Subtrochanteric fractures of the femur that are caused by low-energy
trauma are less common than other proximal femoral fractures are, but they
occur in a similar population of elderly
individuals1.
Because of the difficultyr of maintaining a satisfactory reduction by
nonoperative
means2,3,
most of these injuries are treated with open reduction and internal fixation.
Although it is assumed that the outcome following such fractures is similar to
that following other proximal femoral fractures, with a high mortality rate
and an increased level of social dependence among
survivors4,
prospective evaluation of this subgroup of subtrochanteric fractures has not
previously been undertaken, to our knowledge.
Operative stabilization of these fractures in the elderly is often
technically difficult because of fracture comminution and osteoporosis. The
techniques that are used to stabilize these fractures have evolved over the
last thirty years, from the use of fixed-angle
blade-plates2,5
and condylar screw/plate
systems6-8,
to the use of interlocking
nails9,10,
to the current widespread use of cephalomedullary
nails11-14.
Intramedullary nailing offers considerable biomechanical advantages
compared with open plating techniques, and the use of a closed procedure is
theoretically less invasive, causing reduced intraoperative blood loss and
risk of
infection15.
However, fractures with proximal extension into the piriformis fossa may be
unsuitable for treatment with an intramedullary nail because this site is used
as an entry point for nail
insertion3. Newer
designs of lateral offset trochanteric entry nails, such as the long Gamma
nail (Stryker-Howmedica, Berkshire, United Kingdom), have been designed to be
inserted through a greater trochanteric entry point and offer the possibility
of stabilizing fractures with and without extension into the piriformis
fossa16.
Since 1994, we have used trochanteric-entry cephalomedullary nailing to
stabilize all low-energy subtrochanteric fractures. The aim of the present
study was to review the results of this treatment in a series of cases with
respect to the mortality, the rate of perioperative complications, and the
subsequent level of social dependency.
Between February 1994 and December 2001, we operatively treated a
consecutive cohort of 302 patients who had sustained a nonpathologic
subtrochanteric fracture as a result of a low-energy injury (defined as a fall
from at or below standing height). The patients included 103 men and 199 women
who had a median age of 78.5 years (absolute range, thirty-four to ninety-nine
years; interquartile range, 65.75 to 86 years). A subtrochanteric fracture was
considered to be an injury in which the main fracture line was centered within
the subtrochanteric region (defined as the area extending from the distal
border of the lesser trochanter for a distance of 5 cm, adjusted for
radiographic magnification). We also used this method to treat any fracture
centered within this area that had secondary fracture lines extending
proximally into the intertrochanteric region or piriformis fossa or into the
lesser trochanter as well as any fracture that extended distally from the
subtrochanteric region into the proximal femoral diaphysis. The
first-generation long Gamma nail (Stryker-Howmedica) was used to treat all of
these fractures.
The patients who were managed with this method of reconstruction but who
were excluded from the final series of 302 patients included seventeen
patients who were not local residents of the regional catchment area,
seventy-three patients with high-energy subtrochanteric fractures, and
forty-one patients with fractures due to metastatic tumor. Twenty-two patients
had a subtrochanteric fracture that was related to a previous hip-fracture
implant, and twenty-one patients had a subtrochanteric fracture distal to a
hip that was stiff or deformed as a result of degenerative or other hip joint
disease. These fractures were deemed to be unsuitable for nailing and were
treated with other techniques. Three patients with subtrochanteric fractures
who died preoperatively as a result of medical comorbidities and an additional
four patients who were medically unfit for anesthesia and who were managed
nonoperatively also were excluded.
All fractures were classified according to the Russell-Taylor system
(Fig. 1), which was chosen
because it evaluates the degree of proximal extension of the subtrochanteric
fracture and previously has been shown to serve as a basis for
treatment3. Most
fractures were of the relatively simple Russell-Taylor Type-I configuration,
with smaller numbers of patients having the more complex Russell-Taylor
Type-II configuration. The age and gender distributions of the four fracture
subclassifications were similar (Table
I).
Operative Technique
All 302 operations were performed within the first seventy-two hours after
admission. The procedure was performed by either an attending surgeon (C.M.R.;
141 patients) or by a senior trainee orthopaedic surgeon under the supervision
of the attending surgeon (161 patients). The injured limb was placed in
longitudinal traction on a radiolucent traction table, with the foot secured
in a traction boot. The contralateral limb was abducted, flexed, and
externally rotated to facilitate passage of the image intensifier. An attempt
at closed reduction was made initially in all cases. We initially used
longitudinal traction through the traction table, with adjustment of
rotational alignment with the aid of an image intensifier. An external crutch
was used to support the thigh in twenty-eight cases in which there was
excessive posterior sag of the proximal part of the femur distal to the
fracture. In twenty-three cases, because of excessive displacement,
comminution, or muscle interposition, an adequate closed reduction was not
achieved and the patient underwent open reduction of the fracture. In seven of
the patients with severe displacement, supplementary cerclage wire fixation
was used to provisionally stabilize the fracture prior to nailing.
