Abstract
Background: Persistent pain in the region of implanted hardware
following fracture fixation commonly leads to implant removal. This
prospective study evaluated patient outcomes and pain reduction following
removal of orthopaedic hardware implanted for fracture fixation.
Methods: Sixty patients who had been treated previously for a
fracture and complained of pain in the region of the fracture fixation
hardware constituted the study cohort. Patients were carefully examined by the
treating physician to rule out other causes of pain such as infection and
nonunion. Baseline data were recorded preoperatively. Data obtained
postoperatively at three, six, and twelve months included a visual analog pain
scale score and results on the Short Musculoskeletal Function Assessment
Questionnaire and the Medical Outcomes Study Short Form-36. At the one-year
interval, a patient satisfaction questionnaire was completed and outcomes were
analyzed.
Results: There were no complications associated with implant removal
surgery. Three patients did not have complete follow-up, leaving a total of
fifty-seven patients with complete follow-up. At one year, all patients
indicated that they were satisfied, that they would have the procedure done
again, and that their overall function had improved. The scores for pain on
the visual analog scale decreased from a mean (and standard deviation) of 5.5
± 2.5 before hardware removal to 1.3 ± 1.8 after hardware
removal, with an overall improvement at one year of 76% (p = 0.00001). At one
year, thirty (53%) of the fifty-seven patients had complete resolution of
pain. In addition, the results on the Short Musculoskeletal Function
Assessment Questionnaire showed a 43% improvement from baseline (p = 0.0001),
and the results on the physical component of the Short Form-36 showed a
similar improvement of 40% (p = 0.0001).
Conclusions: Following fracture-healing, removal of hardware is safe
with minimal risk. Improvement in pain relief and function can be
expected.
Level of Evidence: Therapeutic Level IV. See Instructions
to Authors for a complete description of levels of evidence.
Persistent pain in the region of implanted orthopaedic hardware after
radiographic evidence of fracture union commonly leads to implant
removal1-3.
In addition, concerns about systemic and local effects of retained implants
have led many patients to request elective hardware
removal3,4.
It is important to understand what appropriate patient expectations for pain
relief and functional improvement should be.
When hardware is removed for pain relief alone, the results may be
unpredictable and dependent on the type of implant and its anatomic location.
Published studies on hardware removal for pain have all been
retrospective2,5-10.
Given the costs of the procedure as well as time off for postoperative
recovery, the decision to remove hardware has substantial economic
implications1,5.
While these procedures are frequently viewed as simple, they can be
challenging. Furthermore, hardware removal can lead to further complications,
such as neurovascular injury, refracture, a retained implant, or recurrence of
deformity3,11-13.
This study was designed to determine prospectively whether removal of
symptomatic fracture-fixation hardware provides pain relief and functional
improvement of the affected extremity.
Over a two-year period, from 2004 to 2006, sixty consecutive patients
between seventeen and eighty-one years of age (mean, 46.9 years) were
identified at our institutions as having symptomatic hardware following
operative fracture care. All fractures were united radiographically, and the
patients were seen long enough after their injuries to have reached maximal
recovery. The study protocol had institutional review board approval, and all
patients signed an informed-consent form.
All patients were carefully examined by the treating physician to rule out
other causes of pain such as infection, neurogenic origin, and fracture
nonunion. Patients in whom hardware was removed for a known infection or had
been placed with planned later removal were excluded from the study.
All patients had generalized complaints of pain in the region of the
implants. These complaints were suspected by point tenderness over superficial
metal implants on physical examination. Radiographic confirmation was used for
deeper implants. Intramedullary nail-related symptoms were confirmed if
patients complained of pain at the site of the entry portal or pain in the
region of the locking bolts and other causes of pain had been excluded. All
fractures had united both clinically and radiographically at the time of
hardware removal.
The baseline data were recorded preoperatively and included patient age,
sex, medical comorbidities, and level of pain as rated on a visual analog
scale from 0 (no pain) to 10 (the most pain imaginable). Smoking status, body
mass index, insurance status, Workers' Compensation status, and length of time
that the hardware had been in place were also recorded. The Short
Musculoskeletal Function Assessment Questionnaire (SMFA) and the Medical
Outcomes Study Short Form-36 (SF-36), two standardized surveys assessing
functional outcome, were completed at baseline as well as at three, six, and
twelve months
postoperatively2.
