Inclusion criteria limited this study to isolated, displaced talar neck
and/or talar body fractures. Isolated peripheral fractures of the talus such
as lateral process, talar head, colliculi, and osteochondral fractures were
excluded. A total of fifty-three isolated talar neck and/or talar body
fractures in fifty-two patients were treated at our institution between
January 1993 and May 1998. We excluded eight fractures because they were
associated with an ipsilateral tibial, pilon, ankle, navicular, or calcaneal
fracture; three fractures because the patient died prior to hospital
discharge; six nondisplaced fractures because they were treated
nonoperatively; two fractures because an immediate amputation had been done as
a result of severe open injury; and one fracture because an immediate fusion
had been performed as a result of substantial bone loss. A minimum follow-up
period of forty-eight months was required to accurately evaluate the presence
of osteonecrosis and/or arthritis.
Seven patients (seven fractures) were lost to follow-up, leaving twenty-six
fractures in twenty-five patients that met the inclusion criteria (see
Appendix). These consisted of eighteen talar neck fractures, classified
according to the method of Hawkins as modified by
Canale1 (eleven type
2, six type 3, and one type 4), and eight talar body fractures. Of the
eighteen talar neck fractures, sixteen were isolated and two were combined
with a body fracture. According to the Gustilo-Anderson
classification13,
there were seven open fractures (three type 2, two type 3A, one type 3B, and
one type 3C) and nineteen closed fractures. There were ten male patients and
fifteen female patients with an average age of 37.3 years (range, twenty-four
to eighty-three years; median, thirty-eight years). Of the twenty-five
patients, nine (nine fractures) smoked a minimum of one pack of cigarettes per
day.
Eighteen fractures (eighteen patients) were initially seen in the emergency
department at our institution. Four patients (four fractures) were transferred
from outlying facilities, whereas three patients (four fractures) were
referred to our office practice for elective treatment.
Closed fractures were treated surgically as soon as medical clearance was
obtained, and open fractures were operated on within six hours after the
injury (except for one fracture in a patient who was transferred to our
institution late). The patient was positioned supine on a radiolucent table,
and a tourniquet was applied; however, the tourniquet was inflated only if
necessary to control bleeding. An anteromedial, anterolateral, combined
anteromedial-anterolateral, or posteromedial (one fracture) surgical approach
was used for exposure as needed. All open wounds (including those with
vascular injuries) were débrided and irrigated according to a standard
protocol14,15.
Stable fixation was achieved through compression if the fracture pattern
allowed it without shortening, translation, or angulation of the fracture
fragments. If the talar neck was excessively comminuted or if there was
substantial bone loss, position (non-lag)
screws16 were
placed to maintain length. No bone-grafting was employed in this series. A
variety of fixation devices were used to maintain reduction, depending on the
fracture pattern; the devices included 3.5-mm cortical screws, 4.0-mm
cancellous screws, small Herbert screws (Zimmer, Warsaw, Indiana), and
bioabsorbable screws and bio-absorbable pins (Bionx Implants, Blue Bell,
Pennsylvania). Mini-plates were not
used17.
Fracture reduction was rated on postoperative radiographs, including
anteroposterior, lateral, mortise, and Canale
views3. An anatomic
reduction meant that there was no step-off at the neck or body and no
angulation. A nearly anatomic reduction was defined as a 1 to 3-mm step-off of
any fracture fragment or slight varus angulation (=5°). A poor
reduction was an articular or neck mismatch, a step-off or gap of >3 mm, or
neck angulation of >5°.
Patients were kept non-weight-bearing by wearing an off-the-shelf fracture
boot with early foot and ankle motion for ten to twelve weeks, until
radiographs and clinical examination revealed evidence of union.
Anteroposterior, lateral, and mortise radiographs were routinely made at two
weeks, six weeks, ten to twelve weeks, and six months postoperatively.
