J Shoulder Elbow Surg (2009) 18, 612-621

www.elsevier.com/locate/ymse

Functional outcomes after nonoperative management

of fractures of the proximal humerus
Beate Hanson, MD, MPHa, Philipp Neidenbach, MDa, Piet de Boer, BChir, MB, FRCSa,b,
Dirk Stengel, MD, PhD, MSca,c,*

a
AO Clinical Investigation and Documentation, Dübendorf, Switzerland
b
Department of Traumatology, York District Hospital, York, United Kingdom
c
Center for Clinical Research, Unfallkrankenhaus Berlin and University Hospital of Greifswald, Berlin, Germany

Background: Prospective follow-up data after nonoperative treatment for fractures of the proximal
humerus are scarce. We studied functional outcomes and rates of complication and failure after conserva-
tive management of these common injuries.
Materials and methods: Consecutive patients aged older than 18 years presenting to the emergency
department of a large district hospital with an isolated, closed proximal humeral fracture considered suit-
able for functional treatment by the surgeon on charge were enrolled in a prospective, externally monitored
observational study. Surgeons were free to reduce the fracture and to prescribe any type of sling or brace.
Active follow-up after 12 weeks, 6, and 12 months included plain radiographs, Constant score, and Disabil-
ities of Arm, Shoulder and Hand (DASH) score.
Results: We enrolled 160 patients (118 women; mean age, 63.3  14.8 years), and 124 completed 1-year
follow-up. There were 85, 71, and 4 AO 11 A, B, and C fractures, and 75 one-part, 60 two-part, 23 three-
part, and 2 four-part and head-splitting fractures. After 1 year, the mean difference in Constant scores
between the injured and contralateral shoulder was 8.2 (95% confidence interval [CI], 6.0-10.4). The
mean difference in 1-year DASH scores to baseline assessment was 10.2 points (95% CI 7.3-13.1 points).
The risk of delayed and nonunion was 7.0% (95% CI, 3.6%-12.3%). Four patients subsequently underwent
surgical fixation, and 5 had arthroscopic subacromial decompression.
Conclusion: This study may provide reference values for future investigations and stresses ceiling effects
that will make it difficult to demonstrate a significant advantage of surgical over nonoperative treatment in
patients with proximal humeral fractures.
Level of evidence: Level 4; Prospective case series without a control group.
Ó 2009 Journal of Shoulder and Elbow Surgery Board of Trustees.
Keywords: Proximal humeral fracture; functional treatment; Constant score; Disabilities of the Arm,
Shoulder and Hand

*Reprint requests: Dirk Stengel, MD, PhD, MSc, Head, Center for
Clinical Research, Department of Orthopedic and Trauma Surgery,
Unfallkrankenhaus Berlin and University Hospital of Greifswald
Unfallkrankenhaus Berlin, Warener Str 7, 12683 Berlin, Germany.
E-mail address: stengeldirk@aol.com (D. Stengel).
Fractures of the proximal humerus and the humeral
head are most common injuries worldwide.4 The reported
incidence rates range from 21/100,000 person-years in
Japanese men to 221/100,000 person-years in Swedish

