1351

C OPYRIGHT Ó 2019 BY T HE J OURNAL OF B ONE AND J OINT S URGERY, I NCORPORATED

Pediatric Gartland Type-IV Supracondylar Humeral

Fractures Have Substantial Overlap with
Flexion-Type Fractures
Stuart L. Mitchell, MD, Brian T. Sullivan, MD, Christine A. Ho, MD, Joshua M. Abzug, MD,
Micheal Raad, MD, and Paul D. Sponseller, MD, MBA

Background: Knowledge is limited about the diagnosis and treatment of modified Gartland type-IV supracondylar
humeral fractures. We determined the prevalence of type-IV fractures, identified preoperative characteristics associated
with these injuries, and assessed operative treatment characteristics.
Methods: We retrospectively identified patients <16 years of age who underwent operative treatment of a supracondylar
humeral fracture at 2 centers between 2008 and 2016. We compared patient, injury, and treatment characteristics
between type-IV and type-III fracture groups (1:4, cases:controls). Preoperative radiographs were assessed by 4 pediatric
orthopaedists blinded to fracture type. The odds of a fracture being type IV were assessed using univariate logistic
regression for individual radiographic parameters. Significance was set at alpha = 0.05.
Results: Type-IV fractures accounted for 39 (1.3%) of the supracondylar humeral fractures treated operatively during the
study period. A type-IV fracture was associated with the following radiographic parameters: flexion angulation (odds ratio
[OR] = 17; 95% confidence interval [CI] = 4.9 to 59), valgus angulation (OR = 5.6; 95% CI = 1.6 to 20), and lateral
translation (OR = 4.1; 95% CI = 1.6 to 11) of the distal fragment; osseous apposition between the proximal and distal
fragments (OR = 4.0; 95% CI = 1.8 to 9.0); and propagation of the fracture line toward the diaphysis of the proximal
segment (OR = 9.2; 95% CI = 1.6 to 53). We found no significant differences in patient or injury characteristics between
the groups. Compared with type-III fractures, type-IV fractures were treated more frequently with open reduction and
percutaneous pinning (13% compared with 3.8%; p = 0.04) and were associated with longer mean operative time (82 ± 42
compared with 63 ± 28 minutes; p = 0.001).
Conclusions: We identified 5 preoperative radiographic parameters associated with greater odds of a supracondylar
humeral fracture being type IV rather than type III. No patient or injury characteristic differed significantly between the
groups. Substantial overlap likely exists between type-IV and flexion-type fractures. Type-IV fractures were associated with
longer operative time and were treated with open reduction more frequently than were type-III fractures.
Level of Evidence: Therapeutic Level IV. See Instructions for Authors for a complete description of levels of evidence.

S
upracondylar humeral fractures are the most common
elbow fracture in children1,2. They are grouped into
extension and flexion types, according to the direction of
anteroposterior displacement3. Most surgeons use the Wilkins
modification4 of the Gartland classification5 to categorize pediatric
extension-type supracondylar humeral fractures according to the
degree of anteroposterior displacement of the distal fragment, as
viewed on lateral radiographs (type I: nondisplaced; type II: dis-
placed, with intact posterior cortical hinge; or type III: complete
disruption of the posterior cortex)3. Type-III fractures are assumed
to have an intact posterior periosteum that can aid in reduction
maneuvers6. In 2006, Leitch et al.7 further modified the classifi-
cation system to include a variant of type-III fractures with mul-
tidirectional instability resulting from circumferential loss of the
periosteal hinge. This rarer fracture type is referred to as a modified
Gartland type-IV fracture. Type-IV fractures are diagnosed in the
operating room (during reduction attempt[s]/manipulation),
where instability can be assessed without impediments to flexion
or extension, as opposed to preoperative radiographic diagnosis,
which is possible for fracture types I to III.
The standard treatment for type-III and type-IV supra-
condylar humeral fractures is closed reduction and percutaneous

Disclosure: This study was supported in part by a T32 grant (AR067708) from the National Institutes of Health. On the Disclosure of Potential Conflicts of
Interest forms, which are provided with the online version of the article, one or more of the authors checked “yes” to indicate that the author had a relevant
financial relationship in the biomedical arena outside the submitted work (http://links.lww.com/JBJS/F361).