Additional radiography, computed tomography scans, and magnetic resonance
imaging were performed as often as needed but were carried out at least
annually, specifically to evaluate for the development of osteonecrosis and
posttraumatic arthritis. The final follow-up for this study included
anteroposterior, lateral, and mortise radiographs; a clinical evaluation; and
determination of AOFAS (American Orthopaedic Foot and Ankle
Society)18 hindfoot
scores (see Appendix) at a minimum of forty-eight months after the injury.
Posttraumatic arthritis was defined as a decreased joint space,
juxta-articular osteophytes, subchondral sclerosis, and/or subchondral cysts
seen either on plain radiographs or on computed tomography scans.
Osteonecrosis was documented as present or absent as seen on standard
radiographs. Partial or total collapse of the talar body was also noted.
Fractures were given a minimum of six months to unite. If there was no
progression of healing, and no additional evidence of healing on standard
radiographs for three consecutive months thereafter, the fracture was
considered to be a nonunion. All data were obtained from the patient's
private-office medical records chart or radiology jacket. These files
contained not only a complete record of follow-up office visits and
radiographs, but also a complete duplicate of all hospital admission data,
both radiographic and with respect to records documentation. On accepting care
by the physicians at our private facility, our patients are requested to sign
the Authorization to Use Patient Medical Information for Medical Research
Form, which allows use of their data for research purposes. All patients in
this study signed this form. Charts were reviewed retrospectively at the time
of follow-up. Because of the nature of chart review and the sequential
evaluation of radiographs at that visit, the reviewers were not blinded to the
clinical history or the radiographs of the individual patients.
Statistical Analysis
Variables that were analyzed to determine their significance included wound
type (open or closed), fracture type (neck or body), Hawkins/Canale type (1
through 4), comminution, timing of the surgical intervention (within six hours
after the injury or not), surgical approach, quality of reduction, Hawkins
sign, osteonecrosis, union, time to union, posttraumatic arthritis, and the
AOFAS scores including subscores (pain, function, and alignment). The Fisher
exact test, Wilcoxon two-sample test, and Kruskal-Wallis test were performed
to compare these variables and to assess for significance. A p value of
=0.05 was considered significant.
Closed Compared with Open Fracture
The average duration of follow-up was 73.6 months (range, forty-eight to
113 months). Twenty-three of the twenty-six fractures healed after the index
procedure, for an overall union rate of 88%. All nineteen closed fractures
united without an additional operation. The average time to union for these
fractures was 3.1 months. Osteonecrosis developed after seven of the nineteen
closed fractures, whereas posttraumatic arthritis developed after all
nineteen. Osteonecrosis developed after six of the seven open fractures, and
posttraumatic arthritis developed after all seven. Only four of the seven open
fractures went on to uneventful union. Three fractures failed to heal, two of
which were associated with osteomyelitis prior to the final procedure. Of
these three fractures, two were treated with late arthrodesis and one required
a subsequent talectomy to treat a septic nonunion. The patient with the septic
non-union ultimately requested and received a below-the-knee amputation
because of instability and pain. Of the four open fractures that went on to
union, one required additional surgery (bone-grafting). The average time to
union of these four open fractures was 3.9 months. The prevalence of
osteonecrosis was much higher following the open fractures than it was
following the closed fractures (86% compared with 37%); this difference
approached significance (p = 0.073).
Neck Fracture Compared with Body Fracture
With the numbers available, there were no significant differences between
the sixteen isolated neck fractures and the eight isolated body fractures with
regard to union rate, AOFAS hind-foot scores, osteonecrosis rate, or
prevalence of posttraumatic arthritis. The overall union rate for the neck
fractures was the same as that for the body fractures (88%). Isolated neck
fractures complicated by osteonecrosis were associated with a much lower
average AOFAS hindfoot score (46.7 points) than were isolated neck fractures
without osteonecrosis (76 points). The two groups differed significantly with
regard to the total score (p = 0.009) and the subscores for function (p =
0.004) and alignment (p = 0.04). The average AOFAS score did not differ
significantly between isolated body fractures with osteonecrosis (51.4 points)
and those without osteonecrosis (60.3 points) (p = 0.2).
