The study group comprised sixty-four patients who had sustained an
intraoperative fracture of the femur during revision total hip arthroplasty
with a cementless diaphyseal fitting stem between December 1998 and March
2002. This extensively porous-coated stem is designed to obtain a tight distal
diaphyseal, so-called scratch, fit over 4 to 6 cm to achieve bone ingrowth and
distal fixation. The control group consisted of 147 patients who had undergone
revision with use of the same extensively coated stem during the same time
period but had not sustained an intraoperative femoral fracture. Fractures
were identified by a retrospective review of the operative record and verified
by examination of the postoperative radiographs. Permission was obtained from
the university and hospital internal review boards prior to the commencement
of the study.
The patient's medical record was reviewed to identify the side of the
operation, the duration of the hospital stay, the stem diameter and shape
(bowed or straight), the surgical approach (including whether an extended
osteotomy had been
performed16), and
whether the femur had been underreamed. Comorbidity was assessed by summating
scores for coexisting conditions according to the Charlson
index17 and
categorizing them as 0, 1, 2, or =3. The validated Vancouver system for
classifying postoperative periprosthetic
fractures18,19
was adapted for use in the intraoperative period and was employed to classify
the intraoperative
fractures20.
According to this classification system, the femur was divided into three
anatomical areas: A (the pertrochanteric region), B (the diaphysis), and C
(the distal metaphyseal, or supracondylar, region). In each anatomic location,
the fracture was categorized as 1 (a cortical perforation), 2 (an undisplaced
linear crack), or 3 (a displaced or unstable fracture).
Evaluation
Clinical
The potential fracture risk factors that were examined included age,
gender, bowed or straight stem, stem diameter, stem length, surgical approach
(extended or no extended osteotomy), and line-to-line reaming or underreaming
for bowed stems. When straight stems were used, it was routine to underream
the cortex by 0.5 mm. Outcome assessment was based on patient responses to the
Western Ontario and McMaster Universities Osteoarthritis (WOMAC) index, the
Oxford-12 and Short Form-36 (SF-36) questionnaires, and a satisfaction
questionnaire. Of the 211 consecutive patients treated with a revision
operation included in the retrospective analysis of risk factors, 115
(thirty-nine with an intraoperative fracture and seventy-six controls) were
assessed clinically at a minimum of two years (mean, thirty-one months)
postoperatively. The patients with and without a fracture were comparable with
regard to age, gender, and comorbidity scores
(Table I). This cohort of 115
patients was sent questionnaires. Any questionnaires that were not returned
were resent after six weeks, and, if the patient again failed to respond, he
or she was contacted by telephone.
Radiographic
All preoperative radiographs were reviewed, and the degree of preoperative
bone loss was recorded according to a modification of the classification
system of Paprosky and
Burnett21. Type-I
bone defects were minimal, and the femur did not substantially differ from
that encountered during the primary hip arthroplasty. In type-II defects, bone
loss was present principally in the metaphyseal region, with minimal
diaphyseal bone loss. Type-III defects had both metaphyseal and diaphyseal
bone loss, but at least 4 cm of distal fixation was achievable. Type-IV
defects had extensive metaphyseal and diaphyseal bone loss with thin cortices
and a widened canal that precluded reliable press-fit fixation.
All immediate postoperative radiographs were reviewed to characterize the
presence and type of an intraoperative fracture according to the Vancouver
classification system. At the time of follow-up, the cementless stem fixation
was classified radiographically, with the system described by Engh et al., as
bone ingrown; stable fibrous encapsulation; questionable, with signs of
impending instability; or definitely unstable, with implant
migration22. An
individual blinded to the purpose of the study made the assessment. The
prerevision thickness of the mid-diaphyseal femoral cortex and total femoral
width were also recorded.
Statistical Analysis
The primary outcome was function, as measured by the physical function
domain of the WOMAC index. The sample size was calculated with use of a power
of 0.8 (ß = 0.2), a = 0.05, a standard deviation of 18, and a
difference in effect size of 10 points on the WOMAC scale. To fulfill these
parameters, thirty patients were required in the fracture group and sixty, in
the control group.
The primary outcome was analyzed with use of a onetailed t test with a
significance level of p < 0.05. Regression analysis was then used to look
for factors that were predictive of functional outcome. The SF-36, Oxford-12,
and satisfaction questionnaires were also analyzed with use of the Student t
test. A chi-square test or Fisher exact test was employed as appropriate to
analyze associations between categorical variables.
