgical repair of vascular injury is nearly doubled in cases of ballistic
65
injury with fracture compared with cases of ballistic injury without fracture.8,9 Norman et al.9
66
found a threefold increase in the incidence of arterial injuries that did not present with hard signs
67
during physical examination (“subclinical” or “occult” injuries). Furthermore, the presence of
68
fracture renders physical examination more challenging to perform and more difficult to
69
interpret.10 Several possible reasons for the increased incidence of arterial injury in association 3 Page 3 of 30
70
with ballistic fractures compared with other ballistic injuries include creation of a secondary
71
bony missile and tethering of vascular structures to bone.11 Considering that plain radiographs are universally obtained of each patient suspected of
73
having a fracture, this study was undertaken to determine whether the plain radiographs can be
74
used to predict the presence of arterial injury in the setting of a femoral fracture resulting from a
75
civilian gunshot. Considering that the superficial femoral artery is tethered to the femur as it
76
passes through Hunter canal in the adductor hiatus, it is possible the fractures near that location
77
are more likely to cause arterial injury. Therefore, we hypothesized that ballistic femoral
78
fractures located in the distal third of the femur are more likely to be associated with arterial
79
injury.
M
an
us
cr
ip t
72
Patients and methods
82
Study group
te
81
d
80
After approval from our institutional review board, we conducted a review of 2546
84
patients who had sustained gunshot wounds and were admitted to our urban level I trauma centre
85
from 2002 through 2007. Patients were identified with the use of a prospective trauma registry
86
database.
87
Ac ce p
83
Each of the patients selected for the study met the inclusion criterion of having sustained
88
a gunshot wound that resulted in a femoral fracture (n = 153). Eighteen patients were excluded
89
because their plain radiographs did not contain the entire femur for measurement of femoral
90
length, and two patients were excluded because they had sustained ballistic injuries from
91
shotguns. Shotgun injuries were excluded because they are caused by a different mechanism of
92
injury. A shotgun shell contains from a few to hundreds of pellets, leading to a very different
4 Page 4 of 30
injury pattern that is more similar to injury caused by high-velocity ballistics than injury caused
94
by civilian gunshot.40 One of the patients who sustained a shotgun injury also sustained an
95
arterial injury. The 133 remaining patients who sustained femoral fractures as a consequence of
96
civilian gunshot wounds to the lower extremity were identified and comprised the study group.
ip t
93
We reviewed all electronic medical records, including radiology reports, to determine the
98
presence (n = 14, 11%) or absence (n = 119) of arterial injury in each patient. Arterial injury had
99
been confirmed by angiography or computed tomographic (CT) angiography. Twelve of 14
us
cr
97
patients with arterial injury were treated with vascular repair. Of those with arterial injury, the
101
superficial femoral artery was injured in eight (57%), the common femoral artery in one (7%),
102
and the popliteal artery in five (36%). Two patients sustained arterial injuries that were not
103
treated operatively because of minimal arterial wall disruption (intimal defects with intact distal
104
circulation or pseudoaneurysm). Of the 20 excluded patients, two had vascular injuries.
M
d
From our database, we retrieved demographic data on all patients in our study group,
te
105
an
100
including patient age, sex, date and timing of injury, mechanism of injury, Revised Trauma
107
Score (RTS) measured at the scene of injury and measured at admission, Injury Severity Score
108
(ISS), Trauma and Injury Severity Score (TRISS), admission Glasgow Coma Scale score, blood
109
pressure, heart rate, admission and discharge dates, number of days in the intensive care unit,
110
scene time, injury type, method of injury, specific patterns of injury, and disposition. The data
111
were used to ensure that the two study groups were similar demographically.
112
Ac ce p
106
No significant difference was shown between the patients with and without arterial injury
113
regarding age, sex, race, mechanism of injury, diastolic blood pressure, respiratory rate,
114
temperature, RTS at the scene, TRISS, admission Glasgow Coma Scale score, number of days
115
spent in the intensive care unit, or scene time. However, a significant (P < 0.05) difference was
5 Page 5 of 30
116
shown between the two groups regarding systolic blood pressure, heart rate, RTS at admission,
117
ISS, and days in hospital, as might be expected (Table 1).