With the patient under general or spinal anesthesia, and after preparation,
draping, and administration of a single-dose broad-spectrum antibiotic, a
proximal skin incision was made; the incision was centered on the greater
trochanter and extended for approximately 10 cm proximally. Through a starting
point located at the tip of the greater trochanter, a standard olive-tipped
guide-wire was passed across the fracture and into the distal metaphysis.
Fluoroscopy of the distal part of the femur was used to confirm a central
position of the tip of the guide-wire in both the anteroposterior and lateral
projections. The femoral canal was prepared by reaming to 13 or 14 mm distally
and to 17 mm in the proximal region to accept a long Gamma nail of appropriate
length. The nail had a diameter of 11 mm distally, with expansion to 15 mm at
its proximal extent. The length of the nail was determined on the basis of an
intraoperative assessment of femoral length with use of an intramedullary
measuring guide-wire. A solitary proximal lag screw was inserted under
fluoroscopic control with the aid of the radiolucent targeting device. We
attempted to place the lag screw centrally within the femoral head on both the
anteroposterior and lateral projections. Following the recognition in the
mid-1990s of the importance of the tip-apex distance in reducing the risk of
cutout failure of hip
screws17,18,
we routinely attempted to place the lag screw as close as possible to the apex
of the femoral head on both anteroposterior and lateral projections. The
lag-screw angle was either 130° (223 nails) or 135° (seventy-nine
nails). In all cases, static locking of the nail was performed with use of a
set-screw to lock the proximal lag-screw onto the nail. Distal locking screws
were inserted with use of a free-hand technique under fluoroscopic control.
All patients were encouraged to bear full weight beginning on the second
postoperative day unless poor preoperative mobility prevented them from doing
so. All patients were managed with subcutaneous heparin injections until they
were fully mobile. The average operative time was eighty-two minutes (range,
thirty-five to 165 minutes), and the average operative blood loss was 306 mL
(range, 60 to 1200 mL).
Outcome Assessment
Clinical and radiographic documentation of any fracture-related
complications was undertaken by the treating surgeons during routine
outpatient clinic appointments or in the rehabilitation ward if discharge into
the community was delayed. A nonunion was defined as the absence of bridging
of three of four cortices on anteroposterior and lateral radiographs at six
months after the injury, combined with instability resulting from a lack of
osseous consolidation on surgical exploration of the fracture site following
nail removal. A nonunion also was considered to be present if the nail broke
and the fracture displaced more than twelve weeks after the injury. We did not
assess rotational or angulatory malreduction of the fractures radiographically
because of the difficulty of making standing postoperative long-leg
radiographs for these patients. However, we documented the presence of any
clinically important shortening (>2.5 cm) or rotational malalignment
(>20°) in the immediate postoperative period or during follow-up.
Other postoperative complications recorded included reoperations,
complications related to the lag screw, nail penetration or fracture distal to
the nail, operative fracture comminution, superficial infection (defined as
wound inflammation, a bacteriologically confirmed organism on culture of wound
specimens, and no purulent discharge), and deep infection (defined as a
bacteriologically confirmed infection on culture of deep-wound specimens or
purulent discharge from around the nail). In cases in which the patient died
within the first year after the injury, we also recorded the date and cause of
death.
Throughout the period of the review, a dedicated prospective hip-fracture
audit program was instituted in our unit. This program was carried out by
audit workers who independently documented all demographic data and evaluated
patients during the first year after the injury. All patients received a
preoperative functional assessment with use of a standard hip-fracture
assessment questionnaire. In the cases of patients with dementia, this
information was gathered by questioning relatives or caregivers. Surviving
patients underwent additional prospective functional follow-up assessments
between three and four months postoperatively and at one year postoperatively.
Assessments of pain, mental
status19, walking
ability, use of walking aids, level of residence, and level of social
dependence were recorded with use of ordinal categorical scoring systems.
Information regarding readmission to the hospital, reoperation, or additional
treatment also was recorded. Information on patients who were too frail to
attend clinical outpatient review appointments was gathered by means of a
telephone interview or a mailed questionnaire.