J Bone Joint Surg Am. 2019;101:1351-6 d http://dx.doi.org/10.2106/JBJS.18.01178

 1352
TH E JO U R NA L O F B O N E & JO I N T SU RG E RY J B J S . O RG
d
P E D I AT R I C G A R T L A N D T Y P E -IV S U P R A C O N D Y L A R H U M E R A L
V O L U M E 1 01-A N U M B E R 15 A U G U S T 7, 2 019
d d
FRACTURE S

pinning (CRPP) or open reduction and percutaneous pinning
(ORPP)8. Although type-III fractures are more common, type-
IV fractures are often more difficult to reduce because of the
instability caused by the loss of the anterior and posterior per-
iosteal hinges7,9. There is scant literature describing the diagnosis
and treatment of type-IV supracondylar humeral fractures.
The aims of this study were to determine the prevalence
of type-IV fractures; identify preoperative patient, injury, and
radiographic characteristics associated with these injuries; and
assess operative treatment characteristics. The findings of this
study will help surgical teams predict the likelihood of type-IV
fracture, so they can more accurately counsel patients, antici-
pate operative time, and plan for the required surgical equip-
ment for this rare, complex fracture type10.
Materials and Methods
Patient Selection
Patient and Injury Characteristics
We determined patient age, sex, body mass index (BMI), and
injury characteristics (laterality of injury, neurological injury,
vascular injury, open fracture, ipsilateral forearm or wrist fracture,
and compartment syndrome) by reviewing medical records.
Radiographic Parameters
Four fellowship-trained pediatric orthopaedic surgeons, blinded
to fracture type, reviewed preoperative anteroposterior and
lateral radiographs. Radiographic parameters assessed were (1)
angulation of the distal fracture fragment in the sagittal plane
(i.e., flexion or extension), (2) angulation of the distal fracture
fragment in the coronal plane, (3) translation of the distal
fracture fragment in the coronal plane, (4) osseous apposition
(i.e., cortical contact) between of the proximal and distal
fragments on anteroposterior and lateral radiographs, (5)
propagation of the fracture line toward the diaphysis of the

A fter receiving institutional review board approval for this

retrospective review, we queried the patient databases of 2
institutions for all patients <16 years of age who underwent
operative treatment for a modified Gartland type-III or type-IV
supracondylar humeral fracture between 2008 and 2016, using
Current Procedural Terminology and International Classifica-
tion of Diseases, Ninth or Tenth Revision, Clinical Modifica-
tion codes. Operative notes were screened, and fracture type
was identified according to the surgeon’s diagnosis (type III or
IV). We excluded patients who (1) had a history of skeletal
dysplasia or endocrine disorder, (2) were missing preoper-
ative/injury radiographs, (3) had undergone attempted
fracture reduction in the emergency department, (4) had a
transphyseal or adolescent intercondylar distal humeral
fracture pattern, or (5) were missing intraoperative docu-
mentation of fracture type.
We selected a cohort of patients with modified Gartland
type-III fractures (1:4 ratio of cases to such “controls”) using
random sampling to compare with the type-IV fracture cohort.
proximal segment, (6) fracture comminution, and (7) whether
the fracture fragment abutted skin. For each patient, parame-
ters were determined by consensus agreement rating.
Operative Characteristics
We reviewed patient records for operative treatment type (CRPP
or ORPP) and operative time. Operative time was defined as the
time from surgical “time-out” to “anesthesia stop.”
Statistical Analysis
Data were analyzed using chi-square tests or Fisher exact tests
for categorical variables and 2-sample t tests or Mann-Whitney
U tests for continuous variables. Interrater agreement on the
assessment of radiographs was measured using the kappa (k)
statistic and categorized according to the method of Landis and
Koch11. The likelihood of a type-IV fracture for each parameter
was modeled using univariate logistic regression. Odds ratios
(ORs) were estimated and are reported with their associated
95% confidence intervals (CIs). Significance was considered at

TABLE I Patient and Injury Characteristics of 195 Skeletally Immature Patients with a Type-III or IV Supracondylar Humeral Fracture

Parameter Type III (N = 156) Type IV (N = 39) P Value

Patient parameters
Age* (yr) 6.4 ± 2.2 6.5 ± 2.4 0.96
Female sex (no. [%]) 70 (45) 20 (51) 0.47
BMI* (kg/m2) 16 ± 3.1 17 ± 3.3 0.35
Injury parameters (no. [%])
Left side 99 (63) 28 (72) 0.33
Neurological injury 18 (12) 9 (23) 0.06
Ipsilateral forearm or wrist fracture 8 (5.1) 1 (2.6) 0.43
Open fracture 6 (3.8) 0 (0) 0.26
Vascular injury 5 (3.2) 2 (5.1) 0.43
Compartment syndrome 2 (1.3) 0 (0) 0.64

*The values are given as the mean and standard deviation.