118
120
Radiographic Measurement
ip t
119
Two of our co-authors who were blinded to the presence or absence of arterial injury
obtained measurements from the anteroposterior view postoperative radiographs of the femur.
122
All fractures in this study had been treated with intramedullary nail insertion, and the nail was
123
present in all images. Images were viewed with the use of IMPAX for Radiology software (Agfa
124
HealthCare, Greenville, SC) on computerized workstations that are used at our institution for
125
clinical viewing of radiographs on a picture archiving and communication system. The IMPAX
126
Markup Caliper tool (Agfa HealthCare) was used to measure the length of the femur in
127
millimeters.
us
an
M
d
A standardized technique was used to measure each fracture (Fig. 1). One line was drawn
te
128
cr
121
perpendicular to the lesser trochanter and another across the most distal part of the femoral
130
condyles. The length of the femur was recorded as the distance between the two lines. The
131
distance from the lesser trochanter to the proximal and distal fracture lines was also measured. If
132
a single plain radiograph of the entire femur was not available, two plain radiographs that
133
together showed the entire femur were evaluated measuring from the lesser trochanter and the
134
femoral condyles to a stable structure, often a screw, present in both radiographs. The distance to
135
the middle of the fracture was calculated from the measured distance to the proximal and distal
136
fracture lines. All measurements were normalized to the total length of the femur.
Ac ce p
129
137 138
Analysis
6 Page 6 of 30
Statistical analysis was conducted with Microsoft Excel version 2007 (Microsoft,
140
Redmond, WA). Unequal variance Student’s t tests were conducted to compare continuous data,
141
including gunshot wound fracture location and continuous demographic data. For discrete data,
142
Chi-squared or Fisher’s exact test, if the sample size in any box of the contingency table was less
143
than five, was used to assess for significance. A P-value of 6, with a P-value < 0.05. Considering that we found the results to be the same
284
whether the patients with femoral neck and intertrochanteric fractures were included or excluded,
285
for the sake of simplicity, we reported the data from one methodology. We presented the data
286
including the patients with femoral neck and intertrochanteric fractures to eliminate any
287
confusion regarding whether these patients affected our results. This methodology slightly
288
underestimates the location of these fractures by a few millimeters. The fractures are actually a
289
few millimeters more proximal than the lesser trochanter, which makes it more difficult for the
290
location of the fracture in the distal third of the femur to predict arterial injury.
cr
us
an
M
d
te
Ac ce p
291
ip t
278
Although all results were statistically significant at P < 0.05, which is the medical
292
standard for indication of statistical significance, confidence intervals were relatively wide.
293
However, the statistical significance of these data suggests that location of ballistic femoral
294
fractures can be used, with confidence, to predict presence of arterial injury.
295
Our study had several strengths. The results were consistent regardless of which fracture
296
extent (proximal extent, middle of fracture, or distal extent) was analyzed. To our knowledge,
297
this study is the largest investigation of its type in the literature and provides the clinician with a
13 Page 13 of 30
298
useful clinical risk factor to aid in the evaluation of patients with femoral fractures after civilian
299
gunshot wounds.
300
302
Conclusions
ip t
301
The results of the present study indicate that if a gunshot wound-related femoral fracture occurs in the distal third of the femur, it is approximately six times more likely to be associated
304
with arterial injury. Practitioners should be aware that such gunshot wound fractures deserve
305
particular attention, considering the gravity of the complications associated with delayed
306
treatment of arterial injury.
Ac ce p
te
d
M
an
us
cr
303
14 Page 14 of 30
References
308
1. Levy BA, Zlowodzki MP, Graves M, Cole PA. Screening for extermity [sic] arterial injury
309
with the arterial pressure index. The American Journal of Emergency Medicine 2005;23:689−95.
310
2. Payne WK, Gabriel RA, Massoud RP. Gunshot wounds to the thigh: evaluation of vascular
311
and subclinical vascular injuries. The Orthopedic Clinics of North America 1995;26:147−54.
312
3. Sirinek KR, Levine BA, Gaskill HV III, Root HD. Reassessment of the role of routine
313
operative exploration in vascular trauma. The Journal of Trauma 1981;21:339−44.
314
4. Frykberg ER. Arteriography of the injured extremity: are we in proximity to an answer? The
315
Journal of Trauma 1992;32:551−2.