Timing of Surgery
Twelve of the twenty-six fractures received surgical treatment within six
hours after the injury. Overall, the time to surgery averaged eighty-five
hours with a range of two to 504 hours (median, 10.5 hours). With the numbers
available, no significant difference was seen between the six closed fractures
fixed within six hours after the injury and the thirteen closed fractures
fixed more than six hours after the injury with respect to the AOFAS hindfoot
score, nonunion rate, osteonecrosis rate, or prevalence of posttraumatic
arthritis. Open fractures were operated on within six hours more frequently
(six of seven) than were closed fractures (six of nineteen). This difference
was significant (p = 0.03). When open and closed fractures were considered
together, again no significant difference was seen between early and late
fixation with regard to the average AOFAS hindfoot score, nonunion rate,
osteonecrosis rate, or prevalence of posttraumatic arthritis.
Surgical Approach
Comparisons were also made between the surgical approaches used for
internal fixation. Both anteromedial and anterolateral incisions were used to
treat thirteen fractures, and osteonecrosis developed after the treatment of
eight of them. Five of these fractures were open, and osteonecrosis developed
after the treatment of all five. Only an anteromedial incision was used to
treat seven fractures, and osteonecrosis developed after the treatment of four
of them. One of these four fractures was open, and osteonecrosis developed
after the treatment of that fracture. Five fractures were fixed through only
an anterolateral incision; osteonecrosis developed after the treatment of one
closed fracture. The numbers were too small to identify any significant
differences between surgical approaches with regard to osteonecrosis rates or
AOFAS hindfoot scores.
Reduction
Surgical intervention resulted in anatomic reduction of sixteen fractures,
nearly anatomic reduction of five, and poor reduction of five. All eight
noncomminuted fractures were anatomically reduced. The ability to obtain an
anatomic reduction was significantly associated with the amount of comminution
(p = 0.003). With the numbers available, no association was seen between the
quality of the reduction and the fracture type. The quality of the reduction
did significantly affect the total AOFAS score (p = 0.05) as well as the
function (p = 0.05) and alignment (p = 0.006) subscores.
Union Rates
Overall, the union rate was 88% (twenty-three of twenty-six). Closed
fractures were more likely to unite than were open fractures (p = 0.01), and
fractures with a better reduction had a somewhat greater likelihood of uniting
(p = 0.05). Furthermore, there was a trend for the time until union to be
associated with the quality of the reduction in the twenty-three fractures
that went on to union (p = 0.07).
The union rate for the isolated talar neck fractures was 88% (fourteen of
sixteen). Importantly, all thirteen closed displaced talar neck fractures
healed after treatment with standard techniques, regardless of the time delay
until surgical intervention, which was twenty-one days in one case. Closed
fractures united earlier than did open fractures (p = 0.02). The better the
reduction of the neck fracture, the higher the probability of union (p =
0.05). The AOFAS scores were not associated with union or the time to
union.
Arthritis
Posttraumatic arthritis of the subtalar joint was the most common
radiographic finding and was seen in all patients at the time of final
follow-up. Isolated arthritis of the ankle did not develop in any patient.
Posttraumatic arthritis of the subtalar joint alone developed in ten limbs,
traumatic arthritis within both the subtalar and the ankle joint developed in
fifteen, and subtalar, ankle, and talonavicular joint arthritis developed in
one. With the numbers available, there was no association between
osteonecrosis and joint arthritis (p = 0.23). The average AOFAS hindfoot score
was 63.2 points for the patients with involvement of the subtalar joint alone
and 60.4 points for those with combined subtalar and ankle arthritis.
Osteonecrosis
Osteonecrosis was the second most common radiographic finding, seen after
thirteen of the twenty-six fractures (eight neck and five body fractures).