 1353
TH E JO U R NA L O F B O N E & JO I N T SU RG E RY J B J S . O RG
d
P E D I AT R I C G A R T L A N D T Y P E -IV S U P R A C O N D Y L A R H U M E R A L
V O L U M E 1 01-A N U M B E R 15 A U G U S T 7, 2 019
d d
FRACTURE S

a = 0.05 for all analyses. Analyses were performed using SPSS
Statistics for Macintosh software (version 24.0; IBM) and Stata
software (version 15; StataCorp).
Results
Radiographic Characteristics
One patient with a type-IV fracture was excluded from
radiographic analyses because the preoperative lateral radio-
graph was missing. This left 194 fractures for analysis (156
type-III and 38 type-IV fractures).

T hirty-nine children aged 3 to 13 years with a type-IV

supracondylar humeral fracture were included. This re-
flected a prevalence of 1.2% (26 of 2,215) at 1 center, 1.6% (13 of
825) at the other center, and 1.3% overall among all operatively
treated supracondylar humeral fractures during the study
period. The 39 patients with a type-IV fracture were compared
with the control cohort of 156 patients with a type-III fracture.
Patient and Injury Characteristics
There were no significant differences between type-III and
type-IV fracture groups in terms of patient characteristics (age,
sex, BMI) or injury characteristics (laterality of injury, neuro-
logical injury, vascular injury, open fracture, ipsilateral wrist or
forearm fracture, compartment syndrome) (p > 0.05 for all)
(Table I). No type-IV fractures were open injuries, and no
patient with a type-IV fracture had compartment syndrome.
Diagnosis of type-IV fracture was significantly more
likely given the presence of the following: flexion angulation of
the distal fragment (OR = 17; 95% CI = 4.9 to 59), valgus
angulation of the distal fragment (OR = 5.6; 95% CI = 1.6 to
20), lateral translation of the distal fragment (OR = 4.1; 95%
CI = 1.6 to 11), osseous apposition (cortical contact) between
the proximal and distal fragments (OR = 4.0; 95% CI = 1.8 to
9.0), or fracture propagation toward the diaphysis (OR = 9.2;
95% CI = 1.6 to 53) (Table II). There were no significant
associations with comminution or whether the fracture frag-
ment was abutting the skin.
Among type-IV fractures, there was flexion angulation of
the distal fragment in 30% (versus 3% in type-III fractures),
valgus angulation of the distal fragment in 66% (versus 40% in
type-III fractures), lateral translation of the distal fragment in 69%
(versus 41% in type-III fractures), osseous apposition in 64%

TABLE II Univariate Odds of Supracondylar Humeral Fractures Being Type IV (N = 38) Versus Type III (N = 156) on the Basis of
Preoperative Radiographic Parameters*

Parameter OR (95% CI) P Value

Angulation of distal fragment in sagittal plane

Extension Referent Referent
Flexion 17 (4.9-59) <0.001
Neutral 3.4 (0.3-39) 0.33
Angulation of distal fragment in coronal plane
Varus Referent Referent
Valgus 5.6 (1.6-20) 0.008
Neutral 2.3 (0.6-9.3) 0.24
Translation of distal fragment in coronal plane
Medial Referent Referent
Lateral 4.1 (1.6-11) 0.004
None 1.9 (0.7-5.5) 0.25
Osseous apposition between proximal and distal fragments
Absent Referent Referent
Present 4.0 (1.8-9.0) 0.001
Propagation of fracture line toward diaphysis of proximal segment
Absent Referent Referent
Present 9.2 (1.6-53) 0.01
Comminution
Absent Referent Referent
Present 1.8 (0.7-4.5) 0.22
Fracture fragment abutting the skin
Absent Referent Referent
Present 0.8 (0.3-2.1) 0.63

*OR = odds ratio, and CI = confidence interval.