All surgeries were performed on an outpatient basis with the patient under
regional or general anesthesia in the operating room. In patients in whom
infection was suspected on the basis of conditions found at the time of
surgery, deep cultures were obtained.
At one year, a patient satisfaction questionnaire was administered to all
patients. It consisted of three questions: (1) Are you happy that the hardware
was removed? (2) Would you have the surgery performed again? (3) Do you think
your overall function has improved since you underwent this procedure?
Statistical Methods
Standard statistical methods were used. Pain scores (assessed on a visual
analog pain scale) were correlated with the functional outcome scores (SMFA
and SF-36) for all patients. Statistical analysis of functional outcome and
pain scores was performed with use of paired t tests to compare baseline
scores on the pain visual analog scale, SMFA, and SF-36 with those at each
follow-up point. A linear regression model was used to assess whether body
mass index, smoking, comorbidities, insurance status, Workers' Compensation
status, and age had an effect on overall outcome. The level of significance
was set at p < 0.05.
Our power analysis showed that sixty patients would provide >90% power
to detect a difference of 1.5 points on the SMFA between baseline and the
one-year follow-up evaluation. This difference was based a priori on our
belief that a difference of <2 points on the SMFA would indicate a return
to baseline function, while a difference of >2 points would indicate
incomplete recovery. We used 1.5 points as a more conservative estimate.
Between September 2004 and December 2005, sixty consecutive patients who
had reported pain in the region of retained fracture-fixation implants were
identified and recruited. Approximately 30% of the patients had had the
hardware implanted at an outside institution. Fifty-seven patients (95%) who
completed follow-up at all time-points form the basis for this study. One
patient was lost to follow-up after the three-month follow-up visit. One
patient was involved in ongoing litigation and was advised by her lawyer not
to complete the study three months postoperatively. One patient was excluded
after final operative cultures were positive for infection.
The patients enrolled in this study included thirty-three women and
twenty-seven men who were between the ages of seventeen and eighty-one years
(mean, 46.9 years). Fourteen (23%) of the sixty patients had Medicaid
insurance. Five of the sixty patients were involved in a Workers' Compensation
claim. The average time that the metal implant had been in place was 38.6
months (range, six to 216 months; median, fifteen months). The average body
mass index was 26.3 (median, 26.7). Nineteen patients had comorbidities, and
eleven patients were smokers. The implants were in the upper extremity
(shoulder, humerus, olecranon, or forearm) in eight patients and in the lower
extremity (hip, femur, tibia, ankle, or calcaneus) in fifty-two patients.
Demographic data, implant locations, and type of metal implant are listed in
Table I. There were no
intraoperative or postoperative complications associated with implant removal
surgery.
The score for pain on the visual analog scale
(Fig. 1) showed a mean
reduction in pain of 76% from the baseline preoperative value at one year (p =
0.0001). The overall pain scores decreased from a mean (and standard
deviation) of 5.5 ± 2.5 before hardware removal to a mean of 1.3
± 1.8 after hardware removal. Thirty (53%) of the fifty-seven patients
had complete resolution of pain at one year, with a response of 0 on the
visual analog pain scale.
The SMFA results (Fig. 2)
showed an overall improvement of 44% in function from baseline (p = 0.0001).
With a lower SMFA score indicating a higher functioning level, the mean score
was 99.2 ± 31.9 at baseline and improved to 56.0 ± 14.5 at the
final evaluation (p = 0.0001).
The score on the physical component of the SF-36
(Fig. 3) showed a similar
improvement of 39% (p = 0.0001) at one year compared with baseline. A higher
SF-36 score indicates a higher functioning level. The actual mean score
improved from 39.1 ± 10.8 at baseline to 54.2 ± 6.5 at one
year.
The score on the mental component of the SF-36 showed the greatest
improvement at three months, with a 10.5% gain compared with baseline (p =
0.0063). The mean score was 45.5 ± 12.9 before surgery and 50.4
± 10.5 at three months. At one year, the mean score was 48.6 ±
8.04, an overall improvement of only 6.8% compared with baseline and the
difference was not significant (p = 0.108).