Risk Factors for Intraoperative Fractures
All Types of Fractures
Sixty-four (30%) of the 211 patients sustained an intraoperative fracture;
the remaining 147 patients had no fracture (the control group). There was no
significant difference in age, sex, or comorbidity between these two groups
(Table II). According to the
Paprosky and Burnett classification, there was more preoperative bone loss in
the intraoperative fracture group than in the control group (p = 0.0024). The
fracture group also had significantly larger stem diameters (median, 18 mm;
mode, 19.5 mm) than did the control group (median, 16.5 mm; mode, 18 mm) (p =
0.046). In addition, the longest stem (10 in [254 mm]) had been used more
frequently in the fracture group (p = 0.049).
There was no significant difference in the fracture rate between the bowed
and straight stems: thirty-nine (61%) of the sixty-four stems were bowed in
the intraoperative fracture group compared with ninety-six (65%) of the 147
stems in the control group (p = 0.648). A posterior approach was used in 141
(67%) of the 211 procedures and in thirty-nine (61%) of the sixty-four
procedures in the fracture group. The approach was combined with an extended
osteotomy in 102 (48%) of the 211 procedures and in twenty-seven (42%) of the
sixty-four procedures in the fracture group. An anterolateral approach was
used in nine procedures; an anterolateral approach with an extended osteotomy,
in twenty-six; a trochanteric slide, in twenty-four; a transtrochanteric
approach, in six; and a vastus slide, in five.
None of the approaches was associated with an increased rate of
intraoperative fracture (p = 0.600, chi-square test). There was also no
significant association between the fracture rate and the side of the
procedure, preoperative diagnosis, prerevision presence of a cemented stem, or
revision of the acetabular component in addition to the stem.
The intraoperative fractures occurred most frequently during trial or
actual insertion of the stem, with a fracture occurring at this stage in
thirty-seven (18%) of the 211 procedures. The next most frequent time for
intraoperative fractures to occur was during cement removal, with twenty
fractures (13%) occurring during removal of the 160 cemented stems. Three
patients sustained fractures during both cement removal and stem insertion. An
additional four fractures occurred during hip reduction, and six were first
diagnosed on immediate postoperative radiographs. The types of fractures are
summarized in Table III.
Twenty-five of the thirty-nine B2 (minimally displaced diaphyseal split)
fractures propagated or connected with the trochanteric osteotomy.
Twenty-eight B2 fractures occurred during actual or trial insertion of the
stem, eight occurred during cement removal and canal preparation, and three
were only diagnosed on the postoperative radiograph made in the recovery room.
Two of the three patients in whom the fracture was identified on the
radiograph were managed with protected weight-bearing, and one was taken back
to the operating room, where cerclage wires were placed around the femur. Nine
of the eleven B1 (cortical perforation) fractures occurred during cement
removal, and the other two occurred during stem insertion. The methods of
treatment of all fractures are presented in
Table IV.
Diaphyseal Split Fractures
Thirty-nine patients (18%) sustained a type-B2 intraoperative fracture. In
addition, there were three type-C2 fractures that extended down to the
supracondylar region and two type-B3 (unstable) fractures. There was no
significant difference in the average age (p = 0.277) or in the gender
distribution (p = 0.946) between the patients who sustained a B2, B3, or C2
fracture and those with no fracture.
According to the Paprosky and Burnett classification, the group with a
diaphyseal split fracture demonstrated more preoperative bone loss than did
the control group (p = 0.0028). Compared with reaming line to line, the effect
of underreaming the femoral cortex by 0.5 mm was significantly associated with
B2 fractures: underreaming had been done in thirty-two (82%) of the patients
with a B2 fracture compared with fifty-five (37%) of the 147 who did not
sustain a fracture (p = 0.020). The stem diameters were significantly larger
in the group with a diaphyseal split fracture (median, 18 mm; mode, 19.5 mm)
than they were in the control group (median, 16.5 mm; mode, 18 mm) (p =
0.036), and the preoperative mid-diaphyseal cortex of the femur was an average
of 1.1 mm thinner in the patients with a B2 fracture (range, 4 to 11 mm) than
it was in the control group (range, 3 to 15 mm) (p = 0.006). The ratio of the
cortical width to the femoral diameter was also significantly lower in the
B2-fracture group: 0.23 (range, 0.12 to 0.31) compared with 0.25 (range, 0.1
to 0.38) in the control group (p = 0.013).
