An estimated six million fractures occur each year in the United States,
and up to 10% of these may result in
nonunion1. Fractures
account for the largest (24%) total lifetime cost associated with any one
injury type, with over $99 billion in estimated medical costs and productivity
loss in the United
States2. Fracture
care, therefore, is a major health-care priority. As new therapies emerge to
enhance fracture-healing, it is crucial that they be evaluated rigorously to
assess their effect. A critical element in conducting high-quality and
clinically relevant trials to study such therapeutics is choosing the right
outcome to measure.
Unfortunately, assessment of fracture-healing both in clinical practice and
outcomes research is a subjective process without a readily available gold
standard. In the laboratory, fracture-healing is an observable continuum. From
the initial fracture hematoma and inflammatory response, to the soft callus,
hard callus, and remodeling phases, it is difficult to determine at what point
a fracture has "healed." The mechanical environment affects
fracture-healing, with stability favoring intramembranous ossification and
instability favoring endochondral ossification. Therefore, the same fracture
may appear histologically and radiographically different depending on how it
is stabilized. In the clinical setting, it is not an option to routinely
obtain a biopsy specimen from the fracture site or apply expensive
high-resolution imaging modalities; therefore, the clinician must rely on
cruder measures to determine the state of healing. Moreover, it is unclear how
patient-important quality-of-life measures correlate with this biological
process. Fracture-healing is often dichotomized to aid in clinical
decision-making and for the purpose of comparing results of therapies in
studies, but, by simplifying this continuum, there is a substantial risk of
drawing biased conclusions and losing information.
It has been difficult to develop a consistent definition of
fracture-healing that is both clinically and biologically accurate. In an
international survey of over 400 orthopaedic trauma surgeons, it was found
that a wide range of clinical and radiographic criteria were being used to
define tibial fracture union, with the time necessary to declare a nonunion
ranging anywhere from two to twelve
months3. This lack
of agreement among clinicians and researchers poses an obstacle for performing
clinical trials to test therapeutic technologies in fracture repair.
Inaccurate or imprecise outcome measurement threatens the ability to correctly
identify efficacious interventions that promote fracture repair. The purpose
of this review is to describe available methods of assessing clinical
fracture-healing and summarize those that are prevalent in the recently
published literature.
Conventional Radiography
Several measures of radiographic healing are available to clinicians
studying fracture repair, but the oldest and most commonly used method is
conventional radiography. Conventional radiography allows qualitative
assessment of callus formation, cortical bridging, loss of the fracture line,
and trabecular crossing at the fracture site. Its advantages are wide
availability, low cost, and low radiation dosage. Unfortunately, numerous
investigators have found poor correlation between radiographic features and
mechanical
strength4-6.
In testing multiple criteria of diaphyseal tibial fracture-healing on
orthogonal views, cortical bridging was found to be the most reliable, with an
interobserver kappa statistic of 0.75 (95% confidence interval, 0.61 to
0.89)7. This and
other commonly used definitions have been shown to be poor or unreliable with
respect to metaphyseal fracture-healing and varied with respect to type of
fixation8,9.
Absorptiometry and Photodensitometry
Numerous techniques have measured bone mineral density from the absorption
rates of photons directed at a body part, including
photodensitometry10,
single-photon
absorptiometry10-12,
and dual x-ray
absorptiometry13.
While these methods are not considered high-quality imaging techniques like
computed tomography or magnetic resonance imaging, they have certain
advantages, including low cost and utility in the presence of internal
fixation devices. Of these modalities, high-resolution dual x-ray
absorptiometry has been found to have the highest sensitivity (100%) and
specificity (78%) (by sixteen weeks) for diagnosing nonunion of tibial
fractures. While these modalities provide a more quantitative assessment of
fracture-healing, they have not been shown to add to the clinical management
of tibial shaft
fractures14 and are
rarely used for clinical or research purposes.
Bone Scintigraphy
Bone scintigraphy involves the production of a two-dimensional image of the
distribution of radioactivity in tissues after the internal administration of
a radiopharmaceutical imaging agent such as technetium-99m methylene
diphosphonate. Much of the interest in scintigraphy has focused on its
predictive power for the development of delayed union or nonunion early in the
course of fracture-healing (sensitivity 70%, specificity
90%)15.