We also analyzed the outcomes on the pain scale, SMFA, and SF-36 to
determine whether there was a difference between patients with upper extremity
hardware and those with lower extremity hardware. Fifty patients had hardware
in the lower extremity, and seven patients had hardware in the upper
extremity. For the seven patients in the upper extremity group, the average
score on the visual analog pain scale was 4 ± 2.3 at baseline and 1.8
± 3.1 at the one-year follow-up examination, which was an overall
improvement of 55% (p = 0.03).
For the fifty patients with hardware in the lower extremity, the average
score on the visual analog pain scale was 5.6 ± 2.5 at baseline and 1.3
± 1.7 at the one-year follow-up examination, which was an overall
improvement of 77% (p = 0.001) (see Appendix). The patients with hardware in
the upper extremity were less symptomatic initially than the patients with
hardware in the lower extremity, with a lower pain level at baseline (4.0
compared with 5.6) and therefore a smaller reduction in pain at one year (55%
compared with 77%).
The mean SMFA score for the upper extremity group was 76.3 ± 24.1 at
baseline and 49.5 ± 8.1 at the time of the final follow-up, an
improvement of 35% (p = 0.013). For the lower extremity group, the mean score
on the SMFA was 101.9 ± 31.8 at baseline and 57.0 ± 15.1 at one
year, an overall improvement of 44% (p = 0.001) (see Appendix).
The SF-36 physical component score for the upper extremity group showed an
overall improvement of 32% (p = 0.003) at one year. The actual mean score
increased from 44.1 ± 6.3 at baseline to 58.2 ± 3.5 at one year.
For the lower extremity patients, the SF-36 physical component demonstrated a
41% improvement (p = 0.001) at one year. The actual mean score increased from
38.2 ± 11.0 at baseline to 53.7 ± 6.6 at one year (see
Appendix).
In summary, the patients who had an implant removed from a lower extremity
had a slightly larger percentage of improvement in the SF-36 score (41%)
compared with the patients who had an implant removed from an upper extremity
(32%). They also had had a greater baseline disability, which may explain the
degree of improvement. However, the upper extremity group still had slightly
better SF-36 scores than the lower extremity group at one year (58.2 compared
with 53.7).
We also evaluated the impact of body mass index, smoking, comorbidities,
insurance status, Workers' Compensation status, and age on the overall
outcome. With the numbers studied, these potential confounders did not appear
to have an effect on the change in pain or SMFA functional scores between
baseline and any time-point. The one exception was the effect of age on the
SMFA scores. For each one-year increase in age, the difference between the
total SMFA score at baseline and at one year increased by >0.5 point
(ß = 0.66, p = 0.015).
At one year, all fifty-seven patients (100%) responded on the satisfaction
questionnaire that they were satisfied, that they would have the procedure
done again, and that their overall function had improved.
Removal of fracture fixation hardware is a common procedure, accounting for
approximately 5% of all orthopaedic procedures done in the United
States14. We found
that all patients who underwent removal of fracture fixation implants that had
been causing symptoms were ultimately very satisfied with the outcome and
would undergo the procedure again. In addition, all patients had improvement
in all aspects of the SMFA and SF-36 scores as early as three months
postoperatively, and those trends continued to one year. All sixty procedures
were performed on an outpatient basis without complications.
In one study, 5095 implants placed for fracture fixation were removed after
fracture-healing1.
In relation to the number of internal fracture fixations performed, the total
removal percentage was 81%. Removal of the implant accounted for 29% of the
elective procedures and 15% of total orthopaedic procedures performed at that
institution during a seven-year period compared with a rate of 6% of
orthopaedic procedures in all of Finland for the duration of that study.
Rates of implant removal have varied on the basis of the anatomic location
and type of implant. In one study of fifty-five patients undergoing
tension-band wire fixation of olecranon fractures, 61% required revision
surgery for painful
hardware10. In a
retrospective review of surgically treated patellar fractures, nine of
eighty-seven patients underwent removal of symptomatic
hardware5. Neither
study, however, noted pain relief outcomes.
Intuitively, one may assume that superficial implants present a greater
problem with regard to symptoms; however, that does not seem to be borne out
in the literature. Regarding deep implants such as intramedullary nails, a
retrospective review of eighty patients by Dodenhoff et al. noted that eleven
of seventeen patients who underwent implant removal following a healed femoral
fracture experienced pain
relief7. With tibial
intramedullary nails, knee pain has been a common indicator for removal.
