The idea that ultrasounds can heal bone fractures sounds like something from science fiction at the best of times. Fracture healing involves a complex interplay of cellular processes that, if all goes well, are thought to help with the fusing of a broken bone.
Ultrasound is a noise in a frequency that humans cannot hear. It is applied through a probe to the skin surface and is used for many medical applications that we’ve all come to know well. According to the widely publicised evidence, ultrasound causes effects ranging from small temperature changes in the tissue to increased gene expression.
However, the effects of ultrasound have been debated, and there have been several studies over the years, from a variety of backgrounds, showing that certain fractures, specifically non-unions, can show faster healing with the use of ultrasound. There is also evidence that these devices can help in patients who have poor healing potential, including diabetics, smokers and patients taking oral steroid medications.
According to a 2009 study from the University of Missouri: “Fracture healing can be compromised by numerous exogenous and endogenous patient factors. Low-intensity pulsed ultrasound (LIPUS) has been proposed as a modality that may have a benefit for increasing reliable fracture healing as well as, perhaps, increasing the rate of fracture healing.”
The report goes on to state that “research on LIPUS in animal fracture models has demonstrated promising results for acceleration of fracture healing and for promotion of fracture healing in compromised tissue beds.”
A body of work on cellular and animal research has already been published, much of which reveals that LIPUS may indeed be beneficial for accelerating normal fracture healing or for promoting fracture healing in compromised tissue beds.
But a team of North American researchers recruited 501 patients with fractures to the tibia for a study to determine whether this was really the case. This was a much larger trial than most previous ones on the topic, and more rigorously monitored with a large selection of academics form several institutions all taking part. The tests meant that, after standard care to repair their fractures, the patients were randomly assigned to LIPUS (similar to the ultrasound used in foetal monitoring), or they were exposed to a sham treatment. There was little proven effect from the ultrasound group.
Bare bones
Professor Jason Busse, who is based at McMaster University in the city of Hamilton, Ontario, and was the leader of the study, explains how this means there does not appear to be any evidence for ultrasound in the management of tibia fracture (the specific body part the trial focused on, and the most common place it is tried in medical settings) – so what does this mean and where do we go from here?
“About ten years ago,” he explains over the phone from his office, “we published a small, systematic review, which was in the Canadian Medical Association Journal, that looked into LIPUS and fracture healing. Effectively, what we thought at the time, is that there appeared to be a potential value for the therapy in terms of accelerating radiographic healing, but there was uncertainty: the trial groups were small, they had the limitations and there wasn’t a lot of evidence on functional outcomes.
“This got the attention of one of the major industry producers, Smith and Nephew,” he adds. Smith and Nephew (a UK medical equipment producer) asked Busse and the team to design a study that would help resolve the issue of whether or not it was scientifically valid.
“We decided to launch a pilot study that attempted, basically, to do a larger trial,” he continues. “The pilot study
was successful, and then we launched a definitive trial, which was partially funded by the Federal Government of Canada, and also by Smith and Nephew.” This trial was published in the British Medical Journal in October 2016 and delivered the opinion that there was no causation from ultrasound on bone fractures.
The background of using ultrasound as bone fracture treatment is a complicated one, however, and was always made in good faith, though its science was often flawed, says Busse. “The idea was that a fracture might take a long time to heal, and sometimes develop complications such as non-unions [when the fracture fails to completely unite for a long period of time]. And some fractures, such as tibia fractures, have a very long recovery period. If there could be a modality that could accelerate fracture healing or, perhaps, even reduce rates, or complications such as non-unions, that would be something quite important,” Busse explains.
“There was work done that mentioned a couple of ideas; one was a natural stimulation and the other was LIPUS. Tests were applied in laboratory settings and demonstrated some effect on bone cell growth,” he says. “So when you look at cell lines and animal models, there were some preliminary signals that they could be helpful in fracture healing.”
But the preliminary results were not what they found in the end. “These findings then resulted in a couple of small clinical trials that were submitted to the FDA and appeared to show promise in the therapy; however, these did have some important limitations,” notes Busse. “They had only a 50% loss-to-follow-up rate, meaning they only had data on about 70% of the people that they enrolled in the trial – and that potentially presents a serious bias to the results, because the people you lost from the study, may then undo the patterns that you would typically have in a randomised control trial,” he says.
When this happens, it’s easy to see why so many studies have finished with different results over the years. “If you end up losing a lot of people from the treatment group who, for whatever reason, were at a higher risk of delayed recovery, then you might assume the intervention was effective,” he says, explaining the ultimate flaw in the study.
“But only because you lost people in the treatment group that were destined to do less well, and then attributing the fact of the intervention to that it was more an artefact of the loss-to-follow-up, you proportionally bias the treatment to look successful.”
Bad science
According to Busse, the other problem with these two small trials that were used to acquire FDA approval for LIPUS was that they focused on the sole outcome – the results of the radiographic healing.
“Then the patients present to their surgeon and say ‘doctor, you’ve got to help me, I just don’t think my X-rays look nice enough’,” says Busse. “What we really need to know is do they facilitate faster recovery? Do they go back to work faster? Do they have less pain? Can they resume weight bearing on the affected limb – a tibia fracture, for example. These are the things that we need to understand.”
There was an absence in the results about the effect in this trial on the importance of patient-adherent outcomes, and Busse reiterates, “The trials had different limitations: high loss-to-follow-up rates; some did not blind the outcome assessors or the patients to what they were receiving. And so, there was a growing body of evidence in this area that there were a lot of limitations without a real focus on patient-important outcomes,” he explains. This meant that there was a chance that the modality could work for the FDA and its checks, but that there hadn’t been a definitive study that really showed the effect on outcomes or what patients consider important.
“A lot of the background was based on, ‘well, it looks promising on the cells in the test tube; it also looks promising, perhaps, on some animal tissue’, but there wasn’t really this defined trial, so that’s what we did,” Busse says.
What’s interesting, Busse believes, is that this modality has been on the market, since about 1994. “FDA and most
other regulatory bodies really don’t demand evidence that a medical device is effective in order to license it on patient-important outcomes – you simply have to show it doesn’t cause harm to people, and that’s really often as far as you often have to go.”
Industry reaction
Busse’s trial looked at outcomes on the patient and, as noted, they did not find an effect. He says they also failed to find an effect on radiographic healing and have recently completed a systematic review looking at all the ultrasound trials. In that systematic review, they found that it’s trials at high risk of bias have shown the reduction in radiographic healing but, as Busse says, “when you’re restricting analysis to try it at low risk of bias, there does not appear to have been an effect on radiographic healing.”
This is the challenge: the truth of the situation is that obviously the industry is looking for trial-result support to increase the marketing potential of their products. “I understand that,” says Busse, “So I understand the industry sponsor was disappointed [by the results]. They were hoping the trial was going to be positive – it was not.” But it does now mean that the industry must be honest and upfront with the results. More trials should be taken in the future to fully ascertain all remaining questions.
The trial is currently under systematic review and is being prepared to be published in the Canadian Medical Association Journal at some point this year. To Busse, it represents an important lesson in running good tests and trials, always following rigorous procedures and not allowing industry to sway results, no matter what the cause.
However, he does also express disappointment that, to all evidence, ultrasound does not do what it was previously thought to. Ultrasounds do not, we can now say with a high rate of assurance, help heal bone fractures any faster. There have been some signs of it in cell growth in labs, and with animals, but for a regular human with their leg in a cast hoping for an answer to a bone that won’t meld, there is no help here.