Measurements are based on a relationship between early (within fifteen minutes
of injection) uptake at the fracture site and adjacent normal bone between one
and four weeks after injury. However, validation of this modality has proved
to be difficult because measurements have been unreliable in the setting of
internal
fixation16.
Ultrasonography
Ultrasonography is based on images generated from cyclic mechanical
pressure waves transmitted through tissues. Ultrasound is inexpensive,
portable, and lacks ionizing radiation; an increase in the resolution of
images has led investigators to apply ultrasonography to the assessment of
fracture-healing in recent
years17-19.
In one study of forty-seven patients with tibial fractures who underwent
unreamed medullary nailing, ultrasound findings at six to nine weeks had a 97%
positive predictive value (sensitivity 100%, specificity 92%) of union (the
gold standard used was radiographic evidence of healing or lack of need for
additional
surgery)20. While
published reports are promising, ultrasonography continues to be quite
sensitive to overlying soft tissues and operator technique, which have limited
its clinical use.
Computed Tomography
Numerous studies have expounded the benefits of computed tomography in
providing a more accurate assessment of
healing21 and a
higher correlation with fracture
rigidity22.
Moreover, computed tomography may be particularly useful in assessing
metaphyseal and periarticular fracture-healing in situations in which
conventional radiographic findings of callus formation are less
obvious23. Using a
composite gold standard of intraoperative findings or no surgery required
within six months of initial evaluation, one study showed good reliability and
diagnostic accuracy for fracture nonunion (kappa = 0.67, p < 0.0001;
sensitivity = 100%, specificity = 62%), versus plain radiography (kappa =
0.14, p = 0.36; sensitivity = 54%, specificity =
62%)24.
Quantitative computed tomography has expanded on the benefits of computed
tomography technology in allowing a quantitative assessment of callus volume
and density of bone based on Hounsfield units. Early studies that made use of
quantitative computed tomography showed a high correlation with bone mineral
content25,26,
and newer three-dimensional quantitative computed tomography measurement of
true volumetric bone mineral density and callus density allows prediction of
torsional strength and torsional stiffness in the diaphysis of long
bones27.
Experimental applications of high-resolution and microquantitative computed
tomography are discussed further in the review by Kalpakcioglu et al. (pp.
68-78), which
is included in this supplement. At the present time, application of these
techniques in large-scale clinical trials is still limited by cost, difficulty
of accurate quantitative assessment in the setting of implants, and higher
radiation dosage.
Mechanical property testing has been used primarily in the laboratory and
in in vitro settings; several techniques have been introduced for use at the
bedside to assess fracture-healing and will be described briefly here.
Mechanical property assessment falls into two classes: vibrational analysis
and biomechanical testing.
Vibrational Analysis
Vibrational analysis has the advantage of being noninvasive and painless,
and it is comprised of two modalities—resonant frequency analysis and
computerized sonometry. Resonant frequency analysis is based on the tendency
of a material (bone) to oscillate at maximum amplitude with a certain
frequency. As bone heals, the change in elasticity alters this resonant
frequency and can be detected by causing a vibration with an instrumented
hammer and measuring the resultant accelerations elsewhere with an
accelerometer. Resonant frequency analysis has shown a high correlation with
bending rigidity and torsional stiffness when the Young modulus has reached at
least 5% of intact
bone28,29.
However, studies have shown that these measurements are only reliable in the
setting of a subcutaneous
bone30, that the
success of such analysis is dependent upon the location of the fracture, and
that the method is potentially unreliable when applied to rigidly stabilized
fractures31.
Computerized sonometry is based on the measurement of sound transmission
across fracture gaps. Techniques have been described for reliable clinical
monitoring of ultrasound propagation speeds for the monitoring of
conservatively managed long-bone
fractures32. Close
correlation with clinical and radiographic healing as well as findings from
quantitative computed tomography and bone densitometry has been shown in a
study of externally fixed tibial
fractures33. Still,
unreliable measurements can result from a varying thickness of soft tissue and
even slight differences in path length. Thus, sensitivity to interposed
tissues and internal fixation devices make these two modalities uncommon in
studies of fracture repair.