Keating et al.8
showed a 45% rate of complete relief of knee pain in 110 knees after tibial
nail removal. In addition, 35% of the patients experienced partial relief,
while 20% had no relief. In a retrospective review of the cases of 169
patients, Court-Brown et
al.9 noted that 27%
had complete pain relief and 69% had marked relief after nail removal.
However, 3.2% of the patients reported worsening pain after hardware removal.
In another study involving 100 patients, 12% of previously asymptomatic
patients had knee pain develop after tibial nail
removal15.
In a retrospective study published in 2004, Gosling et
al.16 found that
after femoral nail removal in fifty-one patients who had been asymptomatic
preoperatively, ten (20%) had symptoms develop postoperatively. Furthermore,
of the fifty-eight patients who had pain preoperatively, 78% reported
improvement after nail removal, which was similar to the rate in the present
study. Those authors concluded that only symptomatic patients should be
considered for femoral nail removal. In the present prospective study, we had
a very similar overall improvement rate (77%), yet none of the five patients
who had removal of a femoral nail had a new onset of pain postoperatively.
In a recent study that used the same outcome measures as were used in the
present study, Brown et
al.2 retrospectively
examined functional outcomes after internal fixation of ankle fractures and
found lower scores for pain, measured with a visual analog scale, and lower
scores on the SMFA and SF-36 questionnaires for patients with pain overlying
hardware on the lateral side of the ankle. Of the thirty-nine patients who
reported pain, twenty-two underwent removal of hardware, but only eleven had
improved pain relief. According to the authors, the results confirmed
"that removal of hardware after operative treatment of a fracture was
associated with neither an optimal functional result nor a reduction in
long-term complications." This is in contrast to our prospective
results, which showed a 77% reduction of pain.
Richards et al., in
19924, prospectively
observed the removal of metal internal fixation devices in eighty-six
patients. Of those patients, only twenty-seven requested surgery because of
pain. The others were admitted for routine surgery on medical advice. The
authors reported that there was no correlation among symptoms, the length of
time the implant had been in situ, or the location of the implant. Good
results were achieved in 91% of the symptomatic patients on the basis of
patient questionnaires. Overall, the complication rate was 3%, and the authors
concluded that it might be appropriate to leave asymptomatic implants in situ.
While our study group comprised only symptomatic patients, it confirmed the
observations of Richards et al.
Although we did not intend to enroll patients who had an infection around
the hardware, one patient in the study had positive cultures for infection and
was subsequently excluded. This confounding variable may account for some pain
at hardware sites postoperatively.
Age was the only factor that we could identify that was related to
functional outcome scores. This finding may be interpreted to mean that, if
the hardware is truly causing pain, other variables except for increasing age
may have less impact on functional status. This is intuitive because the
overall functional status of an eighteen-year-old patient is different from
that of a seventy-year-old patient.
One limitation in our study is that we did not know the percentage of
patients with a healed fracture who had removal of symptomatic hardware.
Patients who underwent fracture fixation at our institution may have had
follow-up care provided elsewhere, and some were lost to follow-up. We
enrolled and followed only patients with a complaint of pain related to
hardware.
In addition, one of the difficulties in evaluating patients for functional
improvement on the basis of the SF-36 and SMFA questionnaires is the fact that
they may have additional disabilities not related to orthopaedic hardware that
compromise the overall functional and mental scores. Furthermore,
posttraumatic arthritis related to the initial injury may have developed in
some patients. This may explain why the improvement in function based on the
SF-36 and SMFA scores in the present study was not as substantial as the
overall pain reduction.
In our study, implants in all anatomic locations were considered when
patients were enrolled, to evaluate the spectrum of orthopaedic hardware
removal and the potential improvements. We did differentiate, however, between
upper and lower extremity implants. It appears that the patients with an
implant in the upper extremity were less symptomatic since they started from a
lower pain level (4 compared with 5.6), with only a 55% reduction in pain
compared with a 77% reduction for patients with an implant in a lower
extremity. The functional outcomes (SMFA and SF-36) also showed more
improvement in the patients with an implant in a lower extremity. A larger
prospective study would be needed to clarify the outcomes at various anatomic
locations.