1058-2746/2009/$36.00 - see front matter Ó 2009 Journal of Shoulder and Elbow Surgery Board of Trustees.
doi:10.1016/j.jse.2009.03.024
Nonoperative management of proximal humerus fractures
Figure 1 Patient selection procedure and completeness of follow-up.
women.11 According to current projections, every 20th
woman at 65 years of age will sustain a fracture of the
proximal humerus in her remaining life time.1
The availability of modern orthopedic implants, such as
locking plates with excellent biomechanical stability even
in osteoporotic bone, may have contributed to the trend
toward surgical treatment of these fractures. Internal fixa-
tion restores the anatomy of the proximal humerus but
requires hospitalization and is associated with the risk for
implant-related complications. Functional therapy with
short immobilization, followed by an accelerated physio-
therapy protocol, is a simple, convenient, noninvasive,
ubiquitously available, and efficient management option.
Court-Brown et al5-7 observed good to excellent clinical
results (as assessed by the Constant score) in non-
operatively treated impacted valgus and varus fractures.
Apart from these series, however, little is known on phys-
ical recovery and deficits after conservatively treated frac-
tures of the proximal humerus.
We prospectively assessed the function of the upper
extremity in a large cohort of patients who were assigned
to nonoperative treatment of a proximal humeral fracture.
Our primary objective was to obtain robust reference
values for the Constant score and the Disabilities of the
Arm, Shoulder, and Hand (DASH) instrument in this
common clinical setting. Further objectives were to eval-
uate the duration of sick leave, complications, and treat-
ment failures and to model possible risk factors and
predictors of shoulder function 1 year after the fracture
event.
613
Materials and methods
The local Institutional Review Board (York Research Ethics
Committee, UK [Reference No. YREC 02/11/006)] approved the
study, and all patients provided written informed consent.
Study design and entry criteria
All patients with a radiographically proven, closed fracture of the
proximal humerus admitted to the emergency department of a large
district hospital in the United Kingdom, who were considered
suitable for primary nonoperative management by the physician on
charge, were asked to participate in this investigation.
We excluded skeletally immature patients, patients presenting
to the hospital 10 days or more after injury, patients with open
fractures or multiple trauma and preexisting illness affecting the
function of the upper limb, such as multiple sclerosis, paraplegia,
and others. Also excluded were patients with a history of drug or
alcohol abuse and those who were deemed unlikely to cooperate
or attend all scheduled study visits.
Baseline documentation
On admission, patient demographics (ie, gender, age, profession,
dexterity, smoking, concomitant diseases, and medication) and
injury characteristics (ie, accident type, energy level of trauma,
concomitant injuries, fracture classification) were recorded. Patients
were asked to rate their upper limb function 1 week before the
accident to determine their baseline DASH score using the extended
3-modular questionnaire.13 Normalized DASH scores range from
0 (perfect function) to 100 (functionless extremity/joint).
614 B. Hanson et al.

Table I Differences in demographic baseline values between included and excluded patients
Item Clinical journal Study cohort Difference, % (95% CI) P
Patients, No. 172 160
Age, mean (SD), y 67.5 (18.1) 63.3 (14.8) 4.2 (0.6 to 7.8) .022
Male gender, 51 (29.7) 42 (26.3) 3.4 (e6.2 to 13.0) .490
No. (%)
Smokers, No. (%) 40 (23.5) 24 (15.0) 8.5 (0.1 to 17.0) .050
AO type, No. (%) .003
A 106 (61.6) 85 (53.1) 8.5 (e2.1 to 19.1)
B 51 (29.7) 71 (44.4) 14.7 (25.0 to 4.4)
C 15 (8.7) 4 (2.5) 6.2 (1.4 to 11.1)
AO subtype, n (%) .081
A1 39 (22.7) 32 (20.0) 2.7 (e6.1 to 11.5)
A2 25 (14.5) 23 (14.4) 0.2 (e7.4 to 7.7)
A3 42 (24.4) 30 (18.8) 5.7 (e3.2 to 14.5)
B1 38 (22.1) 55 (34.4) 12.3 (21.9 to e2.7)
B2 6 (3.5) 5 (3.1) 0.4 (e3.5 to 4.2)
B3 7 (4.1) 11 (6.9) 2.8 (7.7 to 2.1)
C1 5 (2.9) 0 (0.0) 2.9 (0.2 to 5.7)
C2 5 (2.9) 2 (1.3) 1.7 (e1.4 to 4.7)
C3 5 (2.9) 2 (1.3) 1.7 (e1.4 to 4.7)
CI, Confidence interval.