 1354
TH E JO U R NA L O F B O N E & JO I N T SU RG E RY J B J S . O RG
d
P E D I AT R I C G A R T L A N D T Y P E -IV S U P R A C O N D Y L A R H U M E R A L
V O L U M E 1 01-A N U M B E R 15 A U G U S T 7, 2 019
d d
FRACTURE S

(versus 32% in type-III fractures), and fracture propagation
toward the diaphysis in 11% (versus 1% in type-III fractures).
Interobserver agreement on radiographic parameters
was substantial for translation of the distal fragment in the
coronal plane (k = 0.76); moderate for the fracture fragment
abutting the skin (k = 0.53), angulation of the distal fragment
in the sagittal plane (k = 0.53), osseous apposition between
proximal and distal fragments (k = 0.44), and angulation of the
distal fragment in the coronal plane (k = 0.44); fair for fracture
propagation toward the diaphysis (k = 0.24); and slight for
comminution (k = 0.11) (see Appendix). Significant radio-
graphic parameters, the prevalence of each parameter in the
type-IV fracture cohort, and associated representative images
of each parameter are presented in Figure 1.
Operative Characteristics
The groups differed significantly by treatment type (Table III).
ORPP was performed for a greater percentage of type-IV
fractures compared with type-III fractures (p = 0.042). This
difference was even more pronounced when comparing only
closed fractures (p = 0.009). The mean operative time (and
standard deviation) was significantly longer for all type-IV
fractures (82 ± 42 minutes) compared with type-III fractures
(63 ± 28 minutes) (p = 0.001).

Fig. 1
Significant radiographic parameters with at least moderate interobserver agreement, their prevalence in the type-IV cohort, and representative images. The
images of flexion angulation, lateral translation, and valgus angulation are from 2 patients each. The images of osseous apposition are antero-
posterior and lateral views of the same patient.