Data Analysis
All information that was gathered during the study was prospectively coded
and analyzed with use of the SPSS software package (SPSS, Chicago, Illinois).
The rates of mortality, union, and complications were assessed, and the
functional results at one year after surgery were compared with the preinjury
level of function. Mortality data were rechecked against local census data to
ensure their accuracy. Because of the high mortality rate during the first
year after the injury, it was apparent that a substantial proportion of
patients did not survive long enough for fracture-related complications to
develop. We therefore used survival analysis to assess the overall reoperation
rate and the nail revision rate in the cohort. In this analysis, patients in
whom these complications had not developed were censored from further analysis
at the time at which they died or were lost to follow-up.
Loss to Follow-up and Mortality
Ninety-one (30.1%) of the original cohort of 302 patients did not
complete the final functional assessment at one year. This group comprised
seventy-four patients who died and seventeen patients who were lost to
follow-up during the first year. Although we were unable to assess the
functional or social status of the surviving patients who were lost to
follow-up, none had been readmitted to our unit (the sole source of trauma
care for the local population) for the treatment of complications or for
additional surgery.
The cause of death was attributed to cardiopulmonary or cerebrovascular
disease in sixty-four (86%) of the seventy-four patients who died during the
first year after the injury. None of the deaths occurred during the first
forty-eight hours postoperatively, and none were attributed to direct
complications of the operation. Of the deaths that occurred within the first
year after the injury, forty-nine (66%) occurred within the first three months
after the injury, often during the initial hospitalization.
Complications
Postoperative and fracture-related complications occurred in seventy
(23.2%) of the 302 patients within the first year after the injury. Although
forty-three patients (14.2%) were managed nonoperatively for the treatment of
complications, twenty-seven (8.9%) required reoperation
(Table II). Eighteen (6%) of
the 302 patients required nail revision within the first year, and nine (3%)
underwent reoperation without nail revision
(Table II). Survival analysis,
with censorship of patients at the time of death or loss to follow-up,
revealed that the cumulative rate of reoperation for the treatment of a
fracture-related complication was 7.0% (95% confidence interval, 3.9% to
10.0%) at six months after the injury and 10.3% (95% confidence interval, 6.6%
to 14.0%) at one year (Table
III). The cumulative rate of revision following the original
procedure was 3.8% (95% confidence interval, 1.5% to 6.1%) at six months after
the injury and 7.1% (95% confidence interval, 4.0% to 10.2%) at one year
(Table IV).
Eight of the 250 patients who survived for at least six months after the
injury had a radiographically suspected nonunion as indicated by the absence
of bridging callus of three of four cortices on anteroposterior and lateral
radiographs (seven patients) or nail breakage (one patient). These patients
underwent operative nail removal followed by exploration of the fracture site.
In three patients, the fracture was thought to be solidly united at the time
of exploration and stressing of the site of the fracture, and no additional
action was taken. All three of these fractures subsequently progressed to full
union after additional follow-up. Four patients had definite fracture nonunion
at the time of exploration and were managed with autologous bone-grafting and
exchange nailing. These fractures subsequently united without additional
complications. One patient had nail breakage at the site of entry of the
proximal lag screw into the nail at five months after the injury. The fracture
was not united at the time of exploration, and the nonunion was treated with
insertion of a new nail with autogenous bone-grafting. The fracture
subsequently healed uneventfully. Confirmed nonunion was therefore observed in
only five (2%) of the 250 patients who survived for at least six months after
the injury.
Complications related to the proximal lag screw occurred in twelve (4%) of
the 302 patients. Lag-screw cutout occurred in nine patients. Five of these
patients were managed with revision of the fixation either with another long
Gamma nail or with a dynamic compression screw and plate. The other four
patients were frail and elderly and were managed nonoperatively. Cutout
failure occurred in association with four (2.6%) of the 153 Type-Ia fractures,
four (5.9%) of the sixty-eight Type-Ib fractures, and one (1.6%) of the
sixty-three Type-IIb fractures. Lag-screw penetration into the ace-tabulum
occurred in one patient; the screw was changed to a new, shorter, locked lag
screw, without revision of the nail. Excessive lag-screw back-out (i.e., >5
cm of lag-screw back-out) occurred in two patients because of the failure to
achieve static locking with the set screw. Revision involving reinsertion and
static locking of the lag screw was carried out in one of these cases. The
fracture united in all patients who had a lag-screw complication. One patient
with a Type-Ia fracture had development of osteonecrosis of the femoral head,
without cutout failure; the fracture necessitated revision with a long-stem
cemented total hip replacement after the fracture had healed.