In conclusion, it is important to be able to counsel patients about the
expected level of success when planning the removal of fracture fixation
implants that are presumed to be the cause of discomfort. The high level of
patient satisfaction, the improvement in both pain relief and function, as
well as the overall minimal risk for complications make this particular
procedure an excellent choice for patients who have persistent pain following
uneventful fracture-healing in the region of an orthopaedic implant.
Graphic representations of the scores on the mental and physical components
of the SF-36 and comparisons of scores for patients with hardware in an upper
extremity and those with hardware in a lower extremity 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). ?
Note: The authors thank Uzoma Nwoso, MD, and Jaspal Singh, MD,
for their help in the preparation of this manuscript.
Bostman O, Pihlajamaki H. Routine
implant removal after fracture surgery: a potentially reducible consumer of
hospital resources in trauma units. J Trauma.1996;41:
846-9.41846Â
1996Â
[PubMed][CrossRef] Â
Brown OL, Dirschl DR, Obremskey WT.
Incidence of hardware-related pain and its effect on functional outcomes after
open reduction and internal fixation of ankle fractures. J Orthop
Trauma.2001;15:
271-4.15271Â
2001Â
[CrossRef] Â
Busam ML, Esther RJ, Obremskey WT.
Hardware removal: indications and expectations. J Am Acad Orthop
Surg.2006;14:
113-20.14113Â
2006Â
Â
Richards RH, Palmer JD, Clarke NM.
Observations on removal of metal implants. Injury.1992;23:
25-8.2325Â
1992Â
[PubMed][CrossRef] Â
Smith ST, Cramer KE, Karges DE, Watson
JT, Moed BR. Early complications in the operative treatment of patella
fractures. J Orthop Trauma.1997;11:
183-7.11183Â
1997Â
[PubMed][CrossRef] Â
Jacobsen S, Honnens de Lichtenberg M,
Jensen CM, Torholm C. Removal of internal fixation—the effect on
patients' complaints: a study of 66 cases of removal of internal fixation
after malleolar fractures. Foot Ankle Int.1994;15:
170-1.15170Â
1994Â
[PubMed] Â
Dodenhoff RM, Dainton JN, Hutchins PM.
Proximal thigh pain after femoral nailing. Causes and treatment. J Bone
Joint Surg Br.1997;79:
738-41.79738Â
1997Â
[CrossRef] Â
Keating JF, Orfaly R, O'Brien PJ. Knee
pain after tibial nailing. J Orthop Trauma.1997;11:
10-3.1110Â
1997Â
[PubMed][CrossRef] Â
Court-Brown CM, Gustilo T, Shaw AD. Knee
pain after intramedullary tibial nailing: its incidence, etiology, and
outcome. J Orthop Trauma.1997;11:
103-5.11103Â
1997Â
[PubMed][CrossRef] Â
Romero JM, Miran A, Jensen CH.
Complications and re-operation rate after tension-band wiring of olecranon
fractures. J Orthop Sci.2000;5:
318-20.5318Â
2000Â
[PubMed][CrossRef] Â
Sanderson PL, Ryan W, Turner PG.
Complications of metalwork removal. Injury.1992;23:
29-30.2329Â
1992Â
[PubMed][CrossRef] Â
Beaupre GS, Csongradi JJ. Refracture
risk after plate removal in the forearm. J Orthop Trauma.1996;10:
87-92.1087Â
1996Â
[PubMed][CrossRef] Â
Langkamer VG, Ackroyd CE. Removal of
forearm plates. A review of the complications. J Bone Joint Surg
Br.1990;72:
601-4.72601Â
1990Â
Â
Rutkow IM. Orthopaedic operations in the
United States, 1979 through 1983. J Bone Joint Surg Am.1986;68:
716-9.68716Â
1986Â
[PubMed] Â
Boerger TO, Patel G, Murphy JP. Is
routine removal of intramedullary nails justified. Injury.1999;30:
79-81.3079Â
1999Â
[PubMed][CrossRef] Â
Gosling T, Hufner T, Hankemeier S, Zelle
BA, Muller-Heine A, Krettek C. Femoral nail removal should be restricted in
asymptomatic patients. Clin Orthop Relat Res.2004;423:
222-6.423222Â
2004Â
[PubMed][CrossRef] Â