There was also a trend for the rate of osteonecrosis to be higher after open
fractures than after closed fractures (p = 0.073). Importantly, with the
numbers available, osteonecrosis was not related to the quality of the
reduction (p = 0.31), comminution (p = 0.2), operative approach (p = 0.34),
type of fracture (neck or body) (p = 0.7), or time to surgery (p = 0.7).
Arthritis of more than one joint developed in ten of the thirteen limbs with
osteonecrosis. The average AOFAS hindfoot score was 49.5 points in the group
in which osteonecrosis developed and 72.9 points in the group in which it did
not. As expected, the average total AOFAS score was significantly better when
osteonecrosis was absent (p = 0.009). Both the function and the alignment
AOFAS subscores were also significantly worse when osteonecrosis was present
(p = 0.0013 and p = 0.0151, respectively).
Of the sixteen limbs with a fracture involving the talar neck alone, seven
had osteonecrosis postoperatively; four of the ten type-2 talar neck
fractures, two of the five type-3 fractures, and the one type-4 fracture were
associated with osteonecrosis. Importantly, with the numbers available,
neither the time to surgery (p = 0.6) nor the operative approach (p = 0.53)
was associated with the presence of osteonecrosis. With regard to the accuracy
of the Hawkins sign for predicting the development of osteonecrosis after the
treatment of the sixteen isolated talar neck fractures, the sensitivity was
67% (six of nine), the specificity was 86% (six of seven), and the accuracy
was 75% (twelve of sixteen). Thus, the sign was a good predictor of the
development of osteonecrosis (p = 0.06).
Pain
Pain was the most common subjective finding and was reported by all
patients. According to the AOFAS scoring system, five patients (five
fractures) had severe pain, ten patients (eleven fractures) had moderate pain,
and ten patients (ten fractures) had mild pain. The AOFAS subscore for pain
was not associated with any variable in this study.
Smoking
With the numbers available, no significant difference in the union or
osteonecrosis rate was seen between smokers and nonsmokers.
This study, which included only isolated displaced talar neck and talar
body fractures, confirms previous reports of osteonecrosis and posttraumatic
arthritis as common long-term sequelae. Importantly, in our series, arthritis
surpassed osteonecrosis as the principal long-term finding associated with
these fractures. Posttraumatic arthritis of the subtalar joint developed in
every patient; arthritis of both the subtalar and the ankle joint, in fifteen;
and arthritis of the subtalar, ankle, and talonavicular joints, in one.
In early reports, displaced and nondisplaced talar neck fractures were
often treated nonoperatively or without internal fixation. Osteonecrosis was
found to be the major complication, and it was seen more commonly than
posttraumatic arthritis. Hawkins, in his initial study of fifty-seven talar
neck fractures in 1970, reported an overall prevalence of osteonecrosis of 53%
while only briefly noting posttraumatic arthritis in the form of
"routinely marked decrease in
motion."8 When
only the displaced fractures in Hawkins' series were considered, however, the
osteonecrosis rate increased to 59%. Hawkins also noted that arthritis may be
a result of osteonecrosis. In a 1978 study of the long-term results of
seventy-one talar neck fractures, Canale and Kelly reported a 52% prevalence
of osteonecrosis and stated that posttraumatic arthritis developed in fewer
than one-half of the
limbs1. However, the
osteonecrosis rate associated with only the displaced fractures was 63%. More
than one-half of the fractures in the above two studies were treated
nonoperatively, including thirty of fifty-four Hawkins type-2 fractures and
seven of fifty Hawkins type-3 fractures. (An additional four Hawkins type-3
fractures underwent an open reduction without any internal fixation.) It is
difficult, if not impossible, to compare these results of closed treatment of
displaced talar fractures with the results in current studies because the
current treatment of Hawkins type-2, 3, and 4 fractures is open reduction and
internal fixation.
