Keywords shoulder fractures - complications - conservative treatment - diagnostic imaging
Introduction
Osteoporosis-related fractures affect ∼ 2 million people per year in the United States,
and proximal humeral fractures (PHFs) account for 10% of these injuries.[1 ] Recent studies revealed that nonoperative treatment outcomes are similar to those
obtained with surgery, regardless of the age and fracture pattern of the patient.[2 ]
[3 ]
However, there are still controversies about the main fracture patterns benefiting
from surgical treatment and little evidence regarding prognostic radiographic parameters
for nonoperative treatment.
The Neer, Arbeitsgemeinschaft für Osteosynthesefragen (AO) and binary classification
systems have low reliability and add little to the therapeutic decision-making process.[4 ]
[5 ]
[6 ] Previous studies demonstrate that functional outcomes are influenced by several
fracture-related features that are not evaluated by these classification systems,
such as medial metaphyseal comminution,[7 ] deviation type and degree in coronal and sagittal views, and bone loss due to impaction.[8 ] Other prognostic factors are related to the patient and the injury, such as age,
osteoporosis and time from fracture to treatment.[9 ]
[10 ]
Few studies evaluate different radiographic criteria to predict functional outcomes
from the nonoperative treatment of deviated PHF.[8 ]
[11 ]
[12 ]
[13 ]
[14 ]
[15 ] In addition, no previous study evaluates the Resch criteria and other radiographic
variables in PHF prognosis.
The primary objective of the present study is to describe functional outcomes from
the nonoperative treatment of deviated PHFs using the American Shoulder and Elbow
Surgeons (ASES) score at a 12-month follow-up. The secondary objective is to evaluate
whether different classification systems and initial radiographic measurements are
related to worse outcomes according to the ASES score.
Methods
Study design
We conducted a prospective study involving 40 patients with deviated PHFs submitted
to nonoperative treatment. The patients belong to a randomized study and were treated
from February 2016 to October 2018 at a single center. The protocol was approved by
the Ethics Committee of our service under the opinion number 1.266.876.
Subjects
The inclusion criteria were age ≥ 60 years, trauma within the last 30 days and PHF
with surgical neck involvement and at least one of the following parameters: head-diaphysis
angle with ≥ 20° deviation (in varus or valgus); head-to-shaft translation > 1 cm
(in frontal or sagittal view) and greater and/or lesser tubercle fracture with > 0.5 cm
deviation. Two physicians specialized in shoulder and elbow surgery evaluated these
criteria before the patient was included in the study.
Dislocation-fractures, fractures with no contact between the humeral head and the
diaphysis, and bilateral fractures were not included. Patients with ipsilateral or
contralateral upper limb fractures, neurological injuries diagnosed at physical examination,
other fractures in the affected limb, pathological fractures, bilateral fractures,
previous surgery on the affected shoulder or previous full-thickness tear of one of
the rotator cuff tendons were not included in the present study.
Intervention
The nonoperative treatment consisted of velpeau sling use and early rehabilitation.
No patient underwent closed reduction. Analgesia was standardized with dipyrone (500 mg;
one tablet every 8 hours for 10 days), codeine (30 mg; one tablet every 6 hours for
7 days and then as required, according to pain) and paracetamol (500 mg; one tablet
each 8 hours for 10 days).
Guidance to sling use was standardized and each patient received a printed leaflet
containing the rehabilitation guidelines described below after inclusion in the study.
Home exercises were personally guided by a physical therapist on the 1st day of inclusion in the study.
Elbow, wrist, and hand movements started on the 1st day of inclusion in the study. Sleeping position, personal hygiene, and feeding modes
were advised. Home cervical, scapular, elbow, wrist, and hand exercises were oriented
on the 1st day of inclusion in the study; pendulum exercises were added on the 7th day. In 15 days, passive shoulder exercises for lifting, abduction and rotation were
started. After 30 days, active-resisted elbow exercises, active-assisted and active-free
shoulder exercises, as well as isometric exercises for the rotator cuff, the deltoid
and the scapular girdle were started as tolerated by the patients. Active-resisted
exercises for the rotator cuff, the deltoid and the scapular girdle began at the 45th day. The maximum time indicated for sling use was 30 days after the fracture.