316
5. Frykberg ER. Advances in the diagnosis and treatment of extremity vascular trauma. The
317
Surgical Clinics of North America 1995;75:207−23.
318
6. Seamon MJ, Smoger D, Torres DM, Pathak AS, Gaughan JP, Santora TA, et al. A prospective
319
validation of a current practice: the detection of extremity vascular injury with CT angiography.
320
The Journal of Trauma 2009;67:238−43.
321
7. Dennis JW, Frykberg ER, Veldenz HC, Huffman S, Menawat SS. Validation of nonoperative
322
management of occult vascular injuries and accuracy of physical examination alone in
323
penetrating extremity trauma: 5- to 10-year follow-up. The Journal of Trauma 1998;44:243−52.
324
8. Gahtan V, Bramson RT, Norman J. The role of emergent arteriography in penetrating limb
325
trauma. The American Surgeon 1994;60:123−7.
326
9. Norman J, Gahtan V, Franz M, Bramson R. Occult vascular injuries following gunshot
327
wounds resulting in long bone fractures of the extremities. The American Surgeon
328
1995;61:146−50.
Ac ce p
te
d
M
an
us
cr
ip t
307
15 Page 15 of 30
10. Applebaum R, Yellin AE, Weaver FA, Oberg J, Pentecost M. Role of routine arteriography
330
in blunt lower-extremity trauma. American Journal of Surgery 1990;160:221−4.
331
11. Tejan J, Lindsey RW. Management of civilian gunshot injuries of the femur: a review of the
332
literature. Injury 1998;29(Suppl 1):SA18−22.
333
12. Centers for Disease Control and Prevention. Injury Prevention & Control: Data & Statistics
334
(WISQUARS): Fatal Injury Data. http://webappa.cdc.gov/sasweb/ncipc/mortrate.html. Last
335
accessed, March 21, 2013.
336
13. Asensio JA, Kuncir EJ, Garcia-Nuñez LM, Petrone P. Femoral vessel injuries: analysis of
337
factors predictive of outcomes. Journal of the American College of Surgeons 2006;203:512−20.
338
14. Brown KV, Ramasamy A, Tai N, MacLeod J, Midwinter M, Clasper JC. Complications of
339
extremity vascular injuries in conflict. The Journal of Trauma 2009;66(Suppl 4):S145−9.
340
15. Cone JB. Vascular injury associated with fracture-dislocations of the lower extremity.
341
Clinical Orthopaedics and Related Research 1989;243:30−5.
342
16. McHenry TP, Holcomb JB, Aoki N, Lindsey RW. Fractures with major vascular injuries
343
from gunshot wounds: implications of surgical sequence. The Journal of Trauma
344
2002;53:717−21.
345
17. Miller HH, Welch CS. Quantitative studies on the time factor in arterial injuries. Annals of
346
Surgery 1949;130:428−38.
347
18. Shah DM, Corson JD, Karmody AM, Fortune JB, Leather RP. Optimal management of tibial
348
arterial trauma. The Journal of Trauma 1988;28:228−34.
349
19. Kendall RW, Taylor DC, Salvian AJ, O’Brien PJ. The role of arteriography in assessing
350
vascular injuries associated with dislocation of the knee. The Journal of Trauma 1993;35:875−8.
Ac ce p
te
d
M
an
us
cr
ip t
329
16 Page 16 of 30
20. Brinker MR, Bailey DE. Fracture healing in tibia fractures with an associated vascular injury.
352
The Journal of Trauma 1997;42:11−9.
353
21. Weaver FA, Yellin AE, Bauer M, Oberg J, Ghalambor N, Emmanuel RP, et al. Is arterial
354
proximity a valid indication for arteriography in penetrating extremity trauma? a prospective
355
analysis. Archives of Surgery 1990:125:1256−60.
356
22. Dennis JW, Frykberg ER, Crump JM, Vines FS, Alexander RH. New perspectives on the
357
management of penetrating trauma in proximity to major limb arteries. Journal of Vascular
358
Surgery 1990;11:84−92.
359
23. Atteberry LR, Dennis JW, Russo-Alesi F, Meanwat SS, Lenz BJ, Frykberg ER. Changing
360
patterns of arterial injuries associated with fractures and dislocations. Journal of the American
361
College of Surgeons 1996;183:377−83.