Radiographs were obtained in anterior-posterior projection and
Neer’s view upon admission in the emergency department and
after manipulation. Additional computed tomography (CT) scans
were ordered at the discretion of the treating surgeon. Fractures
were classified according to the AO and Neer scheme by the
physician on charge.19
Surgeons were free to reduce the fracture, if deemed necessary.
Complications with attempted fracture reduction were recorded.
We documented the type and duration of immobilization and
beginning of active assisted and unrestricted mobilization as well
as the need for immediate or later surgical repair.
Follow-up
Patients were actively monitored and physically examined after 3,
6, and 12 months. Primary outcome measures comprised raw
Constant scores and differences to the contralateral, healthy
shoulder.3 Power was instrumentally measured by a spring balance
with the arm held in 90 abduction. The DASH was recorded as
a coprimary end point at the final 12-month visit.
Plain radiographs of the injured shoulder in 2 planes were
obtained to determine fracture healing.
Anticipated complications included loss of reduction, fragment
dislocation, axis deviation, head necrosis, nonunion, impingement
and shoulder stiffness, muscle weakness, neurologic sensations
(ie, tingling or numbness of the ipsilateral hand, brachial plexus
palsy, and others), pain, and death during follow-up. Reported
complications were reviewed by the principal investigator (PdB)
and an independent data safety monitoring board to define
whether complications were related to the fracture, fracture
treatment, or other conditions.
Patients were further interviewed concerning their pain and
shoulder mobility, and clinically examined using 3 power
measurements of both shoulders in abduction to individual
Constant scores of the injured and contralateral shoulder.
Data were entered in a Qualicare 5.1 database (Qualidoc,
Trimbach, Switzerland), and monitored by an independent trial unit.
Statistical analysis
Because of the observational character of this study, we did not
pose a formal null hypothesis. A target sample size of 120 fully
documented patients was deemed to yield sufficient precision of
estimated means and proportions, with standard errors not
exceeding 5%. Assuming a 20% drop-out rate, we planned to
enroll 160 patients. All analyses were made according to the
intent-to-treat principle; that is, all patients who were assigned to
conservative management were included in the final evaluation,
regardless of complications or subsequent surgery.
We calculated means, medians, proportions, and risks with
their appropriate measures of distribution (ie, standard deviations,
ranges), as well as odds ratios (OR) and 95% confidence intervals
(CI). Categoric variables were tabulated with absolute and relative
frequencies. We used the McNemar test for paired dichotomous
variables, the Fisher exact test for independent dichotomous
variables, the Kruskal-Wallis nonparametric analysis of variance
for continuous variables and multiple group comparisons, and the
paired t test for intrasubject comparisons of mean values. All
values of P were interpreted descriptively.
Univariate and multivariate logistic and linear regression anal-
yses were computed to identify variables independently associated
with primary and secondary outcomes. The probability of fracture
consolidation and return to work was estimated by the Kaplan-
Meier product limit method. Because it is unlikely that a fracture
that had been considered united after 6 months will be classified as
a nonunion after 1 year, a last observation carried forward approach
was used in case of missing data at the follow-up examinations. The
log-rank test was applied for comparisons of survival curves
between subgroups of patients. Statistical analyses were conducted
with Stata 10 software (StataCorp, College Station, TX).
Nonoperative management of proximal humerus fractures 615