Outcomes
The primary outcome consisted in clinical evaluation using the ASES score 12 months
after the fracture. Secondary outcomes were the following: Constant-Murley score,
both in absolute values and relative values for the contralateral limb (CRI) and the
Single Assessment Numeric Evaluation (SANE) score. Scores were assessed at 3, 6 and
12 months after the fracture. Clinical scores were applied by an evaluator who did
not participate in the physical rehabilitation or clinical follow-up of the patients.
Radiographs were obtained in four views, including the AP (anteroposterior) view,
Y lateral view, axillary view and Velpeau views, whenever tolerated by the patient.
Images were evaluated using the institutional image storage system (iSite enterprise
4.1, Phillips, Amsterdam, Netherlands). All patients underwent a computed tomography
(CT) scan before study inclusion. The CT results complemented the definition of radiographic
fracture variables, such as classifications and deviation features. Two orthopedists,
members of the Brazilian Shoulder and Elbow Society, with 11 and 12 years of experience,
evaluated all images and consensually determined ratings. The measurements used in
the study were taken by one of these evaluators.
An ultrasound examination was performed 6 months after the fracture to assess rotator
cuff integrity. Ultrasounds were obtained at the radiology department of the institution
by the musculoskeletal radiologist team using a Logiq E9 device (GE Healthcare, Waukesha,
WI, USA) with an ML6-15 linear transducer (6- to 15-MHz linear, ML-6-15, GE Healthcare,
Waukesha, USA). Findings related to rotator cuff tendons were classified according
to the full-thickness tear of one or more tendons.
Complications were noted according to their occurrence, and their total number was
recorded separately for each individual patient. Vicious consolidation was not deemed
a complication. The need for surgical approach and the type of surgery performed were
also recorded. The following complications were analyzed as binary variables (present
or absent) and were considered present if identified at any time during treatment:
recurrent pain with functional limitations requiring additional treatment (clinical
or surgical therapy) after 1 year of treatment; fracture-related clinical complication
or treatment requiring hospitalization; death related to the fracture or treatment;
re-fractures: shoulder stiffness, defined by decreased shoulder range of motion with
sustained functional limitation 6 months after the fracture; complex regional pain
syndrome; heterotopic ossification; glenohumeral osteoarthritis; and osteonecrosis
of the humeral head and pseudoarthrosis, defined by the lack of consolidation 1 year
after the fracture. Ancillary tests (CT, magnetic resonance imaging [MRI] or electroneuromyography)
were requested per clinical suspicion.
Analyzed variables
The analyzed clinical variables were the following: gender, side, dominance, smoking
status, diabetes, time until the beginning of rehabilitation, rehabilitation duration
and number of sessions.
Radiographic variables included Neer and Resch classification systems and presence
of tuberosity fracture (regardless of deviation), metaphyseal comminution, and medial
periosteal lesion. Several radiographic parameters of fracture deviation were evaluated.
The head-shaft angle was assessed as previously published.[16 ] Greater tuberosity deviation was evaluated according to the distance from the apex
of the humeral head at a AP (anteroposterior) view radiograph. The lesser tuberosity
was evaluated on axillary or velpeau views if the fragment was deviated. Other measures
were taken in relation to fragment deviation. All measures were categorized into three
levels according to previously established criteria. [Table 1 ] describes all analyzed radiographic variables.
Table 1
Metaphyseal comminution
No
Yes
Medial periosteal lesion
No
Yes
Tuberosity fracture
No tuberosity fracture
Greater tuberosity fracture
Lesser and greater tuberosities fracture
Greater tuberosity deviation
Up to 2 mm inferior to the top of the humeral head
Up to 2 mm superior to the top of the humeral head
Over 2 mm from the top of the humeral head
Lesser tuberosity deviation
Up to 2 mm
2 to 5 mm
> 5 mm
Diaphyseal head angle, front view
125° to 150°
110° to 124° or 151° to 165°
< 110° or >166°
Diaphyseal head angle, lateral view
No deviation
Moderate (0-20°)
Severe (> 45°)
Diaphyseal deviation, front view
Up to 5 mm
5 to 10 mm
> 10 mm
Diaphyseal deviation, lateral view
Up to 5 mm
5 to 10 mm
> 10 mm
Statistical analysis
Due to the sample size, nonparametric tests were used regardless of data normality.