The length of the stem was not related to the rate of diaphyseal split
fractures, and although 254-mm stems had been used more frequently in patients
with a B2 fracture (in 27% compared with 16% in the control group) this
difference did not reach significance (p = 0.082). Two of the three C2
fractures occurred in patients with a 254-mm stem (p = 0.06). In addition,
with the numbers available, no significant association was found between
intraoperative fracture and preoperative diagnosis, a cemented stem, surgeon,
revision of the acetabular component in addition to the stem, or surgical
approach. The rate of intraoperative diaphyseal split fractures was the same
for the bowed and straight stems (p = 1.000, Fisher exact test) overall as
well as for the 203-mm (8-in) straight and bowed stems (p = 0.962). All three
straight 203-mm stems in the fracture group were inserted with underreaming,
whereas four of the sixteen were inserted with line-to-line reaming in the
control group (p = 1.00). Bowed 203-mm stems were less frequently inserted
with line-to-line reaming in the fracture group (three [20%] of fifteen
compared with twenty-six [43%] of sixty in the control group), but this
difference was not significant with the numbers available (p = 0.084).
Clinical Outcomes
Of the 115 patients who were evaluated two years or more after the surgery,
four had died by the time of follow-up, leaving thirty-nine patients with an
intraoperative fracture and seventy-two control patients. Another eighteen
patients could not be traced or did not complete their questionnaires
adequately, leaving thirty-two patients with an intraoperative fracture and
sixty-one controls. There was no significant difference, with the numbers
available, between these groups with regard to the final WOMAC, Oxford-12, or
SF-36 scores (Table V) (post
hoc power = 0.810).
Radiographic Outcomes
Radiographic follow-up was performed at a minimum of two years
postoperatively for thirty-three (85%) of the thirty-nine patients with an
intraoperative fracture and sixty-one (85%) of the seventy-two patients
without a fracture who were alive and had been followed clinically for a
minimum of two years. The mean duration of radiographic follow-up was 32.1
months (range, 24.8 to 53.4 months). According to the classification system of
Engh et al.22,
thirty-two (97%) of the thirty-three stems in the patients with an
intraoperative fracture had bone ingrowth and one (3%) had stable fibrous
ingrowth. In the control group, nine stems (15%) had stable fibrous ingrowth,
and three (5%) had unstable fibrous ingrowth; two of the stems with unstable
fibrous ingrowth were revised because of thigh pain.
In a series of cementless femoral revisions ranging in size from
seventy-four to 1536 hips, intraoperative fracture rates have ranged from 6.3%
(forty of 630 revisions) to 50% (thirty-seven of seventy-four
revisions)1,2,8,9,23-26.
Paprosky et al. reported a rate of intraoperative split fractures of 8.8%
(fifteen of 170) and a rate of intraoperative cortical perforation fractures
of 5.9% (ten of 170) in their long-term review of cementless femoral revision
arthroplasties27.
Our rates of diaphyseal split fractures and cortical perforation fractures,
although higher than those in the series of Paprosky et al., are within the
range reported in previous series.
Our study demonstrated that the degree of preoperative bone loss is a
principal risk factor for intraoperative fracture. The width of the cortical
wall is also a risk factor, and patients with a low ratio of cortical width to
femoral diameter appear to be at risk for a diaphyseal split fracture. Our
data do not support a previous finding that the revision of a cemented stem
increases the risk of intraoperative fracture compared with that associated
with revision of a primary cementless
stem6. The specific
surgical approach appears to be unrelated to fracture risk. It may be that the
prudent use of an extended trochanteric osteotomy decreases the risk of
fractures, including perforations associated with attempts to remove cement
entirely from the proximal end of the femur.
The use of either a bowed or a straight stem does not appear to be a risk
factor. However, underreaming rather than line-to-line reaming for the
implantation of a bowed stem is a significant risk factor for fracture.
Straight stems have the advantage of being easier to remove should revision
become necessary. It is noteworthy, though, that the two unstable
(grade-322) stems
were in the control group, in patients in whom line-to-line reaming had been
performed. Thus, there is a compromise between the risk of intraoperative
fracture and the risk of an unstable fibrous union. Interestingly, there was a
higher rate of bone ingrowth after an intraoperative B2 fracture, suggesting
that the presence of a stable fracture may stimulate bone ingrowth rather than
hinder it.
When a fracture does occur, the distal end of an extended osteotomy is a
frequent site for distal fracture propagation. We think that a prophylactic
distal cerclage wire or cable should be applied in patients with an extended
osteotomy.
In conclusion, intraoperative split fractures and perforation fractures
during revision cementless total hip arthroplasty are perhaps more common than
previously reported in the literature. The primary intrinsic risk factor is
the degree of preoperative bone loss. What is not as clear is the influence of
such fractures on patient mobilization, functional status, prosthetic
stability, the risk of subsequent fracture, and ultimate revision rates. The
management protocols that we employed do not appear to compromise
intermediate-term function, pain, or stiffness.