smoke, and had a higher prevalence of AO type C fractures

Table II Patient profile of the study cohort
(Table I).
Item No. % or (SD) Of the 160 patients (118 women) included, 139 (86.9%)
Patients 160 100 were available for clinical examination after 3 and 6
Age, mean (SD), y 63.3 (14.8) months, and 124 (77.5%) completed 1 year of follow-up.
Gender One patient died of reasons not attributed to the fracture,
Female 118 73.8 for a mortality rate of 0.6% (95% CI, 0.1%-3.4%).
Male 42 26.3 With a mean age of 63.3  14.8 years and a 3:1
Smokers 24 15.0
predominance of women, this cohort reflected the typical
Working status
Retired/unemployed 103 64.4
spectrum of patients sustaining a fracture of the proximal
Part-time 18 11.3 humerus (Table II). It is noteworthy that 35% were still
Full-time 38 23.8 employed part-time or full-time. Also, more than 40%
Mainly manual work 21 13.1 patients participated in sports.
Affected side Patients’ ratings of DASH scores 1 week before the
Right 78 48.8 injury with a mean of 3.7 (95% CI, 2.2-5.2) indicated
Left 82 51.3 almost perfect function of the affected shoulder and upper
Affected dominant extremity 74 46.3 limb .
Sporting activities 67 41.9 Comorbidities, mostly cardiovascular and respiratory
Cause of accident diseases, were present in 90 patients (56.3%). Conse-
Sport and leisure 69 43.1
quently, about 15% of study participants received aspirin,
Home 65 40.6
Traffic 10 6.3
10% received b-blockers, and 4% took steroids.
Work 8 5.0 The distribution of fractures suited the common clinical
Other 8 5.0 indication for nonoperative treatment, with about half of all
Soft-tissue injury (Oestern-Tscherne) fractures classified as AO type A and undisplaced, ‘‘1-part’’
0 154 96.3 injuries (Table III). However, the cohort also included 23
1 4 2.5 (14.4%) 3-part, and 71 (44.4%) AO type B fractures. A
2 2 1.3 high-energy injury occurred in 18 patients (11.3%). Six
Multiple fractures 8 5.0 patients (3.8%) had concomitant fractures of the distal
High-energy injury 18 11.3 radius (n ¼ 3), the facial bones (n ¼ 3), thumbs or fingers
Concomitant disease (n ¼ 2), and the hip joint (n ¼ 1).
Cardiovascular 46 28.8
Respiratory 17 10.6
Osteoarthritis 11 6.9
Diabetes mellitus 10 6.3
Treatment details
Psychiatric 8 5.0
Gastrointestinal 8 5.0 All participants were treated as outpatients, with initial
Rheumatoid arthritis 4 2.5 immobilization mostly achieved in a cuff and collar (Table
Neurologic 3 1.9 IV). As a rather unusual finding, attempts to reduce the
Medication fracture were made in more than one-third. Patients were
Aspirin 23 14.4 permitted to start assisted shoulder mobilization after 22.8
b-Blockers 16 10.0 days (95% CI, 20.8-24.8 days), and unrestricted mobiliza-
Glucocorticoid/ 6 3.8 tion was allowed after a mean of 35.9 days (95% CI, 33.5-
immunosuppressant
38.5 days).
Inhaler 4 2.5
Oral 2 1.2
In 7 patients (4.4%; 95% CI, 1.8%-8.8%), surgery was
NSAID or COX-II inhibitors 4 2.5 considered necessary to optimize fracture or fragment
positioning (Table V), and 4 (2.5%, 95% CI, 0.7%-6.3%)
COX, cyclooxygenase; NSAID, nonsteroidal anti-inflammatory drug.
underwent the planned procedure within the observation
period. Five patients (3.1%, 95% CI, 1.0%-7.1%) subse-
quently underwent arthroscopic subacromial decompres-
Results sion because of impingement. Three patients received intra-
articular local anesthetics. One patient underwent shoulder
Study profile and patient characteristics manipulation under general anesthesia. Another patient was
offered prosthetic replacement but refused.
Of 332 patients screened at the referring institution, 172 Treatment-related complications included 17 fracture
(51.8%; 95% CI, 46.3%-57.3%) were excluded for various displacements and losses of reduction after initial manip-
reasons (Figure 1) and enrolled in a clinical journal group. ulation (10.6%, 95% CI, 6.3%-16.5%) and 11 episodes
Patients excluded from the study were older, more likely to of impingement and shoulder stiffness (6.9%, 95% CI,
616 B. Hanson et al.

Table III Fracture details (study cohort, n ¼160)

Type AO Total
Neer A1 A2 A3 B1 B2 B3 C2 C3 No. (%) No. (%)
Nondisplaced, 1-part) 17 15 10 27 0 6 0 0 75 (46.9) 75 (46.9)
Anatomic neck, 2-part 0 1 0 0 0 0 0 0 1 (0.6) 60 (37.5)
Fracture dislocation
2-part anterior 10 0 0 0 0 0 0 0 10 (6.3)
2-part posterior 1 0 0 0 0 0 0 0 1 (0.6)
Surgical neck, 2-part 0 7 20 9 0 4 1 0 41 (25.6)
Greater tuberosity
2-part 4 0 0 3 0 0 0 0 7 (4.4)
3-part 0 0 0 15 4 1 0 1 21 (13.1) 23 (14.4)
Lesser tuberosity, 3-part 0 0 0 1 0 0 1 0 2 (3.1)
Greater/lesser tuberosity, 0 0 0 0 1 0 0 0 1 (0.6) 2 (1.3)
4-part
Head-splitting- articular 0 0 0 0 0 0 0 1 1 (0.6)
surface
Total, No. (%) 32 (20.0) 23 (14.4) 30 (18.8) 55 (34.4) 5 (3.1) 11 (6.9) 2 (1.3) 2 (1.3)
85 (53.1) 71 (44.4) 4 (2.5)

at 74.3 (95% CI, 72.0-76.5) and 8.2 (95% CI, 6.0-10.4)