A metaphyseal fracture distal to the tip of the long Gamma nail occurred in
five of the 302 patients following minor falls between one and eight weeks
postoperatively (Fig. 2-A). All
fractures occurred in elderly women (age range, seventy-eight to ninety-three
years) with marked bowing of the distal part of the femur. In all cases,
postoperative lateral radiographs showed the tip of the nail to be abutting
directly against the anterior cortex of the distal part of the femur. An
additional sixteen patients had a similar radiographic appearance of placement
of the nail against the anterior cortex of the distal part of the femur but
did not subsequently sustain a distal fracture. In four patients, the fracture
was treated with revision of the nail with use of a modified long Gamma nail,
which was pre-bent to impart extra anterior bowing, combined with cerclage
wiring (Fig. 2-B); in the
remaining patient, the fracture was treated with open reduction and internal
fixation with use of a 95° dynamic compression screw device, with the nail
left in situ. All patients had subsequent union of both the subtrochanteric
and distal metaphyseal fractures without additional surgery. Protrusion of the
tip of the nail through the anterior cortex of the distal part of the femur
occurred during insertion in two other elderly patients with severe anterior
bowing of the femur. Revision of the fixation was not performed, and the
fracture healed uneventfully in both patients.
Superficial infection occurred in thirty-seven (12.3%) of the 302 patients.
In thirty-five patients, the infection resolved with intravenous antibiotic
therapy. The other two patients underwent exploration for the evaluation of a
suspected deep infection, although none was found. All patients who had a
superficial infection were free of infection at the time of subsequent
follow-up, and none of these patients had development of a nonunion. Five of
the 302 patients had development of a deep infection at the site of the
proximal nail-entry wound within the first ten days after surgery. One patient
with a deep infection, who was unfit to undergo additional surgery, was
managed nonoperatively and the infection was suppressed with long-term oral
antibiotic therapy. The remaining four patients were managed with a protocol
involving initial débridement of the wound, insertion of antibiotic
beads, and intravenous antibiotic therapy dictated by the bacteriological
growth. The fixation was solid in all cases, and no patient had an acute
revision. Repeated inspection of the wounds was carried out at regular
forty-eight-hour intervals, and closure was performed when the wounds were
free of active drainage. Use of this protocol revealed that one patient had an
additional episode of deep infection that required drainage, but all patients
were clinically free of infection by six weeks. All fractures subsequently
united without implant removal or revision of fixation. No patient had
development of sepsis.
We did not encounter any clinically important rotational or shortening
deformity of the femur that required reoperation following nailing.
Preoperatively, eighty-five (28.1%) of the 302 patients had had clinically
important medical comorbidities (predominantly cardiovascular, respiratory,
and renal comorbidities) that required treatment before surgery.
Postoperatively, seventy-six patients (25.2%) required additional supportive
medical treatment for complications (predominantly nosocomial infections,
confusional states, deep venous thrombosis, pulmonary embolus, and
exacerbation of preexisting medical comorbidities).
Functional Outcome (Table
V)
Of the 211 patients who were evaluated with use of an ordinal categorical
scale for pain at one year after the injury, nineteen (9%) had mild pain,
sixty (28.4%) had moderate or tolerable pain, seven (3.3%) had pain that
prevented normal activities, and only two (0.9%) had severe pain that they
described as disabling. At the time of the initial fracture treatment, 133
(44%) of the 302 patients showed evidence of moderate or severe cognitive
impairment; this proportion was similar among the patients who were reviewed
at one year.
There was a trend toward an increased level of social dependency among
surviving patients at one year (Table
V). Of the 209 patients who had been living in their own home
before the injury, 111 (53.1%) were able to return to their own home,
forty-four (21.1%) had died, ten (4.8%) were lost to follow-up, and the
remaining forty-four (21.1%) required residential care.
The percentage of patients who were unable to walk or who required
assistance to walk increased at one year, whereas the percentage of patients
who were able to walk independently in the community or within the home
decreased correspondingly during the same period
(Table V). Of the 178 patients
who previously had been able to walk independently, eighty (44.9%) continued
to be able to do so, fifty-four (30.3%) had decreased walking ability, one
(0.6%) was wheelchair-bound, thirty-six (20.2%) had died, and seven (3.9%) had
been lost to follow-up. The proportion of patients who were fully mobile
without any walking aids decreased, whereas the number who required the use of
one or more walking aids or who were wheelchair-bound or completely immobile
increased (Table V). Of the 158
patients who had been fully mobile without walking aids, fifty-eight (36.7%)
continued to be fully mobile without walking aids at one year, forty-three
(27.2%) required one cane or crutch when walking, fifteen (9.5%) required more
than one walking aid, none were wheelchair-bound or immobile, thirty-four
(21.5%) had died, and eight (5.1%) had been lost to follow-up.