In 1980, Penny and Davis reported an osteonecrosis rate of 48% in
twenty-seven patients with a talar neck fracture but did specify subtalar
joint symptoms as a reason for their poor rate of
success10. Similar
to the findings of
Hawkins8 and Canale
and Kelly1, the
osteonecrosis rate in the study by Penny and Davis increased to 59% when only
displaced fractures were considered. In 1985, Comfort et al. noted an overall
43% prevalence of osteonecrosis following treatment of displaced talar neck
fractures with open reduction and internal
fixation17. They
also noted that subtalar motion decreased by an average of 50%, although only
approximately one-third of patients demonstrated osteophytic changes on
radiographs. The authors of the above two studies were, to our knowledge, the
first to advocate anatomic reduction through open reduction and internal
fixation in order to achieve the best results.
Investigators who have noted much lower osteonecrosis rates than typically
reported in the literature have attributed them to early fixation. Grob et al.
reported an osteonecrosis rate of 16% in a series in which twenty-eight of
forty-one talar fractures were fixed surgically within eight hours after the
injury4. When only
displaced fractures of the neck and body were considered, the prevalence of
osteonecrosis increased to 33%. Frawley et al. also reported an osteonecrosis
rate of 16%, after operating on fourteen of twenty patients within twelve
hours after the
injury3. Seven of
the twenty-eight fractures in that study were not displaced. Elgafy et al.
presumed that the low prevalence of osteonecrosis (16.6%) in their study was
due to early anatomical fixation, although the rate increased to 32% when only
displaced neck and body fractures were
considered7. As
mentioned, Comfort et al. found an osteonecrosis rate of 43% and, although
they noted that the degree of osteonecrosis was not related to a delay in
fracture fixation, they still recommended early
fixation17.
In our study of isolated displaced talar neck and body fractures, the
prevalence of osteonecrosis was 50%. The rate of osteonecrosis was associated
with the Hawkins classification, as was noted in other published
reports1,3,7-9,19.
The rate of osteonecrosis was not related to the timing of fixation but rather
to the initial degree of fracture displacement and the presence of an open
injury. It is probable that modern techniques of fixation offer enough
stability and compression to allow revascularization to occur even after a
delay in treatment. It appears that the severity of the injury itself, and not
a delay in fracture fixation, may ultimately be responsible for higher rates
of osteonecrosis. In addition, although our series confirms an overall high
association between the Hawkins sign and the presence of osteonecrosis, the
Hawkins sign appears not to be a foolproof evaluation tool as our series
included both false-positive and false-negative findings.
More recent studies have demonstrated a higher prevalence of posttraumatic
arthritis than had been previously reported after the treatment of talar
fractures. In addition, posttraumatic arthritis has now been shown to be a
more common finding than osteonecrosis. Szyszkowitz et al., in 1985, reported
an increased prevalence of posttraumatic arthritis after the treatment of
displaced talar fractures, with the arthritis involving the subtalar joint in
74% of their cases and the ankle joint in
52%20. Although
Grob et al. reported a low prevalence of posttraumatic arthritis (37%), they
found it to be much more common than osteonecrosis (16%) when they analyzed
both displaced and nondisplaced
fractures4. In 1995,
Frawley et al. stated that "DJD is the most common and disabling
complication and osteonecrosis is not as common a problem as the literature
suggests."3
Elgafy et al., in 2000, found a 53% prevalence of subtalar arthritis in
patients with a talar
fracture7.
The observations in these recent studies more closely parallel the findings
in our study, in which posttraumatic arthritis was twice as common as
osteonecrosis. However, many of the previous
studies1,3,7-9,21,22
included minor talar fractures, nondisplaced fractures, and associated
ipsilateral peritalar fractures (i.e., ankle, pilon, calcaneal, and navicular
fractures), which may predispose to the development of posttraumatic arthritis
in those joints, even in the absence of a talar fracture. The authors of these
previous studies failed to make a distinction between minor and major
fractures. In addition, minor talar fractures and nondisplaced talar fractures
would likely be associated with lower rates of posttraumatic arthritis than
would displaced neck and body fractures. Because only isolated, displaced neck
and body fractures were included in our study, the prevalence of posttraumatic
arthritis that we reported more accurately reflects the true occurrence of
this sequela following a displaced talar neck or body fracture.