Table IV Treatment details (study cohort, n ¼ 160)
after 1 year (Figure 2). Of note, these differences were only
Item No. %, (SD), gradually influenced by the severity of fracture displace-
(range)
ment (Kruskal-Wallis test, P ¼ .0791; Figure 3).
Immobilization After 1 year, the overall difference in DASH scores
Cuff and collar 110 69.2 compared with baseline assessment was 10.2 points (95%
Broad arm sling 43 27.0 CI, 7.3-13.1 points; paired t test, P < .0001; Figure 4),
Brace 6 3.8
indicating that patients had not fully recovered. Of note,
Length of immobilization
employed patients showed significantly lower differences
Mean (SD), d 24.1 (13.8)
Median (range), d 21 (1e88) between DASH scores before and after injury than part-
Fracture reposition 65 36.9 time or unemployed patients (Kruskal-Wallis test, P ¼
Fracture reposition 8 5.0 .0417; Figure 5).
under anesthesia Differences in DASH scores were comparable among
Treatment complications age groups up to the 70% percentile (< 73 years), whereas
Problems with reposition 4 2.5 Constant scores decreased slightly but monotonously with
Neurologic deficits 4 2.5 increasing age (Figure 6). Patients with fracture-related
Extensive bruising 1 1.2 complications had slightly higher differences in Constant
scores than those without complications (13.5 vs 7.1; mean
3.5%-11.9%). To enhance precision, these events were difference, 6.5; 95% CI, 0.8-12.1; P ¼ .0265). This was
merged to a composite measure for further analyses. mainly caused by episodes of shoulder impingement. No
Logistic regression failed to reveal significant associations other variables had a marked effect on shoulder function
(at conservative P < .1) between certain potential risk after 1 year (Figure 7).
factors and the composite of impingement and displace- As a striking finding, 41 of 42 employed patients
ment, although smokers may have a 2.25-times increased (97.6%) returned to work (95% CI, 87.4%-99.9%). The
risk compared with non-smokers (Table VI). Of the 17 predicted median time off work was 10 weeks (95% CI, 8-
displacements, 9 were later regarded resolved with, and 5 16 weeks for survival time data; Figure 8). There was no
without, persistent damage. Of the 11 patients with difference in sick leave between manual and nonmanual
impingement syndrome, 3 each resolved with and without workers (10 vs 9 weeks; log-rank test P ¼ .3612).
persistent damage. The remaining outcomes were unclear.
Bony union
Functional outcomes
Altogether, 93% of patients showed solid union after 1 year
Constant scores of the injured shoulder and side-to-side of follow-up. Using only the available number of patients at
differences improved steadily over time and were estimated different follow-up intervals for calculation of healing rates,
Nonoperative management of proximal humerus fractures 617

Table V Patients deemed to require or undergoing surgery

Sex Age Profession AO Neer Surgical procedure
M 59 Chartered surveyor A3 Surgical neck, 2-part Planned for pinning
F 62 University lecturer B2 Greater/lesser tuberosity, 4-part Planned for nailing
F 65 Retired A1 Greater tuberosity, 2-part Suturing
F 58 Retired B2 Greater tuberosity, 3-part ORIF
M 72 Retired A1 Greater tuberosity, 2-part Screw fixation,
greater tuberosity
F 63 Retired C3 Greater tuberosity, 3-part ORIF
F 76 Retired B1 Greater tuberosity, 3-part Offered ORIF, but refused
F, Female; M, male; ORIF, open reduction, internal fixation.

Table VI Association between potential risk factors and the composite end point of impingement and secondary loss of reduction
(logistic regression analysis)
Variable No. Impingement Dislocation Composite OR (95% CI) P
Age 0.99 (0.96-1.01) .343
Sex 1.15 (0.47-2.86) .759
Male 42 2 6 8
Female 118 9 11 20
Smoking status 2.25 (0.83-6.10) .109
Smokers 24 3 4 7
Nonsmokers 136 8 13 21
AO type 1.38 (0.61-3.14) .435
A 85 4 9 13
B 71 7 7 14
C 4 0 1 1
Neer type 1.45 (0.84-2.52) .377
1-part 75 9 2 11
2-part 60 0 10 10
3- or 4-part 25 2 5 7
Soft-tissue damage 0.94 (0.11-8.38) .956
Grade 0 154 11 16 27
Grade 1/2 6 0 1 1
Fracture reduction 1.33 (0.59-3.03) .492
Reduction 65 5 8 13
No reduction 95 6 9 15
CI, Confidence interval; OR, odds ratio.