TH E JO U R NA L O F B O N E & JO I N T SU RG E RY J B J S . O RG
V O L U M E 1 01-A N U M B E R 15 A U G U S T 7, 2 019
d d
TABLE III Operative Parameters for Skeletally Immature
Patients with Type-III or IV Supracondylar Humeral
Fractures*
Type-III Type-IV
Parameter Fracture Fracture
All fractures† (no. [%])
Treatment type
CRPP 150 (96) 33 (87)
ORPP 6 (3.8) 5 (13)
Operative time‡ (min) 63 ± 28 82 ± 42
Closed fractures§
(no. [%])
Treatment type
CRPP 147 (98) 33 (87)
ORPP 3 (2.0) 5 (13)
Operative time‡ (min) 64 ± 28 82 ± 42
*CRPP = closed reduction and percutaneous pinning, and ORPP =
open reduction and percutaneous pinning. †Includes 156 type-III
fractures and 38 type-IV fractures. ‡The values are given as the
mean and standard deviation. §Includes 150 type-III fractures
and 38 type-IV fractures.
Discussion
W e identified preoperative radiographic parameters asso-
ciated with greater odds of a supracondylar humeral
fracture being a type-IV rather than a type-III fracture, but we
found no patient factors associated with the diagnosis. In 2006,
Leitch et al.7 first described a multidirectionally unstable
supracondylar humeral fracture variant, in 9 patients. The
authors referred to this as the modified Gartland type-IV
fracture. In our study, the proportion of type-IV supracondylar
humeral fractures treated was relatively small, at only 1.3% of
all operatively treated supracondylar humeral fractures during
the study period. To our knowledge, this represents the largest
cohort of type-IV supracondylar humeral fractures reported to
date, which also highlights the rarity of this injury.
Given the rarity of type-IV supracondylar humeral frac-
tures, determining the true rate of injury is challenging, with
reports ranging from 3% to 9%7,12 of all operatively treated
supracondylar humeral fractures. The variability in frequency of
type-IV fractures reported is likely the result of several factors. The
diagnosis of type-IV fractures is only possible intraoperatively
(during reduction attempt[s]/manipulation in the operating
room), and the inter- and intrarater reliabilities of the diagnosis
are unknown. Diagnosis is a subjective matter that is ultimately
decided by the treating surgeon. A type-IV fracture may be caused
iatrogenically during reduction attempt(s) in the operating room7,
further confounding the diagnosis and rates of occurrence.
Reported rates of neurological injury sustained by patients
with type-IV supracondylar humeral fractures vary widely, from
0% to 36%6,7,9,13. These rates are comparable with the 23% of
patients with at least 1 nerve palsy in our study. Furthermore,
1355
d
P E D I AT R I C G A R T L A N D T Y P E -IV S U P R A C O N D Y L A R H U M E R A L
FRACTURE S
there was no significant difference in the rate of neurological
injury between type-III and type-IV fractures (p = 0.059), which
is also consistent with published comparative data6,13. Also con-
sistent with previous data, there were no differences in patient or
injury characteristics between the 2 fracture types6,7.
No preoperative clinical factors were associated with type-
P Value
IV supracondylar humeral fractures. However, we identified
several radiographic parameters that were associated with type-IV
0.042 fractures on the basis of significantly elevated ORs and prevalence
in ‡30% of type-IV fractures. The most strongly associated
parameter was flexion angulation of the distal fragment, followed
0.001 by valgus angulation, lateral translation, and the presence of
osseous apposition between the proximal and distal fracture
fragments. It is likely that substantial overlap exists between type-
0.009
IV and flexion-type fractures. Although not all type-IV fractures
are flexion-type variants and not all flexion-type fractures are
multidirectionally unstable, our findings suggest that there is
substantial overlap between these 2 rare fracture types.
0.002
Regarding operative characteristics, the longer operative
times and higher frequency of ORPP for type-IV fractures com-
pared with type-III fractures are consistent with previous data
from the largest, to our knowledge, available series6. However,
other groups with smaller sample sizes of type-IV fractures (7 or 8
cases) reported success with CRPP in all cases7,9. Even in our large
cohort of type-IV fractures, ORPP was still a relatively rare
occurrence, at only 13% (n = 5), which may explain why ORPP
was not observed in the smaller series. Our rate of ORPP for type-
IV fractures was even higher than that noted for type-III fractures
when comparing only closed fractures; half of the type-III fractures
that underwent ORPP (3 of 6) were open fractures, which likely
necessitated an open approach for irrigation and debridement and
not necessarily because of the difficulty of fracture reduction.
Our study had several limitations, including those inherent
in a multicenter, retrospective review, such as heterogeneity in
patient population and incomplete or inaccurate data. Addi-
tionally, there are limitations related to studying type-IV supra-
condylar humeral fractures because of the lack of an objective,
“gold standard” diagnostic test. The interrater reliability results
were variable, ranging from substantial for 1 parameter to only
slight for another parameter, but most parameters had moderate
agreement. Nonetheless, several factors were statistically more
likely to be associated with a diagnosis of type-IV rather than
type-III supracondylar humeral fracture. It is important to note
that the k statistic is influenced by the prevalence of a given factor
or variable being considered14. When prevalence is low, as with
propagation of the fracture fragment toward the diaphysis, low k
values do not necessarily reflect poor overall agreement. Lastly, a
multivariable regression model could not be supported with the
current data because of the small sample size of this rare clinical
entity. The translation to other centers and the generalizability of
our findings cannot be confirmed because of this limitation.
Despite these limitations, this study had several strengths.
To our knowledge, this is the largest series of type-IV supra-
condylar humeral fractures and the first study to identify
preoperative factors associated with type-IV supracondylar
humeral fracture. Radiographic parameters were determined
 1356
TH E JO U R NA L O F B O N E & JO I N T SU RG E RY J B J S . O RG
d
P E D I AT R I C G A R T L A N D T Y P E -IV S U P R A C O N D Y L A R H U M E R A L
V O L U M E 1 01-A N U M B E R 15 A U G U S T 7, 2 019
d d
FRACTURE S

by consensus of 4 fellowship-trained observers, increasing the Joshua M. Abzug, MD5

generalizability compared with a single-observer approach. Micheal Raad, MD1
In conclusion, we identified several preoperative radio- Paul D. Sponseller, MD, MBA1
graphic factors that were associated with type-IV supracondylar 1Department of Orthopaedic Surgery, The Johns Hopkins University
humeral fracture. Preoperative identification of this rare fracture School of Medicine, Baltimore, Maryland
type would allow for improved preoperative planning (e.g., antic-
ipation of longer operative time, higher likelihood of ORPP) and 2Department of Orthopaedic Surgery, Texas Scottish Rite Hospital for
more accurate counseling of patients and their families about the Children, Dallas, Texas
likelihood of open reduction when a type-IV fracture is suspected. 3Department of Orthopaedic Surgery, Children’s Medical Center, Dallas, Texas

Appendix 4Department of Orthopaedic Surgery, UT Southwestern Medical Center,

Supporting material provided by the authors is posted Dallas, Texas
with the online version of this article as a data supplement
at jbjs.org (http://links.lww.com/JBJS/F362). n 5Department of Orthopaedic Surgery, University of Maryland Medical
NOTE: The authors acknowledge the contributions of Cheryl Lawing, MD, for the initial project Center, Baltimore, Maryland
conceptualization and imaging review and Richard L. Skolasky, ScD, for data management and
conceptualization of the statistical analysis. Additionally, the authors thank Rachel Box, MS, and
Jenni Weems, MS, for their editorial assistance and help preparing the manuscript for submission. E-mail address for P.D. Sponseller: psponse@jhmi.edu