Subtrochanteric fractures that occur during low-energy trauma are
relatively uncommon in elderly
patients20. Our
results demonstrate that the prognosis associated with these fractures is
typical of that associated with proximal femoral fractures in the
elderly4,21,
with a similar mortality of 24.5% within the first year after the injury and
with an increased level of social dependency and reduced walking ability among
survivors.
Operative intervention for these fractures has become widespread, and,
given the poor prognosis and the expected increases in the level of social
dependency, the ideal implant should restore sufficient stability to allow
early mobilization. In addition, the implant should be associated with a low
rate of complications, especially those necessitating reoperation. Finally,
the implant should be versatile enough to stabilize all fracture types. We
believe that our results demonstrate that trochanteric-entry cephalomedullary
nailing meets these criteria. All patients in the present series were
mobilized with weight-bearing as tolerated in the early postoperative period.
Despite this early mobilization, the rates of infection and mechanical
complications were lower than those previously recorded in association with
other implants (see
Appendix)2,5-14,22-43
and the overall rates of reoperation and implant revision within the first
year after surgery were low. On survivorship analysis, only 7.1% of the nails
had to be revised, and only 2% of the patients who were alive at six months
after the injury had an ununited fracture.
Nail impingement or protrusion through the anterior cortex of the distal
part of the femur was a technical problem that was encountered with use of the
first-generation cephalomedullary nail in this series. This problem resulted
in a fracture distal to the nail in five patients, all of whom required
revision surgery. This complication was reported extensively following the use
of the short Gamma
nail44,45
and was variously attributed to the mechanical properties of the nail,
technical error, and the presence of an intramedullary stress-riser in the
femoral diaphysis. This complication also has been previously noted in small
numbers of patients in other series involving the use of long cephalomedullary
nails13,46.
It was noteworthy that, in all of the patients who sustained a distal
metaphyseal fracture in our series, the nail was seen to be abutting against
the anterior cortex of the distal part of the femur on postoperative lateral
radiographs. It is possible that greater stiffness of the nail, combined with
the greater radius of curvature compared with conventional centromedullary
nails, places this implant at particular risk for this complication,
especially when used in a bowed, osteoporotic femur. Since this complication
was recognized, we have attempted to reduce its risk by ensuring that the
guide-wire, reamers, and nail are sequentially centrally placed on both
anteroposterior and lateral intraoperative image-intensifier views of the
distal part of the femur. In addition, the use of second and third-generation
cephalomedullary nails, with their reduced stiffness and smaller radius of
curvature, may further help to reduce the risk of this complication in the
future.
Lag-screw complications occurred in twelve patients (4%) and often were
asymptomatic because of the low physical demands of these elderly patients.
Nail revision was required in five of the twelve patients, with another two
patients having surgery to change the proximal locking screw. Cutout failure
of hip screws has been strongly associated with poor lag-screw positioning
within the femoral head. Specifically, a tip-apex distance of <25 mm has
been shown to markedly reduce the risk of cutout
failure17,18.
Lack of standardization of our postoperative radiographs prevented detailed
analysis of the effects of tip-apex distance on cutout failure in our series.
Although the risk of cutout failure was low, it would be of interest in the
future to evaluate whether it can be further reduced by ensuring a tip-apex
distance of <25 mm in all patients. In our study, the prevalence of
mechanical failure in patients with the more unstable fractures involving the
piriformis fossa (Type-II injuries) was similar to that in patients with
fractures of a more simple configuration (Type-I injuries).
Our study has several limitations. Although we were able to prospectively
evaluate the functional and social status of our patients within the first
year after the injury, the high mortality rate and the general debility and
poor mental status of many of the surviving patients prevented more detailed
assessment of functional outcome or general health status. In addition, the
lack of a control group prevented us from drawing any firm conclusions with
regard to whether nailing offers definite functional advantages over the use
of other, more traditional, methods for the treatment of these fractures.
Because of the high mortality rate, it is possible that many of our patients
did not survive long enough for implant-related complications to develop.
The present study demonstrates the efficacy and versatility of
trochanteric-entry cephalomedullary nailing for the treatment of all types of
low-energy subtrochanteric fractures, including those with extension into the
piriformis fossa. The use of this implant is associated with an acceptable
rate of perioperative complications and satisfactory functional outcomes given
the poor physical status of many of these elderly patients before the
injury.
A table reviewing the literature regarding complications of the surgical
treatment of subtrochanteric fractures is 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). ?