Continuous variables were presented as mean and standard deviation (SD), median and
interquartile range values. Categorical values were shown as absolute and percentage
values. The sample was calculated for convenience and cases were included sequentially.
Functional scores before and after treatment were compared using Wilcoxon tests. Kruskal-Wallis
tests were performed for univariate analyzes of radiographic variables, and the Friedman
test was used for post-hoc analysis.
Data was analyzed in IBM SPSS Statistics for Windows, Version 20.0 (IBM Corp., Armonk,
NY, USA) adopting a 5% significance level.
Results
Forty-five patients were included for nonoperative treatment. Five patients did not
show up for the first assessment at 3 months, with complete loss to follow-up. Forty
patients with a complete 12-month follow-up were included at the final evaluation.
[Table 2 ] shows general sample data. Most patients were female, with a mean age of 69 ± 6
years old. [Figures 1 ], [2 ] and [3 ] show radiographic and clinical outcomes from one patient. The mean time until the
beginning of physical therapy was 21 ± 12 days; the mean duration of rehabilitation
was 111 ± 51 days, with 18 ± 10 sessions. The mean ASES score in patients who started
physical therapy more than 15 days after the fracture was 72 ± 25 points, whereas
patients who started rehabilitation within 15 days presented a mean ASES score of
79.9 ± 23.4 points.
Table 2
n
%
Gender
Male
7
17.5
Female
33
82.5
Side
Right
13
32.5
Left
27
67.5
Dominant side affected?
Yes
15
37.5
No
25
62.5
Smoking
No
31
77.5
Smoker
7
17.5
Former smoker
2
5.0
Diabetes
No
30
75.0
Yes, non-insulin-dependent
8
20.0
Yes, insulin-dependent
2
5.0
Fig. 1 Initial radiographs in AP (anteroposterior) view (A), lateral (B) and axillary (C)
views. Computed tomography showing the fracture with three-dimensional reconstruction
(D), in axial (E) and sagittal (F) sections.
Fig. 2 At the 12-month follow-up, images show fracture consolidation and varus deviation
in AP (anteroposterior) view (A), profile (B) and axillary (C) views. Clinical outcome
showing active elevation (D), active lateral rotation (E) and active medial rotation
(F).
Fig. 3 Radiography (a) and magnetic resonance imaging (b) demonstrating a case of osteonecrosis
at the humeral head.
The ASES score at 12 months was 77.7 ± 23.2 points for the entire sample, improving
over time. At 12 months, the mean absolute Constant-Murley score was 68.7 ± 16 points,
with 82.6% for the contralateral side. The SANE score at 12 months was 84.8 ± 19 points.
[Table 3 ] shows clinical outcomes.
Table 3
n
Mean
Standard deviation
Median
IQR
ASES
3 months
39
57.7
24.5
56.3
42.8
6 months
39
71.6
24.3
80.7
36.0
12 months
40
77.7
23.2
85.2
40.1
SANE
3 months
39
66.2
22.6
70.0
40.0
6 months
39
82.2
19.4
85.0
22.5
12 months
40
84.8
19.0
90.0
20.0
Constant-Murley
3 months
39
55.6
16.7
60.0
31.0
6 months
40
64.8
15.9
68.5
28.5
12 months
40
68.7
16.0
72.0
24.0
Relative Constant score for the contralateral limb
6 months
39
80.1%
18.7%
87.2%
37.8%
12 months
39
82.6%
23.6%
89.5%
31.3%
Ultrasonography was performed in 37 patients (92.5%). Full-thickness tears of the
rotator cuff were observed in eight subjects. The mean ASES score in patients presenting
cuff tear was 76.9 ± 24.0 points at 12 months, with no statistically significant difference
compared with patients with no cuff injuries (p = 0.188).