The frequency of posttraumatic arthritis has been purported to be higher
after talar body fractures than after neck
fractures6,7,23.
Inokuchi et al. provided definitions of talar neck and body fractures based on
the inferior, not the superior, fracture
line6. Neck
fractures, according to their definition, are extra-articular and between the
posterior and middle facets of the subtalar joint, whereas body fractures are
intra-articular and involve the posterior facet. With the numbers available in
our study, we found no significant difference in the prevalence of
posttraumatic arthritis between displaced neck and body fractures.
Recently, Vallier et al. reported on fifty-six surgically treated talar
body fractures, including thirty-three involving only the talar body (with
twenty-six associated foot and ankle injuries) and twenty-three involving the
neck and body (with sixteen associated foot and ankle
injuries)24. The
authors concluded that the functional outcomes were worse when the talar body
fracture was followed by the development of arthritis or osteonecrosis with
collapse. Although the majority of patients in that study were lost to
follow-up, it is clear that talar body fractures can be devastating injuries
despite excellent fixation techniques and management; however, our data
indicate that the outcomes of these fractures are no better or worse than
those of isolated talar neck fractures.
Overall, the open fractures in our series fared much worse than did the
closed fractures, with a much lower union rate, a higher osteonecrosis rate,
and substantially higher reoperation and infection rates. Other series of
talar fractures have also shown the results associated with open fractures to
be inferior to those associated with closed
fractures1,7,24-26.
We applied the widely used AOFAS scoring system as the standard method for
reporting clinical status in our
patients18. Authors
of recent studies, however, have questioned the value of this
system27. In a
study of ninety-one patients with foot and ankle problems, SooHoo et al. found
that, overall, the AOFAS clinical rating system correlated poorly with the
Short Form-36 (SF-36)
score27. They noted
that the validity of the AOFAS system was limited by an overemphasis on a
single question evaluating pain, while the physical examination parameters
exhibited poor interobserver and intraobserver reliability. They concluded
that the AOFAS instruments were invalid for evaluating health status, and they
suggested that new foot and ankle outcomes instruments be developed. Because
we used this system exclusively in our study, extrapolations made from our
AOFAS scores must be limited. We believe that, in the future, the SF-36 should
be used to obtain clinical outcome data for patients with foot and ankle
problems, at least until a new scoring system can be developed and
validated.
As with all clinical studies, ours had limitations. These include the
retrospective nature of the review, the fact that the reviewers were not
blinded to the patients' clinical or radiographic data, that the radiographic
studies were not performed on a regularly prescribed basis and were not graded
in a blinded fashion, and that a potentially flawed functional instrument (the
AOFAS score) was used to analyze part of the outcome.
In conclusion, on the basis of an overall union rate of 88% and the fact
that all nineteen closed fractures united without additional operations in our
study, we believe that displaced talar neck and/or body fractures should be
treated with open reduction and internal fixation. Although the time to
surgical fixation should be minimized, a delay does not appear to adversely
affect the outcome, specifically with regard to the development of
osteonecrosis. In addition, posttraumatic arthritis, particularly affecting
the subtalar joint, appears to be much more common than osteonecrosis.
Osteonecrosis developed after 50% of the isolated, displaced talar neck and/or
body fractures and probably is a result of the degree of displacement and/or
the associated open injury. The Hawkins sign may not always be a reliable
predictor of the development of osteonecrosis. Open fractures fared
dramatically worse than did closed fractures with regard to nearly all aspects
evaluated. On the basis of the results of this study, we believe that patients
with a displaced fracture of the talus should be counseled that posttraumatic
arthritis and chronic pain are expected outcomes even after anatomic reduction
and with no evidence of osteonecrosis.
Tables showing patient demographics, fracture patterns, and AOFAS outcome
scores are available with the electronic versions of this article, on our web
site at
(go to the article citation and click on "Supplementary Material")
and on our quarterly CD-ROM (call our subscription department, at
781-449-9780, to order the CD-ROM).