Biomechanical Testing
Biomechanical testing involves true measures of stiffness and strength in
bending and in torsion and can be performed in either a direct or an indirect
fashion. Direct testing involves measurement of strain with use of a strain
gauge attached to an external fixator column or implanted into a
conservatively managed fracture under local
anesthesia34-36.
Direct biomechanical testing, in contrast, involves measurement of angulation
at the fracture site secondary to applied load by either radiography or
surface
measurement37,38.
An obvious disadvantage of this technique is the requirement for the removal
of any splint or fixation device and potential secondary discomfort to the
patient. Moreover, neither direct nor indirect testing has been validated in
the setting of commonly used internal fixation devices such as plates or
intramedullary nails.
We systematically reviewed the literature in order to summarize prevalent
methods of assessing outcomes and defining union in long-bone clinical
fracture-healing research. We performed a comprehensive search of MEDLINE and
the computerized electronic journal databases for articles published in three
prominent orthopaedic journals— The Journal of Bone and Joint
Surgery (American Volume), The Journal of Bone and Joint Surgery
(British Volume), and the Journal of Orthopaedic Trauma—from
January of 2000 through December of 2006. Articles that did not explicitly
state the criteria by which fracture union was determined were excluded. Key
word, title, abstract, and medical subject headings were used. Additional
articles were sought by performing a manual search of the table of contents.
Two reviewers independently assessed studies for inclusion eligibility and
extracted data in parallel, with disagreements resolved by consensus.
Despite the numerous modalities available to assess fracture union reviewed
here, we found a combination of conventional radiography and ad hoc clinical
questions to be the most common means of assessing outcome in recently
published clinical studies. We also found a surprising lack of reporting on
reliability for subjective radiographic and clinical measures of union and on
blinding of assessors. While the use of general health or region-specific
outcome instruments appears to be on the increase, only about two-thirds of
studies used them to supplement their assessment of outcome.
Assessing fracture-healing in clinical trials of interventions that are
meant to augment fracture repair remains a challenge. One reason is that we
are attempting to simplify a complex biological continuum that varies
according to fracture location, choice of treatment or fixation, and multiple
host and injury factors. Another issue is that we rely on marginally valid and
unreliable measures both in clinical practice and research. Even when
validated instruments such as the Short Form-36 or Musculoskeletal Functional
Assessment are used, the threshold for establishing efficacy is difficult to
state. Nevertheless, health-related quality-of-life instruments are
increasingly important in assessing outcomes that matter most to patients.
Several factors will be instrumental in conducting better clinical trials
of fracture repair in the future. First, explicitly stating whether a trial is
meant to establish therapeutic efficacy versus effectiveness will help
determine the optimal outcome to measure. This will aid in the decision-making
process with regard to study design, sample size, and power calculations.
Efficacy trials may require finer, more quantitative measures that are not
feasible for effectiveness studies, which intend to generalize to a larger
population. Second, attention to patient-important outcomes ought to increase
with validated general and disease-specific or region-specific instruments
applied whenever possible. Third, while diagnostic accuracy cannot be verified
in most clinical settings, assessor blinding and reliability statistics (i.e.,
kappa or intraclass correlation coefficient) should be provided. This will
inform readers about the potential for measurement bias and how consistent or
precise the used definition is.
In the future, instruments may be developed that capture the multiple
factors that constitute biological and clinical fracture-healing. More
objective assessment of fracture-healing will likely draw on enhanced imaging
technologies and molecular markers that can be used to track the healing
process with greater resolution. Assessing patient-important end points
requires the development and validation of disease-specific measures that
better assess the experience of fracture-healing. Analyzed separately or
combined into a composite definition of objective and subjective parameters,
these future techniques will provide a more robust illustration of the
fracture-healing process. Until such technologies become available and their
use is supported by evidence, investigators will have to choose from the
methods available and report outcomes with an awareness of the inherent
advantages and limitations as reviewed in this article. ?