Eight patients (20%) had complications, including 4 cases of osteonecrosis (10%),
2 cases of pseudarthrosis (5%) and 2 cases of persistent stiffness (5%). No patient
chose for a surgical approach to treat complications. [Figure 3 ] shows an osteonecrosis case and [Figure 4 ] demonstrates a pseudoarthrosis case. The mean ASES score in patients with complications
was 53.8 ± 23.0 points.
Fig. 4 Initial radiograph (A) and at the 12-month follow-up (B) from a patient with humeral
neck pseudoarthrosis. Clinical examination (C) showing maximum active elevation.
The analysis of radiographic variables revealed the negative influence of the following
parameters on the clinical outcome according to the ASES score at 12 months: severity
according to the Neer classification and angular deviation at the coronal plane (measured
by the head-diaphysis angle) and tuberosity fracture. The remaining radiographic variables
did not influence functional outcomes according to the ASES score. [Table 4 ] shows a subgroup analysis for the main variables of the present study.
Table 4
ASES score at 12 months
Mean
Standard deviation
Median
IQR
p-value
Neer classification
2 parts
88.4
15.4
94.2
19.9
3 parts
79.0
21.5
85.2
32.3
4 parts
61.0
28.3
47.5
56.0
0.031
Diaphyseal head angle, coronal view
125° to 150°
93.4
7.8
96.3
11.9
110° to 124° or 151° a 165°
74.5
25.2
78.4
45.4
< 110° or >166°
64.3
25.0
56.5
48.2
0.004
Angular deviation, coronal view
Minimal deviation
84.7
20.4
93.8
26.1
Severe valgus (>166°)
60.0
22.3
58.1
41.9
Severe varus (<110°)
66.2
27.1
56.5
54.9
0.034
Tuberosity fracture
No tuberosity fracture
84.7
19.7
94.2
38.4
Greater tuberosity fracture
82.8
20.3
88.1
27.5
Lesser and greater tuberosities fracture
61.5
26.9
50.4
54.6
0.031
Resch classification
1
94.2
6.4
95.9
12.6
2
88.1
16.8
88.1
NA
3
64.4
27.1
61.5
53.0
4
76.2
23.7
84.9
44.8
0.052
Metaphyseal comminution
No
75.8
23.7
85.4
49.7
Yes
77.7
25.2
85.0
36.7
0.570
Medial periosteal lesion
No
82.9
22.3
93.2
27.3
Yes
71.6
24.7
78.5
51.7
0.168
Discussion
Our results demonstrate that even fractures with significant tuberosity or humeral
head deviation can be treated nonoperatively. The Constant-Murley score for the contralateral
side was 82.6%, consistent with other studies on nonoperative treatment,[8 ]
[11 ]
[12 ]
[13 ]
[14 ] as well as with the surgical treatment using arthroplasty or locked plate fixation.[17 ]
[18 ]
[19 ]
[20 ] Subjective scores (i.e., ASES and SANE) were similar to those reported in systematic
reviews.[2 ] Our results reinforce findings from the Proximal Fracture of the Humerus Evaluation
by Randomization (PROFHER) study and a Cochrane systematic review demonstrating that
nonoperative treatment can provide good functional outcomes even in deviated fractures.[2 ]
[3 ]
In a systematic review, Sabharwal et al.[21 ] obtained similar clinical outcomes with both treatments. However, these authors
demonstrated that, in more complex, four-part fractures, the surgical treatment resulted
in better clinical outcomes and a lower rate of complications. They highlighted the
need for future studies with specific fracture subtypes.
