Fig. 1 Sequential radiographs of a healing fracture illustrate the continuum from definitely broken to definitely healed and the difficulty if identifying exactly when solid union occurs.
Courtesy of Paul Tornetta III, MD


Published 7/1/2008
Annie Hayashi

Fracture healing on trial

By Annie Hayashi

Panel addresses challenges of measuring healing in clinical trials

Because every orthopaedist treats fractures, measuring healing is an important clinical consideration. It’s also crucial for researchers who conduct clinical trials in fracture treatment.

“Although many methods may be used to assess fracture healing in clinical studies,” said Theodore Miclau, MD, “no consensus exists on the most valid and reliable ways to determine union, the most important outcomes, or the most relevant fracture models.”

Citing one study that assessed tibial fracture healing, Dr. Miclau showed the tremendous variation in how respondents measured fracture healing—from callus size to cortical continuity to ability to bear weight and pain at the fracture site.

“When participants were asked to define ‘delayed union,’” responses ranged from one month up to eight months,” explained Dr. Miclau. A literature review showed that “radiographic definition only” was used as a criterion in 37 percent of the articles and that 62 percent used “radiographic and clinical definition.”

“Current orthopaedic literature shows a clear lack of consensus in the clinical and radiographic definitions of fracture healing. Without valid, reliable clinical or radio­graphic measures of union, interpretation of fracture care studies remains difficult,” Dr. Miclau concluded.

Measuring healing in clinical research
“Without a definition of fracture healing, measuring healing in clinical research is difficult,” agreed Paul Tornetta III, MD, who defined healing as a “limb that is strong enough to support function.”

Using radiographs to evaluate fracture healing is difficult. “Clearly at the end, the fracture is healed and at the beginning it is not,” said Dr. Tornetta. “But in the middle, it is unclear when a solid union has occurred (Fig. 1).”

Studies that report a time to union to define fracture healing are also problematic. “Patients don’t always keep appointments,” he noted. “How can one define a time to union when a patient is seen in 8-week intervals?” Instead of using time to union as an endpoint, Dr. Tornetta recommends using “percent united at specific time points.”

Another problem is separating pain and symptoms from fracture union. “How the patient is feeling may determine whether that fracture is going to unite,” he said.

Research requires reproducible measurements. “It’s important to choose a point and, using your experience, predict when union will occur. We can’t ignore the clinical circumstances because the function far outweighs the radiographic parameters,” he concluded.

What outcomes should be measured?
“Research should report clinical outcomes of relevance to treating physicians and patients,” said Marc F. Swiontkowski, MD. Among the clinical outcomes that should be reported in studies are limb length, range of motion, presence or absence of infection, fracture union, and limb strength—as measured by strength testing and sensation, swelling, and cold tolerance. “All can be assessed and are relevant,” according to Dr. Swiontkowski.

Fig. 1 Sequential radiographs of a healing fracture illustrate the continuum from definitely broken to definitely healed and the difficulty if identifying exactly when solid union occurs.
Courtesy of Paul Tornetta III, MD
Tibial shaft fracture
Courtesy of Mohit Bhandari, MD, MS

Functional outcome measures must be validated to ensure that they are measuring what they are designed to measure: patient function. These measures can be strongly affected by confounders such as age, medical comorbidity, morbid obesity, other musculoskeletal and nonmusculoskeletal injuries, educational level, and social support; therefore, information on these important variables should be reported.

Although pain is often conveyed using a visual analog scale (VAS), “One patient’s 2 is another patient’s 10,” he said. He recommended use of the “common pain experience” tool as an adjunct to the VAS. Likewise, “return to work” is often related to the patient’s educational level and job functional demand.

Among the validated instruments Dr. Swiontkowski cited are the “Short form survey instrument-36” (SF-36), “Sickness impact profile” (SIP), “Quality of well being” (QWB), “Disability to the arm, shoulder and hand” (DASH), the Musculoskeletal Functional Assessment index (MFA), and Short Musculoskeletal Functional Assessment index (SMFA).

“So how do we assess fracture outcome? We need to classify the fracture, report clinical outcomes, assess functional outcomes when we have the resources, use validated functional outcome tools, and not forget the confounders that have a major impact on patient outcome,” he concluded.

Conducting clinical trials in fracture healing
“If we were to create the ideal situation,” said Mohit Bhandari, MD, MS, “we would have a high-quality randomized control trial, infrastructure to manage complex coordination, a global network of qualified investigators, research teams with previous experience with multicenter trials, an endless resource of patients for ‘efficient recruitment,’ valid and reliable outcome assessment, and significant levels of funding.

“The reality is that we have a few large trial groups, study designs that are often implemented without consultation with the appropriate people, no ‘gold standard’ for outcomes, sample sizes that are too small, and, not surprisingly, surgeons that question the results,” he continued.

“All too often, we have a false-negative finding indicating there is no difference when, in fact, there was one. If we had just recruited enough patients for a larger sample size, we would have probably validated that difference.”

“Fracture healing is probably more subjective and more prone to bias when it is translated to clinical trials,” he said.

Dr. Bhandari suggested that bias can be potentially limited by blinding radiologists, orthopaedic surgeons, or other reviewers to parts of the radiographic assessment and by using independent outcome adjudication.

“A number of small trial networks are forming around the world. We need to consolidate these into a single global network that includes the United States and Canada.

“China and India have 40 percent of the world’s population. We certainly need to involve these emerging trial centers as we move forward,” Dr. Bhandari concluded.

The symposium Assessment of fracture healing in clinical trials: Methodological challenges was jointly sponsored by the Orthopaedic Research Society (ORS) and the Orthopaedic Trauma Association (OTA) at the 2008 ORS Meeting.

Annie Hayashi is the senior science writer for AAOS Now. She can be reached at