98% showed fracture consolidation. The predicted risk of
delayed and nonunion with nonoperative management of
fractures of the proximal humerus was 7.0% (95% CI,
3.6%-12.3%).
The median time to definite union was estimated at 14
weeks (95% CI, 12-16 weeks), and the time to union or
bridging callus was 13 weeks (95% CI, 12-16 weeks). Not
surprisingly, smoking was a significant risk factor for
nonunion. Smokers had a 5.5-times increased likelihood of
nonunion compared with smokers (20.8% vs 4.5%; OR,
5.5; 95% CI, 1.5-19.9).
Discussion
The present investigation represents one of the largest and
most detailed studies on outcomes after nonoperative
treatment for fractures of the proximal humerus. Active
follow-up and external monitoring ensured completeness of
data sets and unbiased estimates. Demographic baseline
data suggest that the included participants are representa-
tive of patients presenting to emergency departments in
Europe and the United States.
The observed mean difference (10.2  16.5) between
preinjury and 1-year DASH scores was below the threshold
suggested as minimal detectable change.20 Unfortunately,
no minimal detectable change thresholds are currently
available for the Constant score.
The degree of recovery suits the observations from
Court-Brown and others, but is still surprising.5,7,8,15,21,24
One might argue that most patients had less severe injuries
(ie, nondisplaced, or 1-part and 2-part fractures), with
a predictably good prognosis.
618
Figure 2 Box and whisker plot shows steady narrowing of
mean Constant scores of the injured and the contralateral shoulder
over time. The horizontal line in the middle of each box indicates
the median; the top and bottom borders of the box mark the 75th
and 25th percentiles, respectively. The whiskers mark the 90th and
10th percentiles. The circles designate outliers.
Figure 3 Box and whisker plot shows gradual influence of the
Neer classification on differences in Constant scores. The hori-
zontal line in the middle of each box indicates the median; the top
and bottom borders of the box mark the 75th and 25th percentiles,
respectively. The whiskers mark the 90th and 10th percentiles. The
circles designate outliers.
There was evidence of a healthy worker effect, with
fully employed patients showing smaller differences in
before and after DASH ratings. Of note, patients with more
displaced fractures rated their preinjury shoulder function
B. Hanson et al.
Figure 4 Box and whisker plot shows difference between
baseline and 1-year Disabilities of Arm, Shoulder and Hand
(DASH) scores. The horizontal line in the middle of each box
indicates the median; the top and bottom borders of the box mark
the 75th and 25th percentiles, respectively. The whiskers mark the
90th and 10th percentiles. The circles designate outliers.
slightly worse than those with undisplaced fractures. This
stresses the methodologic difficulties of obtaining valid
baseline values after an injury event has occurred.
The finding that age predicted both self-rated and
objective shoulder outcomes is not surprising, but older
patients regained much of their original function after 1
year. Because of compensation mechanisms, upper
extremity function as assessed by the DASH may have been
rated less impaired than shoulder mobility as specifically
measured by the Constant score.
It is widely accepted that fractures with displacement of
more than 1 cm and more than 45 of tilt need to be
surgically fixed. This rule has mainly been derived from
Neer’s classic work in 1970,19 but still needs more support
by scientific evidence.
The severity of fracture comminution and displacement
may have a more significant effect on functional outcomes
than the choice of treatment, with a clear difference in
prognosis between 3- and 4-part fractures, but not between
2- and 3-part fractures.
According to 7 recently published studies, average
Constant scores after surgically treated 2-, 3-part, and 4-
part fractures can be estimated at 82.2 (95% CI, 77.7-86.8),
83.9 (95% CI, 77.7-86.8), and 70.5 (95% CI, 64.9-
76.0).2,9,10,12,14,18,22
The high frequency of closed reductions in this study is
a matter of concern. Attempts to reduce a proximal humeral
fracture may have no, if any, effect on rates of malalign-
ment or functional outcomes. Given the potential risk for
soft tissue and plexus injuries caused by manipulation,
Nonoperative management of proximal humerus fractures
Figure 5 Association between Disabilities of Arm, Shoulder and Hand (DASH) scores and Constant scores and age. Percentiles include