Few investigations evaluated the influence of the classification and deviation criteria
defined by Neer on the nonoperative treatment of PHF.[8 ]
[14 ] Although routinely used as criteria to indicate surgical treatment, a 45° angular
deviation, 1 cm translational deviation and 0.5 cm tuberosity deviation were arbitrarily
defined, as explained by Neer himself in 2002.[22 ]
Our secondary objective was to evaluate not only the deviation criteria described
by Neer, but also several other radiographic parameters. The classification recently
described by Resch et al.[23 ] has better inter- and intraobserver reliability when compared with the Neer system,
but its potential in determining prognosis has not been evaluated yet.[6 ]
Univariate analyses revealed statistically significant, clinically relevant differences
for some radiographic variables, mainly angular deviation, assessed by the head-diaphysis
angle, and greater and lesser tuberosity fracture, regardless of its deviation. The
mean ASES score in patients with greater and lesser tuberosity fractures was 23 points,
a clinically significant relevant finding. Patients with deviated, four-part fractures
according to Neer presented worse outcomes, consistent with Yüksel et al.;[14 ] however, there was no statistically significant difference in two- or three-part
fractures.
The head-diaphysis angle may be difficult to be evaluated on radiographs of acute
fractures and presents variable interobserver correlation.[16 ]
[24 ] Nevertheless, with an adequate radiographic standardization and the potential use
of CT scans, the head-diaphysis angle is an important parameter to aid the decision-making
process, as it directly influences clinical outcomes. Patients with < 110° or > 166°
had 29 points less on the ASES score when compared with subjects with minimal angular
deviation (125° to 150°). Yüksel et al.[14 ] did not observe such influence on the Constant-Murley score, with similar results
between valgus or varus impacted fractures < 110°.
Comparing only the mean head-diaphysis angles can lead to incorrect interpretations
when fractures with varus and valgus deviation are included, as it approximates figures
from opposite ends. The categorized analysis, such as the one performed in our study,
or separated by valgus or varus deviations, avoids evaluation bias. In contrast to
Court-Brown et al.,[12 ] in our sample, subjects with deviated fractures in severe valgus (> 166°) had lower
ASES scores compared to those with severe varus (< 110°), but there was no statistically
significant difference. Consistent with our findings, Foruria et al.[8 ] demonstrated that valgus impacted fractures had a three-fold higher risk of worsening
10 points on the Disability of Arm, Shoulder, and Hand (DASH) score when compared
with varus impacted injuries.
Other factors did not show a statistically significant difference. The most prominent
factor was the presence or absence of metaphyseal comminution, medial periosteal lesion,
and deviations at the sagittal plane. Some studies[25 ]
[26 ] observed worse outcomes regarding stability after plate fixation in patients with
posteromedial metaphyseal comminution, but no study has evaluated the importance of
this finding in the nonoperative treatment.
We were unable to detect a statistically significant difference in functional scores
for the classification described by Resch et al.[23 ] Nevertheless, mean values for subtypes in valgus (type 3) and varus (type 4) were
lower than types 1 and 2.
Rehabilitation-related variables showed no difference in subgroup analyzes either
for time until the start of rehabilitation and its total duration and number of sessions.
As for rotator cuff injuries, there was no relationship between injury and worse clinical
outcomes. Although we did not include patients with previous rotator cuff injury,
it is possible that some cases were asymptomatic lesions, explaining the high rate
observed in our sample. In addition, ultrasound scans in patients with fracture sequelae
may be less accurate due to mobility limitations or deformities.
Rotator cuff tears did not influence the clinical outcome according to the ASES score
in our sample.
Our study has some limitations. Since the sample size is relatively small for multivariate
analysis, these radiographic factors were submitted to a univariate analysis, which
can increase the risk of confounding factors. However, this is a homogeneous sample
consisting of patients with similar features.
As positive points, this is a prospective clinical and radiographic evaluation, in
standardized times, with an evaluator who did not participate in the study and did
not follow-up the patients. All fractures included had significant deviations, which
could be considered an indication for surgical treatment; there were no fractures
with no or minimal deviation in this investigation. Our results can assist the decision-making
process between non-operative and surgical treatment for some specific patterns of
PHFs.
Conclusion
Nonoperative treatment of deviated PHFs in elderly patients had good clinical outcomes.
These outcomes were negatively influenced by the Neer classification, the angular
deviation of the humeral head and the presence of greater and lesser tuberosities
fractures.