362
24. Lohman M, Lauridsen K, Vedel P. Arterial lesions in major knee trauma: pedal pulse a false
363
sign of security? Archives of Orthopaedic and Trauma Surgery 1990;109:238−9.
364
25. Chapman JA. Popliteal artery damage in closed injuries of the knee. Journal of Bone and
365
Joint Surgery British Volume 1985;67:420−3.
366
26. Frykberg ER, Crump JM, Vines FS, McLellan GL, Dennis JW, Brunner RG, et al. A
367
reassessment of the role of arteriography in penetrating proximity extremity trauma: a
368
prospective study. The Journal of Trauma 1989:29:1041−50.
369
27. Frykberg ER, Dennis JW, Bishop K, Laneve L, Alexander RH. The reliability of physical
370
examination in the evaluation of penetrating trauma for vascular injury: results at one year. The
371
Journal of Trauma 1991;31:502−11.
372
28. Bandyk DF. Vascular injury associated with extremity trauma. Clinical Orthopaedics and
373
Related Research 1995;318:117−24.
Ac ce p
te
d
M
an
us
cr
ip t
351
17 Page 17 of 30
29. Manthey DE, Nicks BA. Penetrating trauma to the extremity. The Journal of Emergency
375
Medicine 2008;34:187−93.
376
30. Johansen K, Lynch K, Paun M, Copass M. Non-invasive vascular tests reliably exclude
377
occult arterial trauma in injured extremities. The Journal of Trauma 1991;31:515−9.
378
31. Peng PD, Spain DA, Tataria M, Hellinger JC, Rubin GD, Brundage SI. CT angiography
379
effectively evaluates extremity vascular trauma. The American Surgeon 2008;74:103−7.
380
32. Redmond JM, Levy BA, Dajani KA, Cass JR, Cole PA. Detecting vascular injury in lower
381
extremity orthopedic trauma: the role of CT angiography. Orthopedics 2008;31:761−7.
382
33. Busquéts AR, Acosta JA, Colón E, Alejandro KV, Rodriguez P. Helical computed
383
tomographic angiography for the diagnosis of traumatic arterial injuries of the extremities. The
384
Journal of Trauma 2004;56:625−8.
385
34. Soto JA, Múnera F, Cardoso N, Guarín O, Medina S. Diagnostic performance of helical CT
386
angiography in trauma to large arteries of the extremities. Journal of Computer Assisted
387
Tomography 1999;23:188−96.
388
35. Soto JA, Múnera F, Morales C, Lopera JE, Holquín D, Guarín O, et al. Focal arterial injuries
389
of the proximal extremities: helical CT arteriography as the initial method of diagnosis.
390
Radiology 2001;218:188−94.
391
36. Inaba K, Potzman J, Munera F, McKenney M, Munoz R, Rivas L, et al. Multi-slice CT
392
angiography for arterial evaluation in the injured lower extremity. The Journal of Trauma
393
2006;60:502−6.
394
37. Shayne PH, Sloan EP, Rydman R, Barrett JA. A case-control study of risk factors that predict
395
femoral arterial injury in penetrating thigh trauma. Annals of Emergency Medicine
396
1994;24:678−84.
Ac ce p
te
d
M
an
us
cr
ip t
374
18 Page 18 of 30
38. Gonzalez RP, Falimirski ME. The utility of physical examination in proximity penetrating
398
extremity trauma. The American Surgeon 1999;65:784−9.
399
39. Anderson RJ, Hobson RW II, Padberg FT Jr, Pecoraro JP, DeGroote RD, Jamil Z, et al.
400
Penetrating extremity trauma: identification of patients at high-risk requiring arteriography.
401
Journal of Vascular Surgery 1990;11:544−8.
402
40. Bartlet CS, Helfet DL, Hausman MR, Strauss E. Ballistics and gunshot wounds: effects on
403
musculoskeletal tissues. The Journal of the American Academy of Orthopaedic Surgeons
404
2000;8:21−36.
Ac ce p
te
d
M
an
us
cr
ip t
397
19 Page 19 of 30
405
Figure Legends
406 407
Fig. 1. Radiograph shows method of measuring fractures.
ip t
408
Fig. 2. Frequency distribution for location of distal fracture line in ballistics-related fractures
410
with and without arterial injury.
cr
409
us
411
Ac ce p
te
d
M
an
412
20 Page 20 of 30
412 413 414
Conflict of Interest Statement The authors have no potential conflicts of interest to report. No outside funding was received for this work.