the following age ranges in years: 1 ¼ 21-42; 2 ¼ 43-52; 3 ¼ 53-57; 4 ¼ 58-61; 5 ¼ 62-66; 6 ¼ 67-70; 7 ¼ 71-73; 8 ¼ 74-76; 9 ¼ 77-80;
10 ¼ 81-86.
Figure 6 Box and whisker plot shows the effect of employment
status on Disabilities of Arm, Shoulder and Hand (DASH)
differences on baseline values after 1 year of follow-up, suggestive
of a healthy worker effect (Kruskal-Wallis test P ¼ .0417). The
horizontal line in the middle of each box indicates the median; the
top and bottom borders of the box mark the 75th and 25th
percentiles, respectively. The whiskers mark the 90th and 10th
percentiles. The circles designate outliers.
physicians should critically reconsider the indication for
fracture reduction.17,23
There are obvious limits inherent to the observational
design of this study. First, only half of all eligible patients
were included in the study, introducing selection bias. We
619
did not assess the specific reason for the lack of informed
consent and withdrawals. The monitor queried whether
informed consent had been obtained, and patients without
a completed consent form were excluded according to
protocol, good clinical practice, and Institutional Review
Board mandates. This group might have included patients
who actively refused study participation because they felt
harassed, were too troubled by their acute condition, were
unwilling to attend follow-up (in contrast to those who were
unable to attend follow-up), or simply were not interested
in taking part in a clinical study. On the other hand, it may
include patients who were screened and considered ineli-
gible by the physicians on duty or those who were missed
because of unawareness of the protocol, time constraints,
and other factors.
It is difficult to decide whether the imbalance in the
clinical profile between included and excluded patients was
coincidental or shows the conditional probability of
a patient of being enrolled in the study. Obviously, patients
were not only excluded because they were considered to
require internal fracture fixation. Only 7 of the 15 C-type
cases in the clinical journal underwent surgery, indicating
that variables other than fracture severity contributed to the
selection of patients.
Second, patients were enrolled at a single institution
with distinct treatment standards, compromising the
external validity of our findings. In the light of results from
a recent randomized trial,16 whether the average duration of
immobilization in this study was too long is open to debate.
Third, the study included fractures that might be
considered as surgical cases at institutions that follow
a more invasive treatment philosophy. It is open to debate
whether primary fracture fixation would have avoided the
620
Figure 7 Variables potentially influencing Disabilities of Arm, Shoulder and Hand (DASH) DASH and Constant scores after 1 year of
follow-up.
Figure 8 Kaplan-Meier estimates for the probability of return-
ing to work (dotted line, 95% confidence interval).
17 displacements and losses of reduction. These events,
however, had no effect on functional outcomes. However,
given the serious lack of comparative studies (specifically
randomized trials) that would allow for evidence-based
decision making in the treatment of proximal humeral
fractures, the equipoise principle applied to patients in this
study. There is still much uncertainty about which patients
will benefit from nonoperative treatment, plate fixation,
nailing, or arthroplasty. This demands a thorough trade-off
of the possible benefits and harms of surgical and nonsur-
gical therapy.
Apart from the limitations, this study provides robust
evidence that conservative management of fractures of the
proximal humerus is safe and effective, mainly in AO type
A and B fractures. Conservative treatment may be
expanded even to 3-part fractures.
Employed patients assigned to nonoperative manage-
ment can be counseled about a high probability of
B. Hanson et al.
achieving solid healing, sufficient function, and strength to
return to work after 8 to 16 weeks. Patients should be
advised that smoking may significantly increase the risk of
nonunion and that they should quit smoking until fracture
consolidation. Few fractures primarily regarded suitable for
conservative management later require surgical stabiliza-
tion. Also, the risk of secondary surgical procedures, such
as arthroscopic decompression, is low.
In conclusion, although randomized trials remain the
reference standard for comparing outcomes between certain
treatment options, the noted ceiling effects will make it
difficult to detect substantial advantages of surgical fixation