ORCID iD for S.L. Mitchell: 0000-0001-9066-1088

ORCID iD for B.T. Sullivan: 0000-0003-0072-6891
ORCID iD for C.A. Ho: 0000-0002-6710-6040
Stuart L. Mitchell, MD1 ORCID iD for J.M. Abzug: 0000-0002-9821-7712
Brian T. Sullivan, MD1 ORCID iD for M. Raad: 0000-0001-8167-5808
Christine A. Ho, MD2,3,4 ORCID iD for P.D. Sponseller: 0000-0003-2934-6374

References
1. Omid R, Choi PD, Skaggs DL. Supracondylar humeral fractures in children. J Bone
Joint Surg Am. 2008 May;90(5):1121-32.
2. Skaggs DL, Frick S. Upper extremity fractures in children. In: Weinstein SL, Flynn
JM, editors. Lovell and Winter’s pediatric orthopaedics. 7th ed. Philadelphia: Lip-
pincott Williams & Wilkins; 2014. p 1694-772.
3. Skaggs DL, Flynn JM. Supracondylar fractures of the distal humerus. In: Flynn JM,
Skaggs DL, Waters PM, editors. Rockwood and Wilkins’ fractures in children. 8th ed.
Philadelphia: Wolters Kluwer Health; 2015. p 582-628.
4. Wilkins KE. Fractures and dislocations of the elbow region. In: Rockwood CA, Jr.,
Wilkins KE, King RE, editors. Fractures. 2nd ed, vol 3, fractures in children. Phila-
delphia: Lippincott; 1984. p 363-575.
5. Gartland JJ. Management of supracondylar fractures of the humerus in children.
Surg Gynecol Obstet. 1959 Aug;109(2):145-54.
6. Silva M, Cooper SD, Cha A. The outcome of surgical treatment of multidirec-
tionally unstable (type IV) pediatric supracondylar humerus fractures. J Pediatr Or-
thop. 2015 Sep;35(6):600-5.
7. Leitch KK, Kay RM, Femino JD, Tolo VT, Storer SK, Skaggs DL. Treatment of
multidirectionally unstable supracondylar humeral fractures in children. A modified
Gartland type-IV fracture. J Bone Joint Surg Am. 2006 May;88(5):980-5.
8. Howard A, Mulpuri K, Abel MF, Braun S, Bueche M, Epps H, Hosalkar H, Mehlman
CT, Scherl S, Goldberg M, Turkelson CM, Wies JL, Boyer K; American Academy of
Orthopaedic Surgeons. The treatment of pediatric supracondylar humerus fractures.
J Am Acad Orthop Surg. 2012 May;20(5):320-7.
9. Novais EN, Andrade MA, Gomes DC. The use of a joystick technique facili-
tates closed reduction and percutaneous fixation of multidirectionally unstable
supracondylar humeral fractures in children. J Pediatr Orthop. 2013 Jan;33(1):
14-9.
10. Flynn K, Shah AS, Brusalis CM, Leddy K, Flynn JM. Flexion-type supracondylar
humeral fractures: ulnar nerve injury increases risk of open reduction. J Bone Joint
Surg Am. 2017 Sep 6;99(17):1485-7.
11. Landis JR, Koch GG. The measurement of observer agreement for categorical
data. Biometrics. 1977 Mar;33(1):159-74.
12. Ernat J, Ho C, Wimberly RL, Jo C, Riccio AI. Fracture classification does not
predict functional outcomes in supracondylar humerus fractures: a prospective
study. J Pediatr Orthop. 2017 Jun;37(4):e233-7.
13. Joiner ER, Skaggs DL, Arkader A, Andras LM, Lightdale-Miric NR,
Pace JL, Ryan DD. Iatrogenic nerve injuries in the treatment of
supracondylar humerus fractures: are we really just missing nerve
injuries on preoperative examination? J Pediatr Orthop. 2014 Jun;34(4):
388-92.
14. Feinstein AR, Cicchetti DV. High agreement but low kappa: I. The problems of
two paradoxes. J Clin Epidemiol. 1990;43(6):543-9.