Ac ce p
te
d
M
an
us
cr
ip t
415 416
22 Page 21 of 30
416 417 418
Acknowledgment We thank Senior Editor and Writer Dori Kelly, MA, for editing the manuscript.
Ac ce p
te
d
M
an
us
cr
ip t
419
23 Page 22 of 30
419
Table 1
420
Characteristics of study participants Ballistic fracture with
Ballistic fracture without
arterial Injury
arterial injury
(n = 14)
(n = 119)
ip t
SD
Mean
Age (year)
27.3
9.4
27.3
Height (inch)
70.5
2.9
Weight (pound)
174.6
37.1
14 (100%) 0 (0%)
11 (79%)
Ac ce p
Black
te
Race
0.99
70.3
8.3
0.85
177.9
33.8
0.77
an
us
9.4
0.37
108 (91%) 11 (9%)
101 (85%)
White
3 (21%)
16 (13%)
Unknown
0 (0%)
2 (2%)
RTS_S
SD
d
Female
M
Sex
cr
Mean
Male
P-value*
7
1.5
7.7
0.9
0.15
6.5
1.6
7.5
1
0.04†
24.1
13.7
12.9
7.2
0.009†
TRISS
0.8
0.3
1
0.2
0.1
Days in hospital
15.5
13.5
6.1
15.8
0.03†
RTS_A ISS
24 Page 23 of 30
SD, standard deviation; RTS_S, Revised Trauma Score measured at scene of injury; RTS_A,
422
Revised Trauma Score measured at admission; ISS, Injury Severity Score; TRISS, Trauma and
423
Injury Severity Score
424
*Unequal variance, Student’s t test for continuous data, and Fisher’s exact test for discrete data.
425
†Statistically significant.
cr
ip t
421
Ac ce p
te
d
M
an
us
426
25 Page 24 of 30
426
Table 2
427
Mean location of fracture in femur Ballistic fracture with
Ballistic fracture without
arterial injury (n = 14)
arterial injury (n = 119)
Mean
95% Confidence
Mean
95% Confidence
location
interval
location
Proximal fracture line
0.53
0.48−0.58
0.29
cr
ip t
P-value*
0.27−0.3
0.009†
Middle of fracture line
0.67
0.62−0.73
0.44
0.42−0.46
0.019†
Distal fracture line
0.6
0.55−0.66
0.35−0.38
0.013†
429
†Statistically significant.
us
an
*Unequal variance, Student’s t test.
0.36
M
428
interval
Ac ce p
te
d
430
26 Page 25 of 30
430
Table 3
431
Odds ratios: Does a fracture located in the distal third of the femur predict arterial injury? 95% Confidence interval
P-value*
Proximal fracture line
5.63
1.7−18.6
0.008†
Middle of fracture line
3.96
1.27−12.37
Distal fracture line
6.72
1.78−25.44
†Statistically significant.
cr
433
0.003†
us
*Fisher’s exact test.
0.02†
an
432
ip t
Odds ratio
Ac ce p
te
d
M
434
27 Page 26 of 30
Table 4
435
Diagnostic test characteristics of a fracture in the distal third of the femur: Proximal fracture line
436
present in distal third of femur 95% Confidence interval
Sensitivity
43
20−67
Specificity
88
86−91
Positive predictive value
30
14−47
Negative predictive value
93
90−96
an
us
cr
%
ip t
434
Ac ce p
te
d
M
437 438
28 Page 27 of 30
Table 5
439
Diagnostic test characteristics of a fracture in the distal third of the femur: Distal fracture line
440
present in distal third of femur 95% Confidence interval
Sensitivity
79
50−94
Specificity
65
61−67
Positive predictive value
21
13−25
Negative predictive value
96
91−99
an
us
cr
%
ip t
438
Ac ce p
te
d
M
441 442
29 Page 28 of 30
Ac ce p
te
d
M
an
us
cr
ip t
Figure 1
Page 29 of 30
Ac
ce
pt
ed
M
an
us
cr
i
Figure 2
Page 30 of 30