over conservative management in the studied types of
proximal humeral fractures.
Acknowledgments
The authors, their immediate families, and any research
foundations with which they are affiliated have not
received any financial payments or other benefits from any
commercial entity related to the subject of this article
References
1. Barrett JA, Baron JA, Karagas MR, Beach ML. Fracture risk in the
U.S. Medicare population. J Clin Epidemiol 1999;52:243-9.
2. Calvo E, de M, I, de la Cruz JJ, Lopez-Martin N. Percutaneous fixation
of displaced proximal humeral fractures: indications based on the
correlation between clinical and radiographic results. J Shoulder
Elbow Surg 2007;16:774e81.
3. Constant CR, Murley AH. A clinical method of functional assessment
of the shoulder. Clin Orthop Relat Res 1987:160-4.
Nonoperative management of proximal humerus fractures
4. Court-Brown CM, Caesar B. Epidemiology of adult fractures:
a review. Injury 2006;37:691-7.
5. Court-Brown CM, Cattermole H, McQueen MM. Impacted valgus
fractures (B1.1) of the proximal humerus. The results of non-operative
treatment. J Bone Joint Surg Br 2002;84:504-8.
6. Court-Brown CM, Garg A, McQueen MM. The translated two-part
fracture of the proximal humerus. Epidemiology and outcome in the
older patient. J Bone Joint Surg Br 2001;83:799-804.
7. Court-Brown CM, McQueen MM. The impacted varus (A2.2) prox-
imal humeral fracture: prediction of outcome and results of nonop-
erative treatment in 99 patients. Acta Orthop Scand 2004;75:736-40.
8. Fjalestad T, Stromsoe K, Blucher J, Tennoe B. Fractures in the
proximal humerus: functional outcome and evaluation of 70 patients
treated in hospital. Arch Orthop Trauma Surg 2005;125:310-6.
9. Frangen TM, Dudda M, Martin D, Arens S, Greif S, Muhr G, et al
[Proximal humeral fractures with angle-stable plate osteosynthesiseis
everything better now?]. Zentralbl Chir 2007;132:60-9.
10. Gradl G, Dietze A, Arndt D, Beck M, Gierer P, Borsch T, et al.
Angular and sliding stable antegrade nailing (Targon PH) for the
treatment of proximal humeral fractures. Arch Orthop Trauma Surg
2007;127:937-44.
11. Hagino H, Yamamoto K, Ohshiro H, Nakamura T, Kishimoto H,
Nose T. Changing incidence of hip, distal radius, and proximal
humerus fractures in Tottori Prefecture, Japan. Bone 1999;24:
265-70.
12. Handschin AE, Cardell M, Contaldo C, Trentz O, Wanner GA.
Functional results of angular-stable plate fixation in displaced prox-
imal humeral fractures. Injury 2008;39:306-13.
13. Hudak PL, Amadio PC, Bombardier C. Development of an upper
extremity outcome measure: the DASH (disabilities of the arm,
shoulder and hand) [corrected]. The Upper Extremity Collaborative
Group (UECG). Am J Ind Med 1996;29:602-8.
621
14. Keener JD, Parsons BO, Flatow EL, Rogers K, Williams GR,
Galatz LM. Outcomes after percutaneous reduction and fixation of
proximal humeral fractures. J Shoulder Elbow Surg 2007;16:330-8.
15. Kollig E, Kutscha-Lissberg F, Roetman B, Dielenschneider D,
Muhr G. [Complex fractures of the humeral head: which long-term
results can be expected?]. Zentralbl Chir 2003;128:111-8.
16. Lefevre-Colau MM, Babinet A, Fayad F, Fermanian J, Anract P,
Roren A, et al. Immediate mobilization compared with conventional
immobilization for the impacted nonoperatively treated proximal
humeral fracture. A randomized controlled trial. J Bone Joint Surg Am
2007;89:2582-90.
17. Lill H, Bewer A, Korner J, Verheyden P, Hepp P, Krautheim I, et al
[Conservative treatment of dislocated proximal humeral fractures].
Zentralbl Chir 2001;126:205-10.
18. Linhart W, Ueblacker P, Grossterlinden L, Kschowak P, Briem D,
Janssen A, et al. Antegrade nailing of humeral head fractures with
captured interlocking screws. J Orthop Trauma 2007;21:285-94.
19. Neer CS. Displaced proximal humeral fractures. I. Classification and
evaluation. J Bone Joint Surg Am 1970;52:1077-89.
20. Stiller J, Uhl TL. Outcomes measurement of upper extremity function.
Athlet Ther Today 2005;10:15-7.
21. van den Broek CM, van den BM, Coenen JM, Vegt PA. Displaced
proximal humeral fractures: intramedullary nailing versus conserva-
tive treatment. Arch Orthop Trauma Surg 2007;127:459-63.
22. Voigt C, Woltmann A, Partenheimer A, Lill H [Management of
complications after angularly stable locking proximal humerus plate
fixation]. Chirurg 2007;78:40-6.
23. Zyto K. Non-operative treatment of comminuted fractures of the
proximal humerus in elderly patients. Injury 1998;29:349-52.
24. Zyto K, Ahrengart L, Sperber A, Tornkvist H. Treatment of displaced
proximal humeral fractures in elderly patients. J Bone Joint Surg Br
1997;79:412-7.